Nanofiltration, 2 Volume Set: Principles, Applications, and New Materials [2 ed.] 3527346902, 9783527346905

Citation preview

Nanofiltration

Nanofiltration Principles, Applications, and New Materials Volume 1

Edited by Andrea Iris Schäfer Anthony G. Fane

Second Edition

Nanofiltration Principles, Applications, and New Materials Volume 2

Edited by Andrea Iris Schäfer Anthony G. Fane

Second Edition

Karlsruhe Institute of Technology (KIT) Institute for Advanced Membrane Technology (IAMT) Hermann-von-Helmholtz-Platz 1 76344 Eggenstein-Leopoldshafen Germany

All books published by WILEY-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Prof. Anthony G. Fane

Library of Congress Card No.:

University of New South Wales School of Chemical Engineering Chemical Sciences Bldg. (F10) 1712 Blue Water Lane 2052 Sydney Australia

British Library Cataloguing-in-Publication Data

Editors Prof. Dr. Andrea Iris Schäfer

applied for

A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek

Th Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at . © 2021 WILEY-VCH GmbH, Boschstr. 12, 69469 Weinheim, Germany All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Print ISBN: 978-3-527-34690-5 ePDF ISBN: 978-3-527-82496-0 ePub ISBN: 978-3-527-82497-7 oBook ISBN: 978-3-527-82498-4 Cover Design

Germany

Formgeber, Mannheim,

Typesetting Straive, Chennai, India Printing and Binding

Printed on acid-free paper 10 9 8 7 6 5 4 3 2 1

v

Contents

Volume 1 Foreword (Second Edition, 2020) xiii Foreword (First Edition, 2005) xv Acknowledgements xvii Dedication xxi Introduction xxiii Part I

Principles

1

1

History of Nanofiltration Membranes from 1960 to 1990 3 Charles Linder and Ora Kedem

1.1 1.2 1.3 1.3.1 1.3.2 1.3.3 1.3.4 1.3.5 1.4

Overview 3 Introduction 4 First-Generation NF Membranes 5 Cellulose Acetate Asymmetric Membranes 5 Deficiencies in Cellulosic Membranes 8 Polyelectrolyte Complexes 8 Polyamide Membranes 9 Polysulfones and Other Polymer Membranes 9 Early Studies of Charged Reverse Osmosis (Hyperfiltration) Membranes 10 Dynamic Membranes 10 Polyelectrolyte Membranes 10 Early Models of NF Selectivity 10 Negative Salt Rejection 13 Solutions of One Electrolyte 13 Separation by Negative Salt Rejection 13 Early Development of Industrial NF: Ionic Modification of Asymmetric Cellulose Acetate 13 Early NF Composites 15 General 15

1.4.1 1.4.2 1.5 1.6 1.6.1 1.6.2 1.7 1.8 1.8.1

vi

Contents

1.8.2 1.8.3 1.9 1.9.1 1.9.2 1.9.3 1.10 1.10.1 1.10.2 1.10.3 1.11 1.11.1 1.11.2 1.11.3 1.11.4 1.12

Plasma Polymerization 18 Graft Polymerization 18 NF Composites of the 1980s 18 Piperazineamide Membranes 18 Other NF Interfacially Produced Composites 20 Modification of RO Membrane Composites to Bring Them into the NF Range 20 Composites Produced by Noninterfacial Cross-linking 21 Polyvinyl Alcohol Composites 21 Sulfonated Engineering Plastics as Selective Barriers 22 Polyethyleneimine 22 Chemically Stable NF Membranes 23 Chemically Stable Polymeric Asymmetric Membranes 23 Oxidant and pH-Stable Composite Membranes 23 Solvent-Stable NF Composites 24 Chemically Stable Inorganic NF and Polymeric/Inorganic Hybrids 26 Conclusions 27 Abbreviations 28 References 28

2

Nanofiltration Membrane Materials and Preparation 35 Hanne Mariën, Rhea Verbeke, and Ivo F.J. Vankelecom

2.1 2.2 2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 2.2.6 2.2.7 2.3 2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 2.3.6 2.3.7 2.3.8 2.4 2.4.1 2.4.2 2.5 2.5.1 2.5.2 2.5.3 2.5.4

General Introduction 35 Phase Inversion 36 Introduction 36 Basic Principles 37 Polymer Type 40 Casting Solution 40 Postcasting Evaporation 44 Coagulation Bath 45 Post-treatment 46 Interfacial Polymerization 47 Introduction 47 Support Materials 48 Monomers 49 Monomer Concentrations and Reaction Time 58 Solvent 58 Additives 59 New Approaches 60 Post-treatment 61 Coating 61 Introduction 61 Examples 62 Surface Modification 63 Introduction 63 Plasma Treatment 63 Organic Reactions 63 Polymer Grafting 64

Contents

2.5.5 2.6 2.6.1 2.6.2 2.6.3 2.6.4 2.6.5 2.7 2.7.1 2.7.2 2.7.3 2.8 2.8.1 2.8.2 2.9

Photochemical Modification 65 Ceramic Membranes 65 Introduction 65 General Synthesis Procedure 65 Membrane Types 67 Supports 69 Surface Modification 69 Hollow Fiber Preparation 71 Introduction 71 General Synthesis Procedure 71 Composite Hollow Fiber Membranes 72 Commercial and Novel (SR)NF Membranes 72 Commercial (SR)NF Membranes 72 Novel (SR)NF Membranes 76 Outlook 77 Acknowledgements 78 Abbreviations 78 References 79

3

Nanofiltration Module Design and Operation Tzyy Haur Chong and Anthony G. Fane

3.1 3.1.1 3.1.2 3.1.3 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.2.5 3.2.6 3.3 3.3.1 3.3.2 3.4 3.4.1 3.4.2 3.4.3 3.4.4 3.4.5 3.5 3.5.1 3.5.2 3.5.3 3.5.4 3.6

Introduction 95 Role of the Module 95 Concentration Polarization and Cross-Flow 96 Fouling 101 Module Types and Characteristics 102 Plate and Frame 102 Spiral Wound 103 Tubular 104 Hollow Fiber and Capillary 105 Others 106 Module Characteristics 108 Spiral Wound Module (SWM) – Design Features 110 Feed Channel Spacers 110 Modeling and Optimization 112 Strategies to Improve Control of Concentration Polarization 116 Process Limitation by Concentration Polarization 116 High Shear – Vibrating the Membrane 117 High Shear – Rotor/Stator Modules 119 Two-Phase Flow 119 Unsteady Shear Comparison 120 System Design and Operation 120 System Configurations 121 Diafiltration 124 Reflux-Recycle Cascade (Combining RO and NF) 124 Batch Operation – Energy Saving (Closed Circuit) 126 Conclusions 128 Nomenclature 130 Subscripts 131

95

vii

viii

Contents

Greek Symbols 131 Abbreviations 131 References 131 137

4

Nanofiltration Membrane Characterization Anthony Szymczyk and Viatcheslav Freger

4.1 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6 4.2.7 4.3 4.3.1 4.3.2 4.3.3 4.3.4 4.4 4.5

Introduction 137 Structural Characteristics 139 Microscopy 139 Pore Size 141 Thickness and Morphology of the Active Layer 146 Surface Characteristics 147 Membrane Swelling and Solvent Uptake 149 Chemical Structure 150 Mechanical Properties 153 Charge Related Parameters 154 Electrokinetic Measurements 154 Titration and Ion Exchange 159 Membrane Potential 160 Electrochemical Impedance Spectroscopy 160 Nanofiltration Membranes for Nonaqueous Systems 163 Conclusions 165 Nomenclature 166 Greek Symbols 167 Abbreviations 168 References 168

5

Modeling Nanofiltration of Electrolyte Solutions 183 Andriy Yaroshchuk, Merlin L. Bruening, and Emiliy Zholkovskiy

5.1 5.2 5.2.1 5.2.2

Introduction 183 Basic Equations and Concepts 185 Derivation of Equations 185 Solution of Transport Equations for Macroscopically Homogeneous Membranes: Single Salts and Trace Ions 191 Specification of Phenomenological Coefficients Within the Scope of a Model of Straight, Narrow Capillaries 196 Nanopore Models of NF 197 Steric Exclusion and Hindrance 197 Local Equilibrium Partitioning Mechanisms 200 Solution-Diffusion-Electromigration Models of Nanofiltration 215 An Analytical Solution to Transport of Th ee Ions with Different Charges 215 Determining Single-Ion Permeances Using NF with Trace Ions 220 “Under-Osmotic” Operation 222 Deviations from Local Electrical Neutrality in Ultrathin Barrier Layers 223 Conclusions 228 Acknowledgements 230 Nomenclature 231

5.2.3 5.3 5.3.1 5.3.2 5.4 5.4.1 5.4.2 5.4.3 5.4.4 5.5

Contents

Greek Symbols 233 Abbreviations 234 References 234 6

Chemical Speciation Effects in Nanofiltration Separation 243 T. David Waite

6.1 6.2 6.2.1 6.2.2 6.3

Introduction 243 Chemical Speciation 243 Effect of Ionic Strength on Chemical Speciation 245 Effects of Temperature and Pressure on Chemical Speciation 247 Review of Effects of Solute Size, Charge, and Concentration on Rejection by NF Membranes 249 Solution Processes Influencing Speciation and Rejection 249 Acid–Base Transformations 250 Complexation 257 Precipitation 260 Oxidation–Reduction 264 Adsorption 265 Effect of Concentration Polarization on Speciation and Rejection 267 Exceedance of Solubility Product and Precipitation of Solids 268 Aggregation of Macromolecules and Precipitated Solids 268 Formation of Alternative Complexes and Multinuclear Species 268 Conclusions 269 Nomenclature and Symbols 270 Abbreviations 270 References 271

6.4 6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.5 6.5.1 6.5.2 6.5.3 6.6

7

Fouling in Nanofiltration 273 Andrea I. Schäfer, Nikolaos Andritsos, Anastasios J. Karabelas, Eric M.V. Hoek, René Schneider, and Marianne Nyström

7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.3 7.3.1 7.3.2 7.3.3 7.3.4 7.3.5 7.3.6 7.3.7 7.4 7.4.1 7.4.2 7.4.3

Introduction 273 Fouling Characterization 277 Flux Measurement and Fouling Protocols 277 Normalization of Membrane Performance 279 Water Fouling Potential 280 Membrane Autopsy 284 Fouling Mechanisms 286 Concentration Polarization (CP) 288 Osmotic Pressure 290 Solute Adsorption 291 Gel Layer Formation 292 Cake Formation and Pore Blocking 293 Critical Flux and Operating Conditions 295 Additional Fouling Mechanisms 296 Organic Fouling 299 Introduction and Definition of Organic Fouling 299 Common Organic Foulants 299 Adsorption of Organic Matter 301

ix

x

Contents

7.4.4 7.4.5 7.4.6 7.4.7 7.4.8 7.5 7.5.1 7.5.2 7.5.3 7.5.4 7.5.5 7.6 7.6.1 7.6.2 7.6.3 7.6.4 7.7 7.7.1 7.7.2 7.7.3 7.7.4 7.7.5 7.7.6 7.7.7 7.8 7.8.1 7.8.2 7.8.3 7.8.4 7.8.5 7.8.6 7.9

Gel Layer Formation 304 Cake Formation 304 Pore Blocking/Plugging 305 Impact of Solute–Solute Interactions and Salts 306 Impact of Fouling on Retention 308 Scaling 309 Introduction and Definition of Scaling 309 Solubility and Supersaturation of Salts 312 Common Scalants 314 Characterization of Scales 317 Mechanisms of Scale Formation 317 Colloidal and Particulate Fouling 319 Introduction and Definition of Colloidal and Particulate Fouling 319 Colloid Properties 321 NF Membrane Properties 322 Colloid Transport and Deposition 323 Biofouling 327 Introduction and Definition of Biofouling 327 Biofilm Formation in NF Plants 328 Biofilm Structure 333 Growth of Microbes in Biofilms 333 Sites for Biofouling in Membrane Systems 335 Measuring Microbial Load in Feedwaters and Detecting Biofilms in Membrane Systems 336 Biofouling Management in Membrane Systems 338 Fouling Prevention and Cleaning 339 Pretreatment as Fouling Prevention 339 Membrane Modification for Fouling Prevention 339 Cleaning Methods 341 Determination of Cleaning Effectiveness 346 Examples of Cleaning Applications and Cleaning Process Protocols 351 Regeneration of Cleaning Solutions 353 Conclusions 353 Acknowledgements 353 Nomenclature 354 Greek Symbols 355 Abbreviations 355 References 357

8

Pretreatment and Hybrid Processes 381 Jack Gilron, Marianne Nyström, Jukka Tanninen, and Lena Kamppinen

8.1 8.2 8.2.1 8.3 8.3.1

Introduction 381 Pretreatment – An Overview 382 Importance of Pretreatment in NF 382 Non-membrane Pretreatment Methods 383 Control of Inorganic Precipitation (Scaling) 383

Contents

8.3.2 8.3.3 8.3.4 8.3.5 8.4 8.4.1 8.4.2 8.4.3 8.5 8.5.1 8.5.2 8.5.3 8.5.4 8.6 8.7 8.7.1 8.7.2 8.8

Removal of Colloids and Solids 387 Removal of Organic Substances 387 Biological Fouling Prevention 388 Biological Pretreatment 389 Pretreatment Methods Using Filter Media 391 Conventional Filtration 391 Microfiltration (MF) 391 Ultrafiltration (UF) 393 Nanofiltration as a Pretreatment 395 Pretreatment Before Reverse Osmosis (RO) 395 Pretreatment Before Electrodialysis (ED) 396 Pretreatment Before Ion Exchange (IX) 397 Pretreatment Before Evaporation 398 NF in Hybrids Related to Seawater Desalination 398 NF as Post-treatment and Polishing Technology 403 Purification 403 Fractionation 404 Conclusions 407 Acknowledgements 408 Abbreviations 408 References 409 Volume 2 Foreword (Second Edition, 2020) xvii Foreword (First Edition, 2005) xix Acknowledgements xxi Dedication xxv Introduction xxvii Part II

Applications

419

9

Water Treatment 421 Erich Wittmann, Edvard Sivertsen, and Thor Thorsen

10

Water Reclamation, Remediation, and Cleaner Production with Nanofiltration 451 Yoram Cohen, Jin Y. Choi, and Anditya Rahardianto

11

Nanofiltration in the Food Industry 499 Marie-Pierre Belleville, José Sanchez-Marcano, Gerrald Bargeman, and Martin Timmer

12

Nanofiltration in the Chemical Processing Industry 543 Markus Kyburz, G. Wytze Meindersma, and Gerrald Bargeman

13

Nanofiltration in the Pulp and Paper Industry 599 Mika Mänttäri, Marianne Nyström, Jutta Nuortila-Jokinen, and Mari Kallioinen

xi

xii

Contents

14

Nanofiltration of Textile Dye Effluent 621 Chidambaram Thamaraiselvan and Woei-Jye Lau

15

Nanofiltration in Landfill Leachate Treatment 663 Johannes Meier, Kirsten Remmen, Thomas Wintgens, and Thomas Melin

16

Nanofiltration Bioreactors 691 Luong N. Nguyen and Long D. Nghiem

17

Photocatalytic Nanofiltration Reactors 707 Raffaele Molinari, Pietro Argurio, Lidietta Giorno, Leonardo Palmisano, and Enrico Drioli

18

Nanofiltration in Hydrometallurgy 759 Adrian A. Manis, Karin H. Soldenhoff, Elizabeth M. Ho, and Peter D. Macintosh

19

Trace Contaminant Removal by Nanofiltration 805 Alessandra Imbrogno, Youssef-Amine Boussouga, Long D. Nghiem, and Andrea I. Schäfer

20

Organic Solvent Nanofiltration 889 Torsten Brinkmann and Volkan Filiz

21

Nanofiltration Retentate Treatment 933 Bart Van der Bruggen

22

Renewable Energy-Powered Nanofiltration Bryce S. Richards and Andrea I. Schäfer Part III

961

Future Nanofiltration Materials 1021

23

Carbon Nanotube Composite Materials for Nanofiltration 1023 Francesco Fornasiero

24

Biomimetic Nanofiltration Materials 1057 Mihail Barboiu

25

Novel Polymer-Based Materials for Nanofiltration Mathias Ulbricht

26

Graphene-Based Membranes for Nanofiltration 1125 Wanqin Jin Conclusions and Future Developments 1165 Andrea I. Schäfer Index 1171

1081

v

Contents

Volume 1 Foreword (Second Edition, 2020) xv Foreword (First Edition, 2005) xvii Acknowledgements xix Dedication xxiii Introduction xxv Part I

Principles

1

1

History of Nanofiltration Membranes from 1960 to 1990 3 Charles Linder and Ora Kedem

2

Nanofiltration Membrane Materials and Preparation 35 Hanne Mariën, Rhea Verbeke, and Ivo F.J. Vankelecom

3

Nanofiltration Module Design and Operation Tzyy Haur Chong and Anthony G. Fane

4

Nanofiltration Membrane Characterization Anthony Szymczyk and Viatcheslav Freger

5

Modeling Nanofiltration of Electrolyte Solutions 183 Andriy Yaroshchuk, Merlin L. Bruening, and Emiliy Zholkovskiy

6

Chemical Speciation Effects in Nanofiltration Separation 243 T. David Waite

7

Fouling in Nanofiltration 273 Andrea I. Schäfer, Nikolaos Andritsos, Anastasios J. Karabelas, Eric M.V. Hoek, René Schneider, and Marianne Nyström

8

Pretreatment and Hybrid Processes 381 Jack Gilron, Marianne Nyström, Jukka Tanninen, and Lena Kamppinen

95 137

vi

Contents

Volume 2 Foreword (Second Edition, 2020) xvii Foreword (First Edition, 2005) xix Acknowledgements xxi Dedication xxv Introduction xxvii Part II

Applications

419

9

Water Treatment 421 Erich Wittmann, Edvard Sivertsen, and Thor Thorsen

9.1 9.2 9.3 9.3.1 9.3.2 9.3.3 9.3.4 9.3.5 9.3.6 9.3.7 9.4 9.5 9.5.1 9.5.2 9.5.3

Introduction 421 Overview of Nanofiltration Applications in Drinking Water 421 Plant Design 423 Membrane Selection 423 Nanofiltration Treatment Systems 425 Configurations with Spiral Wound Membranes 425 Configurations with Tubular Membranes 427 Pretreatment for Spiral Wound Membranes 427 Post-treatment 428 Residual Disposal 429 Plant Operation and Monitoring 429 Case Studies 430 Case Study 1: Méry-sur-Oise 430 Case Study 2: Sulfate Removal – Jarny 436 Case Study 3: Pesticide Removal and Softening of a Borehole Water – Debden Road 438 Plants Treating Highly Colored Water 438 Color Removal Using Tubular Membranes 440 Color Removal Using Spiral Wound Membranes 440 Conclusions 446 Abbreviations 446 References 447

9.6 9.6.1 9.6.2 9.7

10

Water Reclamation, Remediation, and Cleaner Production with Nanofiltration 451 Yoram Cohen, Jin Y. Choi, and Anditya Rahardianto

10.1 10.2 10.2.1 10.2.2 10.2.3 10.2.4 10.2.5 10.3

Introduction 451 Reclamation of Municipal Effluent 452 Project Drivers – Water Reclamation 452 Advantages of NF in Municipal Reclamation Applications 458 Process Fundamentals 458 Process Limitations 460 Conclusions 467 Groundwater Remediation 469

Contents

10.3.1 10.3.2 10.3.3 10.4 10.4.1 10.4.2 10.4.3 10.4.4 10.4.5 10.5 10.5.1 10.5.2 10.5.3 10.5.4 10.5.5 10.6

Project Drivers – Recovery of Industrial By-product 469 Project Drivers – Removal of Groundwater Contaminants 474 Conclusions 477 Agricultural Drainage Water 478 Project Drivers – Reduction of Agricultural Drainage Water Salinity 478 Advantages of Low-Pressure RO/NF 478 Process Feature and Fundamentals 478 Process Limitations and Progress 481 Conclusions 483 Industrial Reuse and Cleaner Production 484 Project Drivers – Water Reuse 484 Advantages of NF 484 Process Features and Fundamentals 485 Limitations 488 Conclusions 489 Closure 489 Acknowledgements 490 Abbreviations 490 References 491

11

Nanofiltration in the Food Industry 499 Marie-Pierre Belleville, José Sanchez-Marcano, Gerrald Bargeman, and Martin Timmer

11.1 11.2 11.2.1 11.2.2 11.2.3 11.3 11.3.1 11.3.2 11.3.3 11.3.4 11.4 11.4.1 11.4.2 11.4.3 11.5 11.5.1 11.5.2 11.5.3 11.5.4 11.6 11.6.1 11.6.2 11.6.3

Introduction 499 Applications in the Milk Industry and Whey Processing 502 General 502 Production of Partially Demineralized Whey Concentrate 502 Production of Lactose from Whey Permeate 506 Applications in the Beverage Processing 507 General 507 Concentration of Fruit Juices 508 Production of Beverage with Low Alcohol Content 509 Concentration of Coffee Extract and Decaffeination of Coffee 511 Applications in Sugar Production 511 General 511 Concentration of Dextrose Syrup 511 Concentration of Thin Juice 513 Applications in the Edible Oil Industry 514 General 514 Solvent-Based Degumming 516 Direct Degumming 517 Deacidification 517 Production of Food Ingredients and Nutraceutics 518 General 518 NF for Production of Peptides and Amino Acids 520 NF for Production of Oligosaccharides and Sweeteners 521

vii

viii

Contents

11.6.4 11.7 11.7.1 11.7.2 11.7.3 11.7.4 11.7.5 11.8

NF for Production of Phenolic Compounds 523 Process Water Treatment 525 General 525 Treatment of Ion Exchange Resin Regenerate 526 Filtration and Reuse of CIP Solutions 528 Demineralization of Colored Brine from Anion Exchange Resin Elution Solutions 530 Filtration and Recycling of Process Water 531 Conclusions 531 Abbreviations 532 Nomenclature 532 References 532

12

Nanofiltration in the Chemical Processing Industry 543 Markus Kyburz, G. Wytze Meindersma, and Gerrald Bargeman

12.1 12.2 12.2.1 12.2.2 12.2.3

Introduction 543 Inorganic Chemical Industry 543 Characterization of the Industry 543 NF of Processing of Sodium Chloride Brines 544 Pollution Treatment in the Inorganic Chemical Industry, MLD and ZLD 554 Organic Chemical Industry 558 Characterization of the Industry 558 Potential and Actual Applications for NF in the Organic Chemical Industry 559 Pharmaceutical and Biotechnology Industry 569 Research and Development of Bench Test Technology 569 General Industrial Process Description 571 NF Applications in the Pharmaceutical and Biotechnology Industry 571 Petrochemical Industry 581 Solvent Lube Dewaxing 581 Removal of Contaminants 584 Deacidifying Crude Oil 585 Tertiary Oil Production: SAGD (Steam-Assisted Gravity Drainage) 586 DeSOx /Bio-deNOx Processes 588 Conclusions 590 Acknowledgements 590 Abbreviations 590 References 591

12.3 12.3.1 12.3.2 12.4 12.4.1 12.4.2 12.4.3 12.5 12.5.1 12.5.2 12.5.3 12.5.4 12.5.5 12.6

13

Nanofiltration in the Pulp and Paper Industry 599 Mika Mänttäri, Marianne Nyström, Jutta Nuortila-Jokinen, and Mari Kallioinen

13.1 13.2

Introduction 599 Streams that could be Processed with NF Membranes 600

Contents

13.3 13.4 13.4.1 13.4.2 13.4.3 13.4.4 13.4.5 13.5 13.5.1 13.5.2 13.5.3 13.5.4 13.5.5 13.6

NF Modules and Demands in the Pulp and Paper Industry 605 Examples of Mill-Stage NF Plants 607 Nanofiltration in Upgrading Effluent Quality 608 Nanofiltration in Recirculation of Water at Paper Mill 608 Zero Discharge Mill with Membrane Technologies 609 Reduction of Concentrate Volume with Nanofiltration 609 Recirculation of Waste Streams in Medium Density Fiber Board Mill 609 Pilot and Bench-Scale Systems 610 Purification of Bleaching Effluents 610 Purification of Mechanical Pulp Mill Process Waters 611 Purification of Paper Mill Process Waters 611 NF as Polishing Stage after Biological Wastewater Treatment 613 NF in Recovering Side Products from Spent Cooking Liquors 613 Conclusions and Future Prospects 615 Abbreviations 615 References 616

14

Nanofiltration of Textile Dye Effluent 621 Chidambaram Thamaraiselvan and Woei-Jye Lau

14.1 14.2 14.3

Introduction 621 Textile Wastewater Treatment Overview 623 Membrane-Based Technologies for Textile Wastewater Treatment 630 Ultrafiltration 630 Membrane Bioreactor (MBR) 631 Nanofiltration (NF) 633 Reverse Osmosis (RO) 633 Advances in Nanofiltration Fabrication and Modification 634 TFC Flat Sheet Membranes 634 TFC and Asymmetric Hollow Fiber Membranes 635 Positively and Negatively Charged Membranes 636 Factors Affecting NF Performance 639 Influence of Feed Properties 639 Influence of Membrane Properties 641 Influence of Hydrodynamic Conditions 642 Fouling Control Approaches 644 Integrated Process Involving Nanofiltration 646 Economic Evaluation on Nanofiltration Hybrid Process 651 Conclusions 653 Abbreviations 653 References 653

14.3.1 14.3.2 14.3.3 14.3.4 14.4 14.4.1 14.4.2 14.4.3 14.5 14.5.1 14.5.2 14.5.3 14.6 14.7 14.8 14.9

15

Nanofiltration in Landfill Leachate Treatment 663 Johannes Meier, Kirsten Remmen, Thomas Wintgens, and Thomas Melin

15.1 15.2

Introduction 663 Landfill Leachate 664

ix

x

Contents

15.2.1 15.2.2 15.2.3 15.3 15.3.1 15.3.2 15.3.3 15.3.4 15.3.5 15.3.6 15.4 15.4.1 15.4.2 15.4.3 15.4.4 15.4.5 15.4.6 15.4.7 15.5

Generation of Landfill Leachate 664 Characteristics of Landfill Leachate 665 Legal Standards for Treated Landfill Leachate 668 Overview of Currently Employed Processes 669 Biological Treatment 670 Adsorption 670 Oxidation/Reduction 671 Membrane Processes 672 Concentrate Removal 673 Process Combinations 674 Landfill Leachate Treatment by Nanofiltration 674 General Features of NF in Landfill Leachate Treatment 676 NF as Single Process 676 Biology and NF 677 Biology and NF with Concentrate Treatment by Adsorption/Oxidation 679 RO with Concentrate Treatment by NF, Crystallization, and High Pressure RO 682 Biology, NF, and Adsorption on Powdered Activated Carbon 684 Economics of the Described Processes 686 Conclusions 687 Acknowledgements 688 Abbreviations 688 References 688

16

Nanofiltration Bioreactors 691 Luong N. Nguyen and Long D. Nghiem

16.1 16.2 16.3 16.4 16.5 16.6 16.6.1 16.6.2 16.7 16.8 16.8.1 16.8.2 16.9

Introduction 691 NF-MBR Configurations 692 Wastewater Treatment Applications 693 Removal of Organic Matter and Nutrients 693 Removal of Trace Organic Contaminants 695 Operational Challenges 696 Salinity Build-up 696 Membrane Fouling and Degradation 697 Nutrient Recovery Opportunities 698 Bioprocessing 698 Bioethanol Production 698 Raw Chemical Precursor Production 699 Conclusions 701 Abbreviations 702 References 702

17

Photocatalytic Nanofiltration Reactors 707 Raffaele Molinari, Pietro Argurio, Lidietta Giorno, Leonardo Palmisano, and Enrico Drioli

17.1

Introduction 707

Contents

17.2 17.2.1 17.2.2 17.3 17.3.1 17.3.2 17.3.3 17.3.4 17.4 17.4.1 17.4.2 17.4.3 17.4.4 17.5

Background 708 Basic Principles of Heterogeneous Photocatalysis 709 Factors Affecting the Performance of PMRs 714 Possible System Configurations 718 Membranes as TiO2 Particle Separator: PMRs with Suspended Photocatalyst 719 Membranes as TiO2 Confinement or Carrier: PMRs with Immobilized Photocatalyst 723 Irradiation of Immobilized and Suspended Catalyst in the Photoreactors 725 Investigation on General NF Membranes Behavior 727 Some Applications from Laboratory to Industrial Scale 728 Examples of Contaminant Removal in Pressure-Driven Membrane Photoreactors 728 Examples of Contaminant Removal in Submerged NF Membrane Photoreactors 736 NF Wastewater Treatment in Photocatalytic Membrane Reactors 740 Hint on Processes Alternative to NF in Water Treatment 743 Conclusions 744 Abbreviations 745 References 746

18

Nanofiltration in Hydrometallurgy 759 Adrian A. Manis, Karin H. Soldenhoff, Elizabeth M. Ho, and Peter D. Macintosh

18.1 18.2

Introduction 759 Challenges in the Application of Nanofiltration to Hydrometallurgy 760 Nanofiltration in Copper Hydrometallurgical Processing 762 Nanofiltration of Copper Pregnant Leach Solution 762 Nanofiltration of Copper Electrowinning Bleed 764 Nanofiltration of Tailings Pond Water 764 Application of Nanofiltration in a Copper Smelting Plant 765 Nanofiltration in Uranium Processing 765 Nanofiltration of Uranium In Situ Leach and Heap Leach Solutions 767 Sodium Bicarbonate Recovery from Ion Exchange Eluate 769 Acid Recovery from Ion Exchange/Resin-in-Pulp Eluate 769 Acid Recovery from Solvent Extraction Strong Acid Strip Liquor 772 Sodium Chloride Recovery from Ion Exchange Eluate or Solvent Extraction Strip Liquor 773 Uranium Recovery by Ion Exchange and Nanofiltration from Alkaline Leach Liquor 774 Nanofiltration in Processing of Lithium Brines 776 Nanofiltration of Raw Brine 777

18.3 18.3.1 18.3.2 18.3.3 18.3.4 18.4 18.4.1 18.4.2 18.4.3 18.4.4 18.4.5 18.4.6 18.5 18.5.1

xi

xii

Contents

18.5.2 18.5.3 18.5.4 18.5.5 18.6 18.6.1 18.6.2 18.6.3 18.7 18.7.1 18.7.2 18.7.3 18.7.4 18.8 18.8.1 18.8.2 18.8.3 18.8.4 18.8.5 18.9 18.9.1 18.9.2 18.9.3 18.10

Nanofiltration After Dilution of Evaporated Brine 779 Treatment of Salt Lake Brine by a Combined Nanofiltration and Membrane Distillation Process 779 Treatment of Geothermal Brines 780 Research into New Membranes 780 Nanofiltration in Zinc Processing 780 Nanofiltration in Waelz Oxide Processing 780 Nanofiltration to Concentrate Bioleach Liquor and Recover Indium 782 Nanofiltration to Remove Boron 782 Nanofiltration in Gold Processing 783 Nanofiltration of Gold Cyanide Leach Liquor 783 Nanofiltration Integrated with Activated Carbon Elution and Electrowinning 784 Pressure Oxidation of Gold Ore 784 Treatment of Hypersaline Process Water 785 Other Processing Applications 785 Nanofiltration in Vanadium Processing 785 Nanofiltration of Bayer Process Liquors (Alumina Production) 787 Nanofiltration in Tungsten Processing 788 Nanofiltration in Treatment of Spent Nickel Electrolyte 790 Metal Separation from Phosphoric Acid by Nanofiltration 790 Nanofiltration for Recovery of Critical Materials from Secondary Sources 793 Scandium 793 Rare Earths 794 Germanium and Rhenium 794 Conclusions 794 Acknowledgements 795 Abbreviations 795 References 796

19

Trace Contaminant Removal by Nanofiltration 805 Alessandra Imbrogno, Youssef-Amine Boussouga, Long D. Nghiem, and Andrea I. Schäfer

19.1 19.2

Introduction 805 Occurrence of Trace Contaminants and their Effect on Health and Environment 806 Water Sources 808 Current Regulations and Water Guidelines 810 Endocrine-Disrupting Chemicals (EDCs) 811 Pharmaceutically Active Compounds (PhACs) 815 Pesticides 816 Disinfection By-products (DBPs) 818 Perfluorochemicals (PFCs) 819 Arsenic 820 Fluoride 821

19.2.1 19.2.2 19.2.3 19.2.4 19.2.5 19.2.6 19.2.7 19.2.8 19.2.9

Contents

19.2.10 19.2.11 19.2.12 19.3 19.3.1 19.3.2 19.3.3 19.4 19.4.1 19.4.2 19.5 19.5.1 19.5.2 19.5.3 19.6

Uranium 822 Boron 823 Nitrate 824 Nanofiltration in Water and Wastewater Treatment 825 Removal of Trace Contaminants by Nanofiltration 825 Application of NF for Organic Trace Contaminant Removal in Prospect with Water Guidelines 826 Application of NF for Inorganic Trace Contaminant Removal with Respect to Water Guidelines 831 Removal Mechanisms of Trace Contaminants by Nanofiltration 832 Organic Contaminants 833 Inorganic Contaminants 848 Fouling, Chemical Cleaning, and Aging 857 Impact of Membrane Fouling on Trace Contaminant Removal 857 Impact of Chemical Cleaning on Trace Contaminant Removal and Desorption 859 Impact of Membrane Aging and Defects on Trace Contaminant Removal 861 Conclusions 862 Acknowledgements 863 Nomenclature 863 Abbreviations 864 References 865

20

Organic Solvent Nanofiltration 889 Torsten Brinkmann and Volkan Filiz

20.1 20.2 20.2.1 20.2.2 20.2.3 20.3 20.4

Introduction 889 Membrane Materials 890 Introduction 890 Membrane Preparation and Fabrication 892 Membranes 893 Membrane Modules for Organic Solvent Nanofiltration 903 Modeling and Simulation of the Membrane Module Performance 906 Models for the Description of the Permeation Behavior 907 Process Examples 911 Pharmaceuticals 911 Base Chemicals 912 Organic Solvent Nanofiltration in Combination with Catalytic Processes 913 Conclusions 915 Nomenclature 917 Greek Symbols 918 Superscripts 918 Subscripts 918 Abbreviations 919 References 920

20.5 20.6 20.6.1 20.6.2 20.6.3 20.7

xiii

xiv

Contents

21

Nanofiltration Retentate Treatment 933 Bart Van der Bruggen

21.1 21.2 21.3 21.4 21.5 21.6 21.7

Introduction 933 Disposal Strategies 934 Further Treatment 937 Volume Reduction 942 Resource Recovery Strategies 946 Concentrates as the Target Fraction 949 Conclusions 953 Abbreviations 953 References 954

22

Renewable Energy-Powered Nanofiltration 961 Bryce S. Richards and Andrea I. Schäfer

22.1 22.1.1 22.1.2 22.1.3 22.1.4 22.1.5 22.2 22.2.1 22.2.2 22.2.3 22.2.4 22.2.5 22.3 22.3.1 22.3.2 22.3.3 22.3.4 22.3.5 22.3.6

Introduction 961 Water as a Global Challenge 961 Water Quality Issues 962 Water Energy Nexus 965 Decentralized Treatment Systems 966 Robust Systems for Harsh Environments 967 Renewable Energy-Powered Nanofiltration (RE-NF) Systems 967 Renewable Energy 967 Concept of Renewable Energy-Powered Systems 970 Impact of Membrane Choice 973 Energy Storage and Buffering 973 Overview of Commercially Available RE Membrane Systems 974 Performance of Small-Scale RE-NF Systems 976 System Design 977 Constant Power 977 Safe Operating Window (SOW) 980 Directly Connected Solar-Powered RE Membrane System 982 Directly Connected Wind-Powered RE Membrane System 986 Energy Storage and Buffering for Small-Scale RE Membrane Systems 990 System Performance in Terms of Water Quality 995 Salinity 998 Variation of pH and Speciation 1000 Removal of Trace Contaminants 1002 Removal of Organic Matter 1004 Fouling and Scaling 1006 Water Distribution 1007 Conclusions and Outlook 1008 Acknowledgements 1010 Abbreviations 1011 References 1011

22.4 22.4.1 22.4.2 22.4.3 22.4.4 22.4.5 22.4.6 22.5

Contents

Part III

Future Nanofiltration Materials 1021

23

Carbon Nanotube Composite Materials for Nanofiltration 1023 Francesco Fornasiero

23.1 23.1.1 23.1.2 23.1.3 23.1.4 23.1.5 23.1.6 23.1.7 23.1.8 23.1.9 23.1.10

Carbon Nanotube Membranes 1023 Introduction 1023 Basic Properties of Carbon Nanotubes 1024 Principles: Modeling of CNT Transport 1024 Fabrication of Carbon Nanotube Membranes 1029 Functionalization 1035 Water Permeability 1037 Solute Transport and Retention Properties 1039 Antifouling Properties 1043 Challenges for Fabricating Industrially Viable Membranes 1044 CNT Membrane Potential for Water Purification and Desalination 1046 Acknowledgements 1047 Nomenclature 1047 Greek Symbols 1048 Abbreviations 1048 References 1049

24

Biomimetic Nanofiltration Materials 1057 Mihail Barboiu

24.1 24.2 24.2.1 24.2.2 24.2.3 24.3 24.4 24.4.1 24.4.2

Introduction 1057 Self-organized Hybrid Membranes 1059 Directional Nanochannels for Facilitated Ionic Conduction 1059 Nanochannels for Proton Conduction 1060 Chiral Nanochannels 1064 Adaptive Constitutional Membranes 1066 Artificial Water Channels (AWCs) 1069 Artificial Water Channels in Bilayer Membranes 1069 Biomimetic Membranes Using Aquaporins and Artificial Water Channels 1073 Conclusions 1075 Acknowledgements 1076 References 1076

24.5

1081

25

Novel Polymer-Based Materials for Nanofiltration Mathias Ulbricht

25.1 25.2 25.3

Motivation and Scope 1081 Overview on Fabrication Methods and Building Blocks 1084 Alternative Membrane Polymers in Established Fabrication Processes 1089 Cross-linked Hydrophilic Neutral Polymers 1090

25.3.1

xv

xvi

Contents

25.3.2 25.3.3 25.3.4 25.3.5 25.3.6 25.4 25.4.1 25.4.2 25.5 25.6 25.6.1 25.6.2 25.6.3 25.6.4 25.6.5 25.7 25.8 25.9 25.10

Biopolymers 1091 Polymers with Intrinsic Microporosity (PIMs) 1092 Ion Exchange Polymers 1092 Amphiphilic Copolymers 1093 Well-Defined Amphiphilic Di- and Triblock Copolymers 1094 Alternative Fabrication Processes Based on Macromolecules and Nanoparticles 1096 Layer-by-Layer Deposition with Linear Polymers 1097 Deposition of Star Polymers or Nanoparticles 1098 Alternative Monomers in Established Interfacial Polymerization Fabrication Processes 1102 Alternative Fabrication Processes Based on Small Molecules 1103 Surface Grafting/Pore Filling Polymerization 1105 Molecular Layer-by-Layer Reactive Assembly Toward Polyamide Layers 1106 Synthesis of Covalent Organic Frameworks (COFs) 1107 Assembly and Cross-linking of Amphiphiles to Liquid Crystalline Microporous Phases 1108 Peptide Channels in Self-assembled Diblock Copolymers 1109 Mixed Matrix Composite Membranes 1109 Postmodification 1111 Approaches to Stimuli-Responsive Nanofiltration Membranes 1113 Conclusions 1114 Nomenclature 1116 Abbreviations 1116 References 1117

26

Graphene-Based Membranes for Nanofiltration 1125 Wanqin Jin

26.1 26.2 26.3 26.3.1 26.3.2 26.4 26.5 26.6 26.7

Introduction 1125 Porous Graphene Layer 1127 Assembled Graphene Laminates 1131 Fabrication Methods 1131 Controlling Membrane Nanostructure 1139 Graphene-Based Composites 1148 Transport Mechanisms 1150 Organic Solvent Nanofiltration 1156 Conclusions and Perspectives 1157 Nomenclature 1158 Abbreviations 1158 References 1160 Conclusions and Future Developments 1165 Andrea I. Schäfer Index 1171

xiii

Foreword (Second Edition, 2020)

In the foreword to the first edition of this book, Robert J. (Bob) Petersen wrote a masterful commentary on the beginning of the nanofiltration era. Little did we know then that this terminology would become so widespread in its use today. It is a testament to the wide breadth of applications for the technology. In its infancy, when it was still called “loose membranes,” companies sought applications with higher valued end products. The “single stage seawater reverse osmosis” membrane was an elusive goal of nearly all companies and the government agency Office of Water Research Technology. It was, after all, highlighted in a much publicized speech made by President John F. Kennedy. Many companies fell by the wayside when they could not achieve this sought-after goal. One of the secondary goals was the separation of sugar and salt. We thought this was a plausible separation considering the difference in size and stereochemistry of the disaccharide sucrose and sodium chloride. In those early days, cellulose (both di- and tri-acetate) was the polymer of choice. The techniques employed hearkened of alchemy more than polymer science. Th additives were referred to as “pore formers” rather than surface charge modifiers or other more sophisticated techniques used today. The initial attempts were technically successful, but not economically competitive with existing sugar purification techniques, but they did spawn new applications for membranes. One of the early successful applications was the concentration of cheese whey followed by drum drying, made necessary by the need to alleviate river and surface water pollution. Along the way, successful applications such as separation of enzymes from mother liquor and extraction of proteins from cheese whey became commonplace. The successful removal of proteins from whey made it possible to recover water-soluble protein instead of denatured protein using thermal methods. Th development of these applications also kick-started the use of more sophisticated hardware for the systems – clean-in-place, for example, was essential for food-grade and bio-pharma processing. Tubular membranes were very expensive and energy intensive but the recycling of recovered water-soluble paint in the automotive industry was a well-known exception. Smaller diameter multiple tubular module configurations resulted, which improved energy consumption

xiv

Foreword (Second Edition, 2020)

but were still costly. The advent of hollow fibers greatly improved surface area to volume ratio and spawned a huge application of membranes for wastewater reuse. Th expanded use of nanofiltration for commercial applications and industrial/domestic wastewater treatment has been nothing short of amazing. It now appears possible to solve age old problems such as produced water disposal by combining new membrane science with more conventional processes. Ceramic membranes, once considered overly expensive, are now more frequently included for difficult applications. I have been encouraged by the sophisticated advances in polymer chemistry in recent years. Perhaps today’s bright young minds can delve into the intricacies of polymer click chemistry and develop unique new nanofiltration membranes through bioconjugation science. Th s may afford a pathway to the solution of the ‘big elephant in the room,’ which is membrane fouling. Th future for nanofiltration continues to be promising, challenging, and ever-changing. 10 February 2020 Poway, CA, USA

David H. Furukawa Consultant to Filmtec from 1978, VP Marketing from 1983 (David attended the meeting with Bob Petersen and John Cadotte when the term, Nanofiltration, was first used.)

xv

Foreword (First Edition, 2005)

It was around the end of 1984. A few of us were gathered in the office of FilmTec’s advertising manager to tackle a problem of terminology. What does one call a reverse osmosis process that selectively and purposely allows some ionic solutes in a feed water to permeate through? Th phrase “loose RO” had been used, but it connoted the idea of leaky membranes. FilmTec had moved on to the expression “hybrid RO-UF,” planning to name some products as hybrid RO-UF membranes. However, neither “loose RO” nor “hybrid” translated well into Japanese. According to FilmTec’s Japanese distributor, the latter term carried objectionable overtones in Japanese. Th source of this naming problem was NS-300, a membrane discovered at North Star Research Institute. In 1976, John Cadotte combined piperazine with trimesoyl chloride, alone and blended with isophthaloyl chloride, to produce a series of thin-film-composite membranes with surprisingly high flux. Th se membranes also exhibited high permeability to aqueous chloride ions but high rejection of aqueous sulfate ions. Th membrane was an orphan. Th U.S. government’s Office of Water Research and Technology, which sponsored the research, did not see any particular usefulness of the membrane for its purposes, which were primarily the development of national water resources. But FilmTec took an interest in it, with industrial applications in mind. Among these were salt whey concentration, pulp and paper effluent treatment, and preparation of sulfate-free seawater on oil platforms for secondary oil recovery operations in barium-containing oilfield strata. FilmTec had named its version of the membrane as FT40. Naming the membrane was not an issue. Naming the process was the problem! I remember suggesting that FilmTec adopt the term “nanofiltration” for such processes. Th term had at least some logical basis. First, Sourirajan and Matsuura had calculated the size of a hypothetical capillary pore in annealed cellulose acetate membranes to be about 9 Å to 0.9 nm – in their development of the surface force/capillary flow model of reverse osmosis. Our “loose” membranes would correspondingly have hypothetical capillary pores slightly larger, presumably in the 1.0–1.2 nm range. Second, hyperfiltration was a term often used in early research on reverse osmosis membranes, and was deemed synonymous with reverse osmosis. Why not simply connect “nano” to

xvi

Foreword (First Edition, 2005)

“filtration”? Thi d, “nanograde” solvents were in wide use, and a term incorporating “nano” would connote goodness, purity, quality (Th t’s a suggestion designed to carry great influence with an advertising man!). One of the advantages of working in a small company was the ability to make instant decisions. We left that meeting with a mandate to use nanofiltration in our trade literature and publications. And two of FilmTec’s “FT” membranes were immediately recast as “NF” membranes. It did not occur to me at the time that the term nanofiltration could be easily transliterated into foreign languages. That is, nanofiltration could be used without modification in some languages, and easily adopted into others by minor changes in spelling. Further, nanofiltration as a descriptor carried no “baggage” with it. As a new word, it referred to a particular membrane process for which it was coined, and to no other. Within a few years, other membrane scientists began using the word nanofiltration. Its widespread use today is testament to the need for just such a descriptor in the membrane lexicon. As the body of literature on nanofiltration membranes and processes has expanded, the meaning of the term has necessarily been stretched to accommodate the wide range of features. It is appropriate that this book begins with an effort to define the term. An interesting aspect of nanofiltration membranes is the fact that so many parameters can come into play, when one tries to model and characterize the pressure-driven selectivities of such membranes. Parameters may include, for example, ionic interactions such as Donnan ion repulsion, site sharing phenomena by polyvalent ions in charged membranes, solute–membrane adsorption affinities, and stearic size interactions. Compared to modeling of nanofiltration membrane behavior, modeling the behavior of high rejection reverse osmosis membranes was a comparatively simpler task. Standard reverse osmosis for water purification has matured in many respects, and has become in large part the domain of engineers engaged in issues of yield, consistency, quality, and manufacturing efficiency. Th objective is always the same – make a pure water permeate with the lowest cost. But nanofiltration, in my opinion, remains the most fascinating extension of reverse osmosis technology. Nanofiltration offers to the membrane scientist a variety of membrane possibilities and a plethora of fascinating applications. Reverse osmosis is like the main course of a dinner, like a beefsteak that can be prepared in only a limited number of ways, but satisfies the hunger. Nanofiltration, on the other hand, is like the wine menu accompanying the meal – an opportunity for creativity and exploration. As you explore this book, enjoy the wonderful variety it provides on the subject of nanofiltration. 23 December 2002 Minneapolis, MN, USA

Robert J. Petersen Director of Research (from 1978), Filmtec Corp

xvii

Acknowledgements

This book has been on a 20-year journey that started with an idea in 2000 based on the realization that nanofiltration was not covered in its own right by membrane books existing at the time. Th first edition “Nanofiltration: Principles and Applications” was eventually published by Elsevier in 2005. Fifteen years on, the second edition “Nanofiltration: Principles, Applications, and New Materials” is about to appear with Wiley-VCH. Editing the first book was a challenge for a then very young scientist, while the excitement of working with so many much more established colleagues was immense, providing the driving force. Repeating this effort two decades later brought new challenges. Time and time again we wondered where the energy should come from to complete this task that appeared to be delaying into the infinite. But, we made it! We are deeply indebted to our contributors for their patience and for giving in to our incessant reminders and ultimately delivering the goods. Many of the original authors from the first edition had since retired or changed research direction, which left us with material too good to discard, yet also needing new caretakers to honor these efforts and produce new updated chapters. This was managed on a very individual chapter-by-chapter basis and we have communicated the wishes of those involved carefully. This new edition has been enriched with several new chapters, as well as an entire new section looking at new materials relevant for nanofiltration. While the foundations of the second edition lie firmly grounded in the original work, several key transitions have transpired. Thus, we would like to thank Professor David Waite (UNSW, Australia) for his initial contributions as an editor of the first edition, and our previous publisher Kostas Marinakis (Elsevier, Netherlands) for the long cooperation and handing over to Wiley-VCH in the most cooperative manner. Warm thanks go to Dr. Frank Weinreich (Wiley-VCH) for being available, effective, and a pleasure to deal with, even concerning the most challenging difficulties. Hopefully we can meet in person soon!

xviii

Acknowledgements

Th highlight for this second edition was to be the book launch in the form of an international conference “Nanofiltration 2020” at the Achalm, Germany, in July 2020. Given the global COVID-19 pandemic this event was postponed to July 2021 and again to June 2022, and it appears we may have needed the extra time to finally get published. The generosity of our publisher (Dr. Frank Weinreich, Wiley-VCH) with sponsoring the wine for the event dinner is gratefully acknowledged. We are, of course, looking forward to this event in a very special environment and to celebrate the joint effort of the book, hopefully in person. Th foreword of the first edition was kindly provided by Robert J. (Bob) Petersen, while David H. Furukawa has provided the foreword for the current edition. Both are included in this book. Bob Petersen and David Furukawa are witnesses to the birth of the term nanofiltration. David H. Furukawa was a consultant to Filmtec from 1978, and VP Marketing from 1983, while Robert J. Petersen was the Director of Research (from 1978), Filmtec Corp. Both attended the meeting with John Cadotte when the term, Nanofiltration, was first used. It is a great privilege to have this moment of history present through the forewords of both editions. Most importantly of course, about 70 authors from 16 countries have contributed to this new edition and put up with our requests, reviews, comments, formatting requirements, revising proofs, and (well, at least to an extent!) the deadlines some well into their retirement – all in return for a token honorarium. It has been a great pleasure to be in touch with you over the last several years, many of you several decades now, and we are especially grateful to those who agreed at the last minute to fill in for those who could no longer contribute so that the book could still proceed. Reviewing book chapters is an enormous task and we appreciate the time taken and contributions made by the scientific reviewers in assisting the editors, with the challenge to peer review each chapter by two to three international experts. Great appreciation goes to colleagues and friends, some of whom have reviewed several chapters, some at very short notice, and most generously giving time and ideas. Sadly, our co-author Dr. Martin Timmer and our reviewer Dr. Darrel Alec Patterson passed away – both far too early – during the preparation of the second edition. We will keep both in our memory. Darrel wrote to us in 2006 very poignantly “I very much enjoyed your excellent book “Nanofiltration – Principles and Applications.” It is great to have works from so many well regarded experts in one volume! Just one comment – the quote by anonymous in the Acknowledgements section can be attributed to the late and great Douglas Adams (author of among other things, ‘The Hitchhikers Guide to the Galaxy’ in its various forms) who was quite well known for his lack of regard for deadlines (to the chagrin of his publishers!).” Reading Darrel’s words once again, we realized that the tempo was more sedate in the second edition, and the deadlines seemed to make less of a whooshing sound … or maybe they did, but we did not notice them above the background noise.

Acknowledgements

Than You to the scientific reviewers; Professor Dr. Mathias Ulbricht

University of Duisburg Essen, Germany

Professor Suzana Nunes

KAUST, Saudi Arabia

Professor Dr. Ing. Thom s Wintgens RWTH Aachen, Germany Professor Bart Van der Bruggen

KU Leuven, Belgium

Dr. Anita Buekenhoudt

VITO, Belgium

Professor Long Nghiem

UTS, Australia

Professor Andrew Zydney

The Pennsylvania State University, USA

Professor Viatcheslav Freger

Technion, Israel

Professor Anthony Szymczyk

University of Rennes, France

Professor Menachem Elimelech

Yale University, USA

Professor Alberto Tiraferri

Politecnico di Torino, Italy

Professor Dr. Stefan Panglisch

University of Duisburg Essen, Germany

Professor Maxime Pontie

University of Angers, France

Professor Dr. Ing. Wilhelm Urban

University of Darmstadt, Germany

Dr. Harry Seah

PUB, Singapore

Professor Jack Gilron

Ben Gurion University, Israel

Professor Darrell Alec Patterson†

Bath University, UK

Professor Dr. Jörg Hinrichs

University of Hohenheim, Germany

Michael Wunsch

Hager + Elsässer Water, Germany

Professor Dr. Steffen Schuetz

Mann + Hummel, Germany

Andreas Flach

MFT, Germany

Professor Abdelhadi Lhassani

Université Sidi Mohamed Ben Abdellah, Morocco

Professor HariKrishnan Ramanan

IIT Tirupati, India

Professor Seungkwan Hong

Korea University, South Korea

Professor Emile Cornelissen

KWR, Netherlands

Professor Benoit Teychene

University of Poitiers, France

Professor Katsuki Kimura

Hokkaido University, Japan

Dr. Filicia Wicaksana

University of Auckland, New Zealand

Professor Dr. Bryce Richards

Karlsruhe Institute of Technology, Germany

Professor Chuanfang (Ted) Yang

Chinese Academy of Sciences, China

Professor Sylwia Mozia

West Pomeranian University of Technology Szezecin, Poland

Professor Ivo Vankelekom

KU Leuven, Belgium

Professor Murielle Rabiller-Baudry

University of Rennes, France

Professor John Pellegrino

University of Colorado Boulder, USA

Professor Berrin Tansel

Florida International University, USA

Dr. Pia Lipp

TZW, Germany

Dr. Michael Hirtz

Karlsruhe Institute of Technology, Germany

Professor Chuyang Tang

Th University of Hong Kong, Hong Kong

Dr. Frank Biedermann

Karlsruhe Institute of Technology, Germany

Dr. Francesco Fornasiero

Lawrence Livermore National Laboratory, USA

Professor Tony Fane

UNSW, Australia

xix

xx

Acknowledgements

We trust the effort will inform many new readers interested in nanofiltration while also providing a comprehensive work to the ever-increasing number of nanofiltration experts. We are grateful (in advance) for any comments, discussions, translation requests, and feedback on the book and the topic in general. With the tremendous interest in nanofiltration, from materials development through transport mechanisms to applications, it promises to remain an exciting field. We are looking forward to the next 20 years of nanofiltration and to welcoming many more to the nanofiltration family. There is a lot to be done with and discovered about nanofiltration yet. Enjoy the read! Unfortunately, there is no such membrane that can separate Happiness and Sorrow of our life (Professor Takeshi Matsuura)

xxi

Dedication

Th s book is dedicated to two very special ladies of the broader membrane family Ora Kedem and Miriam Balaban Ora Kedem is an author in this book and a brilliant scientist with an incredible career in membranes – focusing on transport processes as well as desalination technologies. Miriam Balaban is the most successful disseminator and pioneer in publishing in the area of water and desalination – with a passion for reconciliation. Than you for being an incredible inspiration with a life that achieved tremendous success despite – or because – of the very challenges that life happens to provide. Thei lives tell stories of commitment and resilience. Th se strong women are an inspiration to everyone privileged enough to meet them – whether discussing science or life matters – both remaining active well into their nineties. WOW! An inspiration to keep going – no matter what – and to be incredibly grateful for what life has already provided. Not perfect – but a lot better than the conditions of our foremothers, such as Marie Curie and Lise Meitner. Women remain underrepresented in membrane technology – including as authors in this nanofiltration book – as well as in the water industry and in leadership positions as a whole, yet role models like Ora and Miriam clearly show what women are able to achieve when given the opportunity and are brave enough to rise to the challenge. THANK YOU and SHALOM!

xxii

Dedication

(Toulouse, June 2019)

(Tel Aviv, May 2017)

xxiii

Introduction Andrea I. Schäfer Nanofiltration (NF) is a liquid-phase pressure-driven membrane process with separation properties that overlap with both ultrafiltration (UF) and reverse osmosis (RO). Figure 1 indicates the approximate range of solute sizes relevant to the family of liquid-phase, pressure-driven membrane processes from RO to MF in comparison with other membrane processes, where the boundaries between the various processes are not precisely defined. Here, NF can be seen to fill an important gap between UF and RO, where it is able to fractionate ions and retain relatively low molecular weight organic solutes. These are important separations that have commercially significant applications. Notably, as defined in Figure 1 as “micros” and “macros,” NF can retain typical water contaminants such as humic substances almost completely, “micros” such as micropollutants to a significant extent while the retention of salt can be tuned to a great extent with the choice of membrane between loose and tight NF. In the Introduction section to the first edition “Nanofiltration – Principles and Applications” (2005) [1], the question of whether NF deserved to be considered a process in its own right or whether it was really very loose reverse osmosis (RO) or very tight ultrafiltration (UF) was discussed. At that time, this was a moot point, even though NF had just caught up with the volume of publications of reserve osmosis. However, given the unique properties of NF membranes, the separation mechanisms identified, and the application niches that have developed, it is evident that NF fits into a special category. Th first edition of this book helped to define the domain of nanofiltration, and the subsequent decade has confirmed the unique attributes of NF. It is now widely accepted that NF membranes have individual pores, unlike RO membranes that have dynamic “free volume” between polymer domains. Figure 2a shows an atomic force microscopy (AFM) image, possibly the first, of a pore in a NF membrane and Figure 2b shows the measured pore size distribution of that particular membrane [2, 3]. Th nanometer-scale pores allow passage of solvent water (approximate diameter is 0.275 nm) but retain dissolved species – very close to the size of water – on the basis of steric hindrance, electrostatic and dielectric interactions, as well as interactions with the membrane polymer. For example, differences in hydrated ion size and charge

Introduction

Molecular range

Ionic range

Contaminant size Pore size (μm)

0.001

Molecular weight (g/mol)

100

200

0.1

1000 100 000

Aqueous salt

Solutes

Macromolecular Microparticle range range

0.01

1

10

5 00 000

Virus

Bacteria Micro plastics

Proteins Inorganic contaminants Microsolutes Humic acids

MICROS (1 nm) Proteins, humic substances, natural organic matter, biopolymers, nanoparticles, viruses

Membrane bioreactor

Electrodialysis

Membrane distillation

Reverse osmosis Membrane separartion processes

Microfiltration

Nanofiltration Pervaporation

Ultrafiltration

Figure 1 Liquid-phase pressure-driven membrane processes – typical solute separations at the 1 nm solute scale distinguishing ultrafiltration from reverse osmosis.

40

A

35 30

Å 2.00 1.00 0.00

% of pores

xxiv

16 12 Å

20 15 10

8 12 4

(a)

25

8 0

4 0

5

16

0 0.0

Å

(b)

0.5

1.0

1.5

2.0

Pore size (nm)

Figure 2 Pores in NF membranes: (a) AFM image and (b) pore size distribution. Source: Adapted from [2, 3].

provide the mechanisms for the separation of monovalent and multivalent ions that is typical of NF. More generally, the unique separation properties for typical nanofiltration membranes with negative surface charge have been identified previously [1] and characterized by the following: 1. rejection of ions with more than one negative charge (multivalent anions), 3 such as sulfate (SO2 4 ) and phosphate (PO4 ), being virtually complete; 2. rejection of sodium chloride (NaCl) varying from about 70% down to 0%, while even a negative rejection may be observed in mixed systems with multivalent cations;

Introduction

3. rejection of uncharged, dissolved materials and also of positively charged ions in solutions to relate mostly to the size and shape of the molecule in question; and the 4. molecular weight cutoff (MWCO) to be in the range of 150–300 Da. In addition, NF has a growing role in nonaqueous separations that are largely based on size exclusion mechanisms, modified by membrane solute interactions such as swelling. In terms of performance, NF membranes can exhibit water permeabilities, A, in the range of about 5–15 l/m2 h bar, which are up to an order of magnitude higher than RO, but 1 or 2 orders of magnitude less than typical UF. This offers, at appropriate rejection, a significant energy saving compared to RO. This is a major advantage in many applications. Th retention behavior vs. pressure can reflect both the behavior of UF and RO (Figure 3), depending on the membrane and solute. A solute with a loose membrane with partial retention will show a more typical UF behavior where retention may decrease with pressure. For partially retentive UF membranes, and in the absence of fouling, the effect of increased pressure is typically a reduction in the observed retention (Figure 3). In the UF process, solutes are convected through the pores by solvent flow, and this solute transport is exacerbated by concentration polarization (CP), which tends to be much more significant in UF, whereas CP increases with flux and solute size. Th same is true for NF, albeit for different solutes to UF. Tight membranes are typically showing an increase toward an asymptote as in RO. This trend can be explained in terms of the solution–diffusion transport mechanism applied to “nonporous” RO membranes, which assumes that solute and solvent transport are uncoupled; pressure increases solvent water flow, and “dilutes” the solute in the permeate. Th “RO-like” behavior of tight NF is likely a combination of the effect of restricted convection in confined pores and the modest level of CP in typical 100 Tight NF

RO

Retention (%)

80 60 40 Loose NF 20 UF 0 0

5 10 15 Transmembrane pressure (bar)

45

50

Figure 3 Schematic trends of solute retention vs. pressure for ultrafiltration, loose/tight nanofiltration, and reverse osmosis (lines are indicative only).

xxv

Introduction

Number of publications

2500 2000

UF RO NF

1500 1000 500 0 1985 1990 1995 2000 2005 2010 2015 2020 2025 Year

Figure 4 Publication trends for ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO). 250

Number of publications

xxvi

200

Solvent NF Water NF Total NF

150 100 50 0 1985 1990 1995 2000 2005 2010 2015 2020 2025 Year

Figure 5 Publications in Journal of Membrane Science (Elsevier) for nanofiltration (NF) and both Solvent and Water NF.

NF applications, whereas in loose NF, the behavior is much more “UF-like” for certain solutes. Th growing interest in NF is evidenced by research outputs (number of publications) and commercial activity (global market value). Figure 4 shows publication trends from 1987 based on the topic “Nanofiltration” from the Web of Science database, while Figure 5 shows the same trend in the Journal of Membrane Science. Since the first edition (2005), the annual output of NF-related publications has increased by a factor of >4, from about 300 to about 1300 (in 2019). Publications in the Journal of Membrane Science (JMS) reflect these trends with NF (in topic) rising from about 60 in 2005 to about 160 in 2019. Unsurprisingly, the majority of these papers relate to water treatment, which reflects that hardness removal, organic matter (humic substances and

Introduction

Global market value ($ millions)

400

300

Water and wastewater Food and beverages Pharmaceutical and biomedical Other

200

100

0 2012

2013

2014 Year

2019

Figure 6 Global Market for Nanofiltration Membranes, 2012–2019 (US$ millions). Source: Data from BCC Research [4]. Table 1 Global Market for Nanofiltration Membranes with Market Segments (CAGR is the compound annual growth rate [CAGR] with a constant rate of return).

Market segment

2012

2013

2014

2019

CAGR% 2014–2019

Water and wastewater treatment

128.1

141.4

160.8

338.5

16.1

23.8

26.5

30.4

48.8

9.9

7.8

8.4

9.3

22.7

19.5

Food and beverages Pharmaceutical and biomedical Others

13.1

13.9

15.1

35.1

18.4

Total

172.8

190.2

215.6

445.1

15.6

Source: Data from BCC Research [4].

disinfection by-product precursors) removal, and trace contaminants are major concerns in the water industry and NF can remove many of these effectively. An increasing number relate to NF of organic solvents, and this research interest may anticipate a future growth area for NF applications. Commercially, NF is experiencing an equally significant growth in applications reflected in global market value and compound annual growth rate (CAGR). The data in Figure 6 and Table 1 were sourced from a BCC Research report [4] and show • a predicted CAGR from 2014 to 2019 to be >15.0%; • a global market of nearly US$0.5 billion by 2019, of which 75% is for the water industry; • steady growth in applications in the food and beverage sector and the pharmaceutical and biomedical sector. A more recent BCC report [5] predicts a CAGR from 2019 to 2024 of 18.2% and a market of US$518 million in 2019 growing to US$1.2 billion in 2024, of which

xxvii

xxviii

Introduction

75% is for the water industry. Since the first edition in 2005, the NF market has probably grown from 2.5 to 4-fold, and the data provided in Figure 6 and Table 1 predict a very strong future for nanofiltration indeed. Such a promising outlook raises the question what impact this NF “boom” will have. Savings in energy compared to RO will make a positive contribution to carbon emissions, the treatment of wastewaters in particular will reduce pollution and enable the recovery of previous resources, and the development of decentralized water treatment systems can alleviate much water-related illness and death in developing countries. On the other hand, the increase in NF application will cost energy and cause CO2 emissions, will produce concentrates that are to be treated (zero liquid discharge technologies are to date not yet available and/or economical), and, ultimately, the vast nanofiltration market will result in mountains of spend membrane modules to be disposed of. Th se negative features maybe offset by the use of renewable energy-driven NF (see Chapter 22), development of zero liquid discharge technologies (including resource recovery) [6], and strategies to reuse or recycle RO and NF membrane modules [7, 8]. One can only wish that the enormous market and inevitably profits made will be reinvested into meaningful research to develop more efficient technologies that have less of an impact and create a net positive benefit to our environment. A new generation of engineers and researchers may indeed be driven more by environmental impact than financial gain. May this book provide a contribution to teach about nanofiltration and inspire a new generation of exciting applications and developments beyond nanofiltration as we currently know it. Th book is divided into three parts – a fundamental section on principles, an applications section, and a new materials section. Th overview layout for the new book is presented in Figure 7. From a history of nanofiltration chapter (Chapter 1), the book takes the reader to membrane preparation and materials (Chapter 2), module design and operation (Chapter 3), membrane characterization (Chapter 4), NF membrane performance modeling (Chapter 5), solute speciation effects in NF (Chapter 6), and an overview of current understanding of fouling (Chapter 7) to pretreatment processes and process combinations with NF (Chapter 8), which concludes Part 1: Nanofiltration Principles. In Part 2, Nanofiltration Applications, the contents reflect the major, and in some cases potential, applications of nanofiltration. This takes the reader from NF in water treatment (Chapter 9) and water reclamation (Chapter 10) via NF in the food industry (Chapter 11), chemical processing (Chapter 12), pulp and paper (Chapter 13), and textiles (Chapter 14) to landfill leachates (Chapter 15), nanofiltration bioreactors (Chapter 16), photocatalytic reactors (Chapter 17), metal and acid recovery (Chapter 18), trace contaminant removal (Chapter 19), the growing area of nonaqueous applications (Chapter 20), issues of NF retentate treatment (Chapter 21, new), and use of renewable energy to provide power to NF (Chapter 22, new).

Introduction

PART 1: Nanofiltration Principles • • • • • • • •

Chapter 1: History of Nanofiltration Membranes 1960 to 1990 Chapter 2: Nanofiltration Membrane Materials and Preparation Chapter 3: Nanofiltration Module Design and Operation Chapter 4: Nanofiltration Membrane Characterization Chapter 5: Modeling Nanofiltration of Electrolyte Solutions Chapter 6: Chemical Speciation Effects in Nanofiltration Separation Chapter 7: Fouling in Nanofiltration Chapter 8: Pretreatment and Hybrid Processes

PART 2: Nanofiltration Applications • • • • • • • • • • • • • •

Chapter 9: Water Treatment Chapter 10: Water Reclamation, Remediation and Cleaner Production with Nanofiltration Chapter 11: Nanofiltration in the Food Industry Chapter 12: Nanofiltration in the Chemical Processing Industry Chapter 13: Nanofiltration in the Pulp and Paper Industry Chapter 14: Nanofiltration of Textile Dye Effluent Chapter 15: Nanofiltration in Landfill Leachate Treatment Chapter 16: Nanofiltration Bioreactors Chapter 17: Photocatalytic Nanofiltration Reactors Chapter 18: Nanofiltration in Hydrometallurgy Chapter 19: Trace Contaminant Removal with Nanofiltration Chapter 20: Organic Solvent Nanofiltration Chapter 21: Nanofiltration Retentate Treatment Chapter 22: Renewable Energy Powered Nanofiltration

PART 3: Nanofiltration New Materials • • • •

Chapter 23: Carbon Nanotube Composite Materials for Nanofiltration Chapter 24: Biomimetic Nanofiltration Materials Chapter 25: Novel polymer-based materials for nanofiltration Chapter 26: Graphene based nanofiltration

Figure 7 Nanofiltration – Principles, Applications, and New Materials book structure.

In Part 3, New Nanofiltration Materials, four new chapters enrich the book in terms of carbon nanotube composite materials (Chapter 23, new), artificial ion and water channels (Chapter 24, new), novel polymer-based materials (Chapter 25, new), and graphene-based nanofiltration (Chapter 26, new). This leaves the question – what really is nanofiltration? In the foreword of the first edition (see earlier pages), Bob Petersen (former CEO of Filmtec) concluded that “NF offers the membrane scientist a variety of

xxix

xxx

Introduction

membrane possibilities and a plethora of fascinating applications.” It is the aim of this second edition of “Nanofiltration – Principles, Applications, and New Materials” to update the science and engineering of NF membrane technology, summarize the advances over the past decade or two, and look at how this exciting field may change in the coming decades. Enjoy the new and revised journey into Nanofiltration!

References 1 Schäfer, A.I., Fane, A.G., and Waite, T.D. (2005). Nanofiltration: Principles and

Applications. Oxford: Elsevier.

2 Bowen, W.R. and Doneva, T.A. (2000). Atomic force microscopy studies of

3 4 5 6

7

8

nanofiltration membranes: surface morphology, pore size distribution and adhesion. Desalination 129: 163–172. Bowen, W.R. and Welfoot, J.S. (2002). Modelling of membrane nanofiltration – pore size distribution effects. Chem. Eng. Sci. 57: 1393–1407. BCC Research LLC (2014). Global Markets and Technologies for Nanofiltration. Report NAN045B. BCC Research LLC (2019). Global Markets and Technologies for Nanofiltration. Report NAN045C. Lu, K.J., Cheng, Z.L., Chang, J. et al. (2019). Design of zero liquid discharge desalination (ZLDD) systems consisting of freeze desalination, membrane distillation, and crystallization powered by green energies. Desalination 458: 66–75. Lejarazu-Larrañaga, A., Molina, S., Ortiz, J.M. et al. (2020). Circular economy in membrane technology: using end-of-life reverse osmosis modules for preparation of recycled anion exchange membranes and validation in electrodialysis. J. Membr. Sci. 593: 117423. Lawler, W., Bradford-Hartke, Z., Cran, M.J. et al. (2012). Towards new opportunities for reuse, recycling and disposal of used reverse osmosis membranes. Desalination 299: 103–112.

1

Part I Principles

3

1 History of Nanofiltration Membranes from 1960 to 1990 Charles Linder and Ora Kedem Ben-Gurion University of the Negev, Zuckerberg Institute for Water Research, The Jacob Blaustein Institutes for Desert Research, Sede Boqer Campus, Laboratory for Desalination and Water Treatment Research, P.O. Box 653, Beer-Sheva 84990, Israel

1.1 Overview Th s chapter describes the developments in nanofiltration (NF) membranes from the 1960s to the early 1990s that brought NF technology to its current status. NF began as a spin-off of reverse osmosis (RO) and ultrafiltration (UF) and was thus originally known as open RO, loose RO, or tight UF. Th origin of NF membranes – and indeed of most pressure-driven membranes – can be traced back to the late 1950s and the development of the Loeb–Sourirajan (L–S) anisotropic or asymmetric cellulose acetate (CA) membranes for seawater desalination. Th se membranes constituted the basis for modern membrane development in RO and UF. Within a few years, RO composites comprising a submicron coating of a selective film on an asymmetric UF support were developed. Progress in RO and UF technology gave birth to yet another discipline – NF. Th s R&D effort spanned a remarkably short period of time of about 15 years, starting in 1960. In addition, by the early 1970s, a full range of CA asymmetric (or anisotropic) membranes spanning the entire spectrum, from RO through NF to UF, were available. In the search for improved water treatment economics and for other commercial applications, the limitations of CA as a membrane material were, however, quickly revealed. Th se limitations restricted the range of applications and impeded efforts to expand NF into new areas. One approach to overcoming this problem was the development of integrally skinned asymmetric membranes from materials other than CA, such as polyamides, polyethersulfone (PES), polysulfones, chlorinated polyvinyl chloride (PVC), and polvinylidene fluoride (PVDF). Although open NF membranes could be made by this approach, the selectivity/flux combination needed for many applications could not be achieved. The breakthrough in NF took place with the invention of noncellulosic composites based on coating UF supports with a submicron selective barrier by various methods such as interfacial polymerization. Th work on composites started in the 1970s, but composite NF membranes were not widely available until the second half of the 1980s. Another approach, which followed later, was Nanofiltration: Principles, Applications, and New Materials, Second Edition. Edited by Andrea Iris Schäfer and Anthony G. Fane. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

4

1 History of Nanofiltration Membranes from 1960 to 1990

the development of NF ceramic and inorganic membranes. Today, NF has the power to solve many separation problems, but such actual applications are small in number compared to the potential applications that still await improvements in membrane stability, flux, and selectivity. Ongoing developments in NF membrane preparation and materials are described in Chapter 2.

1.2 Introduction Typically, separations of monovalent and divalent salts and organic solutes of molecular weights up to 1000 characterize membrane selectivity between the RO and UF regions. The range of membrane separation characteristics that are covered by this definition are currently known as NF. This term was not coined until the second half of 1980s, but in reality, such membranes already existed in the 1960s, being categorized as open RO, loose RO, intermediate RO/UF, selective RO, or tight UF membranes. Th beginnings of NF are intertwined with the early days of RO, which are vividly described by Loeb in his “Reminiscences and Recollections” [1]. Production of potable water from saline solutions was first demonstrated by Reid and Breton [2], working with Breton, at the University of Florida. They accomplished desalination at a low flux with a cellulose acetate (CA) membrane. Th desalination program at UCLA arrived at the use of commercially available CA membranes from a different starting point: they had been looking for the manifestation of the negative salt adsorption near the water/air interface predicted by the Gibbs equation. In 1959, Loeb and Sourirajan experimented with porous CA membranes obtained from Schleicher and Schuell (S&S), which after being heated under water acted as desalination membranes only if installed in the experimentally determined “right direction.” Loeb considered this behavior as “…the seminal feature leading to the success of RO desalination and (to the) surge of interest in…membrane separation processes.” The big step forward was the dramatic increase in desalination flux with the development of the Loeb–Sourirajan (L–S) membranes. Th y developed casting solutions resulting in anisotropic RO membranes with fluxes 10 times higher than those of the S&S membranes, with equivalent desalination. Th s development was based on the 1936 work of Dobry, who cast CA membranes from an aqueous, saturated solution of magnesium perchlorate [3]. Th mechanism of membrane formation was later termed by Kesting as phase inversion [4]. It was shown by the electron micrograph studies of Riley et al. [5, 6] that such membranes consisted of a thin (less than 1-m) layer on top of a much thicker porous sublayer. Th degree of desalting obtained with CA membranes depended on the conditions of heat treatment used to anneal and further densify the top dense layer. It was realized that the limited rejection observed with partial annealing could be exploited in various applications, later to be called NF. In the early 1970s, CA and other cellulose esters were the standard materials used for making NF membranes, but it rapidly became evident that their lack of chemical and biological stability severely limited the range of water and industrial

1.3 First-Generation NF Membranes

applications. Thus, developments after 1975 concentrated on other materials and other membrane fabrication processes, resulting in a second generation of membranes based on noncellulosic NF composites. In the second half of the 1980s, improvements in the stability, selectivity, and flux of NF membranes were reflected in a growing number of applications. NF was then being accepted as a useful unit operation for the water treatment, dairy, and chemical industries. At that time, the term nanofiltration (NF) was introduced by FilmTec; it was derived from the membrane’s selectivity toward noncharged solutes of approximately 10 Å or 1 nm cutoff. Today, NF membranes are produced in spiral wound, plate and frame, hollow fiber, capillary, and tubular configurations from a range of materials, including cellulose derivatives, synthetic polymers, inorganic materials, and organic/inorganic hybrids. A short history of the developments that have brought us to the present state of art follows. Emphasis is placed on membrane materials, chemistry, and separation mechanisms, with implicit, but full, recognition of the developments in module design and membrane fabrication and applications that have made commercial NF possible.

1.3 First-Generation NF Membranes Remarkably, in the early 1970s, a whole range of membranes including what we now call NF were commercially available. A list of such commercially available membranes taken from a 1972 review chapter by Lonsdale [7] is given in Table 1.1, covering a range of selectivity between RO and UF. As implied by the Table and article, NF membranes were not a distinct group but rather classified as either open RO or tight UF. In addition, the membranes were either asymmetric (anisotropic) or symmetric (isotropic), and the RO or NF membranes were either based on cellulose or polyelectrolyte complex membranes. 1.3.1

Cellulose Acetate Asymmetric Membranes

Th 1964 U.S. patent of Loeb and Sourirajan describes in addition to membranes with 95+% rejection, open RO membranes with rejections in the range of 20–80% [8]. As pointed out in these patents and others, a wide range of open RO selectivities could be achieved by variation of casting solution composition, evaporation period, and annealing (Table 1.2) [9]. Subsequently, other workers also found that, by incorporating additives into the casting solution, CA membranes could be formed over a wide range of molecular weight cutoffs (MWCOs) that extended from tight RO up to UF, including the intermediate NF range [7]. For example, Cohen and Loeb [10] showed how CA membranes could be cast and modified by heat treatment to form either membranes that retain sucrose and multivalent ions with sodium chloride passage or membranes that pass sucrose but retain multivalent inorganic or organic ions. A transmission electron micrograph of a Loeb–Sourirajan asymmetric cellulose acetate membrane shows the characteristic integrally skinned layered upper surface on a porous support

5

Table 1.1 Commercially available loose RO (NF) membranes in 1973. Membrane:

Manufacturer

Type

Chemical

Net pressure

Water flux

Solute

Rejection

NaCl

25

2

composition

psi

l/m d

Loeb–Sourirajan: anisotropic, unannealed

Several

Cellulose acetate

150

20 gfd

Gel cellophane

DuPont, Union Carbide

Homogeneous

100

1.5

Sucrose

15

Polyelectrolyte: anisotropic (Diaflo UM-3)

Amicon

Sodium polystyrene sulfonate-polyvinylbenzyl triethylammonium chloride

100

25

Sucrose

90

Polyelectrolyte: Anisotropic (Diaflo UM-2)

Amicon

Sodium polystyrene sulfonate-polyvinylbenzyl triethylammonium chloride

100

60

Sucrose

50

Anisotropic: Pellicon PSAC

Millipore

Cellulose ester

100

120

Sucrose

40–60

Source: From Lonsdale 1972 [7], Table 8, p. 160 with permission from the John Wiley and Sons.

%

1.3 First-Generation NF Membranes

Table 1.2 Influence of evaporation and annealing temperature on flux/rejection of Loeb–Sourirajan membranes. Membrane casting solution

Evaporation time (min)

Annealing temperature (∘ C)

Flux (GFD)

Rejection (%)

Acetone 45%, formamide 30%, 25% CA

1

23

97

25.2

Acetone 45%, formamide 30%, 25% CA

1

68.5

44

79

Acetone 45%, formamide 30%, 25% CA

1

71

25.6

88

Acetone 45%, formamide 30%, 25% CA

1

74

30

92

DMF75%/CA 25%

8

87.2

55.6

63

DMF75%/CA 25%

8

93

10.8

97

Acetone 64%/ DMF 21% CA 14%

3.5

Unheated

12.4

89

Membrane casting and evaporation step are carried out under ambient conditions. Membrane testing: 600 psi, RT, and 5000 ppm NaCl.

Figure 1.1 Transmission electron micrograph cross section of the skin and upper porous layer of a Loeb–Sourirajan cellulose acetate membrane. Source: McKinney and Rhode [11]. Reproduced with permission of ACS Publictions.

Skin

2 μm

(Figure 1.1) [12]. This new method of phase inversion was found to be a very versatile tool for forming multilayered membrane structures with a controllable wide variety of morphologies and porosities [4]. Thus, in the early 1970s, membranes based on asymmetric CA, covering the NF range, became commercially available [13] from different suppliers, including Patterson Candy International Ltd. (PCI), Westinghouse Electric Corporation, Millipore, and De Danske Sukkerfabrikker (DDS), among others. These companies offered a range of asymmetric CA with cutoffs of, for example, 80%, 50%, 20%, and 0% rejection to NaCl and 95+% for 1000 MW dextran. Th proposed uses were in water softening, fractionation of pharmaceutical fermentation liquors, whey desalting with lactose retention, skim milk concentration, fractionation of sugars, and concentration of antibiotics. One of the first applications was the

7

8

1 History of Nanofiltration Membranes from 1960 to 1990

treatment of drinking water sources with membranes that were relatively nonfouling and had some chorine resistance. In Florida, they were used for water softening as long ago as 1976 [14]. Commercial NF membranes based on CA, alone or in blends, were subsequently put into a variety of uses for water treatment, especially water softening and color removal from surface water. Although asymmetric CA had certain desirable characteristics, such as low fouling for some water sources, relative ease of cleaning, and chorine resistance, the limitations of this membrane material were quickly revealed when improved water treatment economics and other commercial applications were sought. 1.3.2

Deficiencies in Cellulosic Membranes

Th limitations of cellulosic membranes were primarily their poor biological and chemical stability (e.g. hydrolysis of the acetate groups), resulting in continual changes in rejection and flux loss because of compaction. In addition, although it was possible to cast a membrane with any given rejection within the NF range, initial fluxes were often not sufficiently high for many applications. It was quickly realized that while NF had the potential for application in a large number of processes, especially in the chemical industry, fulfilling this potential would, however, require membranes other than CA or polyelectrolyte complexes. To realize the full market potential of NF, the development catch phrase of the second half of the 1970s became stable noncellulosic membranes (produced at least in part by the powerful tool of asymmetric casting). From 1975, membranes with the following characteristics were sought: • Improved solvent, oxidant, pH, biological, and mechanical stability. • Selectivities and fluxes that would facilitate economically favorable replacement of two or more processes with a single process, e.g. the simultaneous concentration and purification of product streams. • Very high retention to organic solutes (e.g. 99+%), low rejections of inorganic salts, and high water flux. • For water softening and purification, higher rejections of divalent salts and organic solutes, monovalent salt passage, high fluxes with good compaction resistance, and chlorine resistance. 1.3.3

Polyelectrolyte Complexes

During the heyday of CA membrane development, Amicon Co. offered, in the 1960s, NF-type anisotropic membranes of polyelectrolyte complexes made by electrostatic interaction between strongly acidic and basic polyanions and polycations, respectively [15]. Invented by Michaels, membranes covering the whole range between RO and UF could be made by this approach. A series of membranes with MWCOs of 1000, 500, and 380 (sucrose) were commercialized for use in the concentration and demineralization of proteins and organic solutes. Th se membranes never achieved the same widespread application as asymmetric CA NF membranes, possibly because of their relatively low mechanical strength, flux loss because of compaction, and variable separation characteristics in high ionic strength solutions [7].

1.3 First-Generation NF Membranes

1.3.4

Polyamide Membranes

Starting in the 1960s, DuPont and Monsanto began using their extensive fiber technology to develop asymmetric hollow fibers of aromatic polyamides for RO seawater desalination [11, 16]. These polyamide membranes could also be made in the NF range by adjustment of the properties of the casting solution [17]. Although relatively hydrophobic, polyamide membranes gave good rejection, but they could not achieve the fluxes needed for many applications, and their chlorine resistance was poor. When more hydrophilic polyamides were used, higher flux was achieved, but it declined steadily under pressure because of compaction. In addition, the selectivities of the more hydrophilic polyamides were often too low. Th introduction of ionic groups into the polymeric structures, for making membranes from aromatic polyamides (which originally gave good selectivity but low flux), improved the permeability but lowered the rejection. In general, casting of asymmetric membranes from polyamides could not be optimized to compete with existing separation processes or with the new technology of composite membranes. 1.3.5

Polysulfones and Other Polymer Membranes

Many other polymeric materials were investigated to make asymmetric RO and NF with improved chemical stability. Th s effort was guided by the electron micrograph studies of Riley et al. [5, 6] on the ultrastructure of anisotropic membranes and by the extensive work carried out on the phase inversion process by Kesting et al. It was shown that almost any polymer that forms a homogeneous solution in a solvent and a homogeneous precipitate could form asymmetric skin structures [4, 18–21]. Asymmetric membranes could be made from polycarbonates, chlorinated PVC, polyamides, polysulfone, PES, polyphenylene oxide, PVDF, polyacrylonitrile (PAN), copolymers of PAN/PVC, polyacetals, polyacrylates, polyelectrolyte complexes, and cross-linked polyvinyl alcohol (PVA). To a certain extent, the performance of the above-mentioned polymers as membrane-forming materials could be correlated with their hydrophobic/hydrophilic balance. Based on this classification, it was rapidly discovered that for asymmetric NF membranes, many hydrophobic polymers had too low flux or lacked selectivity, while hydrophilic polymers lost flux because of compaction. Achieving the optimum degree of cross-linking to prevent the swelling of hydrophilic polymers was also difficult. Open asymmetric NF membranes with a MWCO of 1000 could, however, be made with some hydrophobic polymers such as polysulfone and PES; these membranes demonstrated good chemical and mechanical stability and reasonable flux [22]. They could not, however, be cast into selective NF membranes with lower MWCOs, such as for sucrose, without losing flux. Increasing the hydrophilicity of polyarylether sulfones by sulfonation [23] to improve flux did not work because to achieve the desired flux, the degree of sulfonation had to be increased to the point that reduced rejection. Carboxylation of polysulfone was tried by Guiver et al. [24] as a substitute for sulfonation to give high flux and selectivity with limited swelling. Model et al. [25] used the hydrophilic polymer

9

10

1 History of Nanofiltration Membranes from 1960 to 1990

polybenzimidazole, which could be cast into NF asymmetric membranes with a range of MWCO as a function of casting solution and coagulation bath formulations. In a similar approach, Bayer developed from sulfonated polybenzoxazindione membranes with an MWCO of 300. Th se membranes were, however, not developed commercially for NF, possibly because of the high cost of polymeric materials and/or because membranes with sufficiently high flux could not be made due to compaction.

1.4 Early Studies of Charged Reverse Osmosis (Hyperfiltration) Membranes 1.4.1

Dynamic Membranes

An inexpensive route to producing membranes rejecting salt by Donnan exclusion was envisioned by Kraus and his group working at the Oak Ridge National Laboratory [26]. By depositing polyelectrolytes on a robust support, a charged membrane could be formed. If the membrane became damaged or clogged, it could be removed or regenerated, hence the term dynamic membrane. Salt rejections of 25–85% could be achieved by circulating low concentrations of polymeric electrolytes, such as vinylbenzyl trimethylammonium chloride or polystyrene sulfonic acid, and depositing the polymers on a porous support [26]. As a transport barrier, this type of membrane was classified as NF in terms of its specific water permeability and MWCO [27]. Dynamic membranes were used to recover dyes and sizing materials in the textile industry [28]. 1.4.2

Polyelectrolyte Membranes

Salt exclusion by membranes carrying fixed charges and the general properties of such polyelectrolyte membranes have been well known to physiologists for many decades; such membranes have been discussed by Meyer and Sievers [29] in the 1930s and by Teorell [30] in 1953. Th salt rejection expected in hyperfiltration through collodion membranes, chemically modified for carrying a fixed charge, was later calculated from the Teorell–Meyer–Siever model by Hoffer and Kedem [31, 32]. The expected dependence of rejection on fixed charge density, salt concentration, and valency of the ions was subsequently confirmed experimentally [33, 34]. It was thought that separation between ions of different salt valencies might be a useful concept for water treatment, but this idea was not carried out in practice for a long time. Instead, the development of charged porous membranes resulted in an early industrial application for NF [35].

1.5 Early Models of NF Selectivity Models to interpret NF selectivity performance were proposed and analyzed from the very beginnings of NF applications covering processes where the rejection was dependent on charge/noncharged, molecular size, and concentration.

1.5 Early Models of NF Selectivity

Th models covered a range of selective processes for membranes with an MWCO of 150–1000 (in effect between RO and UF). Membrane selectivities have been interpreted by a number of models, each suited to a particular range. Some of the models were originally developed for RO, but in reality, they were more applicable and easily adapted to NF. Water flow with retention of various solutes has been studied by generations of physiologists because it is a vital function in living organisms. Developed in the early 1950s, models for exclusion by size, such as that elaborated by the group of Renkin [36, 37], may be used for analysis of transport phenomena in synthetic membranes, including those in the upper limit of the NF range. In this type of model, an effective membrane area is defined, depending on the ratio between the molecular radius of the permeant and the pore radius. Th size of the solutes limits both the probability of entrance into the pore – even if the pore radius is larger than the molecular radius – and the rate of movement through the pore. Selective ion transport is another physiological function that is also an important NF characteristic. As already mentioned, Meyer and Sievers [29] and Teorell [30] sought to understand this phenomenon through their fixed charge model. It is clear today that this was a gross oversimplification for biomembranes, but in NF technology, salt exclusion from the membrane as a consequence of fixed charges, the well-known Donnan exclusion, is a basic mechanism of selectivity. The quantitatively predictable features of Donnan exclusion enabled the preparation of charged-ion-rejecting membranes in the early 1970s [34]. It was possible to relate the rejection of ions of different valencies to the known thermodynamic properties of polyelectrolyte solutions, assuming a homogeneous distribution of ions in the pore volume [32]. This assumption is justified for narrow pores having a pore radius smaller than the thickness of the diffuse double layer. As long ago as 1965, Dresner [38] calculated nonhomogeneous ion distribution in wider charged pores. In 1973, Simons and Kedem [39] performed a detailed calculation of rejection in an assembly of rectangular slits in an ion exchange matrix, taking into consideration both the velocity profile and the ion distribution in the pore. Rejection of ions from a mixed electrolyte feed, which was different from the rejection of each salt separately – now a major application of NF – was predicted by Dresner [40]. A major source of basic ideas in membrane development was classic colloid and interfacial science. Th surface tension of salt solutions is higher than that of pure water, and thus, the Gibbs equation relating the interfacial concentration to surface tension predicts a salt-poor region at the air/water interface [41]. With the aim of carrying out surface skimming, the UCLA group on seawater desalination initiated their research effort with the exploitation of this phenomenon in mind. In the 1977 book edited by Sourirajan [12], the performance of the L–S CA membrane and its interpretation are discussed in detail. The well-known preferential sorption capillary flow mechanism is described by Sourirajan in the first chapter (referring to his work of the 1960s): in this work-up, the surface membrane is microporous and heterogeneous; “pore” or “capillary” refers to any connecting void space, regardless of its origin or size; preferential sorption,

11

12

1 History of Nanofiltration Membranes from 1960 to 1990

positive or negative, takes place at the pore wall/fluid interface and the desalted layer is continually removed under pressure. Th Sourirajan concept was subsequently modified by Glueckauf [42], working together with Russel and coworker [43]. In a model calculation presented at the First International Symposium on Water Desalination in 1965, Glueckauf [42] showed that salt exclusion from the interface with a medium having a low dielectric constant is more pronounced in a narrow cylindrical pore than near a flat surface. The e must then be an optimal pore size that is small enough to lead to dielectric exclusion of ions but large enough to allow water flow. His estimate of the optimal pore diameter was less than 6 Å. Water would enter such a narrow space only if the matrix was sufficiently hydrophilic. Glueckauf thus concluded that the special combination of properties of CA that makes it suitable for hyperfiltration is a low dielectric constant and sufficient hydrophilicity. Salt rejection by dielectric exclusion was further elaborated by Bean [44], and his overestimate of rejection was probably because of his neglecting of the screening by the salt itself. Yet another approach was taken by Kraus et al. [45] who considered the membrane as a continuous organic phase that dissolves water but does not dissolve salt. To back up this idea, they measured salt and water distribution with solvents closely related to CA used to make L–S anisotropic RO membranes. In their paper of 1964, when the existence of a thin dense selective layer was “almost certain,” they concluded from their data and theory that “the effective thickness of such a membrane is presumably of the order of 0.1 m.” Th models described above are basically related to one another far more closely than is apparent from their formal presentation. The phenomena of salt-free layers close to the polymer and of low solubility of salt in the membrane phase are both related to a low dielectric constant of the polymer. Th distinction between the pore model and the solution/diffusion model becomes blurred if pores are of molecular dimensions [46, 47]. None of these models explain the unexpected specificity of seawater RO salt rejection. After all the intense efforts of polymer chemists, only very few polymers show the high salt rejection needed. NF membranes can, however, be prepared from a variety of materials. The nonspecific pore models developed for RO do, in fact, work for NF, and similarly, the Spiegler/Kedem flux equations developed for RO are applicable to NF [48]. Moreover, just as NF is a process lying between UF and RO, models from both these areas and their combinations can be used to represent NF performance. Early NF membranes used to separate salts from dyes were based on a combination of size exclusion and fixed charge exclusion (unlike the salts, the large charged dye ions cannot be drawn into the center of the pores). In thin layer polyamide NF membranes with partial salt rejection, dielectric exclusion is probably the major factor enabling separation. More porous NF relies on Donnan exclusion. The currently accepted theory for ion transport in NF membranes seeks to combine the effects of the dielectric constant of the medium and of fixed charges as a function of pore size (see Chapters 4–6 for a more extensive description of NF selectivity models and mechanisms).

1.7 Early Development of Industrial NF: Ionic Modification of Asymmetric Cellulose Acetate

1.6 Negative Salt Rejection 1.6.1

Solutions of One Electrolyte

As mentioned above, the salt rejection of charged membranes can be described by the model (TMS) based on Donnan exclusion. It was, however, realized that salt rejection depends not only on salt distribution but also on the ratios between the mobilities of the ions. For the extreme case of some acid filtered through positively charged membranes, negative rejection, i.e. enrichment in the product solution, was predicted and obtained experimentally [33]. Negative salt rejection is closely related to the so-called anomalous osmosis, leading to volume flow from the concentrate into the dilute solution in the absence of a pressure gradient. This was observed by the pioneers of membrane science, Sollner and coworker [49] and Schloegl [50]. Negative rejection of a single salt-comprising cations and anions of similar mobilities is obtained in “mosaic” membranes containing small regions of anion and cation elements. It was considered for some time that this effect could serve for desalting [51]. The considerable efforts devoted to these systems have been reviewed by Leitz [52]. 1.6.2

Separation by Negative Salt Rejection

Th salt exclusion originally described by Donnan is obtained when a membrane separates a solution containing charged macromolecules and salt from a solution of salt only. At equilibrium, the salt concentration in the mixed “inside” solution is smaller than that in the outside. When the solutions are separate by an ultrafiltration membrane and pressure is applied, salt will be enriched in the product. Such negative salt rejection was predicted and observed by Lonsdale et al. [53] in the hyperfiltration of citrate and chloride and by Akred et al. [54] in the ultrafiltration of gelatin solutions containing calcium or sodium salts. The principle of negative rejection is illustrated in Figure 1.2. Th technically important negative salt rejection from mixtures containing charged molecules of medium molecular weight (200–1000 Da) can be achieved with nanofiltration membranes of suitable cutoff. This is feasible in principle with any type of NF membrane – charged or neutral.

1.7 Early Development of Industrial NF: Ionic Modification of Asymmetric Cellulose Acetate By about 1972, ecological issues were beginning to become a cause for concern to industry, especially industrial entities in Europe that found themselves in heavily populated areas. In activities such as dye production, large quantities of salty dyed polluted water were being discharged into waterbeds and rivers. One solution to this problem was to apply the new membrane technology that was becoming successful in RO applications such as water desalination and in UF applications such as protein separations. However, many industrial waste

13

14

1 History of Nanofiltration Membranes from 1960 to 1990 Feed NaCl

Cs′ Pressure Pfeed

Na+

Multivalent ion Cx-v

From Ref. [55], negative rejection of a permeating salt in the presence of a highly rejected ion, can be defined by the following equations: For permeating salt alone: Rs = (1–F)σ/[(1–σ)F] (1) F = exp(-Jv A), A = (1–σ) Δx/Pav = (1–σ)/P For permeating salt and highly rejected ions: Define: β = (1 + νC′x/c′s)0.5 Then, Rs =1– [(1–σ)β]/(1–σF) (2)

Ppermeate

Permeating NaCl

Cs ″

When only permeating salt is present, then C′x = 0, β = 1 and eqn (2) reduces to eqn (1)

Rs = NaCl salt rejection, σ = reflection coefficient, C′x = concentration of multivalent ion, ν = number of charged groups, C′S = salt concentration

Figure 1.2 Principle of negative salt rejection in the presence of highly rejected ions [55].

streams presented special problems for the new membrane technology: In RO, the salt concentrations were so high that any concentration effort was doomed by uneconomically low flux because of the osmotic forces of the rejected salt. UF was not efficient, as both the dye and the salt permeated through the membrane. One of the efforts to overcome these difficulties led to the founding of a small company – known as RPR – in the early 1970s by Bloch and Kedem, which used open NF asymmetric cellulosic membranes modified with reactive dyes. This approach to membrane modification was an early example of how existing membranes could be chemically modified to achieve valuable separation characteristics. The modification in this case formed charged groups on the pore walls and at the same time stabilized the membrane structure by cross-linking. With these modified membranes, 99% dye rejection was achieved. It was found that these membranes could be advantageously applied not only to wastewater treatment but also to dye production because of the then-amazingly effective purification effect by negative salt rejection. Th s was explained to be a consequence of Donnan equilibrium, as described in Section 1.6. The salt passed through the membrane easily and was, in practice, equilibrated between the feed and permeate. Th feed contained the large impermeable dye anions, small sodium counterions, and a high concentration of salt, while the permeate contained only small ions, which had been pushed into the permeate by the highly rejected dye counterions. Thus, in the late 1970s, a single-unit operation could be used for both concentration and purification. The first tubular pilot units based on modified cellulosic membranes very rapidly became the core of production machines that

1.8 Early NF Composites

concentrated and desalted simultaneously, thus saving vast quantities of the salt need to precipitate the dye and solving the original development problem of reducing salty/dyed discharge. The membranes were used on an industrial scale by dye manufacturers to desalt and concentrate dye solutions. In one process, both a production and an ecology problem had been solved! A formal description of this Donnan effect in dye solutions was published only much later (1989) after it had long been understood and put into practice [55]. The incorporation of a Donnan distribution term into the flux equations of Spiegler and Kedem [48] defined earlier could explain the strong negative salt rejection found in dye processing (shown schematically in Figure 1.2) and its concentration dependence. Negative salt rejection because of the Donnan effect may be observed for any type of NF membrane, whether charged or uncharged. Cellulosic membranes were, however, not chemically stable and suffered from flux decline after short periods. As experience was gained, it was realized that there were still some significant drawbacks associated with the modified CA membranes: they allowed relatively large quantities of dye to pass through them, which constituted an economic loss, and membrane lifetime was not sufficiently long. It was also realized that tighter membranes, which could operate at higher and lower pH values, would have many other uses. Starting in late 1970s, tighter noncellulosic membranes were developed in RPR’s manufacturing company, Membrane Products Kiryat Weizmann (MPW).

1.8 Early NF Composites 1.8.1

General

By 1975, it became apparent that asymmetric NF membranes from a single polymer or polymer mixtures could not give the characteristics of selectivity and flux needed to compete with standard technologies in many applications. Attempts to make asymmetric NF membranes by casting polymers with the hydrophilic/hydrophobic balance of CA did not produce membranes with sufficient flux or rejection if they were too hydrophobic or with flux stability if they were too hydrophilic. Workers such as Rozelle, Cadotte, and Riley and their colleagues had resolved a similar dilemma for RO in the early 1970s with the development of composite membranes [56, 57]. They produced high-salt-rejecting membranes by placing a very thin selective layer over one surface of a finely porous asymmetric UF membrane. Such a composite membrane was produced either by coating the UF membrane with a thin CA film or by carrying out interfacial cross-linking of polyamines (for example, polyethyleneimine [PEI]) with isophthaloyl chloride (IPC), toluene diisocyanate (TDI), or other aromatic cross-linkers. Th latter approach appeared to be the key breakthrough. Th se composite membranes exhibited important advantages over integrally skinned asymmetric membranes in that the selective barrier film and the support could be optimized independently. With this process, a variety of chemical combinations and methods could be used to form thin barrier coatings, including the use of linear and cross-linked

15

16

1 History of Nanofiltration Membranes from 1960 to 1990

polymers, whereas asymmetric membrane formation was limited to processable stiff linear polymers. In Cadotte’s patent of 1977 [58] and NTIS report 1968 of Rozelle et al. [59] on composite membranes, comparison examples were given of other composites made by different cross-linking reagents, which showed significantly lower salt rejections compared to the claimed RO membranes. Th se more open membranes were, however, what we would now term composite NF. In general, it may be said that the composite approach is applied to form a selective layer that is both thin and sufficiently hydrophilic to give high water flux but at the same time cross-linked to the extent required for NF selectivity. In addition to the original method of interfacial cross-linking of polymers for forming composites, other methods have included interfacial polymerization (Section 1.9), plasma polymerization (Section 1.8.2), polymer coating and curing, and surface modification (see Section 1.8.3). Th first composites – RO composites – were based on cellulose nitrate UF supports, but they suffered from the same lack of biological, chemical, and mechanical stability that had limited CA in RO and UF applications. Very early in the 1970s, polysulfone was recognized as the material of choice for UF porous supports because such membranes combined high surface porosity with minimal pore diameter and high chemical and mechanical stability. They could be readily cast and optimized (high degree of porosity and controlled pore size distribution) to give the asymmetric structures for commercially valuable RO or NF composites [56, 57]. A number of other polymeric materials were also investigated [58], such as polycarbonate, chlorinated PVC [56], polyamide, PVDF, PAN and styrene/acrylonitrile copolymers, polyacetals, and polyacrylates. From these investigations, it was found that polymer molecules making stable compaction-resistant supports are inherently stiff chains capable of hydrogen bonding or polar and hydrophobic bonding, giving networks with low chain mobility. UF membranes based on chemically stable aromatic engineering plastics, such as polysulfone or PES, have currently become the standard supports for composite RO and NF membranes. Different UF membrane morphologies could be used for making composites; however, a typical asymmetric polysulfone is shown in Figure 1.3. The membrane comprises an integral “tight” skin layer of about 0.1–0.7 m, a larger pore intermediate sponge layer of 1–5 m, and a 80–100 plus m thick, open support layer with large finger-like pores. Other UF supports have a sponge-like structure instead of fingers in the porous layer. Figure 1.3 Scanning electron micrograph of UF PES membrane used for making NF composites.

1.8 Early NF Composites

If 1960 marked the beginning of the development of asymmetric membranes, then about 1969 marked the beginning of the use of asymmetric UF membranes for making composites. From 1969, Cadotte and Rozelle [58, 60] and then Wrasildo, Riley, and coworkers [61, 62] showed that high-rejection RO composites could be made by interfacial cross-linking of coated hydrophilic polymers such as PEI or polyepiamine with IPC or TDI on polysulfone or chlorinated PVC UF membranes. As a spin-off of this development on single-pass water desalination membranes, Cadotte and others also described open RO or NF membranes. For example, in 1972, open RO (NF) composite membranes were made from the interfacial reaction of low molecular weight polyamines and teraphthaloyl chloride (Figure 1.4) [60]. Similar membranes were also made from mixtures of PEI with different cross-linkers, as described in a 1977 patent (Figure 1.5) [58]. These membranes were not, however, commercially viable and were superseded by the piperazineamide composites described below. 250

1,3 Propanediamine Test condition: 1500 psi/3.5% NaCl

Flux

200 150 100

1,6-Hexadiamine Piperazine

50

PEI

1,4-Benzenediamine

0 8

31

38 74 NaCl rejection (%)

96

Figure 1.4 Flux (gallons/ft2 d) vs. salt rejection (%) for early open RO membranes (NF) prepared by interfacial polymerization of teraphthaloyl chloride with different polyamines (NTIS report No. PB-229337, November 1972) [60].

200 Test conditions: 1500 psi/3.5% NaCl

Flux

150

Glyoxal Oxalyl chlorid

100 Phosgene

50

Glutaraldehyde

IPC

Divinylsulfone

0

15.5

17.8

28.2 66.2 Rejection %

74.5

99.15

Figure 1.5 Flux (gallons/ft2 -d) vs rejection (%NaCl) for different interfacial membranes from US Patent 4, 039, 440 (1977) [58].

17

18

1 History of Nanofiltration Membranes from 1960 to 1990

1.8.2

Plasma Polymerization

Another approach to the preparation of composites, which first appeared in the 1960s, was plasma polymerization of coated films on a microporous support. In his 1977 review of the subject, Yasuda [63] reported the use of plasma polymerization of different monomers (e.g. 4-vinylpyridine, N-vinyl pyrrolidinone, pyridine 1-methyl-2-pyrrolidinone, thiophene, and thiazole) on polysulfone supports to produce very thin selective membranes. Although the main goal was the production of RO membranes, NF membranes produced from a variety of monomers were also prepared. Plasma processes were also used to modify the surfaces of UF membranes to bring them into the NF range. For example, Sano [64] used plasma polymerization with He and H2 on UF PAN to produce commercial RO membranes by sealing and hydrophilizing the membrane surface. This method can be adapted to give NF membranes; for example, Lai and Chao [65] modified Nylon 4 microporous membranes with gas plasma to produce membranes with 74% rejection to NaCl. Polyarylsulfone UF membranes were also plasma-treated by Sano and his coworkers to give 96.3% rejection, where the original membrane had exhibited 0% rejection. Such membranes could also be expected to give NF under different preparation conditions. 1.8.3

Graft Polymerization

Graft polymerization of nonionic or ionic vinyl monomers by a variety of methods, i.e. with ionizing radiation such as gamma rays from 60 Co, by photochemical means, or by chemical initiation on asymmetric membranes has been carried out in an effort to improve RO performance. In the 1960s and 1970s, Stannett et al. [66] grafted styrene on CA membranes. Although the results were not encouraging for RO, they did give, in some cases, membranes with NF properties. The fluxes were, however, too low for the membranes to be of commercial value. In another study, Kesting and Stannett [67] showed that the attachment of acrylic monomers could be used to increase the starting membrane permeability to salt and possibly to bring RO membranes into the NF range. In yet another study, double grafting of PVC films with 2-vinylpyridine and acrylic acid gave membranes with a high flux per unit thickness with NF salt rejections of 65% [68]. However, the absolute flux in the dense films was low, and only by casting asymmetric membranes could this approach be of practical value.

1.9 NF Composites of the 1980s 1.9.1

Piperazineamide Membranes

Commercial NF composites were not generally available until the second half of the 1980s, even though their development began in the late 1970s [28]. One of the first successful approaches was based on the interfacial polymerization of an aqueous piperazine film on a polysulfone UF support by hydrophobic aromatic

1.9 NF Composites of the 1980s

3500 Test conditions: 200 psi/0.5% MgSO4

3000

Flux

2500 2000

0.5% NaCl has 50% rej flux of 1710lmd x

1500 1000 500 0 0

0.1 0.2 0.33 Fraction of TMC in TMC/IPC mixtures

1

Figure 1.6 Flux (l/m2 d) vs TMC/IPC ratio of piperazineamide interfacial membranes with 99 + % rejection to MgSO4 (US Patent 4, 259, 183 (1981)) [69].

cross-linkers. Many such NF products were made by a number of different companies. Polypiperazineamide NF composites were prepared when Cadotte et al. [69] replaced IPC with trimesoyl chloride (TMC), making NF membranes with high MgSO4 rejection (99%) and low NaCl retention (5 l/m2 h bar

Microdyn Nadir

PES

Acid and caustic environments

35–75% (Na2 SO4 )

>1 l/m2 h bar

®

NP030

OSN/SRNF applications Type

Manufacturer

Chemistry

Resistant in

Rejection

Solvent flux

Puramem

Evonik

Polyimide (P84)

Alcohols, aliphatic hydrocarbons, aromatic hydrocarbons, butyl acetate, ethyl acetate, methyl-ethyl-ketone, methyl-tert-butyl-ether

90% for 280 g/mol styrene in toluene

18 l/m2 h toluene at 30 barg

Duramem150

Evonik

Modified polyimide (P84)

Type1: acetone, tetrahydrofuran, methanol, ethanol, methyl-tert-butyl-ether, methyl-ethyl-letone, methyl-isobutyl-ketone, butyl acetate, ethyl acetate Type2: dimethylformamide, dimethylsulfoxide, N-methylpyrrolidone

90% for 150 g/mol styrene in toluene

n.s.

oNF-2

Borsig GmbH

Silicone polymer-based composite type

Alkanes, aromatics, alcohols, ethers, ketones, esters

350 g/mol

n.s.

NF030306

Solsep BV

Silicone-based polymer

Alcohols, esters, ketones, aromatics, chlorinated solvents, reducing atmosphere, THF, aldehydes, crude alkanes, petroleum ethers

99+% for 500 g/mol dye in ethanol

∼2 l/m2 h bar acetone (T = 20 ∘ C) ∼5 l/m2 h bar acetone (T = 80 ∘ C) ∼2 l/m2 h bar veg. Oil (T = 140 ∘ C)

NF010206

Solsep BV

n.s.

Aldehydes, ketones, crude alkanes, acetone, (m)ethanol, IPA, hexane, petroleum ether, ethyl acetate, MEK, methylbenzol, methylchloride, chlorobenzol, tetrachloroethylene

90% for 300 g/mol FA in acetone

n.s

NF010306

SolSep BV

n.s.

Aldehydes, ketones, crude alkanes, acetone, (m)ethanol, IPA, hexane, petroleum ether, ethyl acetate, MEK, methylbenzol, methylchloride,

99+% for 500 g/mol dye in ethanol

∼2 l/m2 h bar acetone (T = 20 ∘ C) ∼2 l/m2 h bar veg. Oil (T = 140 ∘ C)

NF300705

SolSep BV

n.s

Alcohols

95+% for 500 g/mol dye in ethanol

∼2 l/m2 h bar hexane (T = 20 ∘ C) 0.2–0.4 l/m2 h bar heptane

NF070706

SolSep BV

n.s

Aromatics and alkanes (e.g. toluene, hexane, heptane)

90+% for 250 g/mol sterols in heptane

∼1 l/m2 h bar toluene (T = 20 ∘ C) ∼1 l/m2 h bar heptane

PEEK-SepTM

Porogen (air liquide)

PEEK

n.s.

n.s.

n.s.

PEEK5,20 100

NovamemTM

PEEK

n.s.

n.s.

n.s.

PVDF20,100

Novamem

PVDF

n.s.

n.s.

n.s.

n.s.: not specified by manufacturer.

76

2 Nanofiltration Membrane Materials and Preparation

2.8.2

Novel (SR)NF Membranes

Besides tuning membrane performance through the use of novel monomers, additives or by changing reaction conditions or post-treatment steps, as discussed before for both ceramic and polymeric membranes, new generations of NF membranes are getting increased attention. Especially in the field of polymeric membranes, disruptive findings have been reported. Th embedding of (porous) nanoparticles in the thin PA top layer of thin film nanocomposite (TFN) membranes, giving rise to the so-called TFN membranes, has gained a lot of attention lately. Tremendous increases in permeance are observed, with no or only slight losses in rejection. The exact functioning of these membranes is however poorly understood, demonstrated by large variations in final membrane performance as a function of the incorporated amount of filler. It is speculated that the high water permeances are not only caused by low-resistant flow paths through the embedded porous NPs but also by intrinsic variations in the polymer structure because of the presence of the NPs during synthesis [199]. Examples of embedded NPs are ZIFs, MOFs, PIMs, zeolites, carbon (quantum) dots, COFs, graphene oxide, and silica NPs. Th incorporation of aquaporin, a biological membrane protein, into polymeric membranes has also been proposed as a new pathway to overcome the conventional permeability–selectivity trade-off. Aquaporin is intrinsically highly permeable to water and is also able to selectively reject ions, protons, and neutral solutes. One major challenge is to synthesize defect-free membranes and to overcome both membrane and protein stability [6, 200]. CNTs and other synthetic nanochannels, such as m-phenylene ethynylene or peptide nanotubes [201, 202] or cyclic macromolecules [203], have also been explored for NF and RO purposes as they can be functionalized for specific separations and the diameter of the channels can be customized. Even though these materials have demonstrated excellent separation capabilities, both through experiments and modeling, scale-up is still a major hurdle [200]. Novel nonpolyamide polymers for NF have also been explored. Narrow pore size distribution can be achieved through equilibrium self-assembly of lyotropic or thermotropic liquid crystalline mesophases [204] or through block copolymers [205]. Separations through liquid crystalline mesophasic membranes currently already fall into the NF range [200] and block copolymers also represent a scalable alternative to conventional NF membranes as they can be synthesized through phase inversion. However, elucidating the structure–property–performance relationships of these membranes is of paramount importance for large-scale manufacturing [205]. The use of epoxides as a novel reagent for IP has also been reported. The resulting loose NF membrane with a poly(epoxyether) top layer is chlorine and acid-resistant and has enormous tuning possibilities [206]. Coordination between tannic acid, a polyphenol, and Fe(III) can rapidly form a thin film on PES, resulting in a green TFC membrane, which is able to selectively reject endocrine-disruptive compounds and divalent salts [207]. N-methyl-d-glucamine-assisted polydopamine coating of a PES support membrane resulted in high water fluxes, high dye and divalent salt rejection, and high monovalent salt permeation. Furthermore,

2.9 Outlook

chemical stability in acid and caustic conditions makes this membrane an attractive alternative for polyamide-based NF membranes [208]. An overview of reported polymers for phase inversion and reported monomers for IP used for OSN/SRNF applications is given in [187]. Solvent-resistant mixed matrix and TFN membranes can also be used in this field but are not (yet) commercialized. Recent advances with respect to novel SRNF membranes are mainly based on cross-linking existing polymers or the use of completely new chemistries. Th alkyne functionalization of porous polyoxindolebiphenylene, polytriazole, and polybenzimidazole membranes, for example, allows full stability in DMF, even at 140  C, although the performance is currently still in the UF range [209]. Examples of novel chemistries for IP are the reaction between tannic acid or morin hydrate and TPC on cross-linked PAN, resulting in NMP-resistant membranes [210, 211], or between catechin and TPC on a cellulose substrate, resulting in fully bio-derived TFC membranes stable in DMF over 30 days [212]. Th coating of a microfiltration membrane with ultrathin layers of solvated reduced GO allowed ultrafast acetone transport (215 l/m2 h bar), while achieving rejection in the NF range. Stability in methanol and ethanol, as well as in oxidizing, alkaline, and acidic media, was also achieved [213]. In a similar manner, highly laminated GO membranes outperformed state-of-the-art polymeric membranes with regard to methanol permeance and dye rejection [214].

2.9 Outlook NF technology has become a state-of-the art process for separations at industrial level. RO membranes reject almost all solutes, except for a few small neutral organics (e.g. 1,4-dioxanes and N-nitrosodimethylamine), while the rejection of NF membranes is depending on the characteristics of the solute (size, charge, charge density, etc.). In theory, this allows NF membranes to be tailor-made for a specific application; however, the lack of knowledge of the structure–property–performance of the membrane is still the limiting factor to achieve the full potential of NF. Research should therefore focus on the nanoand molecular control of materials to achieve stronger correlations between synthesis conditions and membrane performance. Additionally, a membrane separation process is traditionally seen as the removal of compound X from solvent Y. However, the selective permeation of one solute over another, which requires development of membranes with very narrow MWCO curves, would further open up the application potential of (SR)NF. As for SRNF, its potential in the chemical, food, and pharmaceutical industry is enormous and is steadily getting more attention. As this field has been entered much more recently than the aqueous, many improvements in terms of membrane selectivity and (long-term) stability, as well as solvent permeance can still be made. Obviously, this field is much more complex because each solvent has its own specific interactions with a given solute and membrane material. This tremendously increases the level of complexity and the difficulty to fully grasp membrane structure–performance relationships as compared to aqueous applications, which take place in only one, well-understood “solvent.”

77

78

2 Nanofiltration Membrane Materials and Preparation

Even though many new materials have emerged as possible candidates for (SR)NF applications, their difficult and expensive large-scale synthesis is limiting scale-up. Better understanding the pitfalls that prevent defect-free membrane synthesis at large scale is therefore mandatory to transfer these technologies from academia to industry. In the same regard, more attention should be focused on module and process design, as, in the end, the membrane itself is only one part of an even more complex overall process.

Acknowledgements Th authors are grateful for the financial support from the OT (11/061) funding from KU Leuven, the Nanomexico IWT.150474 project from IWT-STW, an FWO research project (G0D5119N), the I.A.P. – P.A.I. grant (IAP 7/05 FS2) from the Belgian Federal Government, and the long-term Methusalem (CASAS) funding by the Flemish Government. R.V. thanks Research Foundation Flanders for her SB PhD grant (1S00917N).

Abbreviations AEPPS ALD BHTTM CA CFIC CHMA CPDA DABA DGDE DMF DMSO EB HTC ICIC IP IPA IPC LbL MDEOA mm-BTEC M-m-PDA m-PDA M-p-DPA

aminoethyl piperazine propane sulfonate atomic layer deposition 2–2′ -bis(1-hydroxyl-1-trifluoromethyl-2,2,2trifuoroethyl)-4–4′ -methylenedianiline cellulose acetate 5-chloroformyloxyisophthaloyl chloride 1,3-cyclohexanebis(methylamine) 2-chloro-p-phenylenediamine 3,5-diamino-N-(4-aminophenyl) benzamide diethylene glycol diethyl ether dimethylformamide dimethylsulfoxide electron beam cyclohexane-1,3,5-tricarbonyl chloride 5-isocyanatoisophthaloyl chloride interfacial polymerization isopropanol isophthaloyl chloride layer by layer methyl-diethanolamine 3,3′ ,5,5′ -biphenyl tetraacyl chloride 4-methyl-meta-phenylenediamine meta-phenylenediamine 2-methyl-para-phenylenediamine

References

MW MWCO NF NMP N,N ′ -DMMPD NTSC o-PDA PA PAH PAN PBI PEEK PEG PEI PEI-g-SBMA PES PHGH PI PIM PIP p-PDA PSf PTMSP PVAm PVDF PVP RO SPES-NH2 SRNF TEA TEOA THF THM TFC TMC TOC TPC UF

molecular weight molecular weight cut-off nanofiltration N-methylpyrrolidone N,N ′ -dimethyl-m-phenylenediamine naphthalene-1,3,6,-trisulfonyl chloride ortho-phenylenediamine polyamide polyamide hydrazide polyacrylonitrile poly(benzimidazole) poly(ether ether ketone) polyethylene glycol polyethyleneimine polyethyleneimine-g-sulfobetaine methacrylate polyethersulfone polyhexamethylene guanidine hydrochloride polyimide polymer of intrinsic microporosity piperazine para-phenylenediamine polysulfone poly(1-trimethylsilyl-1-propyne) polyvinylamine poly(vinylidene fluoride) polyvinylpyrrolidone reverse osmosis sulfonated cardo poly(arylene ether sulfone) solvent-resistant nanofiltration triethylamine triethanolamine tetrahydrofuran trihalomethane thin film composite trimesoylchloride total organic carbon terephthaloyl chloride ultrafiltration

References 1 Hermans, S., Mariën, H., Van Goethem, C., and Vankelecom, I.F.J. (2015).

Recent developments in thin film (nano)composite membranes for solvent resistant nanofiltration. Curr. Opin. Chem. Eng. 8: 45–54. 2 Morgan, P.W. (1965). Condensation Polymers: By Interfacial and Solution Methods. New York: Interscience Publishers.

79

80

2 Nanofiltration Membrane Materials and Preparation

3 Loeb, S. and Sourirajan, S. (1963). Sea water demineralization by means of

4 5

6 7

8 9

10

11

12 13

14

15

16

17

18

19

an osmotic membrane. In: Saline Water Conversion—II (ed. R.F. Gould), 117–132. American Chemical Society. Verweij, H. (2012). Inorganic membranes. Curr. Opin. Chem. Eng. 1: 156–162. Lu, G.Q., Diniz da Costa, J.C., Duke, M. et al. (2007). Inorganic membranes for hydrogen production and purification: a critical review and perspective. J. Colloid Interface Sci. 314: 589–603. Paul, M. and Jons, S.D. (2016). Chemistry and fabrication of polymeric nanofiltration membranes: a review. Polymer (Guildf ) 103: 417–456. Holda, A.K. and Vankelecom, I.F.J. (2015). Understanding and guiding the phase inversion process for synthesis of solvent resistant nanofiltration membranes. J. Appl. Polym. Sci. 132: 1–17. Mulder, M. (1996). Basic Principles of Membrane Technology. Dordrecht: Kluwer Academic Publishers. Mulder, M.H.V., Hendrikman, J.O., Wijmans, J.G., and Smolders, C.A. (1985). A rationale for the preparation of asymmetric pervaporation membranes. J. Appl. Polym. Sci. 30: 2805–2820. van’t Hof, J.A., Reuvers, A.J., Boom, R.M. et al. (1992). Preparation of asymmetric gas separation membranes with high selectivity by a dual-bath coagulation method. J. Membr. Sci. 70: 17–30. Pinnau, I. and Koros, W.J. (1993). A qualitative skin layer formation mechanism for membranes made by dry/wet phase inversion. J. Polym. Sci., Part B: Polym. Phys. 31: 419–427. Strathmann, H. and Kock, K. (1977). The formation mechanism of phase inversion membranes. Desalination 21: 241–255. Smolders, C.A., Reuvers, A.J., Boom, R.M., and Wienk, I.M. (1992). Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids. J. Membr. Sci. 73: 259–275. Guillen, G.R., Pan, Y., Li, M., and Hoek, E.M.V. (2011). Preparation and characterization of membranes formed by nonsolvent induced phase separation: a review. Ind. Eng. Chem. Res. 50: 3798–3817. Hendrix, K., Van Eynde, M., Koeckelberghs, G., and Vankelecom, I.F.J. (2013). Crosslinking of modified poly(ether ether ketone) membranes for use in solvent resistant nanofiltration. J. Membr. Sci. 447: 212–221. da Silva Burgal, J., Peeva, L.G., Kumbharkar, S., and Livingston, A. (2015). Organic solvent resistant poly(ether-ether-ketone) nanofiltration membranes. J. Membr. Sci. 479: 105–116. Vanherck, K., Vandezande, P., Aldea, S.O., and Vankelecom, I.F.J. (2008). Cross-linked polyimide membranes for solvent resistant nanofiltration in aprotic solvents. J. Membr. Sci. 320: 468–476. Vanherck, K., Cano-Odena, A., Koeckelberghs, G. et al. (2010). A simplified diamine crosslinking method for PI nanofiltration membranes. J. Membr. Sci. 353: 135–143. Valtcheva, I.B., Marchetti, P., and Livingston, A.G. (2015). Crosslinked polybenzimidazole membranes for organic solvent nanofiltration (OSN): analysis

References

20

21

22

23

24

25

26

27

28

29

30

31

32 33

of crosslinking reaction mechanism and effects of reaction parameters. J. Membr. Sci. 493: 568–579. Blanco, J.F., Nguyen, Q.T., and Schaetzel, P. (2002). Sulfonation of polysulfones: suitability of the sulfonated materials for asymmetric membrane preparation. J. Appl. Polym. Sci. 84: 2461–2473. Tang, B., Xu, T., Gong, M., and Yang, W. (2005). A novel positively charged asymmetry membranes from poly(2,6-dimethyl-1,4-phenylene oxide) by benzyl bromination and in situ amination: membrane preparation and characterization. J. Membr. Sci. 248: 119–125. Gupta, K.C. (1997). Synthesis and evaluation of aromatic polyamide membranes for desalination in reverse-osmosis techniques. J. Appl. Polym. Sci. 66: 643–653. Mohamed, N.A. (1997). Novel wholly aromatic polyamide-hydrazides: 6. Dependence of membrane reverse osmosis performance on processing parameters and polymer structural variations. Polymer (Guildf ) 38: 4705–4713. Dvornic, P.R. (1991). Wholly aromatic polyamide-hydrazides. V. Preparation and properties of semipermeable membranes from poly[4-(terephthaloylamino) benzoic acid hydrazide]. J. Appl. Polym. Sci. 42: 957–972. Yoo, S.H., Kim, J.H., Jho, J.Y. et al. (2004). Influence of the addition of PVP on the morphology of asymmetric polyimide phase inversion membranes: effect of PVP molecular weight. J. Membr. Sci. 236: 203–207. Hołda, A.K., Aernouts, B., Saeys, W., and Vankelecom, I.F.J. (2013). Study of polymer concentration and evaporation time as phase inversion parameters for polysulfone-based SRNF membranes. J. Membr. Sci. 442: 196–205. Boussu, K., Vandecasteele, C., and Van der Bruggen, B. (2006). Study of the characteristics and the performance of self-made nanoporous polyethersulfone membranes. Polymer (Guildf ) 47: 3464–3476. See-Toh, Y.H., Ferreira, F.C., and Livingston, A.G. (2007). The influence of membrane formation parameters on the functional performance of organic solvent nanofiltration membranes. J. Membr. Sci. 299: 236–250. Vandezande, P., Li, X., Gevers, L.E.M., and Vankelecom, I.F.J. (2009). High throughput study of phase inversion parameters for polyimide-based SRNF membranes. J. Membr. Sci. 330: 307–318. Kim, I.-C., Yoon, H.-G., and Lee, K.-H. (2002). Formation of integrally skinned asymmetric polyetherimide nanofiltration membranes by phase inversion process. J. Appl. Polym. Sci. 84: 1300–1307. Hendrix, K., Koeckelberghs, G., and Vankelecom, I.F.J. (2014). Study of phase inversion parameters for PEEK-based nanofiltration membranes. J. Membr. Sci. 452: 241–252. Reuvers, A.J. and Smolders, C.A. (1987). Formation of membranes by means of immersion precipitation. J. Membr. Sci. 34: 67–86. Huang, J. and Zhang, K. (2011). The high flux poly(m-phenylene isophthalamide) nanofiltration membrane for dye purification and desalination. Curr. Dev. Wastewater Treat. India 282: 19–26.

81

82

2 Nanofiltration Membrane Materials and Preparation

34 Soroko, I. and Livingston, A. (2009). Impact of TiO2 nanoparticles on

35

36

37

38

39

40

41

42

43

44

45 46

47 48

morphology and performance of crosslinked polyimide organic solvent nanofiltration (OSN) membranes. J. Membr. Sci. 343: 189–198. Vanherck, K., Hermans, S., Verbiest, T., and Vankelecom, I. (2011). Using the photothermal effect to improve membrane separations via localized heating. J. Mater. Chem. 21: 6079–6087. Li, Y., Verbiest, T., and Vankelecom, I. (2013). Improving the flux of PDMS membranes via localized heating through incorporation of gold nanoparticles. J. Membr. Sci. 428: 63–69. Li, Y., Verbiest, T., Strobbe, R., and Vankelecom, I.F.J. (2014). Silver nanoparticles as localized “nano-heaters” under LED light irradiation to improve membrane performance. J. Mater. Chem. A 2: 3182–3189. Kim, I.-C., Lee, K.-H., and Tak, T.-M. (2001). Preparation and characterization of integrally skinned uncharged polyetherimide asymmetric nanofiltration membrane. J. Membr. Sci. 183: 235–247. Hołda, A.K. and Vankelecom, I.F.J. (2014). Integrally skinned PSf-based SRNF-membranes prepared via phase inversion—Part B: Influence of low molecular weight additives. J. Membr. Sci. 450: 499–511. Panda, S.R. and De, S. (2014). Preparation, characterization and performance of ZnCl2 incorporated polysulfone (PSF)/polyethylene glycol (PEG) blend low pressure nanofiltration membranes. Desalination 347: 52–65. Kim, I.-C. and Lee, K.-H. (2004). Effect of poly(ethylene glycol) 200 on the formation of a polyetherimide asymmetric membrane and its performance in aqueous solvent mixture permeation. J. Membr. Sci. 230: 183–188. Soroko, I., Makowski, M., Spill, F., and Livingston, A. (2011). The effect of membrane formation parameters on performance of polyimide membranes for organic solvent nanofiltration (OSN). Part B: Analysis of evaporation step and the role of a co-solvent. J. Membr. Sci. 381: 163–171. Mohamed, N.A. and Al-Dossary, A.O.H. (2003). Structure–property relationships for novel wholly aromatic polyamide–hydrazides containing various proportions of para-phenylene and meta-phenylene units III. Preparation and properties of semi-permeable membranes for water desalination by reverse osmosis. Eur. Polym. J. 39: 1653–1667. Duarte, A.P., Bordado, J.C., and Cidade, M.T. (2007). Cellulose acetate reverse osmosis membranes: optimization of preparation parameters. J. Appl. Polym. Sci. 103: 134–139. Baker, R.W. (2012). Membrane Technology and Applications. Chichester: Wiley. Sairam, M., Loh, X.X., Bhole, Y. et al. (2010). Spiral-wound polyaniline membrane modules for organic solvent nanofiltration (OSN). J. Membr. Sci. 349: 123–129. See Toh, Y.H., Lim, F.W., and Livingston, A.G. (2007). Polymeric membranes for nanofiltration in polar aprotic solvents. J. Membr. Sci. 301: 3–10. Hendrix, K., Vanherck, K., and Vankelecom, I.F.J. (2012). Optimization of solvent resistant nanofiltration membranes prepared by the in-situ diamine crosslinking method. J. Membr. Sci. 421–422: 15–24.

References

49 Struzy´ ̇ nska-Piron, I., Loccufier, J., Vanmaele, L., and Vankelecom,

50

51

52 53

54

55

56

57

58

59

60

61

62

63

I.F.J. (2013). Synthesis of solvent stable polymeric membranes via UV depth-curing. Chem. Commun. 49: 11494–11496. Van den Mooter, P.R., Daems, N., and Vankelecom, I.F.J. (2019). Preparation of solvent resistant supports through formation of a semi-interpenetrating polysulfone/polyacrylate network using UV cross-linking – Part 1: Selection of optimal UV curing conditions. React. Funct. Polym. 136: 189–197. Altun, V., Bielmann, M., and Vankelecom, I.F.J. (2016). EB depth-curing as a facile method to prepare highly stable membranes. RSC Adv. 6: 55526–55533. Macdonald, W. and Pan, C. (1974). Method for drying water-wet membranes. US Patent No 3, 842, 515A. da Silva Burgal, J., Peeva, L., Marchetti, P., and Livingston, A. (2015). Controlling molecular weight cut-off of PEEK nanofiltration membranes using a drying method. J. Membr. Sci. 493: 524–538. Gevers, L.E.M., Aldea, S., Vankelecom, I.F.J., and Jacobs, P.A. (2006). Optimisation of a lab-scale method for preparation of composite membranes with a filled dense top-layer. J. Membr. Sci. 281: 741–746. Vandezande, P., Gevers, L.E.M., and Vankelecom, I.F.J. (2008). Solvent resistant nanofiltration: separating on a molecular level. Chem. Soc. Rev. 37: 365–405. Ghosh, A.K. and Hoek, E.M.V. (2009). Impacts of support membrane structure and chemistry on polyamide–polysulfone interfacial composite membranes. J. Membr. Sci. 336: 140–148. Jimenez-Solomon, M.F., Gorgojo, P., Munoz-Ibanez, M., and Livingston, A.G. (2013). Beneath the surface: influence of supports on thin film composite membranes by interfacial polymerization for organic solvent nanofiltration. J. Membr. Sci. 448: 102–113. Oh, N.-W., Jegal, J., and Lee, K.-H. (2001). Preparation and characterization of nanofiltration composite membranes using polyacrylonitrile (PAN). II. Preparation and characterization of polyamide composite membranes. J. Appl. Polym. Sci. 80: 2729–2736. Lau, W.J., Ismail, A.F., Misdan, N., and Kassim, M.A. (2012). A recent progress in thin film composite membrane: a review. Desalination 287: 190–199. Saha, N.K. and Joshi, S.V. (2009). Performance evaluation of thin film composite polyamide nanofiltration membrane with variation in monomer type. J. Membr. Sci. 342: 60–69. Song, Y., Sun, P., Henry, L.L., and Sun, B. (2005). Mechanisms of structure and performance controlled thin film composite membrane formation via interfacial polymerization process. J. Membr. Sci. 251: 67–79. Roh, I.J., Park, S.Y., Kim, J.J., and Kim, C.K. (1998). Effects of the polyamide molecular structure on the performance of reverse osmosis membranes. J. Polym. Sci., Part B: Polym. Phys. 36: 1821–1830. Epsztein, R., Shaulsky, E., Dizge, N. et al. (2018). Role of ionic charge density in Donnan exclusion of monovalent anions by nanofiltration. Environ. Sci. Technol. 52: 4108–4116.

83

84

2 Nanofiltration Membrane Materials and Preparation

64 Chen, G., Li, S., Zhang, X., and Zhang, S. (2008). Novel thin-film composite

65

66

67

68

69

70 71

72

73

74

75

76

77

78

membranes with improved water flux from sulfonated cardo poly(arylene ether sulfone) bearing pendant amino groups. J. Membr. Sci. 310: 102–109. Wang, H., Li, L., Zhang, X., and Zhang, S. (2010). Polyamide thinfilm composite membranes prepared from a novel triamine 3,5-diaminoN-(4-aminophenyl)-benzamide monomer and m-phenylenediamine. J. Membr. Sci. 353: 78–84. Yu, S., Ma, M., Liu, J. et al. (2011). Study on polyamide thin-film composite nanofiltration membrane by interfacial polymerization of polyvinylamine (PVAm) and isophthaloyl chloride (IPC). J. Membr. Sci. 379: 164–173. Liu, M., Zheng, Y., Shuai, S. et al. (2012). Thin-film composite membrane formed by interfacial polymerization of polyvinylamine (PVAm) and trimesoyl chloride (TMC) for nanofiltration. Desalination 288: 98–107. An, Q.-F., Sun, W.-D., Zhao, Q. et al. (2013). Study on a novel nanofiltration membrane prepared by interfacial polymerization with zwitterionic amine monomers. J. Membr. Sci. 431: 171–179. Jeon, S., Hyung, C., Park, S. et al. (2018). Star polymer-assembled thin film composite membranes with high separation performance and low fouling. J. Membr. Sci. 555: 369–378. Verbeke, R., Gómez, V., and Vankelecom, I.F.J. (2017). Chlorine-resistance of reverse osmosis (RO) polyamide membranes. Prog. Polym. Sci. 72: 1–15. Xu, J., Wang, Z., Wei, X. et al. (2013). The chlorination process of crosslinked aromatic polyamide reverse osmosis membrane: new insights from the study of self-made membrane. Desalination 313: 145–155. Buch, P.R., Jagan Mohan, D., and Reddy, A.V.R. (2008). Preparation, characterization and chlorine stability of aromatic–cycloaliphatic polyamide thin film composite membranes. J. Membr. Sci. 309: 36–44. Shintani, T., Matsuyama, H., and Kurata, N. (2009). Effect of heat treatment on performance of chlorine-resistant polyamide reverse osmosis membranes. Desalination 247: 370–377. Shintani, T., Matsuyama, H., and Kurata, N. (2007). Development of a chlorine-resistant polyamide reverse osmosis membrane. Desalination 207: 340–348. Tang, Y.-J., Xu, Z.-L., Xue, S.-M. et al. (2016). A chlorine-tolerant nanofiltration membrane prepared by the mixed diamine monomers of PIP and BHTTM. J. Membr. Sci.: 374–384. Konagaya, S. and Watanabe, O. (2000). Influence of chemical structure of isophthaloyl dichloride and aliphatic, cycloaliphatic, and aromatic diamine compound polyamides on their chlorine resistance. J. Appl. Polym. Sci. 76: 201–207. Yu, S., Liu, M., Lü, Z. et al. (2009). Aromatic-cycloaliphatic polyamide thin-film composite membrane with improved chlorine resistance prepared from m-phenylenediamine-4-methyl and cyclohexane-1,3,5-tricarbonyl chloride. J. Membr. Sci. 344: 155–164. Ma, T., Su, Y., Li, Y. et al. (2016). Fabrication of electro-neutral nanofiltration membranes at neutral pH with antifouling surface via interfacial polymerization from a novel zwitterionic amine monomer. J. Membr. Sci. 503: 101–109.

References

79 Li, X., Cao, Y., Yu, H. et al. (2014). A novel composite nanofiltration

80

81

82

83

84

85

86

87

88

89

90

91

membrane prepared with PHGH and TMC by interfacial polymerization. J. Membr. Sci. 466: 82–91. Azarteimour, F., Amirinejad, M., Parvini, M., and Yarvali, M. (2016). Organic phase addition of anionic/non-ionic surfactants to poly(paraphenyleneterephthalamide) thin film composite nanofiltration membranes. Chem. Eng. Process. Process Intensif. 106: 13–25. Yu, S., Liu, M., Liu, X., and Gao, C. (2009). Performance enhancement in interfacially synthesized thin-film composite polyamide-urethane reverse osmosis membrane for seawater desalination. J. Membr. Sci. 342: 313–320. Wang, H., Zhang, Q., and Zhang, S. (2011). Positively charged nanofiltration membrane formed by interfacial polymerization of 3,3′ ,5,5′ -biphenyl tetraacyl chloride and piperazine on a poly(acrylonitrile) (PAN) support. J. Membr. Sci. 378: 243–249. Liu, L.-F., Yu, S.-C., Wu, L.-G., and Gao, C.-J. (2006). Study on a novel polyamide-urea reverse osmosis composite membrane (ICIC-MPD): II. Analysis of membrane antifouling performance. J. Membr. Sci. 283: 133–146. Tang, B., Huo, Z., and Wu, P. (2008). Study on a novel polyester composite nanofiltration membrane by interfacial polymerization of triethanolamine (TEOA) and trimesoyl chloride (TMC): I. Preparation, characterization and nanofiltration properties test of membrane. J. Membr. Sci. 320: 198–205. Tang, B., Zou, C., and Wu, P. (2010). Study on a novel polyester composite nanofiltration membrane by interfacial polymerization. II. The role of lithium bromide in the performance and formation of composite membrane. J. Membr. Sci. 365: 276–285. Zhang, Y., Su, Y., Peng, J. et al. (2013). Composite nanofiltration membranes prepared by interfacial polymerization with natural material tannic acid and trimesoyl chloride. J. Membr. Sci. 429: 235–242. Zhou, C., Shi, Y., Sun, C. et al. (2014). Thin-film composite membranes formed by interfacial polymerization with natural material sericin and trimesoyl chloride for nanofiltration. J. Membr. Sci. 471: 381–391. Lee, K.P., Zheng, J., Bargeman, G. et al. (2015). pH stable thin film composite polyamine nanofiltration membranes by interfacial polymerisation. J. Membr. Sci. 478: 75–84. Liu, M., Yao, G., Cheng, Q. et al. (2012). Acid stable thin-film composite membrane for nanofiltration prepared from naphthalene-1,3,6-trisulfonylchloride (NTSC) and piperazine (PIP). J. Membr. Sci. 415–416: 122–131. Yuan, F., Wang, Z., Yu, X. et al. (2012). Visualization of the formation of interfacially polymerized film by an optical contact angle measuring device. J. Phys. Chem. C 116: 11496–11506. Matthews, T.D., Yan, H., Cahill, D.G. et al. (2013). Growth dynamics of interfacially polymerized polyamide layers by diffuse reflectance spectroscopy and Rutherford backscattering spectrometry. J. Membr. Sci. 429: 71–80.

85

86

2 Nanofiltration Membrane Materials and Preparation

92 Roh, I.J., Greenberg, A.R., and Khare, V.P. (2006). Synthesis and characteri-

93

94

95

96

97

98 99

100

101

102

103

104

105

106

zation of interfacially polymerized polyamide thin films. Int. Congr. Membr. Membr. Process. 191: 279–290. Hermans, S., Mariën, H., Dom, E. et al. (2014). Simplified synthesis route for interfacially polymerized polyamide membranes. J. Membr. Sci. 451: 148–156. Wang, X., Yeh, T.-M., Wang, Z. et al. (2014). Nanofiltration membranes prepared by interfacial polymerization on thin-film nanofibrous composite scaffold. Polymer (Guildf ) 55: 1358–1366. Zhang, Y., Benes, N.E., and Lammertink, R.G.H. (2015). Visualization and characterization of interfacial polymerization layer formation. Lab Chip 15: 575–580. Ali, Z., Al Sunbul, Y., Pacheco, F. et al. (2019). Defect-free highly selective polyamide thin- film composite membranes for desalination and boron removal. J. Membr. Sci. 578: 85–94. Hermans, S., Bernstein, R., Volodin, A., and Vankelecom, I.F.J. (2015). Study of synthesis parameters and active layer morphology of interfacially polymerized polyamide-polysulfone membranes. React. Funct. Polym. 86: 199–208. Freger, V. (2005). Kinetics of film formation by interfacial polycondensation. Langmuir 21: 1884–1894. Ghosh, A.K., Jeong, B.-H., Huang, X., and Hoek, E.M.V. (2008). Impacts of reaction and curing conditions on polyamide composite reverse osmosis membrane properties. J. Membr. Sci. 311: 34–45. Jegal, J., Min, S.G., and Lee, K.-H. (2002). Factors affecting the interfacial polymerization of polyamide active layers for the formation of polyamide composite membranes. J. Appl. Polym. Sci. 86: 2781–2787. Kim, I.-C., Jegal, J., and Lee, K.-H. (2002). Effect of aqueous and organic solutions on the performance of polyamide thin-film-composite nanofiltration membranes. J. Polym. Sci., Part B: Polym. Phys. 40: 2151–2163. Khorshidi, B., Thundat, T., Fleck, B.A., and Sadrzadeh, M. (2016). A novel approach toward fabrication of high performance thin film composite polyamide membranes. Sci. Rep. 6: 22069. Mariën, H., Bellings, L., Hermans, S., and Vankelecom, I.F.J. (2016). Sustainable process for the preparation of high-performance thin-film composite membranes using ionic liquids as the reaction medium. ChemSusChem 9: 1101–1111. Morgan, P.W. and Kwolek, S.L. (1996). Interfacial polycondensation. II. Fundamentals of polymer formation at liquid interfaces. J. Polym. Sci., Part A: Polym. Chem. 34: 531–559. Mansourpanah, Y., Madaeni, S.S., and Rahimpour, A. (2009). Fabrication and development of interfacial polymerized thin-film composite nanofiltration membrane using different surfactants in organic phase; study of morphology and performance. J. Membr. Sci. 343: 219–228. Mansourpanah, Y., Alizadeh, K., Madaeni, S.S. et al. (2011). Using different surfactants for changing the properties of poly(piperazineamide) TFC nanofiltration membranes. Desalination 271: 169–177.

References

107 Liu, M., Yu, S., Tao, J., and Gao, C. (2008). Preparation, structure character-

108

109

110

111

112

113

114

115

116 117

118

119

120

121

istics and separation properties of thin-film composite polyamide-urethane seawater reverse osmosis membrane. J. Membr. Sci. Qiu, S., Wu, L., Zhang, L. et al. (2009). Preparation of reverse osmosis composite membrane with high flux by interfacial polymerization of MPD and TMC. J. Appl. Polym. Sci. 112: 2066–2072. Fan, X., Dong, Y., Su, Y. et al. (2014). Improved performance of composite nanofiltration membranes by adding calcium chloride in aqueous phase during interfacial polymerization process. J. Membr. Sci. 452: 90–96. Duong, P.H.H., Daumann, K., Hong, P.Y. et al. (2019). Interfacial polymerization of zwitterionic building blocks for high-flux nanofiltration membranes. Langmuir 35: 1284–1293. Jimenez-Solomon, M.F., Bhole, Y., and Livingston, A.G. (2012). High flux membranes for organic solvent nanofiltration (OSN)—interfacial polymerization with solvent activation. J. Membr. Sci. 423–424: 371–382. Yao, Z., Yang, Z., Guo, H. et al. (2019). Highly permeable and highly selective ultrathin film composite polyamide membranes reinforced by reactable polymer chains. J. Colloid Interface Sci. 552: 418–425. Kim, S.H., Kwak, S.-Y., and Suzuki, T. (2005). Positron annihilation spectroscopic evidence to demonstrate the flux-enhancement mechanism in morphology-controlled thin-film-composite (TFC) membrane. Environ. Sci. Technol. 39: 1764–1770. Duan, M., Wang, Z., Xu, J. et al. (2010). Influence of hexamethyl phosphoramide on polyamide composite reverse osmosis membrane performance. Sep. Purif. Technol. 75: 145–155. Kong, C., Kanezashi, M., Yamomoto, T. et al. (2010). Controlled synthesis of high performance polyamide membrane with thin dense layer for water desalination. J. Membr. Sci. 362: 76–80. Kong, C., Shintani, T., Kamada, T. et al. (2011). Co-solvent-mediated synthesis of thin polyamide membranes. J. Membr. Sci. 384: 10–16. Kamada, T., Ohara, T., Shintani, T., and Tsuru, T. (2014). Optimizing the preparation of multi-layered polyamide membrane via the addition of a co-solvent. J. Membr. Sci. 453: 489–497. Kamada, T., Ohara, T., Shintani, T., and Tsuru, T. (2014). Controlled surface morphology of polyamide membranes via the addition of co-solvent for improved permeate flux. J. Membr. Sci. 467: 303–312. AL-Hobaib, A.S., Alsuhybani, M.S., AL-Sheetan, K.M., and Shaik, M.R. (2016). Reverse osmosis membranes prepared by interfacial polymerization in n-heptane containing different co-solvents. Desalin. Water Treat. 57: 16733–16744. Hermans, S., Dom, E., Mariën, H. et al. (2015). Efficient synthesis of interfacially polymerized membranes for solvent resistant nanofiltration. J. Membr. Sci. 476: 356–363. Karan, S., Jiang, Z., and Livingston, A.G. (2015). Sub–10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation. Science 348: 1347–1351.

87

88

2 Nanofiltration Membrane Materials and Preparation

122 Jimenez-Solomon, M.F., Song, Q., Jelfs, K.E. et al. (2016). Polymer nanofilms

123

124 125

126

127

128

129

130

131

132

133

134

135

with enhanced microporosity by interfacial polymerization. Nat. Mater.: 1–26. Yang, Z., Yao, Z., Guo, H. et al. (2018). Interfacial polymerization with electrosprayed microdroplets: toward controllable and ultrathin polyamide membranes. Environ. Sci. Technol. Lett. Chowdhury, M.R., Steffes, J., Huey, B.D., and Mccutcheon, J.R. (2018). 3D printed polyamide membranes for desalination. Science: 682–686. Shultz, S., Bass, M., Semiat, R., and Freger, V. (2018). Modification of polyamide membranes by hydrophobic molecular plugs for improved boron rejection. J. Membr. Sci. 546: 165–172. Tang, C.Y., Kwon, Y.-N., and Leckie, J.O. (2009). Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: I. FTIR and XPS characterization of polyamide and coating layer chemistry. Desalination 242: 149–167. Jimenez-Solomon, M.F., Bhole, Y., and Livingston, A.G. (2013). High flux hydrophobic membranes for organic solvent nanofiltration (OSN)—interfacial polymerization, surface modification and solvent activation. J. Membr. Sci. 434: 193–203. Dong, X., Zhang, Q., Zhang, S., and Li, S. (2016). Thin film composite nanofiltration membranes fabricated from quaternized poly(ether ether ketone) with crosslinkable moiety using a benign solvent. J. Colloid Interface Sci. 463: 332–341. Joseph, N., Ahmadiannamini, P., Hoogenboom, R., and Vankelecom, I.F.J. (2014). Layer-by-layer preparation of polyelectrolyte multilayer membranes for separation. Polym. Chem. 5: 1817–1831. Siddique, H., Peeva, L.G., Stoikos, K. et al. (2013). Membranes for organic solvent nanofiltration based on preassembled nanoparticles. Ind. Eng. Chem. Res. 52: 1109–1121. Wang, J., Wang, J., Wang, H., and Zhang, S. (2013). Preparation and characterization of positively charged composite nanofiltration membranes by coating poly(ether ether ketone) containing quaternary ammonium groups on polysulfone ultrafiltration membranes. J. Appl. Polym. Sci. 127: 1601–1608. Lang, K., Sourirajan, S., Matsuura, T., and Chowdhury, G. (1996). A study on the preparation of polyvinyl alcohol thin-film composite membranes and reverse osmosis testing. Desalination 104: 185–196. Johnson, R.W., Hultqvist, A., and Bent, S.F. (2014). A brief review of atomic layer deposition: from fundamentals to applications. Mater. Today 17: 236–246. Nikkola, J., Sievänen, J., Raulio, M. et al. (2014). Surface modification of thin film composite polyamide membrane using atomic layer deposition method. J. Membr. Sci. 450: 174–180. Gorgojo, P., Karan, S., Wong, H.C. et al. (2014). Ultrathin polymer films with intrinsic microporosity: anomalous solvent permeation and high flux membranes. Adv. Funct. Mater. 24: 4729–4737.

References

136 Joseph, N., Ahmadiannamini, P., Jishna, P.S. et al. (2015). ‘Up-scaling’ poten-

tial for polyelectrolyte multilayer membranes. J. Membr. Sci. 492: 271–280.

137 Ilyas, S., Joseph, N., Szymczyk, A. et al. (2016). Weak polyelectrolyte multi-

138

139

140

141

142

143

144

145

146

147

148

149

150

layers as tunable membranes for solvent resistant nanofiltration. J. Membr. Sci. 514: 322–331. Johnson, P.M., Yoon, J., Kelly, J.Y. et al. (2012). Molecular layer-by-layer deposition of highly crosslinked polyamide films. J. Polym. Sci., Part B: Polym. Phys. 50: 168–173. Zhou, X., Zhao, Y., Kim, S. et al. (2018). Controlled TiO2 growth on reverse osmosis and nanofiltration membranes by atomic layer deposition: mechanisms and potential applications. Environ. Sci. Technol. 52: 14311–14320. Volkov, A.V., Tsarkov, S.E., Gilman, A.B. et al. (2015). Surface modification of PTMSP membranes by plasma treatment: asymmetry of transport in organic solvent nanofiltration. Adv. Colloid Interface Sci. 222: 716–727. Aerts, S., Vanhulsel, A., Buekenhoudt, A. et al. (2006). Plasma-treated PDMS-membranes in solvent resistant nanofiltration: characterization and study of transport mechanism. J. Membr. Sci. 275: 212–219. Kim, E.-S., Yu, Q., and Deng, B. (2011). Plasma surface modification of nanofiltration (NF) thin-film composite (TFC) membranes to improve anti organic fouling. Appl. Surf. Sci. 257: 9863–9871. Akbari, A., Mohtasham Khani, S., and Mojallali Rostami, S.M. (2016). Second modification of a polyamide membrane surface. J. Appl. Polym. Sci. 133. Yan, F., Chen, H., Lü, Y. et al. (2016). Improving the water permeability and antifouling property of thin-film composite polyamide nanofiltration membrane by modifying the active layer with triethanolamine. J. Membr. Sci. 513: 108–116. Werber, J.R., Bull, S.K., and Elimelech, M. (2017). Acyl-chloride quenching following interfacial polymerization to modulate the water permeability, selectivity, and surface charge of desalination membranes. J. Membr. Sci. 535: 357–364. Byun, H.S., Burford, R.P., and Fane, A.G. (1994). Sulfonation of cross-linked asymmetric membranes based on polystyrene and divinylbenzene. J. Appl. Polym. Sci. 52: 825–835. Chowdhury, S.R., Kumar, P., Bhattacharya, P., and Kumar, A. (2001). Separation characteristics of modified polysulfone ultrafiltration membranes using NOx . Sep. Purif. Technol. 24: 271–282. Li, Y., Su, Y., Zhao, X. et al. (2014). Surface fluorination of polyamide nanofiltration membrane for enhanced antifouling property. J. Membr. Sci. 455: 15–23. Li, X., Cao, Y., Kang, G. et al. (2014). Surface modification of polyamide nanofiltration membrane by grafting zwitterionic polymers to improve the antifouling property. J. Appl. Polym. Sci. 131: 41144. Van der Bruggen, B. (2009). Comparison of redox initiated graft polymerisation and sulfonation for hydrophilisation of polyethersulfone nanofiltration membranes. Eur. Polym. J. 45: 1873–1882.

89

90

2 Nanofiltration Membrane Materials and Preparation

151 Abuhabib, A.A., Mohammad, A.W., Hilal, N. et al. (2012). Nanofiltration

152

153

154

155

156

157 158

159 160

161

162

163

164

165

166

membrane modification by UV grafting for salt rejection and fouling resistance improvement for brackish water desalination. Desalination 295: 16–25. Zhao, Z.-P., Li, J., Chen, J., and Chen, C.-X. (2005). Nanofiltration membrane prepared from polyacrylonitrile ultrafiltration membrane by low-temperature plasma: 2. Grafting of styrene in vapor phase. J. Membr. Sci. 251: 239–245. Schulze, A., Marquardt, B., Kaczmarek, S. et al. (2010). Electron beam-based functionalization of poly(ethersulfone) membranes. Macromol. Rapid Commun. 31: 467–472. Xu, H.-m., Wei, J.-f., and Wang, X.-l. (2014). Nanofiltration hollow fiber membranes with high charge density prepared by simultaneous electron beam radiation-induced graft polymerization for removal of Cr(VI). Desalination 346: 122–130. Ji, J., Trushinski, B.J., Childs, R.F. et al. (1997). Fabrication of thin-film composite membranes with pendant, photoreactive diazoketone functionality. J. Appl. Polym. Sci. 64: 2381–2398. Van Gestel, T., Kruidhof, H., Blank, D.H.A., and Bouwmeester, H.J.M. (2006). ZrO2 and TiO2 membranes for nanofiltration and pervaporation: Part 1. Preparation and characterization of a corrosion-resistant ZrO2 nanofiltration membrane with a MWCO < 300. J. Membr. Sci. 284: 128–136. Li, K. (2007). Preparation of ceramic membranes. In: Ceramic Membranes for Separation and Reaction, 21–57. Wiley. Chen, X., Zhang, W., Lin, Y. et al. (2015). Preparation of high-flux -alumina nanofiltration membranes by using a modified sol–gel method. Microporous Mesoporous Mater. 214: 195–203. Danks, A.E., Hall, S.R., and Schnepp, Z. (2016). The evolution of “sol–gel” chemistry as a technique for materials synthesis. Mater. Horiz. 3: 91–112. Da, X., Wen, J., Lu, Y. et al. (2015). An aqueous sol–gel process for the fabrication of high-flux YSZ nanofiltration membranes as applied to the nanofiltration of dye wastewater. Sep. Purif. Technol. 152: 37–45. Cai, Y., Wang, Y., Chen, X. et al. (2015). Modified colloidal sol–gel process for fabrication of titania nanofiltration membranes with organic additives. J. Membr. Sci. 476: 432–441. Julbe, A., Guizard, C., Larbot, A. et al. (1993). The sol–gel approach to prepare candidate microporous inorganic membranes for membrane reactors. Appl. Catal., A 96: 101. Zeidler, S., Puhlfürß, P., Kätzel, U., and Voigt, I. (2014). Preparation and characterization of new low MWCO ceramic nanofiltration membranes for organic solvents. J. Membr. Sci. 470: 421–430. Guo, H., Zhao, S., Wu, X., and Qi, H. (2016). Fabrication and characterization of TiO2 /ZrO2 ceramic membranes for nanofiltration. Microporous Mesoporous Mater. de Lange, R.S.A., Hekkink, J.H.A., Keizer, K., and Burggraaf, A.J. (1995). Formation and characterization of supported microporous ceramic membranes prepared by sol–gel modification techniques. J. Membr. Sci. 99: 57–75. Nijmeijer, A., Kruidhof, H., Bredesen, R., and Verweij, H. (2001). Preparation and properties of hydrothermally stable. J. Am. Ceram. Soc. 1: 136–140.

References

167 Tsuru, T., Wada, S.I., Izumi, S., and Asaeda, M. (1998). Silica-zirconia mem-

branes for nanofiltration. J. Membr. Sci. 149: 127–135.

168 Puthai, W., Kanezashi, M., Nagasawa, H. et al. (2016). Effect of firing

169

170

171

172

173

174

175

176

177

178

179

180

181

182

temperature on the water permeability of SiO2 –ZrO2 membranes for nanofiltration. J. Membr. Sci. 497: 348–356. Gu, Y., Hacarlioglu, P., and Oyama, S.T. (2008). Hydrothermally stable silica-alumina composite membranes for hydrogen separation. J. Membr. Sci. 310: 28–37. Farsi, A., Malvache, C., De Bartolis, O. et al. (2017). Design and fabrication of silica-based nanofiltration membranes for water desalination and detoxification. Microporous Mesoporous Mater. 237: 117–126. Benfer, S., Popp, U., Richter, H. et al. (2001). Development and characterization of ceramic nanofiltration membranes. Sep. Purif. Technol. 22–23: 231–237. Chen, X., Lin, Y., Lu, Y. et al. (2015). A facile nanoparticle doping sol–gel method for the fabrication of defect-free nanoporous ceramic membranes. Colloids Interface Sci. Commun. 5: 12–15. Tanardi, C.R., Vankelecom, I.F.J., Pinheiro, A.F.M. et al. (2015). Solvent permeation behavior of PDMS grafted -alumina membranes. J. Membr. Sci. 495: 216–225. Zhang, X., Lin, B., Ling, Y. et al. (2010). Highly permeable porous YSZ hollow fiber membrane prepared using ethanol as external coagulant. J. Alloys Compd. 494: 366–371. Rezaei Hosseinabadi, S., Wyns, K., Meynen, V. et al. (2014). Organic solvent nanofiltration with Grignard functionalised ceramic nanofiltration membranes. J. Membr. Sci. 454: 496–504. Kujawa, J., Rozicka, A., Cerneaux, S., and Kujawski, W. (2014). The influence of surface modification on the physicochemical properties of ceramic membranes. Colloids Surf., A 443: 567–575. Tanardi, C.R., Pinheiro, A.F.M., Nijmeijer, A., and Winnubst, L. (2014). PDMS grafting of mesoporous -alumina membranes for nanofiltration of organic solvents. J. Membr. Sci. 469: 471–477. Mustafa, G., Wyns, K., Vandezande, P. et al. (2014). Novel grafting method efficiently decreases irreversible fouling of ceramic nanofiltration membranes. J. Membr. Sci. 470: 369–377. Li, F., Yang, Y., Fan, Y. et al. (2012). Modification of ceramic membranes for pore structure tailoring: the atomic layer deposition route. J. Membr. Sci. 397–398: 17–23. Chen, H., Jia, X., Wei, M., and Wang, Y. (2017). Ceramic tubular nanofiltration membranes with tunable performances by atomic layer deposition and calcination. J. Membr. Sci. 528: 95–102. Chen, H., Wu, S., Jia, X. et al. (2018). Atomic layer deposition fabricating of ceramic nanofiltration membranes for efficient separation of dyes from water. AlChE J. 64: 2670–2678. Velleman, L., Triani, G., Evans, P.J. et al. (2009). Structural and chemical modification of porous alumina membranes. Microporous Mesoporous Mater. 126: 87–94.

91

92

2 Nanofiltration Membrane Materials and Preparation

183 Shang, R., Goulas, A., Tang, C.Y. et al. (2017). Atmospheric pressure atomic

184

185

186

187

188

189 190 191

192

193

194

195

196

197

layer deposition for tight ceramic nanofiltration membranes: synthesis and application in water purification. J. Membr. Sci. 528: 163–170. Chen, X., Zhang, Y., Tang, J. et al. (2018). Novel pore size tuning method for the fabrication of ceramic multi-channel nanofiltration membrane. J. Membr. Sci. 552: 77–85. Cao, G., Brinkman, H.W., Meijerink, J. et al. (1993). Pore narrowing and formation of ultrathin yttria-stabilized zirconia layers in ceramic membranes by chemical vapor deposition/electrochemical vapor deposition. J. Am. Ceram. Soc. 76: 2201–2208. Lin, C.L., Flowers, D.L., and Liu, P.K.T. (1994). Characterization of ceramic membranes II. Modified commercial membranes with pore size under 40 Å. J. Membr. Sci. 92: 45–58. Marchetti, P., Jimenez Solomon, M.F., Szekely, G., and Livingston, A.G. (2014). Molecular separation with organic solvent nanofiltration: a critical review. Chem. Rev. 114: 10735–10806. Darvishmanesh, S., Tasselli, F., Jansen, J.C. et al. (2011). Preparation of solvent stable polyphenylsulfone hollow fiber nanofiltration membranes. J. Membr. Sci. 384: 89–96. Pearce, G. (2007). Water and wastewater filtration: membrane module format. Filtr. Sep. 4: 31–33. Dashti, A. and Asghari, M. (2015). Recent progresses in ceramic hollow-fiber membranes. ChemBioEng Rev. 2: 54–70. Menne, D., Kamp, J., Erik Wong, J., and Wessling, M. (2016). Precise tuning of salt retention of backwashable polyelectrolyte multilayer hollow fiber nanofiltration membranes. J. Membr. Sci. 499: 396–405. Liu, C., Shi, L., and Wang, R. (2015). Crosslinked layer-by-layer polyelectrolyte nanofiltration hollow fiber membrane for low-pressure water softening with the presence of SO42 - in feed water. J. Membr. Sci. 486: 169–176. Veríssimo, S., Peinemann, K.V., and Bordado, J. (2005). Thin-film composite hollow fiber membranes: an optimized manufacturing method. J. Membr. Sci. 264: 48–55. Frank, M., Bargeman, G., Zwijnenburg, A., and Wessling, M. (2001). Capillary hollow fiber nanofiltration membranes. Sep. Purif. Technol. 22–23: 499–506. Sun, S.P., Hatton, T.A., Chan, S.Y., and Chung, T.S. (2012). Novel thin-film composite nanofiltration hollow fiber membranes with double repulsion for effective removal of emerging organic matters from water. J. Membr. Sci. 401–402: 152–162. Roth, H., Luelf, T., Koppelmann, A. et al. (2018). Chemistry in a spinneret – composite hollow fiber membranes in a single step process. J. Membr. Sci. 554: 48–58. Kope´c, K.K., Dutczak, S.M., Wessling, M., and Stamatialis, D.F. (2011). Chemistry in a spinneret-on the interplay of crosslinking and phase inversion during spinning of novel hollow fiber membranes. J. Membr. Sci. 369: 308–318.

References

198 Geise, G.M., Lee, H.S., Miller, D.J. et al. (2010). Water purification by mem-

199

200

201

202 203

204

205

206

207

208

209

210

211

212

branes: the role of polymer science. J. Polym. Sci., Part B: Polym. Phys. 48: 1685–1718. Van Goethem, C., Verbeke, R., Pfanmöller, M. et al. (2018). The role of MOFs in thin-film Nanocomposite (TFN) membranes. J. Membr. Sci. 563: 938–948. Werber, J.R., Osuji, C.O., and Elimelech, M. (2016). Materials for next-generation desalination and water purification membranes. Nat. Rev. Mater. 1: 16018. Hourani, R., Zhang, C., Van Der Weegen, R. et al. (2011). Processable cyclic peptide nanotubes with tunable interiors. J. Am. Chem. Soc. 133: 15296–15299. Zhou, X., Liu, G., Yamato, K. et al. (2012). Self-assembling subnanometer pores with unusual mass-transport properties. Nat. Commun. 3: 948–949. Shen, Y., Si, W., Erbakan, M. et al. (2015). Highly permeable artificial water channels that can self-assemble into two-dimensional arrays. Proc. Natl. Acad. Sci. U.S.A. 112: 9810–9815. Gin, D.L., Bara, J.E., Noble, R.D., and Elliott, B.J. (2008). Polymerized lyotropic liquid crystal assemblies for membrane applications. Macromol. Rapid Commun. 29: 367–389. Zhang, Y., Sargent, J.L., Boudouris, B.W., and Philip, W.A. (2015). Nanoporous membranes generated from self-assembled block polymer precursors: Quo Vadis? J. Appl. Polym. Sci.: 1–17. Verbeke, R., Arts, W., Dom, E. et al. (2019). Transferring bulk chemistry to interfacial synthesis of TFC-membranes to create chemically robust poly(epoxyether) films. J. Membr. Sci. 582: 442–453. Guo, H., PENG, L.E., Yao, Z. et al. (2019). Non-polyamide based nanofiltration membranes using green metal–organic coordination complexes: implications for the removal of trace organic contaminants. Environ. Sci. Technol. 53: 2688–2694. Chen, Y. and He, C. (2017). High salt permeation nanofiltration membranes based on NMG-assisted polydopamine coating for dye/salt fractionation. Desalination 413: 29–39. Pulido, B., Chisca, S., and Nunes, S.P. (2018). Solvent and thermal resistant ultrafiltration membranes from alkyne-functionalized high-performance polymers. J. Membr. Sci. 564: 361–371. Pérez-Manríquez, L., Neelakanda, P., and Peinemann, K.-V. Tannin-based thin-film composite membranes for solvent nanofiltration. J. Membr. Sci. 541, 137–142. Pérez-Manríquez, L., Neelakanda, P., and Peinemann, K.V. (2018). Morin-based nanofiltration membranes for organic solvent separation processes. J. Membr. Sci. 554: 1–5. Abdellah, M.H., Pérez-Manríquez, L., Puspasari, T. et al. (2018). A catechin/cellulose composite membrane for organic solvent nanofiltration. J. Membr. Sci. 567: 139–145.

93

94

2 Nanofiltration Membrane Materials and Preparation

213 Huang, L., Chen, J., Gao, T. et al. (2016). Reduced graphene oxide

membranes for ultrafast organic solvent nanofiltration. Adv. Mater. 28: 8669–8674. 214 Yang, Q., Su, Y., Chi, C. et al. (2017). Ultrathin graphene-based membrane with precise molecular sieving and ultrafast solvent permeation. Nat. Mater. 16: 1198–1202. 215 Nunes, S.P. and Peinemann, K.-V. (2001). Membrane preparation. In: Membrane Technology (eds. S.P. Nunes and K.-.V. Peinemann), 6–11. Wiley-VCH.

95

3 Nanofiltration Module Design and Operation Tzyy Haur Chong 1,2 and Anthony G. Fane 2,3 1 Nanyang Technological University, School of Civil and Environmental Engineering, 50 Nanyang Avenue, 639798, Singapore 2 Nanyang Technological University, Nanyang Environment and Water Research Institute, Singapore Membrane Technology Centre, 1 Cleantech Loop, Cleantech One, 637141, Singapore 3 University of New South Wales, UNESCO Centre for Membrane Science & Technology, Sydney NSW 2052, Australia

3.1 Introduction Th necessary components of a membrane plant are the membranes, the modules that house the membranes, the system (which includes the arrangement of modules, pumps, piping, tanks, controls, monitoring, pretreatment, and cleaning facilities), and the operating concept (continuous or batch, etc.). Th s chapter describes the role of the membrane module and the commonly used types, with particular reference to nanofiltration (NF). The spiral wound module (SWM) is covered in most detail because it is now the most widely used concept for NF. Various arrangements of modules that can be incorporated into the membrane process design are discussed. Operational strategies, such as diafiltration (to improve product recovery or purity) and batch processing, are also described. Some of the recent developments derive from the use of reverse osmosis (RO) membranes but could apply to NF membranes. The emphasis in this chapter is on aqueous processing with polymeric NF membranes. However, there is a growing application of NF to nonaqueous feeds; this is covered in Chapter 12 on Nanofiltration in the Chemical Processing Industry and Chapter 20 on Organic Solvent Nanofiltration. 3.1.1

Role of the Module

Th module is effectively the membrane “housing,” and it has two important roles, supporting the membrane and providing effective fluid management. Membranes are produced in flat sheet or cylindrical formats. The latter format includes tubular membranes (typically >10 mm of inside diameter) and hollow fibers (typically 1.5 mm i.d.) (in this chapter, the term “hollow fiber” is taken to mean both small-bore and capillary hollow fibers). In the flat sheet Nanofiltration: Principles, Applications, and New Materials, Second Edition. Edited by Andrea Iris Schäfer and Anthony G. Fane. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

96

3 Nanofiltration Module Design and Operation

Membrane

Figure 3.1 Solute concentration gradient.

Cw

Cb

Flux J

δ

Cp

and tubular forms, the membranes are not self-supporting and have to be placed against a porous support, which can withstand the applied pressure and provide a duct for permeate removal. Hollow fibers (see Section 3.2.4) can be self-supporting with either external or internal (lumen) feed, referred to as outside-in or inside-out operation, respectively. Mechanical strength considerations, to avoid collapse or bursting, determine the appropriate i.d. or o.d. for a given feed pressure. To date, hollow fiber formats have been less popular for NF because of the difficulties in composite NF membrane preparation using a hollow fiber substrate. However, developments may be changing this. Effective fluid management is crucial in membrane processing. Th hydrodynamic conditions adjacent to the membrane surface determine the extent of concentration polarization (Section 3.1.2), which has a profound influence on membrane performance. Th various module concepts deal with feed-side fluid flow in different ways, which attempt to balance boundary-layer mass transfer and feed channel pressure losses. Fluid management is also important on the downstream, permeate, and side of the membrane. Permeate usually flows through the membrane support material, and the porosity of this material and the length of the flow path determine the downstream pressure losses, which influence the net transmembrane pressures (TMPs). Th implications of this are considered in Section 3.3.2. A third important aspect of fluid management is the requirement to avoid leakage from feed to permeate. Module designs usually involve permanent (glued) and renewable (O-ring) seals. 3.1.2

Concentration Polarization and Cross-Flow

In the pressure-driven, liquid-phase membrane processes, such as NF, solutes and particles in the feed are convected toward the membrane with the solvent (usually water). The accumulation of retained species close to the membrane is known as concentration polarization, which can be represented by the concentration gradient adjacent to the membrane as depicted in Figure 3.1. Th magnitude of concentration polarization is determined by the balance between convection toward the membrane because of solvent flux (J) and back-transport from the membrane to the bulk because of the concentration gradient. Table 3.1 lists the possible back-transport mechanisms, which include molecular diffusion (Eq. (3.1)), interaction-induced migration (electrokinetic effects) (Eq. (3.2)), and shear-induced diffusion (Eq. (3.3)).

3.1 Introduction

97

Table 3.1 Mechanisms of solute depolarization. Mechanism

Species involved

1. Molecular (Brownian) diffusion (mass transfer coefficient)

Ions, macromolecules, small colloids (0.5 m)

Functional relationship

k = D𝛿 = (kB ⋅ T6𝜇rp )𝛿

(3.1)

JV = (D𝛿) ln(VB 𝛿)

(3.2)

(see Ref. [1]) kS = DS 𝛿 = (0.2 Ub ⋅ rp 2 dh )𝛿 (3.3)

Diffusive back-transport is the most common back-transport mechanism and is represented by the mass transfer coefficient, k (=D/𝛿). Th well-known film model [2] is obtained by a boundary-layer mass balance where net convection = back-diffusion, J ⋅ C  J ⋅ Cp = D ⋅ dCdy

(3.4)

which, after integration over the boundary layer (y = 0 to 𝛿 and C = C W to C b ), gives for a fully retained species (C p = 0), J = k ln(CW Cb )

(3.5)

where C b and C W are the concentrations of solute in the bulk and at the membrane surface. Eq. (3.5) shows that flux is directly related to the boundary layer mass transfer coefficient, or for a given flux, the surface concentration C W is exponentially related to the flux to mass transfer coefficient ratio (J/k). Conversely, the ratio, C W /C b , known as the polarization modulus M, is given by, M=

CW = exp(Jk) Cb

(3.6)

Th polarization modulus expresses the degree of concentration polarization in terms of the increase in surface concentration relative to the bulk solution. Th magnitude of M depends crucially on the diffusivity, D, of the retained solute and the flux, which indicates that for typical fluxes and inorganic ions, M is 10. Th significance of the polarization modulus is that as M increases, it can increase solute transmission, osmotic pressure difference (OPD), and fouling, including scaling because of solubility limits. For species that have partial retention (C p > 0), the film model becomes, J = k ln(CW  Cp )(Cb  Cp )

(3.7)

98

3 Nanofiltration Module Design and Operation

Concentration polarization also reduces the separation capabilities of the membrane. Th observed, or apparent, retention is defined as, 𝜎a = (1  Cp Cb )

(3.8)

Th s retention is the value obtained from the analysis of feed and permeate solutions. However, the membrane provides separation based on its intrinsic retention (𝜎 i ) properties and the concentration it “sees,” C W . The intrinsic retention of the membrane relates C p to C W by, 𝜎i = (1  Cp CW ) From Eqs. (3.7)–(3.9), we find, 𝜎i 𝜎a = 𝜎i + (1  𝜎i ) exp(Jk)

(3.9) (3.10)

Thus, the observed retention is less than the intrinsic membrane retention (𝜎 a < 𝜎 i ), and as the concentration polarization increases (as J/k increases), the observed retention becomes lower, and solute transmission increases. In addition, the situation tends to become worse because of fouling if flux is maintained. Th s is because fouling deposits provide an “unstirred” layer adjacent to the membrane, known as cake-enhanced concentration polarization (CECP) (see Section 3.1.3), which causes the effective k value to decrease (i.e. J/k increases). However, in some cases observed in NF, but more typical of ultrafiltration (UF), the fouling may reduce the effective pore size, and this can increase the intrinsic retention and the observed retention. Th implications of the above equations are that the NF membrane performance is directly linked to the prevailing mass transfer coefficient, which is controlled by the feed channel hydrodynamics. Boundary-layer mass transfer is related to design and operating variables by the well-known Sherwood relationship, Sh = a ⋅ Reb ⋅ Scc (dh L)d

(3.11)

Module design and operation determine the coefficients of this equation. Table 3.2 provides details of relevant equations for different geometries at laminar and turbulent flow conditions. The module geometry is characterized by the hydraulic diameter, dh , which is classically defined as (4  cross sectional area/wetted perimeter); Table 3.3 gives dh relationships for geometries of interest. The Sherwood correlations for SWMs are discussed in Section 3.3.1. Th equations in Table 3.2 are not suitable for a priori design because in most applications, there are uncertainties in physical properties (diffusion coefficients) as well as potential solute–solute interactions and fouling phenomena (see Section 3.1.3). However, the relationships are useful in qualitative analysis and for comparisons. For example, the relationships in Table 3.2 for open channels (tubes, fibers, and slits) can be shown to give for laminar and turbulent flow conditions, k = 1.62 Ub0.33 dh0.33 D0.67 L0.33 , laminar

(3.12)

k = 0.023 Ub0.8 dh0.2 D0.67 𝜇0.47 𝜌0.47 , turbulent

(3.13)

3.1 Introduction

Table 3.2 Mass transfer correlations for use in Eq. (3.11). Regime

Geometry

a

b

c

d

Comment

Laminar

Channel or tube

1.62

0.33

0.33

0.33

0.664

0.5

0.5

0.33

100 < ReScdh /L < 5000. Fully developed velocity profile. Developing concentration profile [3]. See Table 3.3 for dh .

Turbulent

Channel or tube

0.023

0.8

0.33

–

0.023

0.875

0.25

–

Laminar

Stirred cell

0.285

0.55

0.33

–

Turbulent

Stirred cell

0.044

0.75

0.33

–

“Entry region.” Developing velocity and concentration profiles. L ≤ 0.029 dh Re [5] Sc ≤ 1 [6]

1 ≤ Sc ≤103 [7]

8  103 < Re < 32  103 [8]

32  103 < Re < 82  103 Re = 𝜌𝜔rsc 2 /𝜇; rsc = radius of cell

Table 3.3 Hydraulic diameters for various geometries. Geometry

Relationship

Comment

Tube or hollow fiber

dh = di

Rectangular (width w, height h)

dh = 2wh/(w + h)

Re turbulent for dt = 3 mm and 1 m/s velocity for water.

dh = 2 h, narrow slit, with w ≫h

Stirred cell

dh → rsc

See Table 3.2

Spacer-filled channel

dh =

4𝜀 [(2h)+(1𝜀)Svp ]

𝜀 = spacer voidage Svp = surface/volume of spacer [4].

Th se equations show that k, and hence flux, varies with cross-flow velocity and has an exponent of 0.33 in laminar flow and 0.8 in turbulent flow. For tubular modules, the flow regime tends to be turbulent (large dh (>10 mm), large cross-flow velocity U b (>1 m/s)), and for hollow fibers, the regime tends to be laminar (small dh ≤ 3 mm). Th benefit of turbulent flow is higher flux and better response to U b . However, the penalty for this is the added channel pressure loss
Pch . The relationship for channel pressure gradient (
Pch /
x) is [9],
Pch 
x  fF ⋅ 𝜌 ⋅ Ub 2

(3.14)

99

100

3 Nanofiltration Module Design and Operation

where f F is the friction factor (f F = C ′ ⋅Ren ). Th value of n is 1.0 for laminar and 0.0–0.25 for turbulent flow [9], so
Pch 
x  Ub 1.0 (laminar) or Ub 1.75–2.0 (turbulent)

(3.15)

A higher energy penalty for turbulent flow is evident. Noting that turbulent flow would occur for dh > 3 mm and cross-flow velocity > 1 m/s NF modules using tubular membranes would tend to operate in the turbulent regime, suitable for niche applications, such as dirty feeds or those with high solid content. Capillary hollow fibers, with dh < 3 mm, would tend to operate at lower Reynolds numbers in the laminar regime. Mass transfer and pressure losses in spacer-filled channels are special cases considered in Section 3.3. Table 3.1 includes back-transport by interaction-induced migration and shear-induced diffusion. Both mechanisms apply to colloidal or particulate species, which could be present in “turbid” feeds in some applications of NF processes (for example, NF combined with a powdered adsorbent [10]). For shear-induced diffusion, Eq. (3.4) becomes, J = ks ln(Cw Cb )

(3.16)

where k s is given by Eq. (3.3) in Table 3.1. For particles with significant charge interaction with the membrane, Eq. (3.2) in Table 3.2 may be applicable. It should be noted that both Eqs. (3.2) and (3.3) show that these mechanisms are enhanced by decreasing boundary layer thickness, 𝛿. This confirms the importance of cross-flow in polarization control. Th effect of particle size on depolarization mechanisms is also significant. For small colloids (0.5 m), shear-induced diffusion applies, and this increases as particle size increases (Eq. (3.3)). Consequently, colloids around 0.1 m experience a minimum in depolarization and are most likely to deposit on the membrane. Th s may be avoided if favorable electrokinetic interactions exist, but deposition would be unavoidable if there are unfavorable electrokinetics (membrane and particle of opposite charge). Th se considerations point to the importance of pretreatment to limit the concentration of potentially fouling colloids (see Chapter 7 on Fouling and Chapter 8 on Pretreatment). Th flux in NF can also be expressed in terms of the osmotic pressure model, which relates driving force and resistance, J=

TMP 
i = A (TMP 
i ) 𝜇Rm

(3.17)

where
i represents the effective OPD across the membrane, and this is related to the osmotic pressure of the solution at the membrane surface and in the permeate, so
i = W  p  M
a

(3.18)

Here, M is the polarization modulus (Eq. (3.6)) and
a is the OPD based on the bulk feed concentration. Th approximation in Eq. (3.18) assumes high retention (p small) and a linear relationship between  and C. Thus, concentration

3.1 Introduction

polarization will increase M and reduce the effective driving force. This is another illustration of the importance of cross-flow and module fluid management. The constant A in Eq. (3.17) represents the membrane permeability coefficient. Typical values for NF are in the range 5.0–10.0 (l/m2 h bar) [11]. Membrane developments that could lead to significantly higher A values bring challenges to module design and operation (see Section 3.4.1). 3.1.3

Fouling

Concentration polarization is a reversible phenomenon, i.e. when flux is reduced to zero, by closing the permeate line, the C W drops back to C b . However, a usual consequence of concentration polarization is that some species deposit irreversibly (not removed by zero flux operation), and this is known as fouling. Detailed discussions of fouling in NF systems can be found in Chapter 7 on Fouling, and a few general comments in the context of NF and modules are given below. To account for fouling, Eq. (3.17) can be modified to give, J=

TMP 
i 𝜇(Rm + Rf )

(3.19)

where Rf is the fouling resistance. Rf can usually be mitigated by controlling concentration polarization via the flux to mass transfer coefficient ratio (J/k), i.e. by limiting flux and/or providing effective fluid management (see Section 3.4). Module design and operation play crucial roles in determining the value of (J/k), and it is a challenge to achieve desired values throughout the module or system of modules. Some module designs are better able to handle fouling feeds and cleaning regimes, and these requirements may be a factor in module selection (see Section 3.2.6). In addition to the fouling resistance, Rf , in Eq. (3.19), the fouling can also reduce the effective driving force because of CECP [12, 13]. A deposit, or cake, on the membrane surface becomes an “unstirred” layer without the influence of surface shear. Th s CECP leads to hindered back-diffusion of retained solute giving higher values of C W and hence higher W , in Eq. (3.18); this is called the cake-enhanced osmotic pressure (CEOP) phenomenon [12]. Equation (3.6) becomes, M = CW Cb = exp (Jkeff )

(3.20)

where k eff combines mass transfer in the boundary layer (k) and the cake (k C ), i.e. 1keff = 1k + 1kC

(3.21)

Under substantial fouling, k ≫ k C , so k eff → k C . Hindered diffusion cake mass transfer is given by; kC = D ⋅ 𝜀C 𝛿C ⋅ 𝜏C

(3.22)

where 𝜀C , 𝛿 C , and 𝜏 C are cake porosity, thickness, and tortuosity respectively. It has been shown for RO that the effect of CEOP can exceed the effect of the fouling resistance [13], and this could also apply in some NF applications.

101

102

3 Nanofiltration Module Design and Operation

Feed

Membrane plate Permeate spacer e eat

rm Pe

Feed spacer

Retentate

Figure 3.2 Plate and frame module.

3.2 Module Types and Characteristics Membranes can be produced in a flat sheet or cylindrical form, and this determines the type of module geometry. In this section, the various modules are described and compared. More details of the most widely used concept, the spiral wound module, are given in Section 3.3. 3.2.1

Plate and Frame

Th plate and frame modules use flat sheet membranes sitting on a plate that provides a porous support for the permeate outlet. Th flow channels are usually thin, 1–3 mm, and sometimes fitted with a mesh-like channel spacer (see Section 3.3.1). The membranes are stacked in flow channels connected in series or parallel. In some cases, the support plates are of disc or elliptical shape with feed flow radially inward or outward or from one side of the elliptical disc to the other. In other cases, the support plates are rectangular with flow from one end to the other (see Figure 3.2). Th mass transfer characteristics of the plate and frame module, in the absence of spacers, are similar to those of a slit channel. In some cases, the ribbed surface of the support plate may provide undulations in the membrane surface, which could enhance boundary layer mass transfer. Plate and frame modules only

3.2 Module Types and Characteristics

provide a modest surface per volume characteristic and membrane replacement tends to be sheet by sheet, which is labor intensive. Another limitation is that in order to contain the relatively high pressures, as often found in NF applications, the modules tend to have heavy duty end plates; the pressure also constrains the diameter or width of the module. A recent development [14] that could overcome this is a lightweight plate and frame module housed in a pressure vessel. Because the module is surrounded by pressurized liquid, it experiences low forces and can be made from low-cost plastic. Plate and frame modules tend to be used for small- to medium-scale applications in niche areas, leachate, hazardous wastes, etc. They have been superseded by the SWM in large applications, such as the dairy industry and water production, such as surface water treatment and groundwater desalination and reclamation. An alternative approach of using flat sheet membranes developed by Rochem Inc [15] has stacks of flat sheet membrane cushions and spacers in a pressure vessel. Th se have been used successfully for highly contaminated leachate processing of streams with suspended solids [10, 16]. Details of NF and leachate are found in Chapter 15 on Nanofiltration in Landfill Leachate Treatment. 3.2.2

Spiral Wound

Th spiral wound module (SWM) uses flat sheet wound around a central tube. Th membranes are glued along three sides to form “leaves” attached to a permeate channel (tube) along the unsealed edge of the leaf. The internal side of the leaf contains a permeate spacer designed to support the membrane without collapsing under pressure. This permeate spacer is porous and conducts the permeate to the permeate tube (see Figure 3.3a). A feed channel spacer (a net-like sheet) is placed between the leaves to define the channel height (typically 1–2 mm) and provide mass transfer benefits (see Section 3.3.1 for more details). The leaves are wound around the permeate tube and given an outer casing (Figure 3.3b). Th flow paths in the SWM are depicted in Figure 3.3c. Pressurized feed flows axially, parallel to the permeate channel, through the thin spacer-filled channels between the membrane leaves. Liquid, permeating through the membrane surface, enters the leaf and flows radially inward, through the spiral wound permeate spacer, to the permeate tube. The permeate spacer is a porous matrix designed to support the membrane without compression and to have a high hydraulic conductivity for permeate flow. The axial pressure losses on the feed and radial pressure losses on the permeate side produce a distribution of TMP drops, which has implications for optimal design (see Section 3.3.2). In order to avoid distortion of the spiral winding because of the axial pressure drop, each module has an antitelescoping end cap, which provides support as well as open flow paths. Some SWM designs have special flow distributors at the upstream face to minimize maldistribution. Th SWM comes in “standard” diameters (2.5, 4, and 8 in.), with 8 in. the most common size for large plant. Th SWM fits into “standard” pressure vessels, which can take several elements connected in series with O-ring seals to prevent by-passing and feed-to-permeate flow. It should be noted that leakage at the permeate tube connections has occasionally been found to impair the integrity of the SWM for pathogen containment [17].

103

104

3 Nanofiltration Module Design and Operation

Outer wrap Feed channel spacer

Feed channel spacer Permeate collection tube Membrane leaf

Membrane Permeate channel spacer

Permeate channel spacer

Permeate collection tube

(b)

(a) Permeate collection tube Antitelescoping device

Outer wrap

d Fee Membrane Feed channel spacer Permeate channel spacer

eate Perm ntate Rete

(c)

Figure 3.3 Spiral wound module. (a) Basic elements; leaves connected to a permeate tube and feed spacers between leaves. (b) Leaves wound around permeate tube and (c) flow paths in SWM.

A recent development, particularly for RO, is the introduction of 16 in. diameter modules with reported cost savings [18]. These modules can be installed vertically, rather than horizontally. In principle, NF membranes could be incorporated into 16 in. modules. The SWM has become the most widely used concept for large-scale UF, NF, and RO, such as the large NF plant treating river water at Mery-sur-Oise, which incorporates over 9000 8 in. spiral modules [19]. More details of the SWM are given in Section 3.3. 3.2.3

Tubular

Tubular modules have the active membrane surface on the inside of the tubes. Typically, the tube diameters are about 10 mm (in the range 5–25 mm). Th modules are similar to the shell and tube heat exchanger (Figure 3.4) with tubes connected in series and parallel. In some designs, the membrane tubes are inserted into perforated metal support tubes, and in other designs, the tubes are self-supporting. Under the latter condition, the burst pressure of the tubes limits NF applications. Tubular modules are also produced in ceramic materials as multichannel monoliths or single tubes, including multichannel NF modules produced by Inopor [20]. Typically, tubular modules are operated in the turbulent flow regime (Re = 𝜌⋅U b ⋅dh /𝜇 > 3000), which provides good control of concentration polarization, but at a relatively high energy cost. This type of module is well suited for “dirty” feeds because it can handle particulates and be

3.2 Module Types and Characteristics

Feed Shell

Membranes (‘Tubes’)

Permeate

Retentate

Figure 3.4 Shell and tube module.

physically cleaned using foam balls, which are passed automatically up the tube to clean the membrane surface. For example, the Fyne process uses tubular NF with automatic foam ball cleaning for chemical-free water treatment of colored waters [21]; more details of this application can be found in Chapter 9 on Water Treatment. Tubular NF modules have several niche applications at the medium scale. 3.2.4

Hollow Fiber and Capillary

Hollow fiber modules (HFM) use membranes that are “self-supporting,” i.e. the walls are strong enough to avoid collapse or bursting. The outer diameters are typically in the range 0.5–1.0 mm with inner “lumen” diameters of 0.2–0.3, the Péclet number should approach or exceed 1. Within the scope of the steric–Donnan exclusion model, obtaining sufficiently large Péclet numbers requires assuming that either the barrier layer thickness is unrealistically large or its porosity is very low (below 7.9

14.2 29.5 24.6 13.66

6.4 Solution Processes Influencing Speciation and Rejection

retained by NF membranes but, in these instances, because the products are likely to be precipitated solids. Reduction of selenate to selenite would be expected to alter the rejection behavior of selenium slightly if undertaken at pH < 7.9 since the product is a monovalent anion compared to the divalent SeO4 2 . In summary, redox transformations may induce dramatic changes in the rejection of the element of interest by a nanofiltration membrane if the product is a solid while the reactant is a dissolved species or vice versa. Possibly less dramatic but still very significant changes in rejection may be observed if the redox transformation results in a product of different charge to the species initially present.

6.4.5

Adsorption

Adsorption of dissolved species to particulates in the feed to a nanofiltration membrane (or possibly to particulates accumulated at the membrane surface) would be expected to alter the rejection characteristics of the entity, particularly if the species in the non-adsorbed state readily passes through the membrane. Consider, for example, the case of U(VI) presented earlier. At pH < 6, the principal uranyl species present in aqueous solution will be UO2 2+ and UO2 OH+ and, in systems in which the CO2 partial pressure is elevated, UO2 CO3 o . Each of these species would be expected to pass relatively easily through a reasonably porous negatively charged nanofiltration membrane as it is unlikely to be retained by either size exclusion or Donnan exclusion effects. As shown in Figure 6.15, addition of amorphous iron oxide to a feedwater containing U(VI) would induce significant adsorption of U(VI) to the oxide surface in the pH region below 6 with the resultant rejection of the uranium-iron oxide assemblage by the membrane. Note that the extent of adsorption depends upon the concentration of uranium present with a higher proportion of U(VI) adsorbed at lower total U(VI) concentrations. Interestingly, while U(VI) exhibits strong adsorption to the iron oxyhydroxide over the pH range 5.5 to around 8, there is a rapid decrease in the affinity of U(VI) for the oxide surface at higher pHs. As shown in Figure 6.16, this tendency of U(VI) to prefer to be in solution at high pH is accentuated on increasing the partial pressure of CO2 . Indeed, it is relatively easy to show, through the application of the so-called “surface complexation” models that account for the solution phase speciation as well as adsorption processes, that the reason for the tendency of U(VI) to remain in solution at higher pH is because of the stability of the uranyl carbonate species UO2 (CO3 )2 2 and UO2 (CO3 )3 4 . As mentioned previously, these negatively charged species are retained by negatively charged NF membranes as a result of Donnan exclusion. Thus, separation of size exclusion and charge effects contributing to rejection of U(VI) by NF membranes may be nontrivial at high pH. It would be expected, however, that any U(VI) adsorbed to particulates would be robustly rejected by the membrane whereas rejection arising from Donnan exclusion could be mitigated to some extent by increasing charge screening through increased ionic strength. Th extent of adsorption of elements such as U(VI) may be influenced by a wide range of interactions both in solution and at the surface. For example,

265

6 Chemical Speciation Effects in Nanofiltration Separation

100

U uptake (%)

80 60 40 20 0

Total U 3

(a)

4

5

6

10–6 mol/l

80 U uptake (%)

10–4 mol/l 10–5 mol/l

100

10–8 mol/l

60 40 20 0 3

4

5

6

(b)

7

8

9

10

pH

Figure 6.15 U(VI) sorption to ferrihydrite for a range of U. (a) Low pH edge and (b) full pH range (Fe = 1 mmol/l, 0.1 M NaNO3 ). Source: Payne 1999 [17]. Reproduced with permission of T. E Payne. 100

80

U uptake (%)

266

p(CO2) = 10–3.5 atm. p(CO2) = 0.01 atm.

60

40

20

0 3

4

5

6

7

8

9

10

pH

Figure 6.16 Uranium adsorption by ferrihydrite at two partial pressures of CO2 . Curves calculated using MINTEQA2 and appropriate “surface complexation” reactions. Experiments were carried out in 0.1 mol/l NaNO3 , U = 1 mol/l, Fe = 1 mmol/l. Source: Payne 1999 [17]. Reproduced with permission of T. E Payne.

6.5 Effect of Concentration Polarization on Speciation and Rejection

the presence of phosphate can enhance the extent of adsorption of U(VI) to mineral surfaces (and hence membrane rejection) in the acidic pH region as a result of the formation of “ternary” surface complexes while organic acids such as citrate act to retain U(VI) in solution through the formation of complexes such as FeOH2 Cit2 UO2 3 (where the Fe is derived from partial dissolution of the iron oxyhydroxide) [17]. Given their high negative charge, such complexes would be expected to be strongly retained by negatively charged NF membranes though, again, such rejection would be influenced by the presence of charge-screening ions. The size of this dissolved species is also approaching that expected to be retained by some NF membranes by size exclusion alone. Indeed, Favre-Reguillon et al. [22] showed that >95% of uranyl ions (let alone iron-citrate-uranium complexes) were rejected by low molecular weight cutoff NF membranes (150–300 Da) though a higher MW cutoff membrane (2500 Da cutoff ) exhibited rejections of 13–40% with the extent of rejection strongly influenced by the concentration of salts in these waters with this effect most likely related to charge shielding by the major ions present.

6.5 Effect of Concentration Polarization on Speciation and Rejection Th rejection of co-ions and the corresponding counterions leads to an accumulation of ions on the feed side of the membrane. Th concentration achieved in this “concentration polarization” layer adjacent to the membrane and the spatial extent of this layer will depend upon a number of intrinsic and operational factors including concentration of co- and counterions in the feed stream, the permeate flux, and the hydrodynamic conditions at the membrane surface (such as cross flow velocity and spatial configuration of the system – see Chapter 3). Th characteristics of the concentration polarization layer would be expected to be relatively dynamic and will alter depending upon the time since membrane cleaning as well as any changes in feed characteristics and/or operational factors (particularly cross flow and permeate flux). While some insight into the extent of change in ionic composition in the vicinity of the membrane as a result of concentration polarization effects is provided by the results of a coupled model of concentration polarization and pore transport of a multicomponent salt mixture in cross flow nanofiltration developed by Bhattacharjee et al. [23], there is a dearth of information on this important issue. Part of the reason for this is obviously the difficulty in obtaining experimental data on the chemical composition of the solution in the vicinity of the membrane with which to validate the available models of this complex process. A variety of speciation changes might be expected to accompany the formation and development of the concentration polarization layer including • exceedance of solubility product and precipitation of solids, • aggregation of macromolecules and precipitated solids, and • formation of alternative complexes and multinuclear species. Each of these possibilities is discussed briefly below.

267

268

6 Chemical Speciation Effects in Nanofiltration Separation

6.5.1

Exceedance of Solubility Product and Precipitation of Solids

Th extent to which specific ions will accumulate in the concentration polarization layer will be highly case specific but it is clear that solids may precipitate at the membrane surface at rates determined by the feed concentration, the extent of rejection, and the hydrodynamic conditions at the membrane–solution interface. Thus, if the major ions in a feed stream were Ca2+ and CO3 2 , and carbonate was rejected because of charge repulsion, calcium ions would also accumulate on the upstream side of the membrane. With the concentration of both ions increasing on continued membrane filtration, exceedance of the ion activity product for calcium carbonate will occur at some point with precipitation likely to soon follow. While it may be possible to either monitor the accumulation of calcium on the feed side of the membrane or model the concentration buildup at the interface, the most useful sensor of speciation change may well be the rapid increase in head loss across the membrane (and reflected in either a reduction in permeate velocity or a need to increase the pressure to maintain a constant permeate velocity). Methods to control inorganic precipitation are described in Chapter 7.

6.5.2

Aggregation of Macromolecules and Precipitated Solids

Th rejection of ions such as Ca2+ as a result of Donnan exclusion may influence the speciation of entities retained because of size exclusion. For example, natural organic matter (NOM) exhibits molecular weights in the range of 1000–30 000 g/mol and will be retained by most nanofiltration membranes on filtration of natural waters. Th s material is relatively highly negatively charged and will tend to initially form a “non-associating” assemblage of molecules near the membrane surface with charge repulsion lowering the tendency for accumulation on the membrane. As divalent calcium ions also accumulate at the interface, they will interact with the NOM forming complexes that are significantly less charged than the unbound NOM. Calcium ions may also act to induce aggregation of NOM assemblages through both charge neutralization and bridging effects. Calcium ions will also lower the effective charge of the membrane surface through both specific binding and electrostatic shielding. Th result of the Ca–NOM and Ca–membrane interactions will be a much enhanced tendency for the NOM to accumulate on the membrane surface. Effects such as this are clearly demonstrated in the results obtained by Seidel and Elimelech [24] and reproduced in Figure 6.17. 6.5.3

Formation of Alternative Complexes and Multinuclear Species

Th presence of increased concentrations of ions in the interfacial region as a result of Donnan exclusion may result in significant differences in species distribution than occurs at the lower solute concentrations typical of the bulk feed solution. These species may possess different charge to those expected to dominate at lower solute concentrations and may, as a result, exhibit quite different membrane rejection behavior. Thus, enhanced formation of the 2- and

6.6 Conclusions

Normalized flux (J/J0)

1.0 0.8 0.6 0.4

Cross flow

0.1 mM Ca2+

0.0

0

1 × 10–3

Cross flow

80.8 cm/s 40.4 cm/s 12.1 cm/s 4.0 cm/s

0.2

–3

2 × 10

–3

3 × 10

4 × 10

Cross flow 0.3 mM Ca2+ –3

0

Cumulative volume (m3)

(a)

40.4 cm/s 12.1 cm/s 4.0 cm/s

1 × 10–3

–3

2 × 10

–3

3 × 10

–3

4 × 10

1.0 mM Ca2+ 0

Cumulative volume (m3)

(b)

80.8 cm/s 40.4 cm/s 12.1 cm/s 4.0 cm/s

1 × 10–3 2 × 10–3 3 × 10–3 4 × 10–3

Cumulative volume (m3)

(c)

Figure 6.17 Effect of cross flow velocity on NOM fouling at various calcium ion concentrations. Results are presented as normalized flux ((J/Jo ) vs. cumulative volume for calcium ion concentrations of (a) 0.1 mM, (b) 0.3 mM, and (c) 1.0 mM. The following conditions were maintained during the fouling experiments: initial permeate flux ((Jo ) of 11.3 m/s, 1.0 mM NaHCO3 , total ionic strength of 10 mM (adjusted by adding NaCl), pH 8  0.1, and temperature of 25  C. Source: Seidel and Elimelech 2002 [24]. Reproduced with permission of Elsevier.

4- charged anionic uranyl carbonate species mentioned earlier may occur on filtration of aqueous streams containing U(VI) as a result of the preferential rejection of CO3 2 species by NF membranes. Another example of particular interest in water treatment is that of aluminum speciation. At concentrations typical of most natural waters and treatment systems, the ionic species Al3+ , AlOH2+ , and Al(OH)2+ dominate over various pH ranges in the acidic region. Th se species would be expected to readily pass through most nanofiltration membranes; however, their accumulation at the membrane surface may occur to satisfy electroneutrality constraints. Under such conditions, formation of the aluminum tridecamer Al13 O4 (OH)24 (H2 O)12 7+ may be favored. This high molecular weight polynuclear complex would accumulate at the membrane surface and lead, eventually, to precipitation of an aluminum oxyhydroxide solid. In addition to the speciation effects mentioned above, the increased concentration of electrolyte ions in the vicinity of the membrane surface as a consequence of concentration polarization will result in increased screening of charge interaction between specific ions in the solution being filtered and charged groups at the membrane surface. Such an increase in screening will result in a reduction in the extent of rejection on the basis of electrostatic effects and increased dependence on factors such as solute size.

6.6 Conclusions In this chapter, we have explored the fact that a variety of different species of the same element may be present in solution with the concentration of the various species dependent upon the specific solution conditions of pH, Eh, temperature and pressure, total component concentration, and ionic strength. We have flagged that each of the species present may exhibit unique rejection behavior on passage through a nanofiltration membrane with the extent of rejection

269

270

6 Chemical Speciation Effects in Nanofiltration Separation

determined, among other factors, by the size and charge of the particular species. Particular attention has been given to the effects of acid–base transformations, complexation, precipitation, oxidation–reduction, and adsorption on species distribution and the resulting membrane rejection behavior. Th ability to predict species distribution using equilibrium concepts has been flagged as have the difficulties posed by concentration polarization effects, which may lead to inhomogeneous distribution of species in the nanofiltration membrane filtration system and lead to unexpected rejection behavior. Of particular note is the dearth of nanofiltration membrane filtration studies in which a sound understanding of chemical speciation behavior is available. An urgent need exists to undertake nanofiltration studies in which solution conditions are controlled such that species distribution information can be ascertained either by modeling or by analytical methods (or, ideally, both). It is recognized that concentration-polarization effects may render clear interpretation difficult but scope exists to control the extent of such effects and, as a result, obtain clear insight into the rejection behavior of specific chemical species.

Nomenclature and Symbols C iep I J K Ka H M NOM pX P R RX T V Zi 𝛾i 𝜁 {i} [i]

concentration isoelectric point ionic strength permeate flux equilibrium constant acidity constant enthalpy molar natural organic matter log X pressure universal gas constant rejection of solute X absolute temperature volume charge of species i activity coefficient of species i zeta potential activity of species i concentration of species i

Abbreviations DMAA NF NOM

dimethyl arsenic acid nanofiltration natural organic matter

References

References 1 Mulder, M.H.V. (2003). Membrane characterization. In: Nanofiltration –

2

3 4

5 6

7

8 9 10 11

12

13 14 15

16

Principles and Applications (eds. A.I. Schäfer, A.G. Fane and T.D. Waite). Oxford: Elsevier, Chapter 5. Bowen, W.R. and Welfoot, J.S. (2003). Modelling the performance of membrane nanofiltration. In: Nanofiltration – Principles and Applications (eds. A.I. Schäfer, A.G. Fane and T.D. Waite). Oxford: Elsevier, Chapter 6. Benjamin, M.M. (2014). Water Chemistry, 2e. Long Grove, IL: Waveland Press. Westall, J.C., Zachary, J.L., and Morel, F.M.M. (1978). MINEQL, a computer program for the calculation of the chemical equilibrium composition of aqueous systems. Technical Note 18, Department of Civil Engineering, Massachusetts Institute of Technology, Cambridge, MA. Allison, J.D., Brown, D.S., and Novo-Gradec, K.J. (1990). MINTEQA2, A Geochemical Assessment Model for Environmental Systems. Athens: USEPA. Waite, T.D. (1989). Mathematical modelling of trace element speciation. In: Trace Element Speciation: Analytical Methods and Problems (ed. G.E. Batley), 117–184. Boca Raton, FL: CRC Press. Pitzer, K.S. (1979). Theory: ion interaction approach. In: Activity Coefficients in Electrolyte Solutions, vol. 1 (ed. R.M. Pytkowicz). Boca Raton, FL: CRC Press. Morel, F.M.M. and Hering, J. (1993). Principles and Applications of Aquatic Chemistry. New York: Wiley-Interscience. Denbigh, K. (1971). Th Principles of Chemical Equilibrium, 3e. London: Cambridge University Press. Simpson, A.E., Kerr, C.A., and Buckley, C.A. (1987). The effect of pH on the nanofiltration of the carbonate system in solution. Desalination 64: 305–319. Richards, L.A., Vuachere, M., and Schäfer, A.I. (2010). Impact of pH on the removal of fluoride, nitrate and boron by nanofiltration/reverse osmosis. Desalination 261: 331–337. Seidel, A., Waypa, J.J., and Elimelech, M. (2001). Role of charge (Donnan) exclusion in removal of arsenic from water by a negatively charged porous nanofiltration membrane. Environ. Eng. Sci. 18: 105–113. Hodgson, T.D. (1970). Properties of cellulose acetate membrane towards ions in aqueous solutions. Desalination 8: 99. Urase, T., Oh, J.I., and Yamamoto, K. (1998). Effect of pH on rejection of different species of arsenic by nanofiltration. Desalination 117: 11–18. Oh, J.I., Urase, T., Kitawaki, H. et al. (2000). Modeling of arsenic rejection considering affinity and steric hindrance effects in nanofiltration membranes. Water Sci. Technol. 42: 173–180. Brandhuber, P. and Amy, G. (2001). Arsenic removal by a charged ultrafiltration membrane – influences of membrane operating conditions and water quality on arsenic rejection. Desalination 140: 1–14.

271

272

6 Chemical Speciation Effects in Nanofiltration Separation

17 Payne, T.E. (1999). Uranium (VI) Interactions with Mineral Surfaces: Control-

18 19

20

21 22

23

24

ling Factors and Surface Complexation Modelling. PhD dissertation. The University of New South Wales, Sydney, Australia. Raff, O. and Wilken, R.-D. (1999). Removal of dissolved uranium by nanofiltration. Desalination 122: 147–150. Chitry, F., Pellet-Rostaing, S., Nicod, L. et al. (2000). New cesium-selective hydrophilic ligands: UV measures of their interactions toward Cs and Cs/Na separation by nanofiltration complexation. J. Phys. Chem. A 104: 4121–4128. Choo, K.H., Kwon, D.J., Lee, K.W., and Choi, S.J. (2002). Selective removal of cobalt species using nanofiltration membranes. Environ. Sci. Technol. 36: 1330–1336. Stumm, W. and Morgan, J.J. (1996). Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters, 3e. New York: Wiley-Interscience. Favre-Reguillon, A., Lebuzit, G., Murat, D. et al. (2008). Selective removal of dissolved uranium in drinking water by nanofiltration. Water Res. 42: 1160–1166. Bhattacharjee, S., Chen, J.C., and Elimelech, M. (2001). Coupled model of concentration polarization and pore transport in crossflow nanofiltration. AlChE J. 47: 2733–2745. Seidel, A. and Elimelech, M. (2002). Coupling between chemical and physical interactions in natural organic matter (NOM) fouling of nanofiltration membranes: implications for fouling control. J. Membr. Sci. 203: 245–255.

273

7 Fouling in Nanofiltration Andrea I. Schäfer 1 , Nikolaos Andritsos 6 , Anastasios J. Karabelas 2 , Eric M.V. Hoek 3 , René Schneider 4 , and Marianne Nyström 5 1 Institute of Advanced Membrane Technology (IAMT), Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany 2 Chemical Process and Energy Resources Institute, Centre for Research and Technology-Hellas, P.O. Box 361, 57001, Thermi, Greece 3 Department of Civil and Environmental Engineering, University of California, Boelter Hall, CA 5731 Los Angeles, USA 4 Departamento de Microbiologia, Instituto de Ciências Biomédicas, Universidade de São Paulo, Brazil 5 Laboratory of Membrane Technology and Technical Polymer Chemistry, Lappeenranta University of Technology, P.O.Box 20, FIN-53851 Lappeenranta, Finland 6 Department of Mechanical Engineering, University of Thessaly, Athinon and Sekeri St., GR 38334, Volos, Greece

7.1 Introduction Fouling is a side product of effective membrane filtration. According to Koros et al. [1] fouling is “the process resulting in loss of performance of a membrane due to deposition of suspended or dissolved substances on its external surfaces, at its pore openings, or within its pores.” Fouling can be described as flux decline that is irreversible and can only be removed by, for example, chemical cleaning [2]. Th s is different from reversible flux decline due to solution chemistry effects or concentration polarization (CP), which is not considered to be fouling and is described in more detail later in this chapter. Fouling of membranes is important as it limits the competitiveness of the process. Costs increase due to an increased energy demand, caused by an increased pressure being required to overcome higher membrane resistance. Additional labor for maintenance and chemical costs for cleaning are necessary and excessive cleaning results in a shorter membrane lifetime. Essential for effective fouling control is a proactive operation of nanofiltration (NF) or reverse osmosis (RO) plants where an early indication of fouling is acted upon and a good identification of the type of fouling is carried out. Staude [3] summarized the possible origins of fouling as follows: • Precipitation of substances that have exceeded their solubility product (scaling) • Deposition of dispersed fines or colloidal matter Nanofiltration: Principles, Applications, and New Materials, Second Edition. Edited by Andrea Iris Schäfer and Anthony G. Fane. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

274

7 Fouling in Nanofiltration

Figure 7.1 Complex deposit of surface water on a membrane. Source: Adapted from Schäfer 2001 [4].

• Chemical reaction of solutes at the membrane boundary layer (e.g. formation of ferric hydroxides from soluble forms of iron) • Chemical reaction of solutes with the membrane polymer • Adsorption of low molecular mass compounds at the membrane polymer • Irreversible gel formation of macromolecular substances • Colonization by bacteria (mostly hydrophobic interactions) Fouling is indeed very complex and an example of such complexity is illustrated in Figure 7.1 with an electron micrograph of a membrane fouled with surface water without pretreatment. The pictures show colloids and organic matter embedded in a gel-like cake layer on top of the membrane that contains all that was retained by the membrane. In Figure 7.2 another surface water deposit on an NF membrane is shown, except that in this case the surface water is pretreated with ultrafiltration (UF) and fouling is dominated by inorganic precipitates. A number of factors contribute to fouling and are strongly interlinked. The main fouling categories are organic, inorganic, particulate, and biological fouling (Table 7.1). Metal complexes (for example Fe, Al, Si) are also important. While research traditionally focuses on one category or fouling mechanism at a time, it is well accepted that in most cases it is not one single category that can be identified. In most real-life applications all four types of fouling occur simultaneously. Scaling and silica fouling originate in general from the concentration of inorganics exceeding the solubility limit (see Section 7.5). This most often occurs in the latter membrane stages due to the increase in retentate concentration. Metal oxides and colloids deposit early in the process as drag forces are relatively high (see Section 7.6). Organic fouling is very specific to the characteristics of the foulant molecules (see Section 7.4). Organic fouling may occur at the beginning as well as at the end stages of the modules depending on the dominating mechanism. Biofouling can be found throughout all filtration stages (see Section 7.7).

7.1 Introduction

Figure 7.2 Scanning electron micrographs (SEMs) of membranes fouled during the filtration of ultrafiltration pretreated surface water and examined in an autopsy (bar length 200 m for all pictures). Source: Photos courtesy of Paul Buijs, GEBetz, Belgium. Table 7.1 Fouling and where it occurs first. Type of foulant

Most susceptible stage of NF/RO

Scaling/silica

Last membranes in last stage

Metal oxides

First membranes of first stage

Colloids

First membranes of first stage

Organic

First membranes of first stage

Biofouling (rapid)

First membranes of first stage

Biofouling (slow)

Th oughout the whole installation

Source: Adapted from Hydranautics Technical Service Bulletin TSB107 in Huiting et al. 2001 [5].

Rapid biofouling can be related to particle attachment, which is found mostly in the first stage, whereas slow biofouling can occur throughout all stages [5]. While in the past bacterial deposition and fouling have often been studied by using latex particles, the adhesive nature of extracellular polymeric substances (EPSs) makes bacteria more adhesive and their deposition mechanism more complex [6]. In order to reduce or eliminate fouling it is necessary to identify the foulants. Th s can be achieved by a characterization of the fouled membrane (membrane

275

276

7 Fouling in Nanofiltration

Table 7.2 Foulants and their control strategies in NF and RO processes. Foulant

Fouling control

General

Hydrodynamics/shear, operation below critical flux, chemical cleaning

Inorganic (scaling)

Operate below solubility limit, pretreatment, reduce pH to 4–6 (acid addition), low recovery, additives (antiscalants) Some metals can be oxidized with oxygen

Organics

Pretreatment using biological processes, activated carbon, ion exchange (e.g. MIEX), ozone, enhanced coagulation, ultrafiltration

Colloids and nanomaterials ( pore size: Pore penetration is not possible and adsorption sites are only available on the membrane surface

Figure 7.7 Simplified diagram of adsorption for different solute to pore size proportions.

Adsorption can be measured using the partitioning coefficient between the membrane and bulk phase, which is defined in Eq. (7.20). (7.20) (lm2 ) M⋅C where is the adsorbed quantity of organic (g/m2 ), M, the molar mass of the adsorbing compound (g/mol), and C, the equilibrium concentration of the solute in the solution (mmol/l). Van der Bruggen et al. [2] observed flux declines of up to 59% with organics in solutions with concentrations of about 1 g/l. Combe et al. [72] also determined K=

291

292

7 Fouling in Nanofiltration

foulant–membrane partitioning coefficients and those researchers also considered the solution volume. Van der Bruggen and Vandecasteele [73] have used an adapted Freundlich equation for the direct description of adsorption from flux decline measurements as per Eq. (7.21). J = Kf ⋅ C n

(7.21) 2

where J is the water flux (l/m h) and K f as well as n are parameters. In filtration, the adsorbed amount can be determined by mass balance using Eq. (7.22) CF VF = A + VP

n ∑ 1

Cpi + Cc Vc

(7.22)

where A is the membrane area (cm2 ), is the amount of solute adsorbed per surface area (ng cm2 ), and n is the number of permeate samples, and C F , C P , C C and V F , V P , V C are concentration and volume of feed, permeate and concentrate, respectively. Using this equation for the determination of adsorption assumes that all solute is adsorbed rather than lost in other ways. For higher concentrations this is more correctly expressed as the amount of deposit. 7.3.4

Gel Layer Formation

Gel formation is considered as the precipitation of organic solutes on the membrane surface. This process usually occurs when the wall concentration due to concentration polarization exceeds the solubility of the organic. Gel formation does not necessarily mean irreversible flux decline and can be imagined as a layer of high viscosity that hinders the flow of low viscosity solute. Th gel polarization model is based on the fact that at steady state conditions flux reaches a limiting value, where an increase in pressure no longer increases the flux. According to the gel polarization model, at this limiting value, the solubility limit of the solute in the boundary layer is reached and a gel formed. For 100% retention, the expression for this limiting flux (J lim ) is described by Eq. (7.23). cG is the gel concentration, beyond which the concentration in the boundary layer cannot increase. c (7.23) Jlim = kS ln G cB Th model does not include membrane characteristics, and tends to predict a lower flux than observed. Tang and Leckie [74] showed that for RO and NF membranes, fouled by humic acid, the limiting flux is independent of membrane properties and is directly proportional to the intermolecular electrostatic repulsive force [75]. This was attributed to the dominance of interactions between

7.3 Fouling Mechanisms

Solute < pore size: Pore penetration is possible and gel formation of a solute occurs on the membrane surface Gel layer and in the pores Solute > pore size: Pore penetration is not possible and gel formation occurs only on the membrane surface

Figure 7.8 Simplified diagram of gel layer formation following adsorption.

deposited foulant and new foulants in the feed [75, 76]. An improvement can be achieved in using DS for the gel layer rather than the bulk solution [58] (Figure 7.8). 7.3.5

Cake Formation and Pore Blocking

Belfort et al. [77] proposed five stages of fouling in microfiltration of macromolecules that are somewhat applicable to NF. These are (i) fast internal sorption of macromolecules, (ii) buildup of a first sublayer, (iii) buildup of multi-sublayers, (iv) densification of sublayers, and (v) increase in bulk viscosity. Th fifth stage can be neglected for dilute suspensions such as surface water. The dependence on particle size can be described as follows: dsolute < dpore : deposit on pore walls, restricting pore size dsolute  dpore : pore plugging or blockage dsolute > dpore : cake deposition, compaction over time Th se principles are illustrated in Figure 7.9. For solutes much smaller than the membrane pores, internal deposition eventually leads to the loss of pores. Solutes of similar size to the membrane pore will cause immediate pore blockage. Particles larger than the pores will deposit as a cake, with the porosity depending on a variety of factors including particle size distribution, aggregate structure, and compaction effects. Th process of small particles adsorbing in the pores may be slow compared to pore plugging, where a single particle can completely block a pore, and therefore flux decline should be more severe for the latter case. If the membrane is nonporous then the deposition of solutes takes place on the membrane surface with smaller solutes generally forming less permeable deposits. Chang and Benjamin [78] pointed out that the mechanisms of organic colloid deposition and gel layer formation require the application of different models, although many authors fail to differentiate between gel and cake formation. Some authors refer to those mechanisms as deposit formation to simplify the issue. Th differentiation is not always simple, especially when considering that aggregation of the gel composites may in fact form a more particulate or colloidal deposit.

293

294

7 Fouling in Nanofiltration

(a)

Pore versus surface fouling Pore adsorption (dsolute> dpore): Colloid or solutes larger than the pores are retained due to sieving effects and form a cake on the membrane surface; depending on pore to particle size ratio flux decline occurs (permeability of the cake layer as well as the cake thickness are important)

(b)

Impact of colloid or solute stability

Stable colloids smaller than the pore size are not retained by membrane, unless adsorbed by the membrane material

Tight aggregates are formed by slow coagulation, are retained and form a cake on the membrane. The aggregate structure may collapse depending on forces on the aggregate and the aggregate stability. Flux through the tight aggregates is usually low unless the aggregates deposit as a porous cake of large particles. Loose aggregates are formed by rapid coagulation and are also retained. Such aggregates form a cake on the membrane. The aggregate structure may collapse depending on forces on the aggregate and the stability of the aggregate. Flux through the open aggregates is high if the structure is maintained during filtration. (c)

Solute-solute interaction

Colloids < pores and stabilized with organics (for example) are not retained by the membrane, unless adsorbed by the membrane material or destabilised with high salt concentrations.

Aggregates with organics adsorbed after aggregation (for example) are fully retained by the membrane, but may penetrate into the upper layer of the membrane. This could also be organics destabilized with multivalent cations. Colloids which are partially aggregated and destabilized such as a variety of solutes that interact with each other in heterogeneous ways in the presence of salts, colloids, and dissolved organics, form small and diverse aggregates which may block pores. Note that colloids can be replaced in such mechanisms with nanoparticles.

Figure 7.9 Colloid – organic fouling mechanisms (a) pore versus surface fouling, (b) impact of colloid or solute stability, and (c) solute-solute interaction (colloids/nanoparticles with organic matter).

7.3 Fouling Mechanisms

7.3.6

Critical Flux and Operating Conditions

Critical flux stems from the concept that the higher the flux the stronger is the drag force toward the membrane (and hence deposition of colloids), the stronger the concentration polarization (and hence the boundary layer thickness and solute concentration), and the higher the compaction of a deposit. The stronger the flux the less dispersible the deposit will be. The original critical flux concept stems from MF and was first introduced by Field et al. [79]. Critical flux is defined as the limiting flux value below which flux decline over time does not occur [74, 79]. Critical flux represents the shift from repulsive interaction (dispersed matter–polarized layer) to attractive interaction (condensed matter–deposit) [80]. Traditionally, critical flux derives from the filtration of particulate matter using porous membranes. Mänttäri and Nyström [12] describe a strong and weak form of critical flux where the strong form describes the flux where the actual flux starts deviating from the CWF, whereas the weak critical flux is the point where flux increase with pressure is no longer linear. This is illustrated in Figure 7.10 where the solid line is the linear dependence of CWF of pressure, while the dashed line is the linear dependence of permeate flux of pressure. The hollow and solid circles show the permeate flux after a stepwise increase and decrease of pressure, respectively. The squares are flux values after filtration at the highest pressure (and hence with significant irreversible fouling). Figure 7.10 Critical flux in NF. Note that lines represent different experiments. Source: Reproduced from Mänttäri and Nyström 2000 [12].

80 Locust bean gum 50 mg/l No precleaning Pure water flux

Flux (l/m2 h)

60 Over the critical flux

40 Weak: the highest flux which is linearly dependent on the pressure

20

Strong: the flux is the same as the pure water flux

0

2

4

6 8 Pressure (bar)

10

12

A number of parameters influence this critical flux. For example, cross flow velocity increases the critical flux while solute concentration decreases it. Repulsion between solute and membrane also increase the critical flux in the case of high molar mass polysaccharides, while in paper industry effluents only weak forms of critical flux were found [12]. Some authors have noted that the thickness of the fouling layer is primarily dependent on the initial flux [39]. Gwon et al. [37] compared NF and RO fouling and found that the fouling layer in NF was mostly organic and could be fully recovered while in RO the fouling was inorganic and organic and could not be recovered. Th fouling at the end of the RO modules was most severe, which indicates the importance of reduced cross flow and increased

295

296

7 Fouling in Nanofiltration

concentration. However, flux at the end of modules would normally be lower than at the module entrance. 7.3.7

Additional Fouling Mechanisms

In NF processes, the retention of ionic species results in a concentrated layer of ions at the membrane surface (known as salt concentration polarization), which creates an osmotic pressure drop across the membrane. Sub-micrometer colloids are highly Brownian, which means that they are influenced by diffusive, as well as convective transport mechanisms. Also, aggregation and deposition of small colloids are strongly influenced by colloidal forces [81]. The polarized layer of rejected ionic solutes exacerbates colloidal fouling of NF membranes by greatly reducing repulsive electrostatic interactions. Moreover, aggregation of organic macromolecules and precipitated salts may occur in the bulk solution near the membrane surface where rejected ionic solute concentration is higher than in the bulk. Th se aggregates may act like very small colloids (typically < 500 nm) and cause severe fouling because they are not removed in the dissolved form by pretreatment. Therefore, feed solution chemistry and membrane ion retention are critical to the formation of colloidal cake layers. Accumulation of rejected dissolved and (organic, inorganic, or biological) colloidal matter at the membrane surface presents the opportunity for additional fouling mechanisms. Th se mechanisms arise from interactions between rejected ions and colloids passing through the concentration polarization layer and at the membrane surface. Th classic picture of this situation is presented in Figure 7.11 where it is assumed that a stagnant cake layer develops with a salt and colloid polarization layer flowing above the cake layer. In addition, analysis of the factors affecting dissolved solute mass transfer reveals one potential interaction between a colloidal cake layer and the salt CP layer. Increasing the bulk flow rate increases the shear rate, which enhances mass transfer (Figure 7.12). However, the most influential variable on mass transfer is the solute diffusivity (k s  D2/3 ). It has been proposed that the mutual diffusion coefficient of rejected salt ions may be hindered within the colloid deposit layers [82–84]. A hindered salt diffusion coefficient has been used to describe colloid cake–CP layer interactions in cross flow RO/NF membrane filtration [83–85]. The result was the elucidation of a single mechanism – “cake-enhanced concentration polarization” – capable of describing the majority of observed flux decline, as well as the observed decline in salt retention due to colloidal fouling of NF (and RO) membranes. Th overall MTC was considered the sum of two MTCs, one describing salt back-diffusion u Polarized layer

y

Cake layer

v(x)

Membrane

Figure 7.11 A schematic of a cross flow NF process showing the development of the cake and concentration polarization layers, and the corresponding permeate flux decline along the axial direction.

7.3 Fouling Mechanisms

U0

Hc

D∞ v(t)

du/dv = γ0 = 6U0/Hc

Hc* = Hc – δc

D* < D∞

v(t) δc

Figure 7.12 Conceptual illustration of hindered mass transfer in cross flow membrane filtration. The tangential flow velocity, U0 and the salt ion diffusion coefficient are critical parameters in determining mass transfer in the salt concentration polarization layer. Tangential flow and salt ion back-diffusion may be locally hindered in the presence of a colloid deposit layer, thus enhancing the membrane surface salt concentration and the resulting transmembrane osmotic pressure.

from the membrane surface through the cake layer, and one through the remainder of the salt CP layer. Incorporating the hindered MTC into Eq. (7.14) and solving for the transmembrane osmotic pressure yields Eq. (7.24): [ ( )] 1  ln(𝜀2 ) J 1
m = fos Cb Ro exp + v𝛿c  (7.24) ks Ds 𝜀 Ds where
𝜋 m * is termed the “cake-enhanced osmotic pressure.” Th term in brackets in Eq. (7.24) comes from considering a thin cake layer, in which the tangential flow field is assumed unchanged by the presence of the cake, and hindered diffusion alone reduces mass transfer [82, 84]. The reduced salt diffusivity in the cake layer is expressed as 𝜀 D /𝜏, with the tortuosity, 𝜏, being approximated as 1  ln(𝜀2 ) [82–84, 86]. The only term on the right-hand side of Eq. (7.21) that is not a known constant or experimentally measurable parameter is 𝜀, the cake layer porosity. The concept of cake-enhanced concentration polarization on flux when operating at constant pressure is illustrated in (Figure 7.13). The data is based on laboratory experiments of Hoek et al. [84, 87]. Initially, the applied pressure and membrane (hydraulic) resistance control pure water flux (a). For a simple electrolyte feed solution (b), an osmotic pressure drop (
𝜋 m ) across the membrane develops nearly instantaneously due to the accumulation of rejected salt ions at the membrane surface. Transmembrane pressure is the difference between applied pressure and the transmembrane osmotic pressure. Immediately after colloidal particles are added to the feed (c) they begin to accumulate on the surface of the membrane and form a “cake” layer. A hydraulic pressure drop forms across the stationary colloid cake layer, which increases as the cake layer thickness increases. More importantly, the concentration of rejected salt ions builds up within in the cake layer because mass transfer

297

298

7 Fouling in Nanofiltration

Cm* Cm Cb

Δπm* > Δπm

Cb

Δπm

Δpc Cp

Cp* > Cp

Applied pressure, ΔP

Δpm(t)

Δπm*(t) Δpc(t)

P

Pure water (a)

Salt added (b)

Particles added (c) Δpc(t) Δπm*(t)

Δπm Time

Figure 7.13 Conceptual illustration of the effect of cake-enhanced concentration polarization on flux when operating at constant pressure based on laboratory experiments of Hoek et al. [84, 87]. Source: Adapted from Hoek et al. 2002 [84] and Hoek and Elimelech 2003 [87].

(back-transport of salt ions) through the cake layer is hindered. Th resulting “cake-enhanced osmotic pressure” (
𝜋 m ) can be 1 order of magnitude greater than the trans-cake hydraulic pressure when membrane salt retention is high. Th greater implication of this finding is that any accumulated mass on the surface of a salt rejecting (NF/RO) membrane may entrap ions, enhancing the transmembrane osmotic pressure. Therefore, cake-enhanced osmotic pressure may play a role in fouling due to the most ubiquitous and recalcitrant foulants in NF processes, namely biofilms, scale, and organic matter [88–98]. Further, the mechanism of salt entrapment within foulant deposit layers helps to explain the commonly observed decline in salt retention associated with NF membrane fouling. Th entrapment of salt ions within the cake layer enhances the membrane surface salt concentration, and therefore, the chemical potential gradient responsible for solute transport through NF membranes. It is possible that fouling by macromolecules with high charge density (e.g. proteins, humic and fulvic acids, etc.) may actually reject salt ions and other dissolved species if they form densely packed cake or gel layer. In such cases, enhanced concentration polarization phenomena may be suppressed. Th se interactions are discussed in more detail in the following sections under organic fouling 7.4.

7.4 Organic Fouling

7.4 Organic Fouling Organics interact with membranes in a number of ways and it is difficult to single out individual interaction mechanisms. Th mechanism is strongly dependent on the organic type and the chemical characteristics of the molecules as well as their affinity towards the membrane material. Some of those solute–membrane interaction mechanisms are described in Chapter 19 for trace contaminants. 7.4.1

Introduction and Definition of Organic Fouling

Organic fouling is the irreversible flux decline due to the adsorption or deposition of dissolved or colloidal organic material. Th s fouling may occur as adsorption at a molecular level or as a monolayer, the formation of a gel on the membrane surface, the deposition or cake formation by organic colloids or the pore restriction and blocking by molecules that can penetrate into the membrane. Such organic fouling can be severe and persistent; for example Roudman and DiGiano [99] reported that even rigorous chemical cleaning failed to remove NOM from NF membranes. 7.4.2

Common Organic Foulants

Organic matter plays an important role in fouling and act in a number of ways. Firstly, organics may adsorb to or deposit on membranes resulting in a variation of the surface characteristics and hence flux and fouling behavior. Secondly, organics may act as a nutrient source for microorganisms and hence facilitate biofouling. Thi dly, organics may adsorb onto colloids, stabilize small colloids, and hence make it more difficult for those colloids to be removed in pretreatment. In fact, in the natural environment colloids commonly have a negative surface charge due to an adsorbed layer of NOM, which can lead to stabilization of the colloids [100, 101]. The degree of colloid stability depends on the amount of organics adsorbed. Organics themselves are also described as “colloids” and hence organic and colloidal fouling categories overlap. In the water and wastewater industry, natural and EfOM are well known and frequently studied foulants [102]. NOM is predominantly composed of so-called humic substances [4]. EfOM is the wastewater equivalent of NOM and contributes to membrane fouling by adsorption, surface accumulation or pore blocking, mostly by the humic fractions and polysaccharides [103]. Depending on the source of NOM, its composition differs and it is indeed the NOM source that determines fouling of membranes [104]. Wiesner et al. [105] identified four NOM categories, which are strong foulants – proteins, aminosugars, polysaccharides, and polyhydroxyaromatics. Polysaccharides were also confirmed as compounds of relevance to fouling in the field of wastewater treatment [56, 106]. Ang and Elimelech [107] observed the interplay of the contribution of a model protein (bovine serum albumin [BSA]) with a model polysaccharide (alginate), while Kimura et al. [108] identified lipopolysaccharides (LPSs) as a cell derived foulant in a membrane bioreactor (MBR). Jarusutthirak et al. [53] further fractionated and characterized EfOM. The following fractions were separated:

299

7 Fouling in Nanofiltration

(i) colloidal EfOM with hydrophilic character composed of polysaccharides, proteins, aminosugars; (ii) hydrophobic EfOM with humic substance characteristics (high aromaticity and carboxylic functional groups); (iii) transphilic EfOM also with humic substance characteristics; and (iv) hydrophilic EfOM containing low molecular mass acids. Lee et al. [41] determined that both the hydrophilic and the hydrophobic fractions adsorbed significantly to UF membranes, whereas transphilic NOM, mostly composed of hydrophilic acids, adsorbed very little. Such a fractionation can be presented in a humification diagram. A diagram for a number of NOM examples from surface waters is presented in Figure 7.14. Th relationship to EfOM is visible in the bottom left corner where the approximate location of fulvic acids from sewage treatment plants is indicated. The different characteristics of EfOM and NOM as well as their fractions reflect in fouling characteristics. Generally, NOM can be described as hydrophobic (mainly humic and fulvic acids) and hydrophilic (mainly polysaccharides and proteins) fractions. Yamamura et al. [111] fractionated NOM into hydrophobic and hydrophilic fractions and identified the hydrophilic fraction, namely carbohydrates and proteins, as the main contributor to irreversible fouling. Khan et al. [116] claimed that even in seawater RO, the conditioning film formation by humic substances is the first step of fouling and it assists microbial adhesion. Nyström et al. [55] have investigated a number of organic molecules toward their fouling characteristics. A type of starch that had higher protein content fouled the membranes very strongly. Fouling of polysaccharides and humic substances showed that when the organics 12

Organics used in this study

10

SAC/OC (l/mgm) Aromaticity

300

IHSSHA IHSSFA Aldrich 100k NOM NOMHA NOMFA NOMHyd

Pedogenic humics

8 Pedogenic fulvics

6

4

Aquagenic fulvics

Other organics

Fulvics from sewage treatment

Seine Main Rhine Karst Kleine Kinzig IHSSFAStd IHSSHAStd

Humification Pathway

2 HS-Hydrolysates

0 250

500

750

1000

1250

1500

1750

Molecular weight (g/mol) Molecularity

Figure 7.14 Humification diagram showing the molecular mass and aromaticity for a number of surface waters as reported by Huber [109] and organic compounds used in water research. Source: Adapted from Huber 1998 [109] and Schäfer et al. 2002 [110], SAC/OC is specific adsorption.

7.4 Organic Fouling

were charged fouling was pH dependent, with the highest amount of fouling occurring when the charge repulsion was lowest [56]. Solute–solute interactions also influence fouling [56]. Lee and Elimelech [112] investigated foulant–foulant intermolecular adhesion forces on organic fouling using AFM. Fouling was reported to be more severe at solution chemistries that resulted in larger adhesion forces, namely, lower pH, higher ionic strength, and presence of calcium ions (but not magnesium ions) [76, 112]. It appears that solution composition plays a bigger role at higher fluxes [76], which may draw into play the solubility and colloidal chemistry at the membrane interface. However, those interactions remain poorly understood and will naturally be very difficult to predict for real waters. A foulant that is of organic nature but with a close relation to biofilm is EPS. EPSs surround microorganisms and may be produced by a biofilm that tends to attach well to surfaces and can cause pore blockage when removed from the bacterial cells, as is described in Section 7.7. Chang and Lee [113] have linked fouling and EPS content in an MBR application and suggested EPS content as a possible feedwater fouling index for wastewater applications. The importance of EPS in MBR fouling was confirmed by Cho and Fane, who determined that fouling occurred in two stages – a gradual deposition of EPS on the (in this case MF) membranes followed by a rapid and sudden stage of biomass growth that required membrane cleaning [40]. Amy and Cho [114] identified polysaccharides as dominant foulants in UF and NF of surface water, although polysaccharide concentration in surface waters would be comparatively low. Kimura et al. [108] established LPS as a significantly more potent foulant than the typical model substances alginate and dextran with a high affinity for membrane (in this case MBR) polymers. Th behavior of foulants in mixtures and ultimately real waters and wastewaters is a complex issue. For example, Mackey [115] studied the fouling of UF and NF membranes by various model compounds, such as polysaccharides, polyhydroxyaromatics, and proteins. Larger compounds (polysaccharides and proteins) caused more fouling, and in mixtures, the fouling further increased. 7.4.3

Adsorption of Organic Matter

Adsorption plays an important role in the fouling of NF membranes by organic compounds. In fact, adsorption is often regarded as the first step in membrane fouling. Adsorption of organic compounds can be considered as the formation of a conditioning film that allows the attachment of bacteria and hence biofouling [116]. Adsorption may also cause pore narrowing and may hence be a precursor to pore plugging. Adsorption of humic substances has, for example, been shown to occur in pores as well as on the membrane surface [72]. Whether adsorption occurs on the membrane surface or in the pores depends on pore size, molecular size and shape, as well as the solution chemistry, which can change the structure and shape of organics [117]. Chang and Benjamin have estimated the thickness of an adsorbed monolayer of NOM to be about 1.6 nm, which is larger than the typical NF pore dimensions. Hence, if discrete pores exist, such adsorption leads to pore restriction or blockage.

301

302

7 Fouling in Nanofiltration

Adsorption of organic molecules into the membrane matrix changes the free volume in the membrane. Depending on the molecule the interaction can either increase or decrease this free volume and hence flux [55]. Nyström et al. [55] showed, for example, that small vanillin molecules caused an increase in flux, if charged. Charged molecules with longer chains did not cause fouling, while proteins caused very strong fouling. Reduced fouling was attributed to a lack of interaction with the membrane due to charge repulsion, while NOM adsorption rendered membranes more hydrophilic and hence increased water permeation [99]. Nikolova and Islam [64] showed that in UF of dextran the adsorbed layer caused most of the flux decline as opposed to osmotic pressure effects, although the adsorption was, in this case, reversible. Adsorption is an equilibrium process between the wall concentration determined by concentration polarization and the adsorbed organics. According to Nikolova and Islam [64] the relationship between adsorption and concentration is linear. Other authors describe the adhesion more mechanistically in that adhesion occurs due to double layer interactions or hydration forces when adsorbing molecules and the membrane reach close enough contact to interact [39]. In order to resolve such mechanisms, time-series image analysis (TSIA) was performed using SEM and AFM for humic substances fouling. Humic substances adsorbed within the first couple of hours of operation, smoothening the surface. This organic “gel” layer then accelerates the fouling formation on the membrane surface by facilitating the attachment of colloids and organic substances [118]. Carlsson et al. [54] performed a study using pulp mill effluent and UF and found that hydrated lignin sulfonates adsorbed to the membrane surface followed by later deposits of cellulosic oligomers. Champlin [119] investigated the impact of NOM adsorption on NF membranes in the presence of particulate matter. NOM adsorption was as high as 12.6% of the available NOM, but interestingly this adsorption is reduced in the presence of particulate matter. The postulated mechanism was that particles were acting as abrasive scouring agents or an adsorbent that competes with the membrane surface for NOM. A similar concept was exploited by Imbrogno et al. [48] where magnetic ion exchange particles that adsorb NOM were able to reduce organic fouling. Adsorption itself can either be the precursor to a more severe fouling layer or cause significant fouling by itself [2]. Adsorption of organic compounds can also alter the membrane surface characteristics (such as increasing hydrophobicity or membrane charge) and hence lead to flux variations. It has been reported that fouling is more severe when nonpolar bonds are formed between the foulant and the membranes as opposed to polar bonds [120]. According to Chon and Cho [121] the adsorption of hydrophilic organic matter and inorganics is the precursor to further organic fouling. Th effect of humic acid on membrane surface charge has been investigated by a number of researchers [122–125] and it has been shown that humic substances influence the surface charge (in general a more negative charge is observed), and the adsorbed organics in fact dominate the surface charge with their functional groups. Th adsorption of organics by hydrophobic membranes is generally higher [72]. The deposition of the more aromatic compounds appears stronger

7.4 Organic Fouling

and can also be facilitated by the presence of calcium [126]. Multivalent ions can play different roles [127]: (i) charge repulsion can be reduced in the presence of the electrolyte; (ii) cations may form bridges between identically charged foulants and membranes; and (iii) the configuration of the foulant molecules is modified. A number of foulant and membrane characteristics are important in adsorption; those are water solubility, dipole moment, octanol water partitioning coefficient, surface charge, hydrophobicity, molecular size/mass, and membrane molecular weight cutoff (MWCO) or pore size [2]. Methods to characterize membranes were described in detail in Chapter 4. Adsorptive fouling of organics does not always decrease as the negative charge and hydrophilicity of a membrane increase. In fact, oxidation of the membrane has been reported to increase negative charge, hydrophilicity, and humic acid adsorption [72]. Jarusutthirak and Amy [103] found that negatively charged membranes adsorbed the hydrophobic fraction of EfOM. Concentration matters, and this influence can be significant in the formation of conditioning films. Nghiem and Schäfer [42] have determined a breakthrough phenomenon for some NF membranes that can be attributed to the adsorption of contaminants at very low (ng/l) concentrations, namely micropollutants. Th amount adsorbed was dependent on whether penetration into the active layer by the contaminants was possible. Adsorption was dependent on the pK a of the contaminants with higher adsorption when the compounds are undissociated. Th s is elaborated in more detail in Chapter 19. Similar trends were observed by Jones and O’Melia [128] when hydrophobic interaction of BSA with membranes decreased as the isoelectric point (IEP) was exceeded. A summary of the adsorbed quantities of trace contaminants in membrane modules is given in Chapter 19. Th actual interaction mechanisms that govern adsorption are not well understood. Freundlich isotherms were found to describe the adsorption of NOM [119]. Roudman and DiGiano [99] suggested the predominance of acid–base interactions and hydrogen bond formation between NOM and membranes. Yamamura et al. [129] investigated the affinity of carbohydrate-like substances (hydrophilic fractions of NOM) to two different MF membranes. Hydrogen bonding was postulated to be the main mechanism resulting in strong adhesion forces between humic substances and PVDF membranes. Hydrogen bonding was also considered as the main reason for the adsorption of biopolymer-NOM on polyamide and polysulfone membranes [130]. This is elaborated in more detail in Chapter 19. To date, it is not easily possible to distinguish between hydrogen bond formation, general hydrophobic interactions, or possibly other more refined supramolecular mechanisms. Th stronger a compound adsorbs to the membrane, the higher the flux decline [2], which has also been attributed to a reduced swelling that increases hydraulic resistance [131, 132]. Partitioning coefficient appears to increase with dipole moment indicating a possible charge interaction. The partitioning coefficient also increases with the octanol water partitioning coefficient (K ow ; log P) and hence hydrophobicity. In other words, hydrophobic interactions between membranes and organics result in flux decline. Water solubility describes the polar character of a molecule, but was not identified to have a correlation with adsorption. Considering the importance of hydrophobic interactions, it is not

303

304

7 Fouling in Nanofiltration

surprising that it is repeatedly reported that hydrophobic membranes foul more [54]. 7.4.4

Gel Layer Formation

As indicated above in Section 7.3.4, gel formation occurs when the solubility of a noncrystalline solute is exceeded. This is often the case when organic molecules flocculate in the presence of salts and at neutral charge conditions [55] such as when the surface concentration increases due to concentration polarization [133]. More details on coagulation are given in Section 7.4.7. Deposits formed on the membrane surface by materials that are too large to penetrate into or through the membrane eventually reach a steady state thickness. Given the importance of concentration polarization, cross flow velocity can be expected to reduce such phenomena [133]. Chang and Benjamin [78] estimated that such a film could grow at a rate of about 0.3 m a day in full-scale systems that remove NOM in water treatment. Th se authors assumed a density of the NOM gel layer of 1 g/cm3 , a water content of 50%, and an NOM carbon content of 50% by mass to calculate that a 0.3 m thick layer requires 75 mg of DOC per m2 . Th s was estimated to be about 1% of the organic carbon that a typical water treatment system would be exposed to. Gel formation was observed by Jarusutthirak et al. [53] in the filtration of EfOM due to the large molecular mass of the colloidal EfOM fraction and the small MWCO of NF. The hydrophobic and transphilic fractions were assumed to cause a gel layer also, initiated by hydrophobic interactions, while charge repulsion reduced fouling by charged molecules. It has been shown that in gel layer formation in MBRs from soluble and colloidal microbial products (SCMPs), strong carboxylic sites are predominantly responsible for gel formation [134]. 7.4.5

Cake Formation

Seidel and Elimelech [133] described fouling of NOM as a combination of permeation drag and calcium binding, and hence a coupled process between hydrodynamics and chemical interactions. Very importantly, those authors have pointed out that permeation drag can overcome repulsive forces of double layers and cause foulant deposition at typical operating conditions. This observation strongly supports the critical flux phenomena from Section 7.3.6. At low flux, below such a critical flux, the repulsion between foulant and membrane may be strong enough to prevent deposition. Calcium can adversely affect some fouling prevention strategies such as cross flow velocity. Hong and Elimelech [127] related solution chemistry with the formation of a membrane deposit with varying characteristics (see Figure 7.15). This illustrates also the change in foulant conformation due to solution chemistry. When the charge of the foulants is low, which for NOM occurs at high ionic strength, low pH, and in the presence of multivalent ions, the NOM is coiled and deposits as a firm cake. If the repulsive forces between the NOM functional groups are enhanced then the cake layer is less sticky and more porous. It should be noted here that the

7.4 Organic Fouling

Chemical conditions

NOM in solution

NOM on membrance surface Compact, dense, thick fouling layer

High ionic strength low pH, or presence of divalent cations Coiled, compact configuration Severe permeate flux decline Loose, spare, thin fouling layer Low ionic strength high pH, and absence of divalent cations

Stretched, linear configuration Small permeate flux decline

Figure 7.15 Effect of solution chemistry on the deposit of NOM on a membrane surface. Source: Reprinted from Hong and Elimelech 1997 [127].

“solution chemistry” refers not only to the feed characteristics but also to the conditions in the boundary layer. Such cake deposits will form where space is available – if pores are large enough then this process is accompanied by pore penetration, restriction, and plugging. When comparing the impact of MWCO, decreasing the NOM molecular weight (using, e.g. oxidation) could effectively reduce the extent of fouling for higher MWCO UF membranes [135]. For lower MWCO membranes surface deposition was the predominant fouling mechanism, where smaller molecules decrease the cake porosity and hence flux. In a study with dairy wastewater, Luo et al. [136] confirmed these findings and observed three stages of fouling: (i) stable flux, (ii) fouling by adsorption of small solutes, and (iii) formation of a cake layer in which molecules and aggregates interact. Park et al. [137] have evaluated the relationship between cake thickness and resistance for three model foulants (alginate, BSA, and HA) using optical coherence tomography (OCT) for brackish water. Th different organic matter types behaved differently in terms of cake fouling resistance and showed no correlation with cake thickness. 7.4.6

Pore Blocking/Plugging

Whether pore blocking occurs is determined mostly by the size of the organic molecules and the pore size of the membranes. Adsorption can play an important role in pore blocking, where pores are initially restricted due to adsorption of molecules, which penetrate into the pores. Th s is also referred to as pore narrowing [138]. Naturally, pore plugging may occur when the retention of solutes is incomplete [64]. Chang and Benjamin [78] stipulated that pore constriction by trapped molecules is the predominant mechanism with NF and tight ultrafiltration membranes, while surface gel is a more important process for looser

305

306

7 Fouling in Nanofiltration

membranes. Th s was confirmed by Cho et al. [139] who described a process of quick flux decline due to pore blockage followed by a gradual narrowing and closing of the remaining pores in UF. Nghiem et al. [140] found that pore blocking by humic substances enhanced the rejection of bisphenol A, which was stronger for the more open pore size NF membranes. Hong and Elimelech [127] observed strong adsorption and pore blocking at low pH for NOM. Jarusutthirak et al. [53] found that the colloidal EfOM fraction was primarily responsible for pore blocking and this mechanism dominated fouling. Pore blocking would be expected to occur for compounds that are small enough to penetrate into the membrane structure and yet large enough to experience hindrance within this structure [2]. Th s effect can be achieved due to the size of the molecule itself or due to solute–solute interactions. 7.4.7

Impact of Solute–Solute Interactions and Salts

Salts in feed solutions, and cations in particular, can have various effects on fouling. Firstly, such cations may cause intermolecular bridging between the organic foulants and the membranes [133]. Secondly, the cations may form complexes with the organics and at higher salt concentrations cause coagulation or precipitation and gel formation [106]. Such interactions are complex and are not yet well understood. Organics can also act as ligands for multivalent cations (for more details see Chapter 6 that deals with speciation), form complexes with specific interactions, and affect the retention and scaling of inorganics. Th s is shown with the example of humic acid and calcium (calcite scale) in Figure 7.16. Calcium and other multivalent cations are well known to increase organic fouling. Divalent ions (Ca2+ and Mg2+ ) may cause membrane fouling mainly by promoting the aggregation of NOM molecules in solution, not by forming “ionic bridges” between membrane surface and NOM [141]. While this was asserted by a modeling study, the actual mechanism may well depend on the actual conditions. Li and Elimelech [142] confirmed the previously suggested mechanism of intermolecular bridge formation of calcium between organic (a)

(b)

(c)

Figure 7.16 Interaction of calcium and humic substances in fouling (a) calcium carbonate (pH 10), (b) calcium carbonate and NOM (pH 10), (c) calcium and NOM (pH 8). Source: Adapted from Schäfer 2001 [4].

7.4 Organic Fouling

foulants and membrane functional group using AFM. Seidel and Elimelech [133] stated that NOM–calcium complexation and aggregation also causes fouling. In contrast, coagulation can modify the interaction between the positively charged calcium ions and negatively charged foulant HA and membrane surface, which affect the mechanism [143]. The interplay between aggregation and scale formation is complicated. Salts also influence solute–solute interactions and may enhance coagulation or aggregation of organics. Wall and Choppin [144] carried out a comprehensive study of humic acids coagulation due to Ca2+ and Mg2+ and established that the Derjaguin–Landau–Verwey–Overbeek (DLVO) theory applies to such organics; in fact even their size distribution changes with time following fractionation. Such a humic colloidal system destabilizes when double layer of individual colloids interact and hence precipitation or coagulation occurs. Th critical coagulation concentration (CCC) for HA was 0.1–1 M (NaCl), 1–5 mM (Ca2+ ), 10–50 mM (Mg2+ ), which is a realistic range for the boundary layer conditions of some membranes. At very high ionic strength (NaCl) the molecular colloids are restabilized and coagulation is prevented. Coagulation also decreased with pH in the range 4–8 rendering the CCC pH dependent (pH 2.5, 4, 7) resulting in CCCs of 1 mM, 100 mM, and 3 M, respectively. Mg2+ was significantly less efficient in coagulating HA than Ca2+ . Such studies on solute–solute interactions shed light on the possible mechanisms observed in NF. Hong and Elimelech [127] confirmed this for NOM fouling in the presence of calcium, where the interaction of those compounds caused the formation of small and coiled macromolecules that deposited at a higher rate. Lee et al. [106] observed severe fouling by a model for polysaccharide (alginate) in secondary wastewater with calcium. Th s fouling was attributed to a dense alginate gel layer caused by calcium–alginate complexation and cross-linking (bridging) of alginate macromolecules by calcium. In dye filtration, Koyuncu and Topacik [65] have found that an increase in ionic strength (NaCl) caused aggregation of the dye and consequently reduced flux decline. Similar results were reported by Schäfer et al. [145] who used ferric chloride (FeCl3 ) as direct pretreatment to NF, and the fouling of calcium and HA was reduced. Th s was explained with binding of HA to FeCl3 flocs as well as iron hydroxides and hence the formation of more porous deposits, unavailability of the HA to form a gel layer, and the removal of larger particulates due to shear forces. Silica colloids is a foulant in UF [146], NF [125, 147], and RO [148] and can have a synergistic effect with organic matter, which results in higher flux decline than what is predicted individually [125, 147, 148]. Contreras et al. [125] examined the mechanisms responsible for such synergy in combined fouling of NF270 with some model organic macromolecules (dextran, Suwannee River HA, alginate, and BSA) with silica. Three mechanisms were involved: (i) increased resistance of the mixed fouling layer, (ii) hindered back-diffusion, and (iii) organic foulant adsorption. The extent of contribution of each mechanism depended on the nature of the organic foulant.

307

308

7 Fouling in Nanofiltration

7.4.8

Impact of Fouling on Retention

Fouling can affect the retention of a membrane. For example, Nyström et al. [55] have demonstrated that the retention of humic acid decreased in the presence of FeCl3 . This was explained with the deposition of a gel layer on the membrane. Seidel and Elimelech [133] confirmed a decreased retention of total dissolved solids (TDSs) due to fouling, in particular at increased calcium concentrations. This was explained with a reduced Donnan charge exclusion. The Donnan effect was described in detail in Chapter 5. The apparent pore plugging of NF membranes by NOM at low pH was reported to cause a decrease in retention [127]. Schäfer et al. [145] established a strong impact of ferric chloride flocs on the retention behavior of NF membranes. Th variations were attributed to the charge of the deposits. Koyuncu and Topacik [65] studied the effect of reactive black dye (991 g/mol) on the retention of inorganic ions by NF. Th dye deposit was also regarded as a porous gel layer that increased the concentration polarization effect and acted like a dynamic membrane. Salt retention decreased with increasing dye concentration, reaching negative retention in some cases. Decrease in salt rejection in NF due to increasing fouling by NOM has been reported [149, 150]. This effect of gel layer on retention can potentially be related to the hindered back-diffusion of salts away from the membrane surface and thus to the enhanced concentration polarization model described in Section 7.3.7. The concept of the formation of an idealized fouling layer which increased retention of compounds smaller than the membrane pore and a cake layer that decreases retention is illustrated in Figure 7.17. Membrane fouling can modify membrane characteristics and subsequently vary retention behavior in both directions. Th s was evidenced for natural

(a)

(b)

Figure 7.17 Schematic of the formation of an idealized fouling layer which increased retention of compounds smaller than the membrane pore (a) and that decreases retention (b).

7.5 Scaling

100 80 Rejection (%)

Figure 7.18 Reduction of the effective pore diameter of membranes and retention of organic compounds due to a fouling layer (calcium and humic substances). Source: Schäfer 2001 [4]. Reproduced with permission of CRC Press.

60

DOC UV

40

Unfouled DOC

20

UV

0 0.1

1

10 Pore diameter (nm)

100

organic matter where fouling (calcium and humic substances) contributed to a reduction of the effective pore diameter of membranes and subsequent shift in retention of organic compounds due to a fouling layer (Figure 7.18). Fouling can enhance retention of microcystins and pharmaceutically active compounds [149, 150]. Jarusutthirak and Amy [103] observed an increase in retention in consequence to the adsorption of EfOM to membranes. Tang et al. [76] reported improved salt rejection by RO and NF membranes, as a result of humic acid fouling, which was attributed to Donnan exclusion by humic substances close to membrane surface. Greater rejection improvement was observed for membranes with rougher surfaces. Jin et al. [151] confirmed such observations. In addition to the increasing negative charge density at the membrane surface, the sealing of molecular-scale defects in the RO thin films by humic acid accumulation was considered as a possible explanation. Plakas et al. [152] observed in NF of humic substances and herbicides (atrazine, isoproturon) an initial decline of herbicides retention up to a certain fouling level, beyond which formation of a denser fouling layer resulted in increased herbicide retention.

7.5 Scaling 7.5.1

Introduction and Definition of Scaling

Scaling remains a limiting factor for NF operation, resulting from the increased concentration of one or more species beyond their solubility limit and their precipitation onto the membranes [105]. Thus, it is essential to operate NF systems at recoveries lower than a “critical value” in order to avoid scaling, unless the water chemistry is adjusted to prevent precipitation. To date, it is not possible to predict the limiting concentration level at which there is a risk of scale formation with a given membrane system and a specific antiscalant treatment with sufficient reliability [153].

309

310

7 Fouling in Nanofiltration

Scaling, also called scale formation or precipitation fouling, occurs in a membrane process whenever the ionic product of a sparingly soluble salt in the concentrate stream exceeds its equilibrium solubility product. The term “membrane scaling” is commonly used when the precipitate formed is a hard scale. Scaling usually refers to the formation of deposits of inverse-solubility salts (CaCO3 , CaSO4 ⋅xH2 O, calcium phosphate, etc.), although this term in general denotes hard, adherent deposits of inorganic constituents of water that formed in situ [154]. As with the other types of fouling, precipitation fouling reduces the quality and the flux of NF permeate and shortens the life of the membrane system. The problem is usually aggravated in attempts to increase the water (permeate) recovery; then the increasing retentate salt concentration leads to supersaturation, in particular very close to the membrane surface. Inorganic scale formation on the membrane may also lead to physical damage of the membranes due to the difficulty of scale removal and to irreversible membrane pore plugging. Inorganic foulants found in NF applications include carbonate, sulfate, and phosphate salts of divalent ions, metal hydroxides, sulfides, and silica. More specifically, the most common constituents of scale are CaCO3 , CaSO4 ⋅2H2 O, and silica, while other potential scaling species are BaSO4 , SrSO4 , Ca3 (PO4 )2 , and ferric and aluminum hydroxides [105, 155, 156]. Reliable prediction of the scaling propensity of water is essential in NF systems in order to maximize recovery and to determine the most efficient scale control method. The main parameters affecting scaling are salt concentration in the concentrate, temperature, fluid velocity, and pH, as well as the type and the material of the membrane. Th precipitation or crystallization of a salt onto a membrane surface involves nucleation and growth from a supersaturated solution. Supersaturation is the thermodynamic driving force for precipitation (or more specifically for the two basic stages of precipitation, nucleation and growth) and it is subsequently discussed in more detail. However, for sparingly soluble salts precipitation seems to be controlled by the kinetics of the process. It is widely accepted that precipitation kinetics is comprised of two main steps [157] either of which may control the process: (1) Nucleation stage: Nuclei (or tiny particles or embryos) are formed at specific sites in pores and at the surface of the membrane. Th s type of nucleation can be characterized as heterogeneous nucleation, as opposed to homogeneous nucleation, which occurs in the absence of a solid interface. A third form of nucleation is the secondary or surface nucleation, resulting from the presence of a crystallization phase in solution (e.g. introduction of seed crystals). Th critical value of supersaturation ratio for the different nucleation processes can be expressed as Sc,homog . > Sc,heterog . > Sc,surface > 1 [158]. In general, nucleation is the most poorly understood step. The rate of nucleation plays an important role in the final scale formation, and antiscalants are usually employed to suppress it. (2) Crystal growth: In the case of surface nucleation, the initial nuclei grow in time to form a thin, sometimes porous, layer. In a simplistic way, “growth units” or scale-forming ions diffuse to the crystal surface and attach themselves to that surface. Often, a delay or induction period exists before

7.5 Scaling

detectable deposits are formed. Th crystal growth process proceeds also in various steps, any of which may control the whole growth process. In the case where nucleation in the bulk dominates, crystal growth takes place in the bulk and the crystalline particles can be deposited onto the membrane surfaces. In an NF process, the highest risk of scaling exists in the concentrate stream at the last section of the membrane system. The withdrawal of the permeate results in an increase in the concentration level of all dissolved species in the concentrate stream and in the establishment of supersaturation of one or more sparingly soluble salts, which subsequently may precipitate. Therefore, it is necessary to estimate the saturation conditions of the concentrate stream throughout a membrane element. These calculations are based on the knowledge of the feedwater composition and of the concentration (or recovery) factor or ratio, CF. For each species i, the latter is defined in Eq. (7.25) CFi =

1  (1  Ri ) 1y

(7.25)

where y is the permeate recovery fraction and Ri the ion rejection factor. For most divalent species in NF systems Ri ranges between 0.9 and 1.0, but for monovalent species a significant fraction passes the membrane. Th concentration of most ions (Ca2+ , Sr2+ , SO4 2 , Cl , etc.) may be estimated as the CF times the feedwater concentration. Th s cannot be applied to all species present in water (e.g. HCO3  , CO2 ), while for SiO2 a correction for the pH change is required. In fact, Eq. (7.25) underestimates the concentrations at the membrane surface, since it does not account for the concentration polarization effect [156, 159]. As desalinated water permeates the membrane, the rejected ions promote development of a boundary layer near the membrane with ionic species concentration greater than that prevailing in the bulk, as illustrated schematically in Figure 7.19. Crystal Applied pressure

Formation of scaling layer

Πf

Water flow

– + – +

+ –

+

– + – +

– + – + – + –

Cation Anion

Concentration polarization layer

+

–

NF/RO Water flux

+ –

+ –

– +

Permeate

Πp

Figure 7.19 Schematic of concentration polarization layer at the membrane surface.

311

312

7 Fouling in Nanofiltration

The efore, at the membrane surface the supersaturation ratio, Sw , and the potential for scaling are greater than in the bulk. This effect is enhanced by increasing permeate flux and decreasing cross flow velocity. Th ratio of concentration at the membrane surface C w to that in the bulk of concentrate C b is called concentration polarization factor, PF. From the well-known film theory, considering total ionic species rejection by the membrane, one obtains Eq. (7.26) PF = Ciw Cib = exp (Jki )

(7.26)

where J is the mean permeate flux and k i the species average MTC at the membrane surface. Th latter can be determined by mass transfer correlations of the usual form, involving Sherwood, Schmidt, and Reynolds numbers. For spacer-filled channels, characteristic of flow passages in RO and NF spiral wound membrane modules, advanced fluid dynamics and mass transfer simulations have been developed and validated by experiments [160, 161]. These studies have led to generalized correlations for the MTC for various spacer types, thus allowing reliable computations of PF, C w , and wall supersaturation (Sw ); the latter is required for determining true scaling potential at the membrane surface. Because of the pivotal importance of the supersaturation concept, a more detailed description is presented below. 7.5.2

Solubility and Supersaturation of Salts

Th phase change associated with precipitation processes can be explained by thermodynamic principles. When a substance is transformed from one phase to another, the change in the Gibbs free energy of the transformation is given by Eq. (7.27)
G = (𝜇2  𝜇1 )

(7.27)

where 𝜇 1 and 𝜇 2 are the chemical potentials of phase 1 and phase 2, respectively. For
G < 0, the transition is spontaneous. The molar Gibbs free energy can be also expressed in terms of activity as Eq. (7.28)
G = RT ln(𝛼𝛼o )

(7.28)

where R is the gas constant, T is the absolute temperature, 𝛼 is the activity of the solute, and o is the activity of the solute in equilibrium with a macroscopic crystal. More specifically, for an ionic substance Mn Xm , which crystallizes according to the reaction in Eq. (7.29) na+ + mXb ↔ Mn Xm (solid)

(7.29)

the thermodynamic driving force for the crystallization either in the bulk or at the membrane surface is defined as the change in the Gibbs free energy of transfer from the supersaturated state to equilibrium in Eq. (7.30) [ a+ n b m ]1(n+m) [ ]1(n+m) (M ) (X ) IAP
G = RT ln = RT ln (7.30) Ksp Ksp

7.5 Scaling

In the above equation, K sp is the thermodynamic solubility product of the phase forming compound and (IAP) is the ion activity product. Quantities in parentheses denote activities of the corresponding ions and can be defined as the supersaturation ratio (S) of the crystalline precipitate as per Eq. (7.31). [ a+ n b m ]1(n+m) [ ]1(n+m) (M ) (X ) IAP S= = (7.31) Ksp Ksp

Concentration

Often in the literature S is written without the exponent. Th activity coefficients can be estimated using various equations applicable to low or high ionic strength. Th development of supersaturation is the driving force for both nucleation and crystal growth. Provided that there is sufficient contact time with a foreign substrate, scale formation may take place. Supersaturation in a membrane system is primarily caused by permeate withdrawal and concentration polarization and, to a lesser extent, by temperature and pH changes. Th solution speciation and the supersaturation ratios of various salts in water are readily computed by various computer codes taking into account all possible ion pairs and the most reliable values for the solubility products and the dissociation constants. In Figure 7.20 a typical solubility diagram for a sparingly soluble salt of inverse solubility (such as calcium carbonate, sulfate, and phosphate) is shown. The solid line corresponds to equilibrium. At point A, the solute is in equilibrium with the corresponding solid phase. Any deviation from this equilibrium position may occur with the increase of solute concentration (isothermally, line AB), with the increase of solution temperature resulting in solubility reduction (at constant solute concentration, line AC), or with varying concentration and temperature (line AD). A solution departing from equilibrium is bound to return to this state through precipitation of the excess solute. For most of the scale-forming sparingly soluble salts, supersaturated solutions may be stable for practically infinite time periods. Th se solutions are called metastable. The e is, however, a threshold in the extent of deviation from equilibrium marked by the dashed line in Figure 7.20, which if reached, wall crystallization (scaling) usually occurs first. Spontaneous bulk precipitation may also occur

B

C D

Labile (precipitation)

A (S = 1)

Metastable Stable region Temperature

Figure 7.20 Solubility–supersaturation diagram of a sparingly soluble salt of inverse solubility.

313

314

7 Fouling in Nanofiltration

with or without a preceding induction period. This range of supersaturations defines the labile region and the dashed line is known as the supersolubility curve. It should be noted that the supersolubility curve is not well defined and depends on several factors such as concentration level of the scale-forming ions, presence of other ions and ionic strength, presence of suspended matter, wall material and roughness, temperature, and pH. Th formation and subsequent deposition of solids occurs only if the solution conditions correspond to the metastable or to the labile region. Below the solubility curve scaling cannot take place. Most membrane suppliers and literature sources set S > 1 as criterion for the onset of scaling. Often this criterion is modified, a little below or a little above unity. Th argument that for S > 1 precipitation is expected is true for the readily soluble salts, where even a small deviation from equilibrium can induce crystallization. However, for most of the sparingly soluble salts responsible for the scaling problems in the NF/RO systems, a significantly higher value of supersaturation ratio (“critical” supersaturation ratio) in the bulk must be exceeded to result in detectable scaling. Th s effect is observed in membrane as well as in non-membrane scaling systems. It has been reported that an RO system exceeded 14 times the BaSO4 equilibrium conditions without scaling problems [162]. A major concern about this “critical” supersaturation ratio is that it is not the same for all scale-forming compounds and that it increases as the solubility of the salts decreases. Consequently, one may expect that this value is higher for the CaCO3 and BaSO4 scaling systems than for the CaSO4 system. In recent years, significant progress has been made toward development of online scale-monitoring techniques (e.g. Refs. [163–166]), which are promising for future applications. Additionally, considering that the presently employed membrane-scale detection techniques lack sensitivity (as discussed in [167]), detailed experimental studies have focused on the early stages of membrane scaling (e.g. Refs. [167, 168]), aiming to improve understanding and develop predictive methods that would allow taking corrective measures in RO and NF plants before the scaling problem is aggravated. Th se studies suggest that in real NF/RO systems (where minute quantities of colloidal and other particles are usually present in feed-waters) a true induction period of scale formation is likely insignificant, and that the rate of scale-particle growth strongly depends on supersaturation ratio S (as one would expect).

7.5.3

Common Scalants

Common types of scalants found in membrane processes are summarized below. It should be noted that deposits forming in membrane modules, as well as in other scaling systems, are rarely homogenous, and in most cases, as seen also in membrane autopsy studies, they consist of a mixture of various sparingly soluble salts and other foulants (e.g. organics, colloids, biofoulants). In brackish and hard waters, CaCO3 and gypsum are the most common scalants for which pretreatment is required.

7.5 Scaling

7.5.3.1

Calcium Sulfate (CaSO4 ) Scale

Th most common form of calcium sulfate scales and the polymorph that precipitates at room temperature is gypsum (calcium sulfate dihydrate, CaSO4 ⋅2H2 O). Gypsum is approximately 50 times more soluble than CaCO3 at 30  C. The effect of temperature (in the range 10–40  C) and of pH on gypsum solubility is rather marginal. One source of sulfate ions in some treated waters is the addition of sulfuric acid to the feedwater in order to control CaCO3 precipitation. Th s method of scale control can lead to calcium (or barium and strontium) sulfate deposition, if excessive amounts of sulfuric acid are used for pH control. For this reason, calculations for assessing the potential for sulfate scaling must be carried out using the analysis of feedwater after acid addition or other pretreatment methods. 7.5.3.2

Calcium Carbonate (CaCO3 ) Scale

Almost all naturally occurring waters contain bicarbonate alkalinity and are rich in calcium, making them prone to scaling problems. The potential for CaCO3 scaling exists for almost all well, surface, and brackish waters. Calcium carbonate forms a dense, extremely adherent deposit and its precipitation in an NF plant must be avoided. It is by far the most common scale problem in several scaling systems, including cooling water systems and oil or gas production systems. Calcium carbonate can exist in three different polymorphs, namely calcite, aragonite, and vaterite, in order of increasing solubility. All three polymorphs have been identified in scales, although vaterite is rather rare. The modynamics predicts that calcite, the least soluble and more stable polymorph, should be the phase favored in the precipitation process. Aragonite is also encountered in certain systems. It has been shown that formation of a particular polymorph depends upon water temperature and chemistry (e.g. pH, ionic strength, presence of other ions/impurities/inhibitors). It is also well known that the presence of magnesium ions, in solutions supersaturated with respect to CaCO3 , favors the precipitation of aragonite and appears to hinder the formation of vaterite. Th tendency to form calcium carbonate can be predicted qualitatively by a plethora of indices derived theoretically or empirically over the past 80 years. The most common indices are the Langelier Saturation Index, the Ryznar Stability Index, and the Stiff and Davis Stability Index, which were originally developed for applications in cooling water systems and in the oil industry. 7.5.3.3

Barium Sulfate (BaSO4 ) and Strontium Sulfate (SrSO4 ) Scale

Th solubility of BaSO4 is much smaller than that of gypsum (K sp = 1.05 ⋅ 1010 mol2 /l2 ) at 25  C [169] and can cause a potential scaling problem in the back-end of the NF/RO systems. Its solubility decreases with decreasing temperature. BaSO4 scale can only be dissolved by crown ethers and concentrated sulfuric acid. Barium ions are seldom reported in analyses of natural waters, and if found, their concentration does not exceed 200 g/L. BaSO4 scale formation is very rare in membrane scaling systems. Out of 150 elements on which autopsy was performed, no instances of barium sulfate were found [170]. Th presence of strontium in many natural waters is more common than that of barium ions. As little as 10–15 mg/l of strontium ions in the concentrate

315

316

7 Fouling in Nanofiltration

may induce SrSO4 scale formation. Barium and strontium sulfates are more commonly encountered in surface waters. 7.5.3.4

Silica Scale

Amorphous silica is one of the major fouling problems in NF/RO systems and in most processes involving water [171]. The silica content in most natural waters can reach 100 mg/l, since silica is one of the primary components of the earth crust. Much has been written about the solubility of amorphous silica in water. Its solubility at room conditions is 100–150 mg/l in the pH range 5–8 and increases significantly with pH at values higher than 9.5. Furthermore, silica solubility increases significantly with temperature. Thus, in usual water treatment operations silica concentration is limited to approximately 120–150 mg/l, whereas the excess precipitates as amorphous silica and silicates. In membrane systems silica scaling has serious consequences; i.e. the cleaning of fouled membranes is costly and not without problems. Th solubility of silica minerals generally decreases with increasing ionic strength, in contrast to the solubility of CaCO3 and sulfate salts. It has been shown [172] that at 25  C and pH 5.0–7.5, the solubility of amorphous silica decreases with the addition of several salts due to the “salting-out” effect of inorganic electrolytes on aqueous silica. This effect is essentially cation dependent and disappears (or better it is reversed) at higher temperatures. Silica scale was found in 66% of about 100 membrane elements treated recently for autopsy [170]. Iron and aluminum were present in 88% and 75%, respectively, of the membranes scaled with silica. 7.5.3.5

Calcium Phosphate Scale

Calcium phosphate scale has become more common in membrane systems as autopsies on membrane elements have shown [170, 173, 174]. This can be attributed to the tendency to treat wastewaters, which are rich in phosphates, and to the use of phosphorus containing antiscalants, injected in the form of phosphonates and other organic phosphorus compounds. Th concentrate may become supersaturated with respect to at least four calcium phosphate phases (as in calcium phosphate scale formation in other systems), although no single phase has been identified in the autopsy studies. It is often assumed that these phases are amorphous calcium phosphate (ACP, stoichiometry corresponding to Ca3 (PO4 )2 ⋅xH2 O), dicalcium phosphate dihydrate (DCPD, CaHPO4 ⋅2H2 O), octacalcium phosphate (OCP), (Ca8 H2 (PO4 )6 ⋅5H2 O), and hydroxyapatite (HAP, Ca5 (PO4 )3 OH), the least soluble phase. It is generally agreed that the formation of HAP from a highly supersaturated solution at neutral pH is usually preceded by ACP or other precursor phases, while the presence of ions may affect the polymorph precipitated. Owing to the presence of other ions in the feedwater, defect apatite can be also formed. Th solubility of calcium phosphates strongly depends on solution pH and, consequently, acid addition alleviates the calcium phosphate scaling problem. Other parameters affecting the scaling tendency of calcium phosphates include the supersaturation ratio, temperature, and ionic strength.

7.5 Scaling

7.5.4

Characterization of Scales

Th techniques for analysis of the crystalline deposits (as in the case with other types of deposits) are not simple and not standardized. Unfortunately, the interior of membrane elements in a real plant is not accessible for examination by naked eye or even by an optical microprobe. Direct scale characterization can only be accomplished by membrane destruction. On the other hand, several small-scale laboratory setups have been used, capable of monitoring the scale formation process by taking and analyzing images of the membrane surface (e.g. Refs. [167, 175, 176]). Deposit characterization is an important step in the autopsy study of a degraded membrane module. Th membrane scales can be characterized by a variety of techniques, the most common of which are briefly mentioned here. Visual and microscopic inspection (SEM) of the scaled surface may comprise the first step of characterization. Energy-dispersive spectroscopy (EDS) is usually employed in conjunction with the SEM system to determine elemental composition, while nuclear magnetic resonance spectroscopy may determine the chemical structure of the scales. The spectroscopic techniques of FTIR, FT-Raman, and XRD can be used to yield quantitative and qualitative results of the scale composition and the dominant crystalline phases. Finally, X-ray photoelectron spectroscopy (XPS) analysis can be used in order to determine the properties of the surface layers of scales. An experimental membrane system can also be used to determine how the scales form and the effectiveness of the various antiscalants; several such setups have been employed in recent years (e.g. Refs. [175, 177–179]). 7.5.5

Mechanisms of Scale Formation

Th great complexity of the scale formation process is a direct consequence of the large number of species usually present in a real system and of the plethora of possible physical mechanisms. Th latter may include mass, momentum, and heat transfer, as well as chemical reactions at the equipment surfaces. Furthermore, the diversity of fluid composition of the various waters treated in membrane systems and the variation of processes taking place along the flow path make difficult the generalization of both the mechanisms responsible for scale formation and the preventive measures. The e are two main mechanisms to explain flux decline in a membrane system due to the formation of crystalline matter: filter cake formation and surface blockage (e.g. Refs. [178, 180]). Th former involves crystalline particles formed in the bulk of the solution that are deposited onto the membrane to create usually a porous, not very coherent, soft layer. According to the cake formation model, the deposit layer has a constant porosity, its thickness increases with time, and flux declines due to growth of the layer. In the mechanism of surface blockage, isolated “islands” of crystals or deposits are initially formed on the exposed membrane surface, which further grow with time, laterally and normally to the surface, to form a continuous and coherent layer. Consequently, the flux would steadily decline as the sections covered by these “islands” would be inaccessible for water permeation. Th two

317

318

7 Fouling in Nanofiltration

main mechanisms of flux decline stem from the different forms of nucleation occurring in the membrane system. As discussed in Section 7.5.2, wall or surface nucleation takes place (for most precipitating species) at lower supersaturation ratios than those needed for nucleation in the bulk (homogeneous, secondary). Consequently, at relatively high supersaturation with respect to a certain salt, bulk nucleation would dominate, resulting in cake formation. On the other hand, at lower supersaturation ratios, membrane fouling would proceed via the growth of crystalline islands. In both mechanisms, an induction period may precede the scale formation process. Th rate of scale formation is determined by several factors, such as the level of supersaturation, the water temperature, the flow conditions, and the surface roughness and material of the substrate. A critical factor in the whole process is the adherence of the deposits to the surface. If adherence is poor, the deposits might be removed by the fluid flow. If adherence is strong, the initial crystals grow laterally and perpendicularly to form a coherent scale layer [181]. Sometimes a third step, the stage of recrystallization or aging, is recognized in the scaling process. Gilron and Hasson [180] and Brusilovsky et al. [177], investigating the CaSO4 scaling system, demonstrated that the flux decline in an RO unit was due to blockage of the membrane surface by lateral growth of the deposit (surface or heterogeneous crystallization). Th CaSO4 scale crystals rested at the edges against the membrane, and they were tightly packed with a tendency to grow outwards (“radiate”) from various growth sites. Th morphology of these scales strongly supports the assumption of a surface crystallization mechanism. In addition, Hasson and coworkers developed a flux decline model based on the surface blockage and involving the lateral spread of a single crystal layer. Lee and Lee [178] examined the effect of operating conditions on scale formation in an NF unit. Th se investigators found that both mechanisms are operative and that the operating conditions (i.e. pressure and cross flow rate) play an important role. Surface crystallization is favored at a low cross flow velocity and a high operating pressure. Le Gouellec and Elimelech [179, 182] investigated the presence of several species and of antiscalants to combat gypsum scale formation in a small recirculating unit. No definite conclusions could be drawn on the scaling mechanisms, but the presence of bicarbonate, magnesium ions, and humic acid showed a tendency to retard the formation of gypsum nuclei. Moreover, a model was developed for predicting the required antiscalant dosage to control gypsum scale in NF systems. A similar mechanism as that described for the CaSO4 by Hasson and coworkers has also been found for the CaCO3 system in a once-through laboratory NF/RO unit [183, 184]. Figure 7.21a presents a typical flux decline curve due to calcium carbonate scale formation on an NF membrane. In the same figure, SEM images of scaled membranes are presented, at different run times, depicting the growth of the scale layer with time. In this particular run (and almost in all runs) the permeate flux was rather constant for an initial period of four to eight hours before declining mainly due to scaling. In the rather limited number of these tests (of maximum duration 15 hours), significant membrane scaling was observed to occur at Sb > 3.3 (based on bulk conditions with reference to calcite) or at LSI > 0.9. Comparing these results with scaling experiments in tubes [185]

7.6 Colloidal and Particulate Fouling

it may be observed that scaling on membranes occurs at lower supersaturation, obviously due to the concentration polarization effect. SEM micrographs at various run times reveal that even when the CaCO3 crystals apparently cover about 40% of the membrane (Figure 7.21c), no detectable flux decline is recorded. This observation somehow contradicts the notion that surface blockage is eventually the main mechanism of membrane flux decline. Recent detailed studies on incipient CaCO3 and CaSO4 membrane scaling in spacer-filled channels [167, 168] tend to confirm the above observations, showing measurable crystalline deposits even at very small bulk supersaturations as well as insignificant induction time.

7.6 Colloidal and Particulate Fouling 7.6.1

Introduction and Definition of Colloidal and Particulate Fouling

Th term, colloidal and particulate fouling, refers to loss of both flux and salt retention due to accumulation of retained colloidal and particulate matter on the membrane surface. Colloids are defined as fine suspended particles in the size range of a few nanometers up to a few micrometers [186], and by nature can be referred to in the lower range as nanoparticles. Colloidal matter is ubiquitous in natural waters, as well as many industrial, process, and wastewaters [187]. Examples of common colloidal sized foulants include inorganic (clays, silica, salt precipitates, and metal oxides), organic (aggregated natural and synthetic organics), and biological (bacteria and other microorganisms, viruses, LPSs, and proteins) matter. A review of colloidal and particulate fouling studies reveals that the foulants of greatest concern for NF separations are colloidal sized substances consisting of silica, organics, metal oxides (specifically iron and manganese), and microorganisms [127, 188–196]. Champlin [119] recommended removing particles down to 1 m in size, although this may not be sufficient to avoid fouling. Conventional processes used to pretreat NF feedwaters fail to remove sub-micron colloids, and even MF/UF processes sometimes fail to remove all colloids below a few hundred nanometers in diameter. Further, the elevated concentration of rejected ionic constituents in the vicinity of the membrane screens electrostatic interactions, which may encourage aggregation of dissolved (organic) matter into colloidal sized particles. Th importance of particle–membrane and particle–particle interactions during colloidal fouling is realized when considering the influence of salt retention and concentration polarization on the solution chemistry in the vicinity of the membrane surface. Electrokinetic properties of colloids and membranes are strongly dependent on pH, ionic strength, and the presence of multivalent ions [81]. Therefore, distinguishing the fundamental physicochemical properties of colloids and membranes is critical to understanding colloidal fouling. Th summary that follows provides brief descriptions of key colloid and membrane properties, transport and deposition, formation of colloid deposit layers, and mechanisms of colloidal fouling in NF.

319

7 Fouling in Nanofiltration

1.0

J/Jo (–)

320

0.8

Pict. b

0.6 P = 10.2 atm pH = 8.1 LSI = 0.9

0.4 0 (a)

5 μm

Pict. c

2

4

Pict. d 6

8 10 Time (h)

12

14

16

5 μm

1h (b) 20 μm

10 μm

4h (c) 20 μm

20 μm

15 h (d)

Figure 7.21 Flux decline curve (a) and SEM images at various times (1, 4, and 15 hours) of scaled TFC-S membranes at pH = 8.1. Source: Tzotzi et al. 2007 [184]. Reproduced with permission of Elsevier.

7.6 Colloidal and Particulate Fouling

7.6.2

Colloid Properties

Colloidal matter is typically charged in aqueous electrolyte solutions. The surface charge on colloids arises from a variety of mechanisms including differential ion solubility (e.g. silver salts), direct ionization of surface groups (typically, –COOH, –NH3 , or –SO3 H), isomorphous substitution of surface ions from solution (e.g. clays, minerals, oxides), anisotropic crystal lattice structures (especially in clays), and specific ion adsorption [81, 186, 187, 197]. Th surface charges contribute to electric (please update throughout and in the abbreviations) double layer (EDL) interactions, which typically determine colloid aggregation and deposition phenomena [81, 198, 199]. The specific property of colloids used to quantify the relative magnitude of EDL interactions is the surface (zeta) potential, 𝜁 , which is commonly determined by measuring the electrophoretic mobility of colloids in a suspension and computing 𝜁 from an appropriate theory [200]. It is well known that solution pH and ionic strength directly influence the zeta potential, and thus greatly influence colloidal interactions. It has been shown that the surface charge properties of colloids can dramatically influence colloid-cake layer structure (porosity) and hydraulic resistance [201–207]. Colloid size and shape also contribute to the hydraulic resistance to permeation they impose when accumulated in a cake layer [204, 205, 208]. While colloids are often modeled as spherical, they may be spheroidal (microbes), crystalline (metal salt precipitates), plate-like (clays), or macromolecular (organic aggregates, proteins). In many natural waters, the range of polydispersity in colloid and particle size, shape, and electrokinetic character make it quite difficult to accurately describe with any tractable modeling approach. Therefore, an average, “spherical” hydrodynamic diameter is determined from dynamic light scattering or potentiometric methods and used in conjunction with a measured average particle zeta potential to predict the influence of colloidal interactions. A unique size fractionation water quality analysis for real agricultural drainage water sampled from the Alamo River in Imperial Valley, California, is provided in Table 7.3. Agricultural drainage water is a valuable alternative water source being considered for reuse in many parts of the world; however, desalination is typically required since it maybe brackish [209–211]. Depending on the intended reuse application RO or NF is considered for desalination. Th data shown in Table 7.3 was obtained following standard methods for all analyses; raw water was allowed to settle for 24 hours in a cold room (5  C) and then sequentially filtered under vacuum through 8.0, 1.0, 0.4, and 0.1 m polycarbonate, track-etched membranes. Th size was determined by dynamic light scattering and confirmed with a Coulter Counter (only for raw, 8 and 1 m fractions). Measured electrophoretic mobilities were converted to zeta potential via the Smoluchoski equation [81]. The results shown are unpublished and are intended only to qualitatively illustrate the physical and chemical properties of natural colloidal matter. In addition, an analysis such as this could be used to justify the selection of a pretreatment process to remove colloidal foulants prior to desalination.

321

322

7 Fouling in Nanofiltration

Table 7.3 Size fractionation water quality determination for agricultural drainage water.

Filter size

Raw

8.0 m

1.0 m

0.4 m

0.1 m

pH (–)

EC (mS/cm)

TDS (mg/l)

Turbidity (NTU)

TOC (mg/l)

Solids (mg/l)

Size (𝛍m)

8.17

3.322

2256

108.0

21.0

2750

9.89 11.6

8.48

3.232

2198

7.540

18.4

2326

8.51

3.303

2254

1.380

13.4

2268

8.51

3.158

2148

0.321

12.3

2151

0.29 11.5

8.59

3.140

2106

0.175

2107

0.00

9.11

ZP (mV)

2.01 14.4

1.74 14.5

7.47

DOC (after 0.22|m filter) = 11.04 mg/l Bacteria count = 160–420; enterococcus = 0–30 cfu/ml. Column headings are EC = electrical conductivity, TDS = total dissolved solids, Turb. = turbidity, TOC = total organic carbon, solids = total solids by gravimetric analysis, size = hydrodynamic diameter, and ZP = zeta potential. Source: Data provided by E.M.V. Hoek (unpublished).

7.6.3

NF Membrane Properties

Physical and chemical properties of NF membranes (i.e. permeability, salt retention, “pore” size, etc.) also contribute to the rate and extent of colloidal fouling [72, 83–85]. The high hydraulic resistance of NF membranes enables substantial colloid cake layers to form before fouling is detected, and retention of ionic solutes exacerbates colloidal fouling by screening electrostatic interactions. It has been demonstrated that NF membrane surface properties (i.e. zeta potential, roughness, hydrophobicity) are strongly correlated to the initial rate of colloidal fouling [51, 72, 212–219]. Figure 7.22 illustrates the range of surface morphologies that may exist for commercially available thin-film composite NF membranes. NF membrane properties are discussed in Chapter 2, while their individual contributions to NF colloidal fouling are described below in Section 7.6.3.

Figure 7.22 Field-emission scanning electron microscopy (FESEM) images of two commercially available NF membranes with RMS roughness values of (HL) 12.8 nm and (NF70) 56.5 nm. Additional (HL) and (NF70) membrane properties include zeta potentials of 18 and 25 mV (at 10 mM and pH 7), contact angles of 51.9 and 51.7 , hydraulic resistances of 3.26 ⋅ 1010 and 3.13 ⋅ 1010 Pa s/m, and salt retentions of 35 and 83% (at 50 l/m h and 10 mM NaCl), respectively. Source: Hoek 2002 [83].

7.6 Colloidal and Particulate Fouling

Table 7.4 Surface properties of typical polyamide thin-film composite membranes. Membrane (name)

Ra (nm)

Rq (nm)

Rm (nm)

SAD (%)

NF270

5.2

6.0

63

0.3

SG

9.1

13.1

161

2.3

HL

10.5

15.9

200

ESPA

31.8

40.0

469

5.3 58

AK

33.3

42.2

403

40

CPA3

33.5

45.8

482

31

LFC1

34.7

44.9

368

27

NF90

37.9

48.7

415

17

XLE

43.4

56.7

560

30

ZPa) (mv)

29 21 10 34 23 22 7 32 25

𝜽w ( ∘ )

69 63 40 41 66 69 60 38 58

Ra = average roughness, Rq = RMS roughness, Rm = max roughness, SAD = surface area difference, ZP = surface (zeta) potential, and 𝜃 w = pure water contact angle. a) At pH 7, 10 mM NaCl.

Analyses of numerous additional polyamide thin-film composite membranes reveal a consistent set of physical and chemical properties. Table 7.4 presents atomic force microscope roughness analyses of nine different membranes, along with experimentally determined surface (zeta) potentials and pure water contact angles. The zeta potential was determined from a streaming potential analyzer (EKA, Brookhaven Instruments) following methods described elsewhere [220]. Th data indicate a range of surface roughness that varies (on average) between a few nanometers and 50 nm and with some features on the order of half a micrometer. Surface area difference (SAD) is an indication of the increase in surface area (over a flat plane of equal projected area) due to the roughness of the surface. SAD is a standard AFM roughness analysis statistic and is also known as Wenzel’s roughness ratio [221]. Membrane surface (zeta) potentials range between 20 and  35 mV at neutral pH in a 10 mM NaCl electrolyte. One membrane has a significantly lower zeta potential (LFC1), ostensibly to lower the fouling potential of the membrane as it is often referred to as a “low fouling composite” by the manufacturer (Hydranautics, San Diego, CA). Th nine membranes samples exhibit a range of “wettabilities” as depicted by the pure water contact angles. Th se contact angles were determined by the sessile drop technique at room temperature with low relative humidity. 7.6.4

Colloid Transport and Deposition

Th key to understanding colloidal fouling is to understand the fundamental transport processes by which particles are brought to the surface of the membrane, how they deposit or attach, and why they accumulate in the form of a cake. Particle transport and deposition in fluids can be described by the convective diffusion equation [222], which in its general form is given by Eq. (7.32): 𝜕c +⋅J=Q (7.32) 𝜕t

323

324

7 Fouling in Nanofiltration

where c is the particle concentration, t is the time, J is the particle flux vector, and Q is a source or sink term. The particle flux vector J is given by Eq. (7.33) J = D ⋅ c + uc +

D⋅F c kT

(7.33)

where D is the particle diffusion tensor, u is the particle velocity induced by the fluid flow, k is Boltzmann’s constant, T is absolute temperature, and F is the external force vector. Th terms on the right-hand side of Eq. (7.30) describe the transport of particles induced by diffusion, convection, and external forces, respectively. In colloidal fouling of membranes, the relevant external forces are colloidal and gravitational, as described in Eq. (7.34) F = FG + Fcol

(7.34)

where FG is the gravitational force and Fcol represents the colloidal forces acting between the suspended particles and the collector surface. Th gravitational force is usually negligible for the sub-micron colloidal systems encountered in NF operations. The colloidal force can be derived from the gradient of the total interaction potential, 𝜙T , as in Eq. (7.35) Fcol = 𝜙T

(7.35)

Within the framework of the traditional DLVO theory [198, 199], 𝜙T is the sum of van der Waals and electrical double layer (EDL) interactions. A theoretical study by Song and Elimelech [223] utilized the general form of the confection–diffusion equation to investigate colloidal deposition onto permeable (membrane) surfaces. Numerical simulations demonstrated that the initial rate of particle deposition was mainly controlled by the interplay between permeation drag and EDL repulsion. Subsequent experimental studies have confirmed the role of permeation drag and EDL repulsion on the initial rate of colloid deposition [204, 205]. While the above formulation is fundamentally correct, direct solution of such equations is impractical for all but academic purposes. Cohen and Probstein [224] provided a facile, but approximate, approach toward quantifying the impact of the physicochemical properties of stable colloids (described via the DLVO approach) on colloid cake layer formation. In this classic paper, the authors studied the rate of flux decline of RO membranes due to iron oxide nanoparticles. Th y assumed that at least a monolayer of coverage by the positively charged colloidal foulants would coat the negatively charged membrane surface, but the similar charge of subsequently deposited foulants and the foulant-coated membrane surface would result in repulsive (electrostatic) interactions. Th authors provided “order-of-magnitude” approximations for the particle fluxes resulting from permeate convection, Brownian diffusion, lateral (inertial) lift, shear induced diffusivity, and repulsive interfacial forces. Th conclusion was that the net deposition rate must be determined by a balance between permeate convection and the interfacial flux due to repulsive electrostatic interactions, all other diffusive or convective fluxes being negligible.

7.6 Colloidal and Particulate Fouling

Goren [225] performed a detailed theoretical analysis of hydrodynamic interactions between colloidal particles and membrane surfaces occurring as particles are convected toward the membrane under the force of permeation drag. The net effect of these hydrodynamic interactions is to increase the effective drag force on a particle as it approaches the membrane surface over that predicted by the Stokes equation. Goren’s analysis yields a correction factor that increases dramatically as a particle approaches a membrane surface and is a complex function of the particle size, membrane resistance, and separations distance. So, in addition to bulk convective and diffusive interactions and interfacial physicochemical interactions, it is (theoretically) important to consider the impact of interfacial micro-hydrodynamic interactions on colloidal fouling. Following the approach of Cohen and Probstein [224] described above, an order-of-magnitude analysis can be performed employing the following representative operating conditions and membrane surface properties and water quality data from Table 7.4: flux of 11 gfd (5  106 m/s), cross flow velocity of 0.5 m/s (Re = 1000), membrane resistance of 3  1013 m1 , membrane surface zeta potential of 20 mV, foulant zeta potentials and sizes from Table 7.3. Figure 7.23 plots the theoretical foulant fluxes due to permeate convection (bulk value – dashed line with open blue diamonds, and Goren corrected value – solid line with solid blue diamonds) and the various back-transport mechanisms of shear induced diffusion, lateral (inertial) lift, Brownian diffusion, and interfacial (DLVO) forces. The conclusion is that without accounting for the correction to permeation drag, the back-transport of colloidal sized foulants by both shear induced diffusion and interfacial forces is estimated to be orders of magnitude larger than the flux due to permeation, and no foulant deposition 0

log (J) (m/s)

–2

Perm-Goren Shear Brownian Interfacial Inertial Lift Perm-bulk

–4 –6 –8 –10 –7

–6

–5

–4

–3

log (ap) (m)

Figure 7.23 Plot of various colloidal foulant fluxes assuming representative NF membrane properties, operating conditions, and foulant properties (size and zeta potential take from Table 7.3). The data labels “Perm-Bulk” and “Perm-Goren” indicate fluxes of foulant particles toward the membrane surface due to permeate convection – using Stokes’ drag (bulk) and Goren’s drag, respectively. Back transport mechanisms of Brownian diffusion (“Brownian”), shear induced diffusion (“Shear”), lateral inertial lift (“Inertial Lift”), and DLVO (“Interfacial”) are plotted for comparison.

325

326

7 Fouling in Nanofiltration

would be anticipated. By applying the hydrodynamic correction provided by Goren the permeate convection flux is increased by several orders of magnitude and prevails over the back-transport fluxes. Analyses of this nature are at best order-of-magnitude approximations and have not been systematically tested at any scale, but by considering all known transport mechanisms a better understanding of colloidal fouling may be accessible. Th extent to which membrane surface properties influence long-term fouling (i.e. through cake formation) is unknown because it is unclear how membrane properties might affect subsequent particle deposition once the membrane is covered with a thin layer of particles. Wiesner et al. [226] provided one of the earliest known studies on membrane filtration of coagulated colloidal suspensions. Th effect of colloid stability on cake layer structure (porosity) and permeate flux decline for both stable and unstable colloids were compared experimentally and theoretically. Subsequent studies, both theoretical and experimental, have confirmed the importance of colloid stability, colloid and membrane surface properties, and colloidal hydrodynamics on cake formation and permeate flux decline in various membrane filtration processes [77, 203, 205, 206, 227, 228]. Although DLVO interactions enable a large amount of experimental aggregation and deposition data to be explained, additional short-range colloidal interactions must also be considered. Such non-DLVO forces include repulsive hydration interactions (due to oriented water molecules adsorbed at each interface), attractive hydrophobic interactions (because of the relatively strong affinity of water to itself compared to that between water and most solid matter), and repulsive steric interactions (from deformation or penetration of adsorbed polymers) [229]. The existence of these non-DLVO forces has long been recognized, but it 5.00

2.50

0

2.50

5.00

2.50

0 5.00 μm

Figure 7.24 Atomic force microscope images of NF membranes challenged with a 0.0002% (v/v) suspension of 100 nm spherical silica colloids at 10 mM NaCl, 1 ⋅ 105 m/s flux, 19.2 cm/s cross flow velocity, 25  C, and pH 7. The filtration experiment lasted only 30 seconds. After filtration, the membranes were removed, rinsed in a particle free electrolyte, and allowed to dry before Tapping ModeTM AFM imaging in air with a silicone nitride cantilever tip [83]. The circled area indicates a large cluster of colloids deposited in the valley of the rough NF membrane. Source: Hoek 2002 [83].

7.7 Biofouling

has only recently been demonstrated experimentally that non-DLVO interactions contribute significantly to colloidal fouling of polymeric membranes [218]. Other experimental studies suggest that surface roughness [203, 206, 219, 230] influences the initial rate of colloid deposition onto membrane surfaces. Figure 7.24 shows AFM images of the NF membranes from Figure 7.22 after filtering a colloidal suspension under the same physicochemical conditions. Th membrane on the right had significantly more particles deposit on the surface and the colloids appear to have deposited preferentially in the valleys of the rough surface. Additional, model calculations supported the preferential deposition of colloids in the valleys of rough membranes [83].

7.7 Biofouling 7.7.1

Introduction and Definition of Biofouling

Biofouling is caused by bacteria and, to a lesser degree, fungi [231] by two distinctively different mechanisms. Cells from these organisms that get physically deposited in the fouling layer contribute to particulate fouling in a manner analogous to that of deposition of inanimate organic or inorganic particles. However, it is the ability to actively influence the constitution of the fouling layer by metabolic activity that distinguishes biological fouling by microbial cells from all other fouling mechanisms [232]. Th least invasive scenario is one where the cells metabolize at low intensity at maintenance metabolic rates and consume mainly the internal metabolic reserves, producing little interaction with the immediate environment of the cell [233, 234]. Although largely restricted to the cell, maintenance metabolism may profoundly alter cell surface composition [235] and thus the interactions of the cell with other foulants in the layer. Cells that metabolize external nutrients have a more pronounced impact on the constitution of fouling layers. Most nutrients that support heterotrophic microbial metabolism and growth in the environment are derived from polymers such as lipids, proteins, polysaccharides, or nucleic acids (DNA or RNA), but heterotrophic bacteria can only take up monomers or small oligomers as carbon sources. Polymers have to be broken down extracellularly into monomers or small oligomers in order to be transported into the microbial cell, which is accomplished by excreted extracellular hydrolytic enzymes [236]. The enzymatic breakdown of foulant layer macromolecules will also affect the viscosity, porosity, and general physical organization of this complex maze of particles and macromolecules [237]. Th diffusion of hydrolytic enzymes into the foulant layer extends the zone of impact of a cell’s metabolism considerably beyond its immediate surroundings. Microbial metabolism in the fouling layer may convert non-fouling substances such as glucose into strong foulants such as glycosphingolipids [238]. The most dramatic microbial impact, however, occurs when the cells grow inside the fouling layer, forming microbial biofilms [232]. Cell growth may alter profoundly both the three-dimensional geometric arrangement of foulants inside the layer and its composition. Cell division and volume increase cause displacement of

327

328

7 Fouling in Nanofiltration

foulants in the vicinity of the organisms, while the consumption of organic or inorganic foulants and the production of metabolites produced by the cells change the composition of the foulants. Biofilms may harbor crystallization nuclei and thus accelerate inorganic fouling as well [239]. Filamentous fungi belonging to the genera Penicillium and Aspergillus often degrade glue lines that separate feed from permeate channels and thus disrupt the integrity of membrane modules. Biofilm bacteria are capable of biodegrading membrane polymers such as cellulose acetate [240]. 7.7.2 7.7.2.1

Biofilm Formation in NF Plants Accumulation of Cells at Interfaces

Biofilm formation can be divided into two phases, the first one encompassing the approach of the cells to the substratum surface, ending with their irreversible adhesion, and the second one comprising the growth and proliferation of adhered cells. Membrane plants offer a great diversity of impermeable and permeable surfaces for microbial colonization, and the mechanisms of transport of cells to surfaces differ between these two types of substrata. In the case of impermeable substrata, the transport of microbes to the surface occurs by passive diffusion, gravitational settling, or active movement (motility), depending on the rheological characteristics of fluid flow. At close proximity to the surface, the further approach of cells to substrata becomes controlled by long-range forces, usually repulsive electrostatic forces and attractive van der Waals forces [241, 242]. The distance where these forces take over is variable and depends on the electrostatic charge of both cell and substratum surfaces as well as on the ionic strength and composition of the liquid medium, as described by the DLVO theory [243]. From this moment onwards further developments depend on the properties (surface charge, hydrophobicity, etc.) of the interacting microbial and substratum interfaces, which consist of the native bacterium or substratum surfaces modified by adsorption of a conditioning film of organic and inorganic constituents of the liquid medium [35, 244]. Short-range interaction forces take over at a distance of 3–5 nm, when adhesive contacts are established between the cell and substratum interfaces. Cells become initially immobilized at a very small distance from the substratum surface by weak adhesive contacts, and small disturbances may cause their release back into the medium. Th s reversible adhesion eventually transitions into an irreversible adhesion with the formation of strong adhesive bonds between the cell and the conditioned membrane surface. Cell approach to permeable membrane surfaces is much more direct and intense, since cells are dragged to the surface with the liquid stream that traverses the membrane. In fact, cells accumulate in large quantities on the membrane due to physical rejection. For example, if one assumes a microbial cell count of 105 cells/ml in feedwater, then the amount of cells rejected will range from 1.4  105 to 3.4  105 cm2 /h at design fluxes of 13.6 l/m2 h (8 gfd) and 34 l/m2 h (20 gfd), respectively. Th concentration polarization model forecasts back-diffusion of part of these cells into the feedwater stream due to

7.7 Biofouling

the concentration gradient buildup at the rejection layer. It is doubtful, however, that this back-diffusion mechanism would represent an effective means for cells exiting the fouling layer, because the close contact between individual cells and their compaction in the rejection layer would facilitate their aggregation into larger clumps, which diffuse at considerably smaller rates compared to individual cells. It is likely that hydrodynamic drag may represent the main mechanism for removal of non-adhered cells from the foulant layers in membrane filtration. Another important issue for consideration is the nature of the surfaces that cells are deposited to in the case of NF membranes, which show effective rejection of larger inorganic ions and of most organic molecules in the feedwater. Microbial cells do not contact directly membrane surface polymers when rejected but become embedded in the concoction of inorganic and organic molecules accumulated at the membrane. Since small molecular weight organics and inorganics are numerically dominant in water samples, it is likely that they will make up most, if not all, of the initial fouling layer on the membrane surface [245]. It is also important to realize that these rejected surface films are not true conditioning films, which are assembled from dissolved organics and inorganics adsorbed on non-permeable surfaces, whereas rejected material may be retained in the foulant layer even when it has no affinity for the membrane surface. Biofilm growth in real membrane plants may not always start from individual cells. Microbial biofilms are ubiquitous in natural environments and grow in many parts of the plant prior to the NF membrane elements. Biofilms shed continuously cells and, randomly, biofilm fragments in a process called sloughing [246]. These biofilm components are transported downstream in the plant with the feedwater and may collect on impermeable surfaces or on membranes. Biofilm formation starting from biofilm fragments may lead to faster build up of biofouling on membrane or spacer surfaces inside NF elements. 7.7.2.2

Biofilm Development

Biofilms constitute a self-replicating fouling layer. Once cells are firmly immobilized either by attachment to a solid impermeable substratum or entrainment in a foulant layer deposited on a membrane, these cells may start to grow by the production and release of daughter cells [247]. The daughter cells move away from each other [248] and, most relevant for biofilm buildup, form microcolonies where microorganisms are held together by a cohesive layer of glycocalyx secreted by the cells. Eventually, the impermeable surface becomes covered with a large number of microcolonies, each microcolony representing an assemblage of microbes of the same strain or species. On permeable surfaces such as membranes, cell proliferation and microcolony formation do not depend on membrane surface adhesion, since there is no need for microcolonies to resist lateral shear forces, and fluid flow compresses the cells against the membrane surface. On permeable membranes, the initial microcolony forming cells are probably recruited from the first batch of microbes rejected at the membrane interface, which become buried deeper in the deposit and remain there for long periods, since there is no backwash cycle to dislodge them from that position in NF. Cell growth and matrix production, however, may occur throughout the fouling layer where individual cells are trapped by a complex network of

329

330

7 Fouling in Nanofiltration

(a)

(b)

(c)

(d)

Figure 7.25 Membrane biofouling. (a) Fouled upper and lower membrane leafs of a feed channel. Section at the center of the leafs with lesser fouling probably because of higher cross flow velocity in this area. (b) Scanning electron micrograph of a fouled membrane surface. Note bacteria embedded in an amorphous matrix. (c) Membrane element fouled by fungi. (d) Fungal mycelia on the fouled membrane surface.

organic and inorganic macromolecules rejected by the membrane. Owing to restrictions on lateral spread, microcolonies will eventually coalesce into multispecies biofilms [232]. Biofilms in channels with water flow will grow until their thickness reaches the point where the shear force of the moving water tears away the upper parts of the biofilm structure. Biofilms will block water channels if shear forces are not strong enough to disrupt these. An example of a fouled spiral wound membrane module is shown in Figure 7.25 and the fouling of spacers is depicted in Figure 7.26. More details on modules, spacers, and their configurations were given in Chapter 3. Generalized de novo biofilm formation occurs only once in an NF membrane element, either at start-up of the plant or when a new element is installed in replacement of an old one. Biofilm growth on replacement membranes may be accelerated by transfer of biofilm fragments from preceding membranes or equipment in the system. During operation, new surfaces for colonization in membrane elements may become available after cleaning, or when biofilm fragments detach by sloughing. There is no information available in the scientific literature on how such spaces that become vacant in a biofilm environment get recolonized.

7.7 Biofouling

(a)

(c)

(b)

Clean feed channel

(d)

Clean feedchannel Feedchannel with fouled membranes only

Feedchannel with fouled membranes and feedspacers

(e)

Δp small TMP Δp small Δ TMP high

Δp high Δ TMP high

(f)

Figure 7.26 Feedspacer fouling. (a) Pressure marks of antitelescoping devices at element exit caused by high feedchannel pressure drop. (b) Spacer extruded because of feedspacer fouling. (c) Feedspacer with firmly attached fouling layer. (d) Schematic representation of consequences of membrane and feedspacer fouling on transmembrane pressure and feedchannel pressure drop. (e) Feedspacer imprint on fouled membrane surface. (f ) Permeate side fouling due to mechanical damage of membrane surface by feedsapcer displacement.

On solid substrata such as feed channel spacers, it is likely that a preexisting neighboring biofilm will spread and take over the new surface niche, rather than it being colonized by a newly adhered organism. On permeable membrane surfaces the free space will be quickly covered by a fouling layer of rejected material from

331

332

7 Fouling in Nanofiltration

the feedwater. Cells embedded deeper in this layer and closer to the membrane are likely to remain there for long periods of time, which should increase the likelihood of them developing stable biofilms in those locations. Whether it is the growth of these new arrivals or the ingrowth of adjacent biofilms that ends up occupying the free space is not known. Houari et al. [249] reported biofilm growth kinetics for modules retrieved from the first stage of one of the largest NF plants in operation at Mery sur Oise (see also Chapter 9). After seven days there were many adherent cells and a few microcolonies visible on the membrane surface, with an average microcolony height of 32 m. Biofilm development was still patchy after 80 days, when average biofilm height reached 54 m. Biofilms grew little in height thereafter, stabilizing on average thicknesses of 61–64 m after 475 and 717 days, respectively. Th data, however, show that cleaning in place (CIP) was performed after about 250 and 500 days of operation. Amide I and II band intensities as surrogates of protein content stabilized after 80 days, suggesting little increase in microbial biomass thereafter. Polysaccharide FTIR band intensity increased rapidly in the first 80 days and steadily thereafter, suggesting continuing production of matrix even after the cell biomass reached a climax. Schneider et al. [250], who performed end of life analysis of fouling layers on all elements of a two-stage RO unit treating surface water, reported uniform foulant layer mass and total microbial cell counts throughout the system. 7.7.2.3

Biofilm Species Composition

Biofilm species composition is dynamic, with continuous incorporation of new microbes, algae, fungi, and protozoa [251]. A large diversity of fungi and bacteria have been isolated from membrane biofilms [252]. Fungal genera recovered from fouled cellulose acetate membranes include Acremonium, Candida, Cladosporium, Rhodotorula, Trichoderma, Penicillium, Phialophora, Fusarium, Geotrichum, Mucorales, and others. Bacterial genera isolated from membrane biofilms include Acinetobacter, Arthrobacter, Pseudomonas, Bacillus, Flavobacterium, Micrococcus, Micromonospora, Staphylococcus, Chromobacterium, Moraxella, Alcaligenes, Mycobacterium, Lactobacillus, etc. It is not known whether these organisms colonize membranes in a random sequence, or whether particular microbial species are always involved in primary colonization. Ridgway [251] reported that Mycobacterium sp. were the initial colonizers of TFC RO membranes installed at Water Factory 21 in Orange County. The range of organisms identified in biofouling studies of RO and NF membranes differs between studies, suggesting that the species composition on membrane biofilms varies between sites. Newer investigations based on direct extraction and analysis of nucleic acids are uncovering a large number of uncultivated microbial and fungal strains in membrane biofilms [253, 254]. There are very few studies about the origin of the organisms that form biofilms on membranes. Th se cells arrive at their location by transport in the feedwater. Most NF systems require some form of pretreatment of the feedwater (see Chapter 8). Incorrectly operating or planned pretreatment stages may represent

7.7 Biofouling

a significant source of biofilm bacteria in membrane installations. Surfaces in pretreatment systems such as ion exchangers, sand filters, granulated activated carbon filters, degasifiers, cartridge filters, holding tanks, and piping are all excellent sources of biofilm-forming organisms on the feed side of membranes in RO or NF systems. Biofilms on permeate spacers originate from bacteria introduced into those locations during manufacturing or from microbes that reach the permeate through holes in the membrane. Indeed, some microorganisms in membrane fouling layers are able to degrade cellulose acetate membrane materials [9]. 7.7.3

Biofilm Structure

Th coalescence of individual microcolonies or columns does not usually result in the coverage of the substratum surface by a compact gel-like biofilm, since even mature biofilms are crisscrossed by a network of channels, which allow access of nutrients and removal of waste products from within the slime layer [255, 256]. Microbial populations inside biofilms are often stratified. Aerobes colonize the surfaces of biofilms and of channels inside the microcosms, but high oxygen consumption and diffusion-limited supply of this electron acceptor usually limits the depth to which aerobes can grow to about 100–200 m from the biofilm surface [257]. Deeper layers inside biofilms are colonized by anaerobic organisms, including denitrifiers, sulfate-reducers, and methanogens [257]. Such stratification in an NF plant may be found in biofilms grown on impermeable substrata such as pipes, reservoir walls, and feedwater channel spacers. The flux of feedwater across biofilms on the permeable membranes may enhance the transport of oxygen into the structure to the extent of preventing the development of anaerobic niches. Th biofilms on the NF membranes of the Mery sur Oise plant had a maximum thickness of 65 m and were most likely aerobic. Biofilms in the different sections of the plant had different elasticity and viscosity properties [258]. Biofilms on membranes may become anaerobic when treating feedwaters in wastewater reuse applications. 7.7.4

Growth of Microbes in Biofilms

Microbial growth requires a supply of macro- and micronutrients for the synthesis of biomass, the most important of which are C, N, P, S, K, Mg, Ca, and Fe. Microbes also require a source of energy to drive metabolism. Heterotrophic microorganisms derive their energy from the oxidation of organic carbon, whereby electrons abstracted from the organic substrates are channeled to the respiratory chain to produce proton gradients across the cell membrane, which are in turn consumed to produce ATP. Terminal electron acceptors, which abstract the spent electrons from the respiratory chain, need to be available in abundance in order for growth to proceed at pace. In the case of NF biofouling, where microbial growth occurs on surfaces inside dark feedwater channels, the most important terminal electron acceptors are oxygen, followed by nitrate, sulfate, or iron(III), which would drive microbial growth in the microaerophilic and anaerobic portions of the biofilms.

333

334

7 Fouling in Nanofiltration

NF feeds sourced from treated wastewater, surface reservoirs or rivers, and shallow aquifers usually contain sufficient macro- and micronutrients to support rapid microbial growth. Nutrient supply may be more limited in deep aquifer feedwaters. Most organic carbon for microbial growth in natural waters is available as polymers. In marine environments, these polymers are mainly degradation products of phytoplankton and consist of 25–50% proteins, 5–25% lipids, and up to 40% carbohydrates [259–261]. Freshwater NOM was described in detail in Section 7.4 and contains about 10% proteins, 30–50% polysaccharides, some lipids, and a significant proportion (15–25%) of lignin-derived humic substances [261]. Monomers are often present only at very low concentrations. Interestingly, in natural waters, large molecular weight organic carbon is a better carbon and energy source for microbial growth than low molecular weight organics [262]. Polymers need to be hydrolyzed extracellularly in order to become available for microbial metabolism. Organic carbon for microbial growth may be inadvertently produced in pretreatment stages, for example, when the chlorine used to improve flocculation kills microbial cells with the consequent release of easily degradable organics [263]. Water treatment additives such as coagulants, antiscalants, or pH control agents may add substantial amounts of biodegradable organic carbon to NF feedwater [264, 265]. Additional sources of organic carbon for microbial growth in fouling layers are metabolites produced by microbes and biopolymers released by death and lysis of cells due to mechanical shear at the membrane surface [28, 266]. Th quantification of the amount of organics present in feedwater available for microbial growth is a difficult task. Ordinary chemical analysis such as total organic carbon (TOC) or DOC determination with high temperature catalytic oxidation [267] are relevant for determining the potential for organic membrane fouling, but not for assessment of the biological growth potential. Owing to the huge diversity of both large and small molecular weight hydrophilic and hydrophobic organics amenable to microbial breakdown in NF feedwaters, chemical analysis of biologically degradable organic carbon is not possible in practice. Instead, bioassays are employed to quantify the amount of carbon that can be utilized by bacteria for growth, AOC, in the samples. AOC assays may be performed with pure culture microbes [268] or, better, via direct assessment of the growth of the indigenous microbiota by flow cytometry [269]. The question of whether a threshold concentration for AOC exists, below which membrane biofouling will not impact membrane operation, is currently a hot topic of research. Some authors suggest that AOC below 30 g/l may be effective, in conjunction with P limitation [270], to control biofouling on spiral wound membranes [271]. However, there may be no safe lower limit for AOC to entirely eliminate biofilm growth, since RO and NF permeates contain sufficient nutrients for biofilm growth [264, 272]. Microbial growth inside the biofilm occurs in diffusion-limited conditions. Th pores of the glycocalyx limit the access of large molecules to cells inside the biofilm and create a tortuous diffusion path for small molecules between the biofilm–liquid interface and the cells embedded in the biofilm matrix [257, 273]. Every microbe inside a biofilm experiences unique environmental conditions, since all nutrients and metabolites are subject to diffusion-limited

7.7 Biofouling

transport [274]. Growth rates of microbes within biofilms are nowhere near the maximum growth rates achievable in well mixed media with balanced nutrient composition, except for growth of initial colonizers directly exposed to liquid [275]. Part of the organic carbon metabolized by the cells is converted into extracellular matrix polymers, the glycocalyx. Th s glycocalyx contains between 50 and 90% of the organic carbon of biofilms and is composed primarily of exopolysaccharides, but it includes other materials of biological origin such as proteins, DNA, RNA, and lipids [237]. The glycocalyx may function as a sponge and adsorb nutrients present in very small concentrations in the aquatic phase [232]. The combined action of rejection of salts at the membrane surface and their entrapment inside the biofilm glycocalyx may lead to enhanced osmotic pressure on the membrane surface and thus reduce flux [276]. It also assures that neighboring organisms remain locked in their relative positions for prolonged periods of time, thus fostering cooperative breakdown pathways [277]. 7.7.5

Sites for Biofouling in Membrane Systems

From the intake through pretreatment to any particular membrane filtration unit, membrane plants offer a vast array of surfaces for microbial colonization. Substrata for biofilm growth may range from concrete in intake structures or water tanks, to steel and a variety of polymeric materials in piping, pretreatment, and membrane filtration units. Many surfaces are non-permeable solids, which may be hydrophilic (steel, concrete) or hydrophobic (polymers). Large area permeable polymer surfaces for microbial growth become available pre-NF when MF or UF membrane units are deployed in pretreatment trains. Th spiral wound NF elements harbor by far the largest surface area available for microbial colonization in an NF membrane plant. Each element of a standard commercial 8′′ NF element offers approximately 37 m2 permeable membrane and 79 m2 of impermeable feed spacer surface for microbial colonization. Th s assumes a feed spacer being 0.86 mm thick, at a 2.6 ⋅ 2.6 mm mesh, 100% of feed spacer surface area available for bacterial colonization, and a total feed spacer area equivalent to half that of the membrane area since feed spacers are sandwiched in between two membrane sheets. Even larger surface areas are available on the permeate side of the NF membrane, with its extensive network of pores in the polysulfone and polyester membrane mechanical support structures and in permeate channel spacers. Owing to the very large surface areas and many points of stagnant or low turbulent flow, some authors consider feed spacers to be the prime site for biofouling in the feed channel of spiral wound modules [278–280]. Vrouwenvelder et al. [280] demonstrated that transmembrane flux was not essential for biofouling buildup in the feed channel, and fouling rates were similar for systems operating with and without flux. Th s compelling dataset suggests that biofilms grown on feed spacers are more critical to membrane performance than membrane surface biofilms. Several factors may contribute to quicker biofilm growth on feed spacers: (i) membrane biofilms are embedded in a rejection layer containing many growth-inhibiting substances such as rejected salt [276]; (ii) feed spacer biofilms

335

336

7 Fouling in Nanofiltration

are exposed in a turbulent flow environment, where substrate and electron acceptor supplies are rapidly and continuously replenished at the biofilm/feedwater interface; (iii) substrate and electron acceptor mass transport per unit area is smaller on membrane biofilms than on feed channel spacers [281]; (iv) membrane fouling layers contain a significant proportion of non-biofilm loose cells deposited by rejection, which are not embedded in a matrix, which consume substrates, and may be washed away by liquid flow; biofouling growth on spacers is catalyzed by biofilm cells, and loose cells will be removed from the biofilm by turbulent drag; and (v) biofilm compression in NF and RO systems is much more severe on membrane than on spacer surfaces [282]. Th specific hydraulic resistance of NF membranes is approximately 2 ⋅ 1013 m1 , while biofilm resistances range from 6 ⋅ 1012 to 5 ⋅ 1013 m1 [283]. Clearly, biofilms have little influence on the increase of hydraulic resistance of the membrane surface, but in the feed channel, where open channel hydraulic resistance is 0, biofilm clogging will cause a resistance increase of several orders of magnitude, which translates into a rise in feed channel pressure drop. Spacer biofouling effectively turns the membrane channel into a porous medium, with most feedwater flowing at high velocity through narrow preferential flow channels that remain open due to the strong rise in local turbulence, since membrane systems are usually managed to produce a constant product flow, and pump pressure is increased when hydraulic resistance to filtration in the system rises (Figure 7.25a). Membrane surfaces in sectors with fouled feed spacers will receive much less feedwater flow with consequent reduction of the surface scouring efficiency in these sections. Increased pumping pressures to compensate reduction of product water flow may lead to feed spacer extrusion at the element outlet (Figure 7.26b). In the past, when anti-telescoping devices were not standard fittings in spiral wound elements, mechanical collapse of modules by telescoping was not uncommon in membrane units experiencing severe biofouling issues [252, 284]. Membrane elements received for autopsy in one of the author’s laboratory frequently show pressure marks of the anti-telescoping devices on the back end of the element (Figure 7.26a). Such spacer extrusion will most certainly tear apart the polyamide or other types of thin NF separation layers in the membrane sheet, thus compromising the integrity of the filtration barrier (Figure 7.26f ). 7.7.6 Measuring Microbial Load in Feedwaters and Detecting Biofilms in Membrane Systems Biofouling prediction and risk analysis requires reliable methods to estimate the microbial burden in feedwater as well as biofilm formation on membrane surfaces. Both tasks are very challenging. Microbes in feedwater samples occur as individual, planktonic cells and in a variety of aggregates. Microbial aggregates may be produced by coagulation or co-adhesion type of phenomena, where cells become attached to each other by adhesive interactions, or represent sloughed biofilm samples, where cells are embedded in a self-produced matrix. Cell density in aggregates is usually orders of magnitude larger than in liquid. Counting planktonic cells is straightforward, but clusters or suspended biofilm samples need to

7.7 Biofouling

be disaggregated before analysis. Standard microbial burden analysis methods usually do not incorporate efficient disaggregation preparative steps in the procedures, and the ensuing errors in analysis are significant. Direct counting on a microscope may at least allow one to detect aggregates present in a sample, but accurate counting of cells in the aggregates is impossible without disaggregation. In growth media, aggregates containing tens of thousands of cells will produce just a single colony count. Unfortunately, many membrane biofouling studies still rely on colony counts for microbial analysis, which demonstrably recover between 3 and 5 orders of magnitude less cells than present in the samples [285]. New and extremely powerful microbial cytometric analysis techniques are replacing traditional time-consuming, cumbersome, and inaccurate procedures for characterization of the microbial burden in NF feedwaters or in biofouling biofilms. Flow cytometry in particular has benefitted greatly from the reduction in price and drastic simplification of equipment operation [286]. The advantage of this technique is that it can be combined with a vast array of cell markers and thus produce a detailed physiological profile of the capabilities of the individual members of a population while counting the cells [269]. There is, however, a need to develop and validate better biofilm disaggregation protocols to make this liquid-based analysis technique more effective for biofouling characterization. Biofouling development in membrane plants is monitored indirectly via both transmembrane pressure drop and feed channel pressure drop [287]. Flux decline in itself is not a specific indicator for biofouling, since any type of deposit on the membrane surface that increases hydraulic resistance to fluid flow will cause a flux decline. Flux decline caused by biofouling usually occurs in two stages. Colonization and growth of a microbial biofilm on a clean membrane causes an initial strong flux decline, which is followed by a second slower phase where flux declines in an almost asymptotic manner, probably because of ripening of biofilm and other foulant layers [249]. Feed channel pressure drop is a much more specific indicator for biofouling, since there are no reports of purely chemical fouling of feed spacers increasing pressure drop in the feed channel. Most probably, the turbulent flow in the feed channel prevents a strong buildup of non-biological foulants on spacer surfaces. The cohesive biofilm matrix is the distinctive feature that allows biofouling to establish and remain firmly anchored on feed spacer surfaces. The e are many techniques available to study biofilms, but the vast majority require the destruction of the membrane element or removal of the membrane from a flow cell prior to analysis [288]. Inline fouling monitoring would be desirable for better biofouling management in membrane plants. Ideally, the fouling monitors would be inserted on feed spacers or membrane surfaces inside the elements and transmit their signals to external transducers for analysis in real time. Th only technology capable of nondestructive analysis in whole spiral wound elements, NMR microscopy, is too bulky and too expensive to be used continuously on commercial installations [289]. Several alternative techniques, including OCT [290], planar oxygen sensing optodes [291], photointerrupt sensor arrays [292], surface enhanced Raman spectroscopy [293], optical sensors [294], and more [295] are under development for real time biofouling monitoring on at least side streams of NF plants. Recently, a number of different flow cell configurations

337

338

7 Fouling in Nanofiltration

have been developed and validated for precisely replicating filtration conditions inside commercial modules for side stream investigations [132, 279, 296]. Membrane autopsy is a last resort destructive procedure to investigate the reasons for loss of membrane performance. Autopsies may also be conducted during operation to check for the efficiency of fouling mitigation procedures [297, 298]. An autopsy allows applying very sophisticated chemical [299, 300] and microbiological [301] characterization techniques to identify the causes of fouling. 7.7.7

Biofouling Management in Membrane Systems

Successful biofouling management in NF filtration plants requires a holistic approach and needs to encompass the entire process from intake to the membrane unit [302]. In the past, conventional pretreatment of feedwater for NF mirrored standard practice in surface water treatment and included coagulation/flocculation/multimedia filtration followed by chemical feedwater conditioning. Cartridge filters installed immediately before the membrane unit prevent membrane damage by particles. Chemical conditioning includes the addition of pH control agents, antiscalants. Continuous dosing of oxidizing biocides such as chlorine at low concentrations has proved effective for biofouling control in chlorine-resistant UF and MF membranes, but cannot be realized for NF membranes, which include oxidation-labile polyamide in their formulation. Chlorine quenchers such as metabisulfite or activated carbon filters have, therefore, to be incorporated into pretreatment to prevent polyamide membrane oxidation. Such a conventional pretreatment scheme is not effective for feedwaters with high biofouling potential [263]. In the past 20 years, MF or UF pretreatment prior to NF has become popular. While effective in removing the microbial burden in the raw water, these microporous membranes do not eliminate AOC from the feedstream. AOC removal is best achieved with one of the many biological filtration reactor configurations available in the industry [303]. Essential to effective biofouling management is tailored design and operation of the membrane unit as summarized by Vrouwenvelder et al. [304]. The basic concept underlying these innovative spiral wound membrane plant design concepts is to uncouple the front end of the plant, where biofouling is most pronounced and most acute, from the back end where chemical fouling becomes an issue. Such a design allows implementing optimized operational procedures, including cleaning regimes, for each section of the plant. A design allowing for low linear velocities of between 0.07 and 0.09 ms1 in the lead membrane elements prevents high feed channel pressure drop in this section of the plant most prone to biofilm development [305]. Periodic feed flow reversal helps reduce feed channel pressure drop even more by periodically subjecting the feedspacer and membrane biofilms to feast (fresh feedwater)–famine (feedwater from a previous membrane element) cycles. Sensitive pressure drop monitoring in the lead sections [287] in conjunction with preventative cleaning cycles helps maintain biofouling to a minimum and limits its spread to the next filtration stages [250]. Conventional design spreads the biofilm fragments detached in the

7.8 Fouling Prevention and Cleaning

front end elements during cleaning to the back elements in the pressure vessel and to the succeeding stages in a multistage installation. Membrane cleaning will have to be periodically conducted to recover performance. Th major challenge for biofilm removal is the effective disruption of the biofilm matrix and removal of biofilm fragments from the feedwater channel. Di Martino et al. [297] reported that membranes remained covered by a thin biofilm after chemical cleaning. Chemical cleaning diminished significantly the colony counts and ATP content on membrane surfaces, but total cell counts remained almost constant [306]. The cells that remained at the membrane surface were either dead or injured, demonstrating a great potential for regrowth in the system [307–309]. An increase in transmembrane flux after chemical cleaning, therefore, may not necessarily reflect good removal of biofilm bacteria from membrane surfaces; it may be due solely to disruption of biofilm structure resulting in increased water permeability [310]. Contact of biocides from cleaning solutions to bacterial cells inside biofilms requires disruption of the biofilm matrix. Oxidizing agents have a low efficiency in biofilm control, since these substances are largely consumed in the oxidation of glycocalyx compounds and do not reach the cells. Biocides have to be used together with compounds such as detergents, chaotropic agents, and chelating agents, which effectively disrupt the glycocalix structure and allow the biocides to act upon the cells directly [311]. Repeated use of the same cleaning solution against a particular biofilm may result in the selection of resistant strains and thus decrease the biofilm’s susceptibility to the cleaning agent. Effective biofilm removal, therefore, depends on the periodic change of the cleaning and sanitizing solutions.

7.8 Fouling Prevention and Cleaning 7.8.1

Pretreatment as Fouling Prevention

In NF normally frequent cleaning is avoided by using different types of pretreatment. For example, particulate matter is aggregated and settled until an almost particle-free feed is achieved. MF and UF may be more effective in removing such particulates than conventional treatment, but small colloidal matter may still permeate. Such pretreatment results in a very low (less than 3) SDI, which was described in Section 7.2.3.2. A success story was reported by Gwon et al. [37] who attributed the absence of biofouling to pretreatment with UF. In pretreatment also biocides and chlorine pretreatment can be used to avoid biofouling. More details on pretreatment in NF can be found in Chapter 8 and related review articles [312, 313]. 7.8.2

Membrane Modification for Fouling Prevention

An alternative to pretreatment in order to avoid fouling and subsequent cleaning is to modify the membrane or the membrane surface in particular. Normally, the modification aims to produce a more hydrophilic membrane [314], a more resistant membrane, or sometimes a more charged membrane. Th problem with

339

340

7 Fouling in Nanofiltration

modification is that in some cases the modification material takes up space in the membrane polymer, and thus the flux decreases. The modification is, therefore, often a compromise between fouling and pure water flux. Lee et al. [41] have studied the effects of anionic and cationic surfactants on UF membranes used for NOM filtration. While anionic surfactants had no effect on flux or retention, cationic surfactants decreased flux, which was accompanied by an increase in NOM retention. In contrast, Combe et al. [72] determined that anionic surfactants reduced adsorption of humic acids considerably. An attempt to make modifications of NF membranes (NF 270) has been performed by Belfer and Gilron [315] (see Table 7.5). Th membranes have been modified in situ (also spiral elements) with acrylates or other similar monomers. Th cross-linking agent was Ethylene Glycol Dimethacrylate (EGDMA) and the reaction time 15–60 minutes. It can be seen that the flux has decreased somewhat due to the modification, while fouling has equally decreased (measured as the difference in the pure water flux before [PWFb ] and after the experiment) especially with the hydroxymethyl ester of methacrylic acid (HEMA)-modified membranes. A similar procedure for RO membranes has been reported by Gilron et al. [316]. Coating membranes with a layer of pre-adsorbed polymer may sterically prevent the foulants from entering into the membrane matrix. Th effectiveness of such a treatment depends on membrane characteristics. Hydrophobic membranes generally are more susceptible to adsorption and fouling and hence making such a membrane more hydrophilic is likely to improve performance [138]. Kilduff et al. [13] have developed a technique to modify the surface of NF membranes using UV irradiation and UV-assisted graft polymerization. Such treatment resulted in increased hydrophilicity of the membranes possibly due to the formation of Table 7.5 NF270 membranes modified in situ by redox-initiated graft polymerization with potassium persulfate/potassium metabisulfate as the initiator (0.005–0.03 M).

Modification

Pure water Permeate flux flux (l/m2 h) (l/m h) (0–3 h) R, UV (%) R, TOC (%) Fouling (%)

Original

205

132

51

80

MA-1 M; 30% PEGMA

65

40

92.7

80

—

AA-1 M

226-183

130

97

83

13

148

90.2

69

22.5 4.9

MA-0.5 M; PEGMA-0.002 M 232

21.6

HEMA-1 M

156

123

98

80

HEMA-0.5 M

173

120

96.8

72

1.4

HEMA-0.2 M

204

118

96.6

72 74

1.4

HEMA-0.2 M (circulated cell) 175

125

97.7

82

0.6

Retention (R) of UV-absorbing compounds (UV) and of total organic carbon (TOC) and fouling in NF of paper machine clear filtrate. Modification agents used: acrylic acid (AA), methacrylic acid (MA), polyethylene glycol ester of methacrylic acid (PEGMA), hydroxymethyl ester of methacrylic acid (HEMA). Source: Adapted from Belfer and Gilron 2002 [315].

7.8 Fouling Prevention and Cleaning

surface hydroxyl groups. Less fouling, however, came at the cost of decreased retention. A further method to enhance the fouling resistance of polyamide RO and NF membranes was improved by grafting poly(ethylene glycol) diglycidyl ether (PEGDE) to the surface of the membrane. Th improvement was attributed to the steric hindrance of poly(ethylene glycol) (PEG) chains preventing a close approach of foulants to the membrane surface. The modification resulted in an increase in NaCl rejection and a flux decrease [317]. Zwitterionics have been used to control membrane surface charges and have been used to modify the fouling resistance of membranes [318–323]. Such deposition can be temporary and can be repeated during operation. Th impacts on membranes are very compound specific and can change permeability and rejection and modify the adhesion of organics, and thus enhance the organic and surfactant fouling resistance of the membrane [318]. Bengani et al. [324] used zwitterions as the membrane selective layer and demonstrated a high resistance to irreversible fouling oil-in-water emulsion and a protein solution. Matin et al. [325] used an initiated chemical vapor deposition (iCVD) technique to deposit 2-hydroxyethyl methacrylate-co-perfluorodecyl acrylate (HEMA-co-PFDA) copolymer on RO to create an amphiphilic surface. By this surface modification the biofouling of the membrane in cross flow at high pressure and salinity conditions was slowed down considerably. Immobilization of the photocatalyst, currently dominated by TiO2 , onto membranes for photocatalytic degradation of contaminants and their intermediate products and giving self-antibiofouling activity to the membrane has become a topic of interest in recent years [326, 327]. However, TiO2 photocatalysts mainly absorb UV photons, while solar light contains only about 5% UV light. The efore, extending the spectral response of TiO2 materials to visible light has become an important field of research [328, 329]. Moustakas et al. [327] coated a UF membrane with modified nanostructured titania (m-TiO), which had extended visible light absorption up to 566 nm. 7.8.3 7.8.3.1

Cleaning Methods Physical Cleaning Methods

While the main focus in this section is on chemical cleaning, some form of physical cleaning is generally part of a cleaning protocol and hence a brief overview is provided. Physical cleaning generally uses mechanical forces to remove foulants [330] and such methods include (i) backflush/forward flush/reverse flush, (ii) scrubbing (e.g. using foam balls for tubular modules), (iii) air sparging, (iv) CO2 back permeation, (v) vibrations, and (vi) sonication. While a number of researchers have used permeate backwash for TFC membranes, this is somewhat surprising as the risk of damage to the thin active layer in backwash operation is considerable. However, Chen et al. [330] have reported beneficial effects of such backflushes presumably due to the disruption of the foulant layer, which was subsequently removed by a forward flush. Sonication is a relatively novel method for membrane cleaning, although ultrasound is commonly used in membrane autopsies to remove the fouling deposits

341

342

7 Fouling in Nanofiltration

from the membranes for chemical analysis. Lim and Bai [331] have used sonication in MF and found that the technique was very efficient in removing cake deposits, but not effective in removing pore blockages. This resulted in a decrease in cleaning efficiency over time as mechanisms such as pore plugging became more predominant. Hence, it was required to combine sonication with backwashing as well as a chemical cleaning protocol. 7.8.3.2

Chemical Cleaning Agents and Processes

Chemical cleaning relies on chemical reactions to break bonds and cohesion forces between membranes and foulants [330]. Such chemical reactions include (i) hydrolysis, (ii) saponification, (iii) solubilization, (iv) dispersion, (v) chelation, and (vi) peptization [332]. In many cases, NF membrane manufacturers co-operate with a cleaning agent manufacturer to establish the most suitable cleaning process for the membranes produced by the membrane manufacturer. However, as described above, the cleaning protocol is not only membrane specific but also foulant dependent. Some typical producers of formulated cleaning agents are Diversey-Lever A/S, Henkel-Ecolab GmbH & CO, Ondeo Nalco Ltd., and Novadan A/S. [333, 334], and several authors have listed cleaning mixtures or protocols [21, 37, 39, 330, 335, 336]. Li and Elimelech [337] established in a very fundamental study that cleaning can only be effective when calcium ion bridging could be removed by the chemical cleaning agent. Th authors used sodium dodecyl sulfate (SDS) and ethylene diamine tetra acetic acid (EDTA) for those studies with organic foulants and multivalent ions. Alkaline cleaning is often the most important as many foulants, especially in natural waters or wastewaters, are of organic nature or inorganic colloids may be coated by organics. Alkaline cleaning aims to remove organic foulants from the surface of the membrane and from the pores of the membrane. Th high pH is usually a result of using sodium hydroxide and sodium carbonate containing cleaning solution. In most cases, then a surfactant is included in the formulated cleaning agent. Th s surfactant emulsifies fat containing particles and prevents the foulant from adhering back on the surface. Th surfactant is mostly anionic or nonionic and acts together with the alkaline agent (caustic) to remove the foulant. In many cases, some sequestering agent such as EDTA is added to the formulation to remove multivalent ions such as calcium and magnesium. Ang et al. [338] showed that when membranes are fouled by organic matter in presence of calcium ions, NaOH solution is not a suitable cleaning agent. Addition of EDTA to the NaOH solution, as cleaning agent, increased the efficiency of cleaning because it is able to remove complexed calcium by a ligand-exchange reaction and disrupt the complex in the gel layer. At higher pH more carboxylic groups of EDTA are deprotonated, increasing the chelating ability [338, 339]. It appears that alkaline cleaning is often the most effective cleaning step to remove organic matter [37]. Acid cleaning aims at removing precipitated salts (scaling) from the surface of the membrane and from the “pores.” Th acid procedure can be the most important cleaning step in RO as the scaling problem occurs in connection with salt

7.8 Fouling Prevention and Cleaning

retention. Often the acid used is nitric acid at a pH of 1–2. In many cleaner formulations citric acid has been used as well as phosphonic and phosphoric acid. Th extensive use of nitric acid depends on its fairly mild oxidizing ability. In the acid cleaner formulations, detergents, cationic or nonionic, as well as some sequestering agents can also be present. Salt cleaning in the form of common inert salts (e.g. NaCl, NaNO3 , and seawater or brines from seawater desalination plants) were introduced as effective cleaning agents for UF and NF/RO membranes fouled by gel-forming organic foulants [340–342]. These reagents are more economically feasible and environmentally acceptable compared to acid, alkaline, or enzymatic cleaning. However, it has been shown that salt cleaning is more efficient when the major foulants are hydrophilic rather than hydrophobic substances. Th s efficiency loss with hydrophobic fouling was attributed to less swelling of the hydrophobic fouling layer upon exposure to the salt solution. Th s results in less diffusion of sodium ions into the layer and hence, lower ion exchange between sodium and bound calcium ions, which is thought to be the main mechanism of freeing calcium ions from the cross-linked gel network and opening the structure of the fouling layer [341]. Ang et al. [343] showed that in cleaning of RO membranes fouled by wastewater treatment plant effluent, addition of NaOH to NaCl cleaning solution can enhance the cleaning efficiency up to 94%, compared to 65% in the case of the individual salt solution without NaOH as NaOH loosens the fouling layer. Enzymatic cleaning is used increasingly where some enzymes can take very high temperatures (70–90  C) even though in most cases their optimal temperature is much lower. Enzymes can often be used when a more neutral pH for cleaning is required, when biofouling is expected or when polysaccharides are the typical foulants. Extracellular substances are often secreted by the biofouling microbes. The nature of the polysaccharide is important to determine the effectiveness of in enzymatic cleaning. Enzymes are mostly very specific in their action and are, therefore, selected for specific foulants or when other cleaning agents do not help. Special caution has to be taken so that the enzyme cannot attack the membrane itself [344]. Khan et al. [345] used protease, lipase, dextranase, and polygalacturonase (PG) based enzymes, at neutral pH, to clean RO membranes. Protease- and lipase-based enzymes were most effective in removing most cells and proteins from the surface. At optimum incubation periods the RO surface properties remained close to the original condition. Biocides for control of microorganisms and biofouling: Zeiher and Yu [346] describe three terms of importance in the control of biological fouling, namely sanitization, which describes a cleaning process with antimicrobial characteristics. A “3 log” or 1000-fold reduction in microbial counts in achieved. Disinfecting is the process in which microorganisms are destroyed, inactivated, or removed in the order of magnitude of a “6 log” or 1 000 000-fold reduction in counts. Lastly, sterilizing means making a system free of all living cells, viable spores, viruses, and sub-viral agents capable of replication. According to Zeiher and Yu [346] it is sterilizing that is desired in membrane systems. Th use of biocides and their common concentrations are summarized in Table 7.6. It should be noted here that many NF membranes are not chlorine resistant

343

344

7 Fouling in Nanofiltration

Table 7.6 Typical biocide concentrations for RO sanitizing. Biocide

Dosage

Comments

Chlorine

0.1–1.0 mg/l

CA membranes and other chlorine resistant only

Peracetic acid

0.02–1.0%

pH of neat product is 3–4

Formaldehyde

0.5–3.0%

Carcinogen

Glutaraldehyde

0.5–5%

Not recommended

Isothiazolone

0.01–0.15%

Slow

Quaternary amines

0.01–1%

Not recommended

DBNPA

Up to 200 mg/l

Fast, easy disposal

Bisulfite

1.5% (preservative) or as needed for Cl2 removal

Preservative, biostatic, Cl2 scavenger

CA, cellulose acetate. Source: Adapted from Zeiher and Yu 2000 [346].

and application of biocides should be in consultation with the membrane manufacturer to avoid membrane damage. For example, Staude indicates that ozone destroys some polymeric membranes, while the resistance to various disinfectants is membrane dependent [3]. Some of those biocides have been reported to cause membrane swelling, which loosens foulants attached to the membranes and hence increases cleaning efficiency [330]. 7.8.3.3

Choice of Cleaning Method

According to Chen et al. [330] the selection of appropriate cleaning protocols is usually based on a trial and error approach. This means testing various cleaning protocols that have been selected by rule of thumb and experience for presumed foulants. If foulants have not been identified, assumptions based on feedwater characteristics need to be made. The different categories of foulants have been described in detail in Sections 7.4, 7.5, 7.6 and 7.7. In terms of the contributions of such foulant categories to fouling, van Hoof et al. [170] have concluded from extensive membrane autopsy surveys that worldwide about 50% of the foulants are of organic nature. This organic foulant fraction is higher in Europe than it is in the United States. With climate change and rising temperatures this organic contribution is bound to increase. Ferric oxide and silica are the next most common foulants followed by alumina, calcium phosphate, calcium carbonate, and calcium sulfate. Silica is apparently more abundant in the United States and calcium phosphate in the United Kingdom. Th effective selection of a cleaning agent is usually preceded by determination of the foulant using feed analysis or membrane autopsy. However, this procedure has limitations in that if several foulants are identified cleaning protocols may become extensive. For this reason, Luo and Wang [21] have optimized a CIP method and established that it is sufficient to remove selected essential foulants as subsidiary foulants may be removed simultaneously. Weis et al. [347] have trialed various cleaning protocols as a function of membrane characteristics

7.8 Fouling Prevention and Cleaning

100

Permeate flux (l/m2 h)

Vendor recommended physical cleaning procedure Optimized physical cleaning procedure

80 60 40 20 0

0

2

4

6 Time (h)

8

10

12

Figure 7.27 Impact of cleaning procedure optimization on flux. Source: Reprinted from Chen et al. 2003 [330].

and established that the choice of cleaning agent was instrumental in achieving a steady state flux value. No doubt the nature and complexity of fouling can make it very difficult to find the ideal cleaning agent. For this reason, Chen et al. [330] have developed a methodology that applies a statistically designed approach (factorial design) to cleaning optimization. The impact of this optimization on an example UF membrane is shown in Figure 7.27 and a significantly increased productivity results. Th cleaning process to be adapted depends on the type of foulant, the tolerance of the membrane toward the suggested cleaning agent, and the regulations applying to cleaning agents for a particular application. For instance, directives exist dealing with issues such as the formulations that are acceptable, and those certainly differ from country to country (e.g. EU and the United States have different norms). Cleaning is in most cases performed using caustic, acid, or enzyme solutions, mixtures of those, combinations of additives, or proprietary commercial cocktails. For sanitation, often chlorine gas is required in some of the treatment steps, although formaldehyde has also been reported. Th effectiveness of such cleaners is usually offset by the damage caused to the membrane materials and hence membrane durability is an important consideration in order not to adversely affect membrane lifetime. 7.8.3.4

Determination of Cleaning Requirement and Frequency

Generally in industry the cleaning interval is designed in such a way that cleaning takes place when a certain amount of flux is lost, although there is evidence that cleaning at an early fouling stage is more efficient than when the fouling is well established and the fouling layer compacted [348]. A flux loss of 10–30% is usually the highest allowable decrease in flux. Another option is to clean on a regular basis, for instance, once a week or less frequently depending on the fouling situation of the process. In some cases, NF can be used almost without cleaning if the operating conditions are such that only subcritical fluxes are used. An

345

346

7 Fouling in Nanofiltration

example of this is NF in humic water treatment in Norway at low flux and low temperature [349]. In most cases, NF processes need less cleaning than UF and MF. Th reason for this is that the common and usually detrimental pore plugging in UF and MF is less important in NF. On the other hand, cleaning is needed more often in NF than in RO due to the more open structure of NF membranes. Owing to the similarities of RO and NF often the same types of cleaning agents and cleaning processes are used. In most cases, the cleaning protocols are dependent on the fluids to be processed by using NF (foulant types). Th cleaning interval depends both on the foulant amounts on the membranes (mostly measured as increased pressure to keep up constant flux) and on the fact that the membranes need to be cleaned and disinfected at regular times (daily in the dairy industry). In the dairy industry fouling and cleaning of MF and UF membranes and visualization of fouling and cleaning efficiency have been reported. Most of the principles are thus reviewed. 7.8.4 7.8.4.1

Determination of Cleaning Effectiveness Water Productivity and Membrane Resistance

The e are different ways to establish how effective a particular cleaning protocol is. One option is to measure the CWF and compare the CWF before and immediately after filtration to determine if flux has recovered due to cleaning. Th s can be done in situ in the process. The recovery of the original steady state process flux is, of course, the most natural way to see that cleaning has been successful [335]. Flux recovery (FREC) can be calculated as shown in Eq. (7.36) FREC =

J0C J0

(7.36)

where J 0C is the clean water flux after cleaning and J 0 the clean water flux of the virgin, unfouled membrane [335]. Alternatively, the effectiveness can also be represented by CWF recovery [330] as in Eq. (7.37) JRecovery =

J0 J0C

(7.37)

Th variation of flux that has been illustrated in Figure 7.3 shows the impact of fouling and cleaning on flux. Th impact of several successive cleaning steps to fill recovery is shown in Figure 7.28 with an example of an UF membrane fouled with proteins, lipids, and carbohydrates. The cleaning was performed with a sequence of (i) rinse wash (water), (ii) an alkaline cleaning (NaOH 0.5 wt%), (iii) a protease detergent (0.75 wt%), and (iv) a sodium hypochlorite (150 mg/l) solution. Th sodium hypochlorite is used as a sanitizing agent and its cleaning effectiveness is attributed to its ability to cause membrane swelling [39]. In accordance with the resistance in series model, this can be depicted as a variation of membrane resistance as shown in Figure 7.29. From this graph, conclusions can be drawn regarding the nature of the foulants and the effectiveness of cleaning. A reduction of resistance by rinsing indicates a loosely associated deposit such as a concentration polarization or a loose gel layer or

7.8 Fouling Prevention and Cleaning

100 Flushing

80 Water flux recovery (%)

Figure 7.28 Flux recovery in the case of successive cleaning steps. Source: Figure reprinted from Sayed Razavi et al. 1996 [39].

Sodium hypochlorite

60 Protease detergent

40 Sodium hydroxide 20

0

Rinse wash 0

15

30

45

60

Time (min)

cake. The resistances are the intrinsic hydraulic membrane resistance (RM ), residual resistance after cleaning (RRES ), resistance after filtration (in this case UF; RUF ), reversible fouling resistance (RRF ), irreversible fouling resistance (RIF ), the total fouling resistance (RF ), and hydraulic resistance of the cleaned membrane (RCW ). Cleaning can be assumed to be complete when RCW  RM , allowing for experimental error [336]. In the case of presentation as resistances, cleaning efficiency can be determined as a function of resistance (Eq. (7.38)): ERW =

RIF  RRES ⋅ 100 RIF

(7.38)

R (m–1)

Rinsing

Cleaning Rrf

Rf

Ruf Rif Rres

Rc Rcw

Rm t (min)

Figure 7.29 Resistances in filtration, rinsing, and cleaning. Source: Adapted from Argüllo et al. 2003 [336].

347

348

7 Fouling in Nanofiltration

7.8.4.2

Foulant Content of Cleaning Solutions

Other than measuring CWF, the variation of the composition of the cleaning solution can be investigated to determine cleaning efficiency. Often changes are visible, such as precipitates when scaling, or a dark yellow or brown color, when NOM is removed. Chemical analysis of cleaning solutions can quantify such observations. For example, Liikanen et al. [335] measured pH, turbidity, color, total solids (TS), and cations in the cleaning solutions. Conducting a mass balance and comparing the amounts removed in the cleaning solution to the amount remaining on the membranes gives further information not only on cleaning efficiency but also on the reversibility of certain foulants. Gwon et al. [37] found that calcium and iron were found predominantly in acid cleans and silica in alkaline cleans. Further, iron was most resistant to removal and adhered strongly to the membranes throughout the cleaning process. This was established with a sonication technique as described in Section 7.2.4. 7.8.4.3

Membrane Surface Investigation

As a further option, the membrane surface can be examined after cleaning to determine if all contaminants have been removed. It has, however, been observed that despite complete flux recovery, not all foulants are removed. Available methods are similar to those used for membrane autopsies such as streaming potential measurements, contact angle methods, FTIR, or SEM. Th se characterization methods are mostly destructive methods. A combination of several methods is the best way to analyze the cleaning efficiency [350–352]. 7.8.4.4

Impact of Cleaning on Permeate Quality and Membrane Retention

Permeate quality and retention are affected by fouling as summarized in Section 7.4.8 and in Chapter 19. According to Liikanen et al. [335] who performed permeate analysis for TOC, UV absorbance (254 nm), pH, alkalinity, hardness, and conductivity, the permeate conductivity generally increased after cleaning. Acidic cleaning assisted in restoring the ion retention of membranes. Simon et al. [353] reported that alkaline cleaning impact on pore size and rejection performance of NF90 membrane was negligible. However, because of having a loose and thin active layer NF270 membrane showed a small increase in pore size and a significant increase in its permeability and salt passage due to alkaline cleaning. As a consequence, cleaning may result in either a restoration or decrease in retention. For example, Chen et al. [330] reported a 10% increase in TDS retention after cleaning. Other important parameters in membrane cleaning are the wash water usage and loss of production. Wash water usage can be represented as the volume of wash water used per total volume of water produced, taking into account that wash water is often membrane permeate and hence product [330] and loss of production is calculated by multiplying the time for cleaning with the average water flux during operation. For the calculation of environmental impact, both water consumption and the generation of a potentially hazardous waste stream need to be considered.

7.8 Fouling Prevention and Cleaning

7.8.4.5

Influence of Operating Parameters on Cleaning Efficiency

Duration: It appears from the literature that shorter filtration cycles (and hence more frequent but shorter cleaning procedures) are beneficial as the fouling layers compact with time and become more difficult to remove. Further, the degree of fouling is an important parameter in recovery during cleaning, which supports the argument for more frequent cleaning (see also Figure 7.27) [330]. Temperature: In general, cleaning efficiency increased with temperature but increases are limited by the heat tolerance of the membranes [335]. It is a rule in cleaning processes to clean at the same or higher temperature as the NF process has been operating. If cleaning is undertaken at a lower temperature there is a risk that the foulants will readsorb on the membrane once normal processing is continued. Optimal cleaning results have been obtained repeatedly in literature (for UF) at a temperature of 50  C [354–356], and also in NF this temperature seems to be quite good to use if the membranes are tolerant to such elevated temperature. A higher temperature could give even better cleaning results, but membranes that can endure temperatures between 70 and 90  C remain scarce. The importance of a high enough temperature in cleaning has two reasons, (i) enhanced removal of foulants and (ii) removal of heat sensitive microbes. Another possibility to circumvent this problem is to give the membranes a short heat shock. Inorganic membranes can endure this type of heat treatment, but availability and applications of inorganic NF membranes remain limited. Pressure and air/water backwashing: In most cases, it has been shown that a high pressure is not beneficial when cleaning. Especially with porous membranes, the pressure pushes the foulants deeper into the membrane, which is also true with open NF membranes. Th applied pressure also causes compaction of the fouling layer [39]. It appears most beneficial to let the membranes soak in the cleaning solution and then transport it out of the module using as little pressure as possible. However, usually a compromise between pressure and flow velocity has to be made in order to get the best fouling removal efficiency. In UF and MF, cleaning is often enhanced by back pulsation, but this is not a possibility in NF due to the membrane and module structures used. For example, with TFC membranes the active layer would be damaged during backwash due to the lack of a support layer, and air backwash is not possible as air cannot penetrate through the small pores in NF. As a solution to this problem, osmotic backwashing has been suggested and investigated in some studies [7, 357, 358]. When osmotic pressure at feed side exceeds the applied hydraulic pressure across the membrane, osmotic backwashing takes place. There are different ways to accomplish the conditions of osmatic backwash: reducing the pressure of feed side; increasing the pressure of permeate side; or injecting a high concentration solution into the feed channel [357]. In fact, osmotic backwash occurs spontaneously when NF is operated discontinuously [68]. 7.8.4.6

Impact of Cleaning on Membrane Durability

NF membranes are generally somewhat less durable than other types of membranes used. UF and MF membranes can be made of PVDF, Teflon,

349

350

7 Fouling in Nanofiltration

polypropylene, polysulfone, and other very strong and resistant materials. In most cases, it has not been possible to make NF membranes of these materials. Such materials, however, are commonly used as supports for the active layer. NF membranes are most often made of aromatic amides, e.g. poly-(piperazine amide) (see Chapter 2). NF membranes have different resistance to chemicals, heat, and pH. Hence, cleaning processes need to be substantially different [359] for a given polymer type. Membrane durability can be examined by measuring pure water flux and salt retention. A combined increase in flux and decrease in salt retention after cleaning as compared to the virgin state of the membrane most likely indicates reduced membrane integrity. A method to diagnose membrane integrity issues was proposed by Niewersch et al. [360], while Laser-Induced Breakdown-Detection (LIBD) has been developed as an inline tool for membrane integrity monitoring [361]. Apart from membrane integrity, membrane aging needs to be considered when developing optimized cleaning/disinfection protocols for membranes. Membrane aging is defined as “aging of the materials which constitute the membrane” [362]. The results of membrane aging are to increase the frequency of cleaning step, changing the physical–chemical properties of the membrane such as elasticity and surface zeta potential, changing of the membrane selectivity, and loss of membrane integrity [131, 362]. Acid/alkali resistance: The alkaline resistance of normal NF membranes goes to about pH 11 and the acid resistance to around pH 1 for the best membranes, depending on the material. In many cases, NF membranes can withstand higher or lower pH values for shorter times, especially if their temperature limits are not exceeded. In fact, a high pH cleaning can often increase the membrane capacity because the high pH modifies the membrane to give a higher flux without a decrease in retention [363]. However, Simon et al. [364] showed that of three different NF membranes, NF270, NF90, and TFC-SR100, the NF270 was the one most influenced by alkaline cleaning. Cleaning of NF270 by alkaline solution resulted in a significant increase in the permeability and a high decrease in the rejection of inorganic salts and trace organic contaminants, while acidic cleaning of NF270 caused only a small decrease in its water permeability. This was attributed to the loose structure of NF270 membrane with a thinner active layer compared to the other two studied membranes. Similar results were obtained by Kallioinen et al. [365] using Desal-5 DL, XN45, and NF270 membranes. A short time exposure of polyamide NF to alkaline conditions causes a rapid decrease in retentions of glucose and MgSO4 and by extending the exposure time a subsequent steady decrease was observed. Many industries use NF membranes for fractionation of their process or effluent streams. Th obstacle has been the high or low pH, which the membranes cannot tolerate. For this reason, there are several incentives to manufacture NF membranes that can operate in a broader pH range, aiming at an increase in 1–2 units in each direction [366]. The durability of the membranes is checked by characterization of retention of glucose/sucrose and salts and by inspection of the surfaces for cracks by using SEM.

7.8 Fouling Prevention and Cleaning

120 XN-40 Desal-5 DL TS-80

Flux (l/m2 h)

100

1

80

2

60

3

40 20 0 10

20

30 40 50 Temperature (°C)

60

70

Figure 7.30 Influence of temperature on fluxes during the filtration of 250 mg/L glucose solution. (1) Temperature was increased from 24 to 65  C, (2) temperature was decreased from 65 to 37  C, and (3) change in flux after alkaline cleaning. Source: Adapted from Mänttäri et al. 2002 [367].

Temperature resistance: NF membranes are not normally very tolerant to heat – up to a limit. Most NF membranes can endure around 40  C and according to some reports their stability extends to 50–60  C. It is of great importance that the membranes can tolerate at least 50–70  C in processes that demand elevated temperatures, and also in cleaning, which would allow them to be sanitized. Some studies have been carried out on the heat stability of typical NF membranes, as summarized in Figure 7.30 [367]. Generally, with increasing temperature flux increases while retention decreases (retention not shown). In some cases, decreasing temperature causes the membranes to become tighter (even after a short heat treatment at 65  C), and hence the flux of such membranes decreases and retention increases (see, e.g. membrane XN-40 in Figure 7.28). Alkaline cleaning then again increases flux. 7.8.5 Examples of Cleaning Applications and Cleaning Process Protocols It is best to divide the cleaning procedures according to what foulants are to be removed or the types of process streams that are filtered. Hence, the cleaning protocols used by a number of example applications is described in Sections 7.8.5.1, 7.8.5.2, and 7.8.5.3. In industrial processes, cleaning is generally performed as a CIP procedure that commences automatically either at set time intervals, when transmembrane pressure in constant flux applications reaches a critical limit or when flux has decreased below the tolerance level in constant pressure filtration. 7.8.5.1

Food Industry

In the food industry, especially in the dairy industry, there are regulations that membranes should be cleaned and sanitized daily. Th s requirement comes from

351

352

7 Fouling in Nanofiltration

the fact that these food products are very sensitive to microbial growth. Owing to this cleaning requirement, membranes in such applications are operated at higher fluxes as fouling prevention is not a priority. In the dairy industry the foulants are mainly proteins, salts, and sugar or their degradation products. For removal of foulants an alkaline cleaning cycle is needed and for salt removal (calcium salts) an acid cleaning step. Normally, this is carried out in three steps: alkaline, acid, and alkaline. Th protocol varies from process to process but it is often more or less standardized also including sanitation. An example of such a cleaning procedure in the food industry (soy flour extract) was given in Figure 7.29 [39]. Most membranes are cleaned according to a CIP procedure [368–371]. Alternatively, enzymatic cleaning has been trialed, but was described as having a low cleaning efficiency and long cleaning times. Some proteins are not removed with enzymes [336]. 7.8.5.2

Water and Wastewater Treatment

In most typical water and wastewater applications cleaning is not carried out on a daily basis and the NF process aims at as few cleanings as possible, which means that the processes are run at lower average fluxes. In cases where the waters contain some organics such as NOM or humic acids, usually alkaline cleaning is required. For example, Li and Elimelech [142] have investigated a number of cleaning agents to remove deposits of HA and calcium. It was found that EDTA was most effective and recovered 100% of the flux when applied at pH 11, while NaOH was ineffective. SDS was effective independent of pH when applied above its critical micelle concentration (CMC). In contrast, Lee et al. [41] found that SDS was ineffective for NOM foulants whereas 0.1 M NaCl was relatively effective compared to more common cleaning agents such as surfactants. Caustic solutions were effective at removing hydrophobic foulants, while hydrophilic NOM fractions were more difficult to remove. In an extensive study determining the efficiency of 13 cleaning schemes, Liikanen et al. [335] determined that Na4 EDTA was very effective, although it was pointed out that this may be dependent on the membrane type, which implies that every membrane may require a cleaning optimization for a particular feedwater. Hong and Elimelech [127] who confirmed the effectiveness of EDTA postulated that EDTA removes the calcium from a solution and in this way reduces (or in the specific case reverses) fouling. Th s makes EDTA an effective agent not only for cleaning but also for pretreatment. Roudman and DiGiano [99] used a commercial inorganic caustic detergent (MC-3) at pH 10.3 and ultrapure water rinses, which could not remove NOM deposits. 7.8.5.3

Desalination and Other Industries

In desalination acid cleaning is the most important. Often the alkaline and the acid cycles are not run directly after each other, but maybe one is run more often than the other; the frequency usually depends on pretreatment and local demands. Th cleaning interval is at least one week and in some cases the process can be run without cleaning for several months [335, 348, 349, 372]. One of the highly polluted industrial effluents is olive mill wastewater. Ochando-Pulido et al. [373] developed a successful method to clean a fouled RO

Acknowledgements

membrane used in filtration of pretreated olive mill wastewater. Th method resulting in maximum cleaning efficiency consisted of acid cleaning with citric acid (0.1% [w/v] solution) followed by alkaline-detersive cleaning with NaOH and SDS (solution of 0.1% [w/v] of each reagent) and then performing turbulent tangential velocity over the membrane at temperature ranging from 30 to 35  C for 20–25 minutes. 7.8.6

Regeneration of Cleaning Solutions

Regeneration of cleaning solutions is an important issue, not only because of economic concerns, but also for environmental reasons. Unfortunately, Argüllo et al. [336] have determined that independent of the initial concentration, about 30% of activity is lost in each cleaning cycle for enzyme cleaners used for whey fractionation in UF. In fact, NF was trialed to treat the cleaning solutions to recover and reuse the acid or caustic fractions in the process. Th concentrate would contain the foulants (which, for example, could be regenerated in the dairy industry as animal food) and the permeate would contain, e.g. the alkali/acid and the other parts of the formulated cleaning agent. This permeate would then be concentrated by using RO [374–376].

7.9 Conclusions Th s chapter has provided a comprehensive overview of fouling characteristics, common foulants, fouling characterization, and membrane autopsy as well as a review of current models. A detailed description of the main fouling categories, namely organic fouling, scaling, colloidal and particulate fouling, and biofouling, was followed by a brief description of cleaning methodologies. While fouling has always been a primary part of membrane research and almost always found in the company of effective membrane filtration, it is clearly membrane cleaning where substantial progress will be made in future research.

Acknowledgements Shyam S. Sablani from the Sultan Qaboos University in Oman is acknowledged for providing Figure 7.6. Linda Dudley from Ondeo-Nalco, UK, is thanked for supplying extensive materials on membrane fouling, scaling, and cleaning. Paul Buijs, GEBetz, Belgium, supplied Figure 7.2 and Jack Gilron, Ben Gurion University, Israel, Figure 7.4, which is much appreciated. Dr. Azam Jeihanipour and Yang-Hui Cai are acknowledged for their contribution with the literature update of the chapter, while Dr. Alessandra Imbrogno, Luiza Von Sperling, Rafdian Nahri, Yang-Hui Cai, and Thais La Costa (all IAMT) helped with redrawing of figures.

353

354

7 Fouling in Nanofiltration

Nomenclature A C CB C BL CF CF CP CW D
G

P
𝜋 m * Ds F IAP J JC J0 k K ks k sp M PF Q R RA RC RCP RCW Re RF RG RIF RM RO ROBS RP RRES RRF S t u

membrane surface area equilibrium concentration of the solute in the solution solute concentration in the bulk solution solute concentration in the boundary layer solute concentration in the feed concentration factor or ratio solute concentration in the permeate solute concentration at the membrane/water interface particle diffusion tensor Gibbs free energy osmotic pressure difference across the membrane transmembrane pressure cake-enhanced osmotic pressure solute diffusion coefficient external force vector ion activity product water flux water flux after cleaning initial pure water flux Boltzmann’s constant partitioning coefficient between membrane and solution phase mass transfer coefficient of the solute thermodynamic solubility product of the phase forming compound molecular mass of the solute concentration polarization factor volume flow rate gas constant resistance due to adsorption resistance due to cake formation resistance due to concentration polarization resistance of the membrane after cleaning Reynolds number total fouling resistance resistance due to gel formation irreversible fouling resistance intrinsic membrane resistance real retention observed retention resistance due to internal pore fouling residual resistance after cleaning reversible fouling resistance supersaturation ratio time duration particle velocity induced by the fluid flow

Abbreviations

V x y

volume distance from the membrane surface permeate recovery fraction

Greek Symbols 𝛼 𝛼O 𝛿
𝜋 m * 𝜀 𝜙T 𝜂T 𝜇 𝜈 𝜏

activity of the solute activity of the solute in its crystal stage boundary layer thickness cake-enhanced osmotic pressure cake layer porosity total interaction potential quantity of organic adsorbed onto the membrane surface viscosity of water at temperature T chemical potential of the solute velocity of water (normal to the membrane surface) membrane tortuosity

Abbreviations ACP AFM AOC ATP ATR-FTIR BFR BSA CWF CA CCC CFU CIP CMC CP DBNPA DCPD DLVO DOC EC EDS EDTA EDX EGDMA EPS FA FR

amorphous calcium phosphate atomic force microscopy assimilable organic carbon adenosinetriphosphate attenuated total reflectance Fourier transform infrared biofilm formation rate bovine serum albumin clean water flux cellulose acetate critical coagulation concentration colony forming unit cleaning in place criticial micelle concentration concentration polarization 2,2-dibromo-3-nitrilopropionamide dicalcium phosphate dihydrate Derjaguin–Landau–Verwey–Overbeek dissolved organic carbon electrical conductivity electron dispersive spectra ethylene diamine tetra acetic acid energy dispersion of X-ray spectroscopy ethylene glycol dimethacrylate extracellular polymeric substance fulvic acid flux reduction

355

356

7 Fouling in Nanofiltration

FESEM FTIR HA HAP HEMA HIM HPC ICP-MS iCVD IEP IR-IRS LIBD LPS LSI MBR MF MFI MIEX MTC MWCO NF NOM NPD NSP NTU OCP OCT PEG PEGDE PGEMA PF PG PWFb PVDF RMS RO SAD SCMP SDI SDS SEM SUVA TCF TDC TDS TEM TFC

field emission scanning electron microscopy Fourier transform infrared humic acid hydroxyapatite hydroxymethyl ester of methacrylic acid helium ion microscopy heterotropic plate count inductively coupled mass spectrometry initiated chemical vapour deposition isoelectric point infrared internal reflection spectroscopy laser-induced breakdown-detection lipopolysaccharide Langelier saturation index membrane bioreactor microfiltration modified fouling index magnetic ion exchange mass transfer coefficient molecular weight cut-off nanofiltration natural organic matter normalized pressure drop normalized salt passage nephelometric turbidity unit octacalcium phosphate optical coherence tomography poly(ethylene glycol) poly(ethylene glycol) diglycidyl ether polyethylene glycol ester of methacrylic acid permeate flux polygalacturonase pure water flux before polyvinylidene difluoride root mean square reverse osmosis surface area difference soluble and colloidal microbial product silt density index sodium dodecyl sulfate scanning electron microscopy specific UV absorbance temperature correction factor total direct cell count total dissolved solid transmission electron microscopy thin-film composite

References

TOC TS TSIA UF XRD XPS ZP

total organic carbon total solids time-series image analysis ultrafiltration X-ray diffractometry X-ray photoelectron spectroscopy zeta potential

References 1 Koros, W.J., Ma, Y.H., and Shimizu, T. (1996). Terminology for membranes

2

3 4

5 6 7

8 9

10

11

12

13

and membrane processes - IUPAC recommendations 1996. J. Membr. Sci. 120: 149–159. Van der Bruggen, B., Braeken, L., and Vandecasteele, C. (2002). Flux decline in nanofiltration due to adsorption of organic compounds. Sep. Purif. Technol. 29: 23–31. Staude, E. (1992). Membranen und Membranprozesse - Grundlagen und Anwendungen. Weinheim: VCH. Schäfer, A.I. (2001). Natural Organic Matter Removal using Membranes: Principles, Performance and Cost. Boca Raton, FL: CRC Press. ISBN: 1-58716-093-5. Huiting, H., Kappelhof, J.W.N.M., and Bosklopper, T.G.J. (2001). Operation of NF/RO plants: from reactive to proactive. Desalination 139: 183–189. Fane, A.G., Beatson, P., and Li, H. (2000). Membrane fouling and its control in environmental applications. Water Sci. Technol. 41: 303–308. Sagiv, A., Avraham, N., Dosoretz, C.G., and Semiat, R. (2008). Osmotic backwash mechanism of reverse osmosis membranes. J. Membr. Sci. 322: 225–233. Grose, A., Smith, A., Donn, A. et al. (1998). Supplying high quality drinking water to remote communities in Scotland. Desalination 117: 107–117. Choi, J.-H., Dockko, S., Fukushi, K., and Yamamoto, K. (2002). A novel application of a submerged nanofiltration membrane bioreactor (NF MBR) for wastewater treatment. Desalination 146: 413–420. Boerlage, S.F.E., Kennedy, M.D., Dickson, M. et al. (2002). The modified fouling index using ultrafiltration membranes (MFI-UF): characterisation, filtration mechanisms and proposed reference membrane. J. Membr. Sci. 197: 1–21. Roorda, J.H. and van der Graaf, J.H.J.M. (2001). New parameter for monitoring fouling during ultrafiltration of WWTP effluent. Water Sci. Technol. 43: 241–248. Mänttäri, M. and Nystrom, M. (2000). Critical flux in NF of high molar mass polysaccharides and effluents from the paper industry. J. Membr. Sci. 170: 257–273. Kilduff, J.E., Mattaraj, S., Pieracci, J.P., and Belford, G. (2000). Photochemical modification of poly(ether sulfone) and sulfonated poly(sulfone)

357

358

7 Fouling in Nanofiltration

14

15 16

17 18

19 20

21 22

23

24

25

26 27

28

29

nanofiltration membranes for control of fouling by natural organic matter. Desalination 132: 133–142. Pendergast, M.T.M., Nygaard, J.M., Ghosh, A.K., and Hoek, E.M.V. (2010). Using nanocomposite materials technology to understand and control reverse osmosis membrane compaction. Desalination 261: 255–263. Braghetta, A., DiGiano, F.A., and Ball, W.P. (1997). Nanofiltraion of natural organic matter: pH and ionic strength effects. J. Environ. Eng. 123: 628–641. Wang, J., Mo, Y., Mahendra, S., and Hoek, E.M.V. (2014). Effects of water chemistry on structure and performance of polyamide composite membranes. J. Membr. Sci. 452: 415–425. DiGiano, F.A., Arweiler, S., and Riddick, A.J.J. (2000). Alternative tests for evaluating NF fouling. J. Am. Water Works Assoc. 92: 103–115. DiGiano, F.A. (1996). Evaluation of natural organic matter fouling in bench-scale, batch recycle tests of nanofiltration. In: ICR Workshop on Bench-scale and Pilot-Scale Evaluations. Cincinnati, OH: AWWA. ASTM-D4516 (2010). Standard practice for standardizing reverse osmosis performance data. ASTM International. ASTM-D6161 (2010). Standard terminology used for microfiltration, ultrafiltration, nanofiltration and reverse osmosis membrane processes. ASTM International. Luo, M. and Wang, Z. (2001). Complex fouling and cleaning-in-place of a reverse osmosis desalination system. Desalination 141: 15–22. Vrouwenvelder, J.S. and van der Kooij, D. (2002). Diagnosis of fouling problems of NF and RO membrane installations by a quick scan. Desalination 153: 121–124. Escobar, I.C. and Randall, A.A. (2001). Assimilable organic carbon (AOC) and biodegradable dissolved organic carbon (BDOC): complementary measurements. Water Res. 35: 4444–4454. Meylan, S., Hammes, F., Traber, J. et al. (2007). Permeability of low molecular weight organics through nanofiltration membranes. Water Res. 41: 3968–3976. Schäfer, A.I., Shen, J., and Richards, B.S. (2018). Renewable energy powered membrane technology: removal of natural organic matter and fluoride in Tanzanian communities. NPJ Nat. Clean Water 24: 1–10. Shaalan, H.F. (2002). Development of fouling control strategies pertinent to nanofiltration membranes. Desalination 153: 125–131. Boerlage, S.F.E., Kennedy, M.D., Aniye, M.P. et al. (2000). Modified fouling indexUltrafiltration to compare pretreatment processes of reverse osmosis feedwater. Desalination 131: 201–214. Fonseca, A.C., Summers, R.S., Greenberg, A.R., and Hernandez, M.T. (2007). Extra-cellular polysaccharides, soluble microbial products, and natural organic matter impact on nanofiltration membranes flux decline. Environ. Sci. Technol. 41: 2491–2497. Koo, C.H., Mohammad, A.W., Suja, F., and Meor Talib, M.Z. (2013). Use and development of fouling index in predicting membrane fouling. Sep. Purif. Rev. 42: 296–339.

References

30 ASTM D8002 (2015). Standard test method for modified fouling index

(MFI-0.45) of water. ASTM International.

31 Reiss, C.R. and Taylor, J.S. (1995). Prediction of nanofilter productivity

32

33 34

35

36 37

38 39

40 41

42 43

44

45

46

decline using fouling indices. In: Membrane Technology Conference (ed. AWWA), 433–448. Reno, NV: AWWA. Van der Kooij, D., Veenendaal, H.R., Baars-Lorist, C. et al. (1995). Biofilm formation on surfaces of glass and teflon exposed to treated water. Water Res. 29: 1655–1662. Hambsch, B. (1992). Untersuchungen zu mikrobiellen Abbauvorgängen bei der Uferfiltration, in. Karlsruhe: Universität Karsruhe. Vrouwenvelder, J.S. and van der Kooij, D. (2001). Diagnosis, prediction and prevention of biofouling of NF and RO membranes. Desalination 139: 65–71. Sadr Ghayeni, S.B., Beatson, P.J., Schneider, R.P., and Fane, A.G. (1998). Adhesion of waste water bacteria to reverse osmosis membranes. J. Membr. Sci. 138: 29–42. Carlson, G. and Silverstein, J. (1998). Effect of molecular size and charge on biofilm sorption of organic matter. Water Res. 32: 1580–1592. Gwon, E.-m., Yu, M.-j., Oh, H.-k., and Ylee, Y.-h. (2003). Fouling characteristics of NF and RO operated for removal of dissolved matter from groundwater. Water Res. 37: 2989–2997. Nghiem, L.D. and Schäfer, A.I. (2006). Fouling autopsy of hollow-fibre MF membranes in wastewater reclamation. Desalination 188: 113–121. Sayed Razavi, S.K., Harris, J.L., and Sherkat, F. (1996). Fouling and cleaning of membranes in the ultrafiltration of the aqueous extract of soy flour. J. Membr. Sci. 114: 93–104. Cho, B.D. and Fane, A.G. (2002). Fouling transients in normally sub-critical flux operation of a membrane bioreactor. J. Membr. Sci. 209: 391–403. Lee, H., Amy, G., Cho, J. et al. (2001). Cleaning strategies for flux recovery of an ultrafiltration membrane fouled by natural organic matter. Water Res. 35: 3301–3308. Nghiem, D.L. and Schäfer, A.I. (2002). Adsorption and transport of trace contaminant estrone in NF/RO membranes. Environ. Eng. Sci. 19: 441–451. Belfer, S., Fainshtain, R., Daltrophe, Y. et al. (2003). Surface modification of NF membrane to increase fouling resistance in operation on tertiary municipal effluent. In: Israeli Desalination Society Conference (ed. IDS). Haifa Israel. Farooque, A.M., Hassan, A., and Al-Amoudi, A. (1999). Autopsy and characterisation of NF membranes after long-term operation in a NF-SWRO pilot plant. In: IDA World Congress on Desalination and Water Reuse, 77–87. San Diego, CA: International Desalination Association. Butt, F.H., Rahman, F., and Badhuruthamal, U. (1997). Hollow fine fibre vs. spiral-wound reverse osmosis desalination membranes Part 2: Membrane autopsy. Desalination 109: 83–94. Kim, K.J., Chen, V., and Fane, A.G. (1994). Characterization of clean and fouled membranes using metal colloids. J. Membr. Sci. 88: 93–101.

359

360

7 Fouling in Nanofiltration

47 Kim, K.J., Dickson, M.R., Chen, V., and Fane, A.G. (1992). Applications of

48

49

50

51

52

53

54

55 56

57 58

59

60 61

62

field-emission scanning electron-microscopy to polymer membrane research. Micron Microscopia Acta 23: 259–271. Imbrogno, A., Tiraferri, A., Abbenante, S. et al. (2018). Organic fouling control through magnetic ion exchange-nanofiltration (MIEX-NF) in water treatment. J. Membr. Sci. 549: 474–485. Rabiller-Baudry, M., Le Maux, M., Chaufer, B., and Begoin, L. (2002). Characterisation of cleaned and fouled membrane by ATR-FTIR and EDX analysis coupled with SEM: application of UF of skimmed milk with a PES membrane. Desalination 146: 123–128. Bowen, W.R., Hilal, N., Lovitt, R.W., and Wright, C.J. (1998). Direct measurement of interactions between adsorbed protein layers using an atomic force microscope. J. Colloid Interface Sci. 197: 348–352. Bowen, W.R. and Doneva, T.A. (2000). Atomic force microscopy studies of membranes: effect of surface roughness on double-layer interactions and particle adhesion. J. Colloid Interface Sci. 229: 544–549. Bowen, W.R., Doneva, T.A., and Stoton, A.G. (2002). The use of atomic force microscopy to quantify membrane surface electrical properties. Colloids Surf., A 201: 73–83. Jarusutthirak, C., Amy, G., and Croue, J.-P. (2002). Fouling characteristics of wastewater effluent organic matter (EfOM) isolates on NF and UF membranes. Desalination 145: 247–255. Carlsson, D.J., Dal-Chin, M.M., Black, P., and Lick, C.N. (1998). A surface spectroscopic study of membranes fouled by pulp mill effluent. J. Membr. Sci. 142: 1–11. Nyström, M., Kaipia, L., and Luque, S. (1995). Fouling and retention of nanofiltration membranes. J. Membr. Sci. 98: 249–262. Mänttäri, M., Puro, L., Nuortila-Jokinen, J., and Nyström, M. (2000). Fouling effects of polysaccharides and humic acid in nanofiltration. J. Membr. Sci. 165: 1–17. Carman, P.C. (1938). Fundamental principles of industrial filtration - a critical review of present knowledge. Trans. IChemE 16: 168–188. Bowen, W.R. and Jenner, F. (1995). Theoretical descriptions of membrane filtration of colloids and fine particles - an assessment and review. Adv. Colloid Interface Sci. 56: 141–200. Wijmans, J.G., Nakao, S., van den Berg, F.R. et al. (1985). Hydrodynamic resistance of concentration polarization boundary layers in ultrafiltration. J. Membr. Sci. 22: 117–134. Ho, W.S.W. and Sirkar, K.K. (1992). Membrane Handbook. New York: Chapman & Hall. ISBN: 0-412-98871-2. Sherwood, T.K., Brian, P.L.T., Fisher, R.E., and Dresner, L. (1965). Salt concentration at phase boundaries in desalination by reverse osmosis. Ind. Eng. Chem. Fundam. 4: 113–118. Sablani, S.S., Goosen, M.F.A., Al-Belushi, R., and Wilf, M. (2001). Concentration polarization in ultrafiltration and reverse osmosis: a critical review. Desalination 141: 269–289.

References

63 Guillen, G. and Hoek, E.M.V. (2009). Modeling the impacts of feed spacer

64

65 66 67 68

69

70 71

72 73

74

75

76

77

78

79

geometry on reverse osmosis and nanofiltration processes. Chem. Eng. J. 149: 221–231. Nikolova, J.D. and Islam, M.A. (1998). Contribution of adsorbed layer resistance to the flux decline in an ultrafiltration process. J. Membr. Sci. 146: 105–111. Koyuncu, I. and Topacik, D. (2002). Effect of organic ion on the separation of salts by nanofiltration membranes. J. Membr. Sci. 195: 247–263. Avraham, N., Dosoretz, C., and Semiat, R. (2006). Osmotic backwash process in RO membranes. Desalination 199: 387–389. Sagiv, A. and Semiat, R. (2005). Backwash of RO spiral wound membranes. Desalination 179: 1–9. Cai, Y.-H. and Schäfer, A.I. (2019). Renewable energy powered membrane technology: impact of solar irradiance fluctuation on direct osmotic backwash. J. Membr. Sci. 598: 117666. Dana, A., Hadas, S., and Ramon, G.Z. (2019). Potential application of osmotic backwashing to brackish water desalination membranes. Desalination 468: 114029. Bacchin, P., Aimar, P., and Field, R.W. (2006). Critical and sustainable fluxes: theory, experiments and applications. J. Membr. Sci. 281: 42–69. Van der Bruggen, B., Daems, B., Wilms, D., and Vandecasteele, C. (2001). Mechanisms of retention and flux decline for the nanofiltration of dye baths from the textile industry. Sep. Purif. Technol. 22–23: 519–528. Combe, C., Molis, E., Lucas, P. et al. (1999). The effect of CA membrane properties on adsorptive fouling by humic acid. J. Membr. Sci. 154: 73–87. van der Bruggen, B. and Vandecasteele, C. (2001). Flux decline during nanofiltration of organic components in aqueous solution. Environ. Sci. Technol. 35: 3535–3540. Tang, C.Y. and Leckie, J.O. (2007). Membrane independent limiting flux for RO and NF membranes fouled by humic acid. Environ. Sci. Technol. 41: 4767–4773. Tang, C.Y., Kwon, Y.-N., and Leckie, J.O. (2009). The role of foulant–foulant electrostatic interaction on limiting flux for RO and NF membranes during humic acid fouling—theoretical basis, experimental evidence, and AFM interaction force measurement. J. Membr. Sci. 326: 526–532. Tang, C.Y., Kwon, Y.-N., and Leckie, J.O. (2007). Fouling of reverse osmosis and nanofiltration membranes by humic acid—effects of solution composition and hydrodynamic conditions. J. Membr. Sci. 290: 86–94. Belfort, G., Davis, R.H., and Zydney, A.L. (1994). The behavior of suspensions and macromolecular solutions in crossflow microfiltration: review. J. Membr. Sci. 96: 1–58. Chang, Y.-J. and Benjamin, M.M. (2003). Modeling formation of natural organic matter fouling layers on ultrafiltration membranes. J. Environ. Eng.: 25–32. Field, R.W., Wu, D., Howell, J.A., and Gupta, B.B. (1995). Critical flux concept for microfiltration fouling. J. Membr. Sci. 100 (3), 259–272.

361

362

7 Fouling in Nanofiltration

80 Van der Bruggen, B., Mänttäri, M., and Nyström, M. (2008). Drawbacks of

81

82

83

84

85

86 87

88

89 90

91

92

93

94

applying nanofiltration and how to avoid them: a review. Sep. Purif. Technol. 63: 251–263. Elimelech, M., Gregory, J., Jia, X., and Williams, R. (1995). Particle Deposition and Aggregation: Measurement, Modeling, and Simulation. Jordan Hill, Oxford: Butterworth-Heinemann Ltd. Hoek, E.M.V., Kim, A.S., and Elimelech, M. (2004). Influence of crossflow membrane filter geometry and shear rate on colloidal fouling in reverse osmosis and nanofiltration separations. Environ. Eng. Sci. 19: 357–372. Hoek, E.M.V. (2002). Colloidal fouling mechanisms in reverse osmosis and nanofiltration. In: Chemical Engineering, 207. New Haven, CT: Yale University. PhD thesis. Hoek, E.M.V., Kim, A.S., and Elimelech, M. (2002). Influence of crossflow membrane filter geometry and shear rate on colloidal fouling in reverse osmosis and nanofiltration separations. Environ. Eng. Sci. 19: 357–372. Hoek, E.M.V. and Elimelech, M. (2002). Cake-enhanced concentration polarization: a major fouling mechanism for salt rejecting membranes. Environ. Sci. Technol. 37: 5581–5588. Boudreau, B.P. (1996). The diffusive tortuosity of fine-grained unlithified sediments. Geochim. Cosmochim. Acta 60: 3139–3142. Hoek, E.M.V. and Elimelech, M. (2003). Cake-enhanced concentration polarization: a new fouling mechanism for salt rejecting membranes. Environ. Sci. Technol. 37: 5581–5588. Gutman, J., Fox, S., and Gilron, J. (2012). Interactions between biofilms and NF/RO flux and their implications for control—a review of recent developments. J. Membr. Sci. 421–422: 1–7. Tho pson, J., Lin, N., Lyster, E. et al. (2012). RO membrane mineral scaling in the presence of a biofilm. J. Membr. Sci. 415–416: 181–191. Radu, A.I., Vrouwenvelder, J.S., van Loosdrecht, M.C.M., and Picioreanu, C. (2010). Modeling the effect of biofilm formation on reverse osmosis performance: flux, feed channel pressure drop and solute passage. J. Membr. Sci. 365: 1–15. Herzberg, M. (2010). Osmotic effects of biofouling in reverse osmosis (RO) processes: physical and physiological measurements and mechanisms. Desalin. Water Treat. 15: 287–291. Vrouwenvelder, J.S., Picioreanu, C., Kruithof, J.C., and van Loosdrecht, M.C.M. (2010). Biofouling in spiral wound membrane systems: three-dimensional CFD model based evaluation of experimental data. J. Membr. Sci. 346: 71–85. Picioreanu, C., Vrouwenvelder, J.S., and van Loosdrecht, M.C.M. (2009). Th ee-dimensional modeling of biofouling and fluid dynamics in feed spacer channels of membrane devices. J. Membr. Sci. 345: 340–354. Herzberg, M., Kang, S., and Elimelech, M. (2009). Role of extracellular polymeric substances (EPS) in biofouling of reverse osmosis membranes. Environ. Sci. Technol. 43: 4393–4398.

References

95 Chong, T.H., Wong, F.S., and Fane, A.G. (2008). The effect of imposed flux

96

97

98

99

100 101

102 103

104

105

106

107

108

109 110

on biofouling in reverse osmosis: role of concentration polarisation and biofilm enhanced osmotic pressure phenomena. J. Membr. Sci. 325: 840–850. Huertas, E., Herzberg, M., Oron, G., and Elimelech, M. (2008). Influence of biofouling on boron removal by nanofiltration and reverse osmosis membranes. J. Membr. Sci. 318: 264–270. Herzberg, M. and Elimelech, M. (2008). Physiology and genetic traits of reverse osmosis membrane biofilms: a case study with Pseudomonas aeruginosa. ISME J. 2: 180–194. Kim, A.S. and Chen, H. (2006). Diffusive tortuosity factor of solid and soft cake layers: a random walk simulation approach. J. Membr. Sci. 279: 129–139. Roudman, A.R. and DiGiano, F.A. (2000). Surface energy of experimental and commercial nanofiltration membranes: effects of wetting and natural organic matter fouling. J. Membr. Sci. 175: 61–73. Tiller, C.L. and O’Melia, C.R. (1993). Natural organic matter and colloidal stability: models and measurements. Colloids Surf., A 73: 89–102. Beckett, R. and Le, N.P. (1990). The role of organic matter and ionic composition in determining the surface charge of suspended particles in natural waters. Colloids Surf. 44: 35–49. Al-Amoudi, A.S. (2010). Factors affecting natural organic matter (NOM) and scaling fouling in NF membranes: a review. Desalination 259: 1–10. Jarusutthirak, C. and Amy, G. (2001). Membrane filtration of wastewater effluents for reuse: effluent organic matter rejection and fouling. Water Sci. Technol. 43: 225–232. Huang, H., Lee, N., Young, T. et al. (2007). Natural organic matter fouling of low-pressure, hollow-fiber membranes: effects of NOM source and hydrodynamic conditions. Water Res. 41: 3823–3832. Wiesner, M.R., Cark, M.M, Jacangelo, J.G., et al. (1992). Committee report: membrane processes in potable water treatment. J. Am. Water Works Assoc. 84: 59–67. Lee, S., Ang, W.S., and Elimelech, M. (2006). Fouling of reverse osmosis membranes by hydrophilic organic matter: implications for water reuse. Desalination 187: 313–321. Ang, W.S. and Elimelech, M. (2007). Protein (BSA) fouling of reverse osmosis membranes: implications for wastewater reclamation. J. Membr. Sci. 296: 83–92. Kimura, K., Kakuda, T., and Iwasaki, H. (2019). Membrane fouling caused by lipopolysaccharides: a suggestion for alternative model polysaccharides for MBR fouling research. Sep. Purif. Technol. 223: 224–233. Huber, S.A. (1998). Organics - the value of chromatographic characterisation of TOC in process water plants. Ultrapure Water: 16–20. Schäfer, A.I., Mauch, R., Fane, A.G., and Waite, T.D. (2002). Charge effects in the fractionation of natural organics using ultrafiltration. Environ. Sci. Technol. 36: 2572–2580.

363

364

7 Fouling in Nanofiltration

111 Yamamura, H., Okimoto, K., Kimura, K., and Watanabe, Y. (2014).

112

113

114

115

116

117 118

119

120

121

122

123

124

125

126

Hydrophilic fraction of natural organic matter causing irreversible fouling of microfiltration and ultrafiltration membranes. Water Res. 54: 123–136. Lee, S. and Elimelech, M. (2006). Relating organic fouling of reverse osmosis membranes to intermolecular adhesion forces. Environ. Sci. Technol. 40: 980–987. Chang, I.-S. and Lee, C.-H. (1998). Membrane filtration characteristics in membrane-coupled activated sludge system - the effect of physiological states of activated sludge on membrane fouling. Desalination 120: 221–233. Amy, G. and Cho, J. (1999). Interactions between natural organic matter (NOM) and membranes: rejection and fouling. Water Sci. Technol. 40: 131–139. Mackey, E.D. (1999). Fouling of ultrafiltration and nanofiltration membranes by dissolved organic matter. In: Environmental Science and Engineering. Houston, TX: Rice University. PhD thesis. Khan, M.T., Manes, C.-L.de.O., Aubry, C. et al. (2013). Kinetic study of seawatesr reverse osmosis membrane fouling. Environ. Sci. Technol. 47: 10884–10894. Ghosh, K. and Schnitzer, M. (1980). Macromolecular structures of humic substances. Soil Sci. 129: 266–276. Monruedee, M., Sarp, S., Lee, Y.G., and Kim, J.H. (2012). Time-series image analysis for investigating SWRO fouling mechanism. Desalin. Water Treat. 43: 212–220. Champlin, T.L. (2000). Using circulation tests to model natural organic matter adsorption and particle deposition by spiral-wound nanofiltration membrane elements. Desalination 131: 105–115. Bouchard, C.R., Jolicoeur, J., Kouadio, P., and Britten, M. (1997). Study of humic acid adsorption on nanofiltration membranes by contact angle measurements. Canad. J. Chem. Eng. 75: 339–345. Chon, K. and Cho, J. (2016). Fouling behavior of dissolved organic matter in nanofiltration membranes from a pilot-scale drinking water treatment plant: an autopsy study. Chem. Eng. J. 295: 268–277. Elimelech, M., Chen, W.H., and Waypa, J.J. (1994). Measuring the zeta (electrokinetic) potential of reverse osmosis membranes by a streaming potential analyzer. Desalination 95: 269–286. Childress, A.E. and Deshmukh, S.S. (1998). Effect of humic substances and anionic surfactants on the surface charge and performance of reverse osmosis membranes. Desalination 118: 167–174. Childress, A.E. and Elimelech, M. (1996). Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes. J. Membr. Sci. 119: 253–268. Contreras, A.E., Kim, A., and Li, Q. (2009). Combined fouling of nanofiltration membranes: mechanisms and effect of organic matter. J. Membr. Sci. 327: 87–95. Schäfer, A.I., Fane, A.G., and Waite, T.D. (1998). Nanofiltration of natural organic matter: removal, fouling and the influence of multivalent ions. Desalination 118: 109–122.

References

127 Hong, S. and Elimelech, M. (1997). Chemical and physical aspects of natural

128

129

130

131

132

133

134

135

136

137

138

139

140

141

organic matter (NOM) fouling of nanofiltration membranes. J. Membr. Sci. 132: 159–181. Jones, K.L. and O’Melia, C.R. (2001). Ultrafiltration of protein and humic substances: effect of solution chemistry on fouling and flux decline. J. Membr. Sci. 193: 163–173. Yamamura, H., Kimura, K., Okajima, T. et al. (2008). Affinity of functional groups for membrane surfaces: implications for physically irreversible fouling. Environ. Sci. Technol. 42: 5310–5315. Gutierrez, L., Aubry, C., Valladares Linares, R., and Croue, J.-P. (2015). Natural organic matter interactions with polyamide and polysulfone membranes: formation of conditioning film. Colloids Surf., A 477: 1–8. Robinson, S., Abdullah, S.Z., Bérubé, P., and Le-Clech, P. (2016). Ageing of membranes for water treatment: linking changes to performance. J. Membr. Sci. 503: 177–187. Dreszer, C., Wexler, A.D., Drusová, S. et al. (2014). In-situ biofilm characterization in membrane systems using Optical Coherence Tomography: formation, structure, detachment and impact of flux change. Water Res. 67: 243–254. Seidel, A. and Elimelech, M. (2002). Coupling between chemical and physical interactions in natural organic matter (NOM) fouling of nanofiltration membranes: implications for fouling control. J. Membr. Sci. 203: 245–255. Wang, X.-m. and Waite, T.D. (2009). Role of gelling soluble and colloidal microbial products in membrane fouling. Environ. Sci. Technol. 43: 9341–9347. Winter, J., Uhl, W., and Bérubé, P.R. (2016). Integrated oxidation membrane filtration process – NOM rejection and membrane fouling. Water Res. 104: 418–424. Luo, J., Ding, L., Wan, Y. et al. (2012). Fouling behavior of dairy wastewater treatment by nanofiltration under shear-enhanced extreme hydraulic conditions. Sep. Purif. Technol. 88: 79–86. Park, J., Lee, S., You, J. et al. (2018). Evaluation of fouling in nanofiltration for desalination using a resistance-in-series model and optical coherence tomography. Sci. Total Environ. 642: 349–355. Lindau, J. and Jönsson, A.-S. (1999). Adsorptive fouling of modified and unmodified commercial polymeric ultrafiltration membranes. J. Membr. Sci. 160: 65–76. Cho, J., Amy, G., and Pellerino, J. (1999). Membrane filtration of natural organic matter: initial comparison of rejection and flux decline characteristics with ultrafiltration and nanofiltration membranes. Water Res. 33: 2517–2526. Nghiem, L.D., Vogel, D., and Khan, S. (2008). Characterising humic acid fouling of nanofiltration membranes using bisphenol A as a molecular indicator. Water Res. 42: 4049–4058. Ahn, W.-Y., Kalinichev, A.G., and Clark, M.M. (2008). Effects of background cations on the fouling of polyethersulfone membranes by natural organic

365

366

7 Fouling in Nanofiltration

142 143

144 145

146 147

148

149

150

151

152

153 154 155 156 157 158

matter: experimental and molecular modeling study. J. Membr. Sci. 309: 128–140. http://www.yale.edu/env/elimelech/index_research.htm (accessed 15 Janauary 2004). Ang, W.L., Mohammad, A.W., Benamor, A., and Hilal, N. (2016). Hybrid coagulation–NF membrane processes for brackish water treatment: effect of pH and salt/calcium concentration. Desalination 390: 25–32. Wall, N.A. and Choppin, G.R. (2003). Humic acids coagulation: influence of divalent cations. Appl. Geochem. 18: 1573–1582. Schäfer, A., Fane, A., and Waite, D. (2000). Direct coagulation pretreatment in nanofiltration of waters rich in organic matter and calcium. Desalination 131: 215–224. Taheri, A.H., Sim, L.N., Haur, C.T. et al. (2013). The fouling potential of colloidal silica and humic acid and their mixtures. J. Membr. Sci. 433: 112–120. Li, Q. and Elimelech, M. (2006). Synergistic effects in combined fouling of a loose nanofiltration membrane by colloidal materials and natural organic matter. J. Membr. Sci. 278: 72–82. Kimura, K., Okazaki, S., Ohashi, T., and Watanabe, Y. (2016). Importance of the co-presence of silica and organic matter in membrane fouling for RO filtering MBR effluent. J. Membr. Sci. 501: 60–67. Teixeira, M.R. and Sousa, V.S. (2013). Fouling of nanofiltration membrane: effects of NOM molecular weight and microcystins. Desalination 315: 149–155. Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J. et al. (2009). Influence of membrane fouling by (pretreated) surface water on rejection of pharmaceutically active compounds (PhACs) by nanofiltration membranes. J. Membr. Sci. 330: 90–103. Jin, X., Jawor, A., Kim, S., and Hoek, E.M.V. (2009). Effects of feed water temperature on separation performance and organic fouling of brackish water RO membranes. Desalination 239: 346–359. Plakas, K.V., Karabelas, A.J., Wintgens, T., and Melin, T. (2006). A study of selected herbicides retention by nanofiltration membranes—the role of organic fouling. J. Membr. Sci. 284: 291–300. Hasson, D., Drak, A., and Semiat, R. (2001). Inception of CaSO4 scaling on RO membranes at various water recovery levels. Desalination 139: 73–81. Cowan, J.C. and Weintritt, D.J. (1976). Water-Formed Scale Deposits. Houston, TX: Gulf Publ. Co. Faller, K.A. (1999). AWWA Manual M46 - Reverse Osmosis and Nanofiltration. In: AWWA. Rautenbach, R. and Albrecht, R. (1989). Membrane Processes. New York: Wiley. Mullin, J.W. (1992). Crystallisation, 3e. London: Butterworths. Dirksten, J.A. and Ring, T.A. (1991). Fundamentals of crystallization: kinetic effects on particle size distributions and morphology. Chem. Eng. Sci. 46: 2389–2427.

References

159 Sutzkover, I., Hasson, D., and Semiat, R. (2000). Simple technique for mea-

160

161

162

163

164

165

166

167

168

169 170 171 172

173 174

suring the concentration polarization level in a reverse osmosis system. Desalination 131: 117–127. Koutsou, C.P., Yiantsios, S.G., and Karabelas, A.J. (2009). A numerical and experimental study of mass transfer in spacer-filled channels: effects of spacer geometrical characteristics and Schmidt number. J. Membr. Sci. 326: 234–251. Koutsou, C.P., Yiantsios, S.G., and Karabelas, A.J. (2007). Direct numerical simulation of flow in spacer-filled channels: effect of spacer geometrical characteristics. J. Membr. Sci. 291: 53–69. van de Lisdonk, C.A.C., Rietman, B.M., Heijman, S.G.J. et al. (2001). Prediction of supersaturation and monitoring of scaling in reverse osmosis and nanofiltration membrane systems. Desalination 138: 259–270. Zhang, Z., Bright, V.M., and Greenberg, A.R. (2006). Use of capacitive microsensors and ultrasonic time-domain reflectometry for in-situ quantification of concentration polarization and membrane fouling in pressure-driven membrane filtration. Sens. Actuators, B 117: 323–331. Lyster, E., Au, J., Rallo, R. et al. (2009). Coupled 3-D hydrodynamics and mass transfer analysis of mineral scaling-induced flux decline in a laboratory plate-and-frame reverse osmosis membrane module. J. Membr. Sci. 339: 39–48. Uchymiak, M., Bartman, A.R., Daltrophe, N. et al. (2009). Brackish water reverse osmosis (BWRO) operation in feed flow reversal mode using an ex situ scale observation detector (EXSOD). J. Membr. Sci. 341: 60–66. Cobry, K.D., Yuan, Z., Gilron, J. et al. (2011). Comprehensive experimental studies of early-stage membrane scaling during nanofiltration. Desalination 283: 40–51. Mitrouli, S., Karabelas, A.J., Karanasiou, A., and Kostoglou, M. (2013). Incipient calcium carbonate scaling of desalination membranes in narrow channels with spacers—experimental insights. J. Membr. Sci. 425–426: 48–57. Karabelas, A.J., Karanasiou, A., and Mitrouli, S.T. (2014). Incipient membrane scaling by calcium sulfate during desalination in narrow spacer-filled channels. Desalination 345: 146–157. Smith, R.M., Martell, A.E., and Motekaites, R.J. (1996). NIST critically selected stability constants of metal complexes database. Gaithersburg, MD. van Hoof, S.C.J.M., Minnery, J.G., and Mack, B. (2001). Performing a membrane autopsy. Desalin. Water Reuse 11: 40–46. Koo, T., Lee, Y.J., and Sheikholeslami, R. (2001). Silica fouling and cleaning of reverse osmosis membranes. Desalination 139: 43–56. Marshall, W.L. and Warakomski, J.M. (1980). Amorphous silica solubilities II: effects on aqueous salt solutions at 250 C. Geochim. Cosmochim. Acta 44: 915–924. Butt, F.H., Rahman, F., and Baduruthamal, U. (1997). Characterization of foulants by autopsy of RO desalination membranes. Desalination 114: 51–64. Zach-Maor, A., Semiat, R., Rahardianto, A. et al. (2008). Diagnostic analysis of RO desalting treated wastewater. Desalination 230: 239–247.

367

368

7 Fouling in Nanofiltration

175 Lyster, E., Kim, M.-m., Au, J., and Cohen, Y. (2010). A method for evaluating

176

177 178

179

180 181 182

183

184

185

186 187 188 189

190 191

192 193

antiscalant retardation of crystal nucleation and growth on RO membranes. J. Membr. Sci. 364: 122–131. Bartman, A.R., Lyster, E., Rallo, R. et al. (2011). Mineral scale monitoring for reverse osmosis desalination via real-time membrane surface image analysis. Desalination 273: 64–71. Brusilovsky, M., Borden, J., and Hasson, D. (1992). Flux decline due to gypsum precipitation on RO membranes. Desalination 86: 187–222. Lee, S. and Lee, C.H. (2000). Effect of operating conditions on CaSO4 scale formation mechanism in nanofiltration for water softening. Water Res. 34: 3854–3866. Le Gouellec, Y.A. and Elimelech, M. (2002). Calcium sulfate (gypsum) scaling in nanofiltration of agricultural drainage water. J. Membr. Sci. 205: 279–291. Gilron, J. and Hasson, D. (1987). Calcium sulphate fouling of reverse osmosis membranes: flux decline mechanism. Chem. Eng. Sci. 42: 2351–2360. Söhnel, O. and Garside, J. (1992). Precipitation. Basic Principles and Industrial Applications. Oxford: Butterworth-Heinemann. Le Gouellec, Y.A. and Elimelech, M. (2002). Control of calcium sulfate (gypsum) scale in nanofiltration of saline agricultural drainage water. Environ. Eng. Sci. 19: 387–397. Pahiadaki, T., Tzotzi, C., Andritsos, N. et al. (2005). An experimental study of CaCO3 scaling of RO and NF membranes. Proceedings of the IDA Conference on Desalination and Water Reuse, Paper No. SP05-124, Singapore. Tzotzi, C., Pahiadaki, T., Yiantsios, S.G. et al. (2007). A study of CaCO3 scale formation and inhibition in RO and NF membrane processes. J. Membr. Sci. 296: 171–184. Andritsos, N., Karabelas, A.J., and Koutsoukos, P.G. (1997). Morphology and structure of CaCO3 scale layers formed under isothermal flow conditions. Langmuir 13: 2873–2879. Myers, D. (1999). Surfaces, Interfaces, and Colloids, 2e. New York: Wiley. Hiemenz, P.C. and Rajagopalan, R. (1997). Principles of Colloid and Surface Chemistry, 3e. Monticello, NY: Marcel Dekker. I0824793978. Liew, M.K.H., Fane, A.G., and Rogers, P.L. (1997). Fouling of microfiltration membranes by broth-free antifoam agents. Biotechnol. Bioeng. 56: 89–98. Kelley, S.T. and Zydney, A.L. (1997). Protein fouling during microfiltration: comparitive behavior of different model proteins. Biotechnol. Bioeng. 55: 91–100. Hagen, K. (1998). Removal of particles, bacteria and parasites with ultrafiltration for drinking water treatment. Desalination 119: 85–91. Fu, L.T. and Dempsey, B.A. (1996). Optimizing membrane operations with enhanced coagulation of TOC for drinking water treatment. Abstracts of Papers of the American Chemical Society, 212, 13-ENVR. Porter, J.J. (1998). Recovery of polyvinyl alcohol and hot water from the textile wastewater using thermally stable membranes. J. Membr. Sci. 151: 45–53. Potts, D.E., Ahlert, R.C., and Wang, S.S. (1981). A critical review of fouling of reverse osmosis mebranes. Desalination 35: 235–264.

References

194 Tchobanoglous, G., Darby, J., Bourgeous, K. et al. (1998). Ultrafiltration as an

195

196 197 198

199 200 201 202

203

204

205

206 207 208

209

210 211

advanced tertiary treatment process for municipal wastewater. Desalination 119: 315–321. van Houtte, E., Verbauwhede, J., Vanlerberghe, F. et al. (1998). Treating different types of raw water with micro- and ultrafiltration for further desalination using reverse osmosis. Desalination 117: 49–60. Vera, L., Villarroel-Lopez, R., Delgado, S., and Elmaleh, S. (1997). Cross-flow microfiltration of biologically treated wastewater. Desalination 114: 65–75. Stumm, W. and Morgan, J.J. (1996). Aquatic Chemistry, 1e. New York: Wiley-Interscience. Derjaguin, B.V. and Landau, L.D. (1941). Theory of the stability of strongly charged lyophobic sols and of the adhesion of strongly charged particles in solutions of electrolytes. Acta Physicochim. URSS 14: 733. Verwey, E.J.W. and Overbeek, J.T.G. (1948). Theory of the Stability of Lyophobic Colloids. Amsterdam, NL: Elsevier. Ottewill, R.H. and Shaw, J.N. (1972). Electrophoretic Studies On Polystyrene Latices. J. Electroanal. Chem. 37: 133–142. Bacchin, P., Aimar, P., and Sanchez, V. (1995). Model for colloidal fouling of membranes. AlChE J. 41: 368–376. Waite, D.T., Schafer, A.I., Fane, A.G., and Heuer, A. (1999). Colloidal fouling of ultrafiltration membranes: impact of aggregate structure and size. J. Colloid Interface Sci. 212: 264–274. Zhu, X.H. and Elimelech, M. (1997). Colloidal Fouling of reverse osmosis membranes - measurments and fouling mechanisms. Environ. Sci. Technol. 31: 3654–3662. Hong, S., Faibish, R.S., and Elimelech, M. (1997). Kinetics of permeate flux decline in crossflow membrane filtration of colloidal suspensions. J. Colloid Interface Sci. 196: 267–277. Faibish, R.S., Elimelech, M., and Cohen, Y. (1998). Effect of interparticle electrostatic double layer interactions on permeate flux decline in crossflow membrane filtration of colloidal suspensions: an experimental investigation. J. Colloid Interface Sci. 204: 77–86. Zhu, X.H. and Elimelech, M. (1995). Fouling of reverse-osmosis membranes by aluminum-oxide colloids. J. Environ. Eng., ASCE 121: 884–892. Yiantsios, S.G. and Karabelas, A.J. (1998). The effect of colloid stability on membrane fouling. Desalination 118: 143–152. Kwon, D.Y., Vigneswaran, S., Ngo, H.H., and Shin, H.S. (1997). An enhancement of critical flux in crossflow microfiltration with a pretreatment of floating medium flocculator/prefilter. Water Sci. Technol. 36: 267–274. Abulnour, A.G., Sorour, M.H., and Talaat, H.A. (2003). Comparative economics for desalting of agricultural drainage water (ADW). Desalination 152: 353–357. Reimann, W. (2002). Treatment of agricultural wastewater and reuse. Water Sci. Technol. 46: 177–182. Talaat, H.A., Sorour, M.H., Rahman, N.A., and Shaalan, H.F. (2003). Pretreatment of agricultural drainage water (ADW) for large-scale desalination. Desalination 152: 299–305.

369

370

7 Fouling in Nanofiltration

212 Riedl, K., Girard, B., and Lencki, R.W. (1998). Influence of membrane struc-

213

214

215

216

217

218

219

220

221 222 223 224 225 226 227

228

ture on fouling layer morphology during apple juice clarification. J. Membr. Sci. 139: 155–166. Carter, J.W., Hoyland, G., and Hasting, A.P.M. (1974). Concentration polarisation in reverse osmosis flow systems under laminar conditions. Effect of surface roughness and fouling. Chem. Eng. Sci. 29: 1651–1658. Hirose, M., Ito, H., and Kamiyama, Y. (1996). Effect of skin layer surface structures on the flux behaviour of RO membranes. J. Membr. Sci. 121: 209–215. Kim, J.Y., Lee, H.K., and Kim, S.C. (1999). Surface structure and phase separation mechanism of polysulfone membranes by atomic force microscopy. J. Membr. Sci. 163: 159–166. Knoell, T., Safarik, J., Cormack, T. et al. (1999). Biofouling potentials of microporous polysulfone membranes containing a sulfonated polyether-ethersulfone/polyethersulfone block copolymer: correlation of membrane surface properties with bacterial attachment. J. Membr. Sci. 157: 117–138. Louey, M.D., Mulvaney, P., and Stewart, P.J. (2001). Characterisation of adhesional properties of lactose carriers using atomic force microscopy. J. Pharm. Biomed. Anal. 25: 559–567. Brant, J.A. and Childress, A.E. (2002). Assessing short-range membrane-colloid interactions using surface energetics. J. Membr. Sci. 203: 257–273. Hoek, E.M.V., Hong, S., and Elimelech, M. (2001). Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes. J. Membr. Sci. 188: 115–128. Walker, S.L., Bhattacharjee, S., Hoek, E.M.V., and Elimelech, M. (2002). A novel asymmetric clamping cell for measuring streaming potential of flat surfaces. Langmuir 18: 2193–2198. van Oss, C.J. (1994). Interfacial Forces in Aqueous Media. New York: Marcell Dekker, Inc. Song, L.F. and Elimelech, M. (1995). Particle deposition onto a permeable surface in laminar-flow. J. Colloid Interface Sci. 173: 165–180. Song, L. and Elimelech, M. (1995). Theory of concentration polarization in crossflow filtration. J. Chem. Soc., Faraday Trans. 91: 3389–3398. Cohen, R.D. and Probstein, R.F. (1986). Colloidal fouling of reverse-osmosis membranes. J. Colloid Interface Sci. 114: 194–207. Goren, S.L. (1973). The hydrodynamic force resisting the approach of a sphere to a plane permeable wall. J. Colloid Interface Sci. 69: 78–85. Wiesner, M.R., Clark, M.M., and Mallevialle, J. (1989). Membrane filtration of coagulated suspensions. J. Environ. Eng., ASCE 115: 20–40. Bowen, W.R., Mongruel, A., and Williams, P.M. (1996). Prediction of the rate of cross-flow membrane ultrafiltration: a colloidal interaction approach. Chem. Eng. Sci. 51: 4321–4333. Bowen, W.R. and Sharif, A.O. (1998). Hydrodynamic and colloidal interactions effects on the rejection of a particle larger than a pore in microfiltration and ultrafiltration membranes. Chem. Eng. Sci. 53: 879–890.

References

229 Israelachvili, J.N. (1992). Intermolecular and Surface Forces, 2e. London: Aca-

demic Press.

230 Elimelech, M., Zhu, X., Childress, A.E., and Hong, S. (1997). Role of mem-

231 232 233

234 235

236

237 238

239

240 241

242

243 244

245

brane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes. J. Membr. Sci. 127: 101–109. Flemming, H.C., Schaule, G., Griebe, T. et al. (1997). Biofouling—the Achilles heel of membrane processes. Desalination 113: 215–225. Flemming, H.C., Wingender, J., Szewzyk, U. et al. (2016). Biofilms: an emergent form of bacterial life. Nat. Rev. Microbiol. 14: 563–575. Bergkessel, M., Basta, D.W., and Newman, D.K. (2016). The physiology of growth arrest: uniting molecular and environmental microbiology. Nat. Rev. Microbiol. 14: 549–562. Van Bodegom, P. (2007). Microbial maintenance: a critical review on its quantification. Microb. Ecol. 53: 513–523. Saini, G., Nasholm, N., and Wood, B.D. (2011). Implications of growth and starvation conditions in bacterial adhesion and transport. J. Adhes. Sci. Technol. 25: 2281–2297. Laspidou, C.S. and Rittmann, B.E. (2002). A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass. Water Res. 36: 2711–2720. Flemming, H.C. and Wingender, J. (2010). The biofilm matrix. Nat. Rev. Microbiol. 8: 623–633. Haas, R., Gutman, J., Wardrip, N.C. et al. (2015). Glycosphingolipids enhance bacterial attachment and fouling of nanofiltration membranes. Environ. Sci. Technol. Lett. 2: 43–47. Hou, Y.L., Li, J.X., Gao, Y.N. et al. (2010). Monitoring of CaSO4 deposition behaviors on biofilm during nanofiltration via ultrasonic time-domain reflectometry (UTDR). Water Sci. Technol. 61: 2853–2861. Murphy, A.P., Moody, C.D., Riley, R.L. et al. (2001). Microbiological damage of cellulose acetate RO membranes. J. Membr. Sci. 193: 111–121. Bos, R., Van Der Mei, H.C., and Busscher, H.J. (1999). Physico-chemistry of initial microbial adhesive interactions - its mechanisms and methods for study. FEMS Microbiol. Rev. 23: 179–229. Marshall, K.C., Stout, R., and Mitchell, R. (1971). Mechanism of initial events in the sorption of marine bacteria to surfaces. J. Gen. Microbiol. 68: 337–348. Hermansson, M. (1999). The DLVO theory in microbial adhesion. Colloids Surf., B 14: 105–119. Schneider, R.P. (1996). Conditioning film-induced modification of substratum physicochemistry - analysis by contact angles. J. Colloid Interface Sci. 182: 204–213. Heffernan, R., Habimana, O., Semião, A.J.C. et al. (2014). A physical impact of organic fouling layers on bacterial adhesion during nanofiltration. Water Res. 67: 118–128.

371

372

7 Fouling in Nanofiltration

246 Xavier, J.d.B., Picioreanu, C., and van Loosdrecht, M. (2005). A general

247 248 249

250

251

252

253

254

255 256 257

258

259

260 261

description of detachment for multidimensional modelling of biofilms. Biotechnol. Bioeng. 91: 651–669. Marshall, K.C. (1986). Adsorption and adhesion processes in microbial growth at interfaces. Adv. Colloid Interface Sci. 25: 59–86. Power, K. and Marshall, K.C. (1988). Cellular growth and reproduction of marine bacteria on surface-bound substrate. Biofouling 1: 163–174. Houari, A., Seyer, D., Kecili, K. et al. (2013). Kinetic development of biofilm on NF membranes at the Méry-sur-Oise plant. France, Biofouling 29: 109–118. Schneider, R.P., Ferreira, L.M., Binder, P. et al. (2005). Dynamics of organic carbon and of bacterial populations in a conventional pretreatment train of a reverse osmosis unit experiencing severe biofouling. J. Membr. Sci. 266: 18–29. Ridgway, H.F. (1987). Microbial fouling of reverse osmosis membranes: genesis and control. In: Biological fouling of Industrial Water Systems: A Problem Solving Approach (eds. M.W. Mittelman and G.G. Geesey), 138–193. Water Micro Associates: San Diego, CA. Ridgway, H.F. and Flemming, H.-C. (1996). Membrane biofouling. In: Water Treatment Membrane Processes (eds. J. Mallevialle, P.E. Odendaal and M.R. Wiesner), 6.1–6.62. New York: McGraw Hill. Chen, C.L., Liu, W.T., Chong, M.L. et al. (2004). Community structure of microbial biofilms associated with membrane-based water purification processes as revealed using a polyphasic approach. Appl. Microbiol. Biotechnol. 63: 466–473. Hong, P.Y., Moosa, N., and Mink, J. (2016). Dynamics of microbial communities in an integrated ultrafiltration–reverse osmosis desalination pilot plant located at the Arabian Gulf. Desalin. Water Treat. 57: 16310–16323. Costerton, J.W., Lewandowski, Z., DeBeer, D. et al. (1994). Biofilms, the customized microniche. J. Bacteriol. 176: 2137–2142. Costerton, J.W., Stewart, P.S., and Greenberg, E.P. (1999). Bacterial biofilms: a common cause of persistent infections. Science 284: 1318–1322. Kühl, M. and Jorgensen, B.B. (1992). Microsensor measurements of sulfate reduction and sulfide oxidation in compact microbial communities of aerobic biofilms. Appl. Environ. Microbiol. 58: 1164–1174. Houari, A., Seyer, D., Couquard, F. et al. (2010). Characterization of the biofouling and cleaning efficiency of nanofiltration membranes. Biofouling 26: 15–21. Bar-Zeev, E., Passow, U., Romero-Vargas Castrillón, S., and Elimelech, M. (2015). Transparent exopolymer particles: from aquatic environments and engineered systems to membrane biofouling. Environ. Sci. Technol. 49: 691–707. Leenheer, J.A. and Croué, J.P. (2003). Characterizing aquatic dissolved organic matter. Environ. Sci. Technol. 37: 18A–26A. Nebbioso, A. and Piccolo, A. (2013). Molecular characterization of dissolved organic matter (DOM): a critical review. Anal. Bioanal.Chem. 405: 109–124.

References

262 Amon, R.M.W. and Benner, R. (1996). Bacterial utilization of different size

classes of dissolved organic matter. Limnol. Oceanogr. 41: 41–51.

263 Schneider, R.P., Ferreira, L.M., Binder, P., and Ramos, J.R. (2005). Analysis of

foulant layer in all elements of an RO train. J. Membr. Sci. 261: 152–162.

264 Hong, S., Escobar, I.C., Hershey-Pyle, J. et al. (2005). Biostability character-

265

266

267

268

269

270 271

272

273

274 275

276 277 278

ization in a full-scale hybrid NF/RO treatment system. J. Am. Water Works Assoc. 97: 101–124. Vrouwenvelder, J.S., Manolarakis, S.A., Veenendaal, H.R., and Van Der Kooij, D. (2000). Biofouling potential of chemicals used for scale control in RO and NF membranes. Desalination 132: 1–10. Habimana, O., Semião, A.J.C., and Casey, E. (2014). Upon impact: the fate of adhering pseudomonas fluorescens cells during nanofiltration. Environ. Sci. Technol. 48: 9641–9650. Aiken, G., Kaplan, L.A., and Weishaar, J. (2002). Assessment of relative accuracy in the determination of organic matter concentrations in aquatic systems. J. Environ. Monit. 4: 70–74. Van Der Kooij, D., Visser, A., and Hijnen, W.A.M. (1982). Determining the concentration of easily assimilable organiccarbon in drinking water. J. Am. Water Works Assoc. 74: 540–545. Prest, E.I., Hammes, F., Kötzsch, S. et al. (2016). A systematic approach for the assessment of bacterial growth-controlling factors linked to biological stability of drinking water in distribution systems. Water Sci. Technol.: Water Supply: 865–880. Vrouwenvelder, J.S., Beyer, F., Dahmani, K. et al. (2010). Phosphate limitation to control biofouling. Water Res. 44: 3454–3466. Weinrich, L., LeChevallier, M., and Haas, C.N. (2016). Contribution of assimilable organic carbon to biological fouling in seawater reverse osmosis membrane treatment. Water Res. 101: 203–213. Park, S.K. and Hu, J.Y. (2010). Assessment of the extent of bacterial growth in reverse osmosis system for improving drinking water quality. J. Environ. Sci. Health. Part A Toxic/Hazard. Subst. Environ. Eng. 45: 968–977. Bryers, J.D. and Drummond, F. (1998). Local macromolecule diffusion coefficients in structurally non-uniform bacterial biofilms using fluorescence recovery after photobleaching (FRAP). Biotechnol. Bioeng. 60: 462–473. Stewart, P.S. and Franklin, M.J. (2008). Physiological heterogeneity in biofilms. Nat. Rev. Microbiol. 6: 199. Liu, S., Gunawan, C., Barraud, N. et al. (2016). Understanding, monitoring, and controlling biofilm growth in drinking water distribution systems. Environ. Sci. Technol. 50: 8954–8976. Herzberg, M. and Elimelech, M. (2007). Biofouling of reverse osmosis membranes: role of biofilm-enhanced osmotic pressure. J. Membr. Sci. 295: 11–20. Nadell, C.D., Drescher, K., and Foster, K.R. (2016). Spatial structure, cooperation and competition in biofilms. Nat. Rev. Microbiol. 14: 589–600. Araújo, P.A., Kruithof, J.C., Van Loosdrecht, M.C.M., and Vrouwenvelder, J.S. (2012). The potential of standard and modified feed spacers for biofouling control. J. Membr. Sci. 403-404: 58–70.

373

374

7 Fouling in Nanofiltration

279 Bucs, S.S., Radu, A.I., Lavric, V. et al. (2014). Effect of different commer-

280

281

282

283 284 285

286

287

288 289

290

291

292

293

294

cial feed spacers on biofouling of reverse osmosis membrane systems: a numerical study. Desalination 343: 26–37. Vrouwenvelder, J.S., Graf von der Schulenburg, D.A., Kruithof, J.C. et al. (2009). Biofouling of spiral-wound nanofiltration and reverse osmosis membranes: a feed spacer problem. Water Res. 43: 583–594. Saeed, A., Vuthaluru, R., and Vuthaluru, H.B. (2015). Impact of feed spacer filament spacing on mass transport and fouling propensities of RO membrane surfaces. Chem. Eng. Commun. 202: 634–646. Valladares Linares, R., Fortunato, L., Farhat, N.M. et al. (2016). Mini-review: novel non-destructive in situ biofilm characterization techniques in membrane systems. Desalin. Water Treat. 57: 22894–22901. Dreszer, C., Vrouwenvelder, J.S., Paulitsch-Fuchs, A.H. et al. (2013). Hydraulic resistance of biofilms. J. Membr. Sci. 429: 436–447. Flemming, H.-C. (1995). Biofouling bei Membranprozessen. Berlin: Springer-Verlag. Hoefel, D., Grooby, W.L., Monis, P.T. et al. (2003). Enumeration of water-borne bacteria using viability assays and flow cytometry: a comparison to culture-based techniques. J. Microbiol. Methods 55: 585–597. Hammes, F. and Egli, T. (2010). Cytometric methods for measuring bacteria in water: advantages, pitfalls and applications. Anal. Bioanal.Chem. 397: 1083–1095. Vrouwenvelder, J.S., van Paassen, J.A.M., Kruithof, J.C., and van Loosdrecht, M.C.M. (2009). Sensitive pressure drop measurements of individual lead membrane elements for accurate early biofouling detection. J. Membr. Sci. 338: 92–99. Al-Juboori, R.A. and Yusaf, T. (2012). Biofouling in RO system: mechanisms, monitoring and controlling. Desalination 302: 1–23. Graf von der Schulenburg, D.A., Vrouwenvelder, J.S., Creber, S.A. et al. (2008). Nuclear magnetic resonance microscopy studies of membrane biofouling. J. Membr. Sci. 323: 37–44. Fortunato, L., Bucs, S., Linares, R.V. et al. (2017). Spatially-resolved in-situ quantification of biofouling using optical coherence tomography (OCT) and 3D image analysis in a spacer filled channel. J. Membr. Sci. 524: 673–681. Farhat, N.M., Staal, M., Siddiqui, A. et al. (2015). Early non-destructive biofouling detection and spatial distribution: application of oxygen sensing optodes. Water Res. 83: 10–20. Tung, K.L., Damodar, H.R., Damodar, R.A. et al. (2012). Online monitoring of particle fouling in a submerged membrane filtration system using a photointerrupt sensor array. J. Membr. Sci. 407–408: 58–70. Cui, L., Chen, P., Zhang, B. et al. (2015). Interrogating chemical variation via layer-by-layer SERS during biofouling and cleaning of nanofiltration membranes with further investigations into cleaning efficiency. Water Res. 87: 282–291. Strathmann, M., Mittenzwey, K.H., Sinn, G. et al. (2013). Simultaneous monitoring of biofilm growth, microbial activity, and inorganic deposits on surfaces with an in situ, online, real-time, non-destructive, optical sensor. Biofouling 29: 573–583.

References

295 Janknecht, P. and Melo, L.F. (2003). Online biofilm monitoring. Rev. Environ.

Sci. Biotechnol. 2: 269–283.

296 Vrouwenvelder, J.S., van Paassen, J.A.M., Wessels, L.P. et al. (2006). The

297

298

299

300

301

302

303

304

305 306

307

308 309

310

membrane fouling simulator: a practical tool for fouling prediction and control. J. Membr. Sci. 281: 316–324. Di Martino, P., Doumeche, B., Galas, L. et al. (2007). Assessing chemical cleaning of nanofiltration membranes in a drinking water production plant: a combination of chemical composition analysis and fluorescence microscopy. Water Sci. Technol.: 219–225. Xu, P., Bellona, C., and Drewes, J.E. (2010). Fouling of nanofiltration and reverse osmosis membranes during municipal wastewater reclamation: membrane autopsy results from pilot-scale investigations. J. Membr. Sci. 353: 111–121. Jeong, S., Naidu, G., Vollprecht, R. et al. (2016). In-depth analyses of organic matters in a full-scale seawater desalination plant and an autopsy of reverse osmosis membrane. Sep. Purif. Technol. 162: 171–179. Pontié, M., Rapenne, S., Th kkedath, A. et al. (2005). Tools for membrane autopsies and antifouling strategies in seawater feeds: a review. Desalination 181: 75–90. Dixon, M.B., Lasslett, S., and Pelekani, C. (2012). Destructive and non-destructive methods for biofouling analysis investigated at the Adelaide Desalination Pilot Plant. Desalination 296: 61–68. Zhao, D. and Yu, S. (2015). A review of recent advance in fouling mitigation of NF/RO membranes in water treatment: pretreatment, membrane modification, and chemical cleaning. Desalin. Water Treat. 55: 870–891. Basu, O.D., Dhawan, S., and Black, K. (2016). Applications of biofiltration in drinking water treatment - a review. J. Chem. Technol. Biotechnol. 91: 585–595. Vrouwenvelder, J.S., Van Loosdrecht, M.C.M., and Kruithof, J.C. (2011). A novel scenario for biofouling control of spiral wound membrane systems. Water Res. 45: 3890–3898. van Paassen, J.A.M., van der Meer, W.G.J., and Post, J. (2005). Optiflux : from innovation to realisation. Desalination 178: 325–331. Vrouwenvelder, J.S., Manolarakis, S.A., van der Hoek, J.P. et al. (2008). Quantitative biofouling diagnosis in full scale nanofiltration and reverse osmosis installations. Water Res. 42: 4856–4868. Gilbert, P., Maira-Litran, T., McBain, A.J. et al. (2002). The physiology and collective recalcitrance of microbial biofilm communities. Adv. Microb. Physiol.: 203–256. Oliver, J.D. (2005). The viable but nonculturable state in bacteria. J. Microbiol. 43: 93–100. Dixon, M.B., Qiu, T., Blaikie, M., and Pelekani, C. (2012). The application of the Bacterial Regrowth Potential method and Flow Cytometry for biofouling detection at the Penneshaw Desalination Plant in South Australia. Desalination 284: 245–252. Vrouwenvelder, J.S. and van der Kooij, D. (2001). Diagnosis, prediction and prevention of biofouling on NF and RO membranes. Desalination 139: 65–71.

®

375

376

7 Fouling in Nanofiltration

311 Rowley, L.H. (1994). Eleven biocides are investigated for microbial efficacy

312

313 314 315 316

317

318

319

320 321

322

323

324

325

326

and membrane compatibility to use at the Yuma Desalting Plant. Desalination 97: 35–43. Shahalam, A.M., Al-Harthy, A., and Al-Zawhry, A. (2002). Feed water pretreatment in RO systems: unit processes in the Middle East. Desalination 150: 235–245. Jamaly, S., Darwish, N.N., Ahmed, I., and Hasan, S.W. (2014). A short review on reverse osmosis pretreatment technologies. Desalination 354: 30–38. Maartens, A., Jacobs, E.P., and Swart, P. (2002). UF of pulp and paper effluent: membrane fouling-prevention and cleaning. J. Membr. Sci. 209: 81–92. Belfer, S. and Gilron, J. (2002). Modification of Commercial NF Membranes. EU-report. Gilron, J., Belfer, S., Väisänen, P., and Nyström, M. (2001). Effects of surface modification on antifouling and performance properties of reverse osmosis membranes. Desalination 140: 167–179. Van Wagner, E.M., Sagle, A.C., Sharma, M.M. et al. (2011). Surface modification of commercial polyamide desalination membranes using poly(ethylene glycol) diglycidyl ether to enhance membrane fouling resistance. J. Membr. Sci. 367: 273–287. Azari, S. and Zou, L. (2012). Using zwitterionic amino acid l-DOPA to modify the surface of thin film composite polyamide reverse osmosis membranes to increase their fouling resistance. J. Membr. Sci. 401–402: 68–75. Ji, Y.-L., An, Q.-F., Zhao, Q. et al. (2012). Novel composite nanofiltration membranes containing zwitterions with high permeate flux and improved anti-fouling performance. J. Membr. Sci. 390–391: 243–253. Lewis, A.L. (2000). Phosphorylcholine-based polymers and their use in the prevention of biofouling. Colloids Surf., B 18: 261–275. Meng, J., Cao, Z., Ni, L. et al. (2014). A novel salt-responsive TFC RO membrane having superior antifouling and easy-cleaning properties. J. Membr. Sci. 461: 123–129. Shafi, H.Z., Khan, Z., Yang, R., and Gleason, K.K. (2015). Surface modification of reverse osmosis membranes with zwitterionic coating for improved resistance to fouling. Desalination 362: 93–103. Wang, J., Wang, Z., Wang, J., and Wang, S. (2015). Improving the water flux and bio-fouling resistance of reverse osmosis (RO) membrane through surface modification by zwitterionic polymer. J. Membr. Sci. 493: 188–199. Bengani, P., Kou, Y., and Asatekin, A. (2015). Zwitterionic copolymer self-assembly for fouling resistant, high flux membranes with size-based small molecule selectivity. J. Membr. Sci. 493: 755–765. Matin, A., Shafi, H., Wang, M. et al. (2016). Reverse osmosis membranes surface-modified using an initiated chemical vapor deposition technique show resistance to alginate fouling under cross-flow conditions: filtration & subsequent characterization. Desalination 379: 108–117. Athanasekou, C.P., Morales-Torres, S., Likodimos, V. et al. (2014). Prototype composite membranes of partially reduced graphene oxide/TiO2 for photocatalytic ultrafiltration water treatment under visible light. Appl. Catal., B 158–159: 361–372.

References

327 Moustakas, N.G., Katsaros, F.K., Kontos, A.G. et al. (2014). Visible light

328

329 330

331 332 333 334 335

336

337

338

339

340

341 342

343

active TiO2 photocatalytic filtration membranes with improved permeability and low energy consumption. Catal. Today 224: 56–69. Fagan, R., McCormack, D.E., Dionysiou, D.D., and Pillai, S.C. (2016). A review of solar and visible light active TiO2 photocatalysis for treating bacteria, cyanotoxins and contaminants of emerging concern. Mater. Sci. Semicond. Process. 42 (Part 1): 2–14. Fujishima, A., Zhang, X., and Tryk, D.A. (2008). TiO2 photocatalysis and related surface phenomena. Surf. Sci. Rep. 63: 515–582. Chen, J.P., Kim, S.L., and Ting, Y.P. (2003). Optimization of membrane physical and chemical cleaning bu a statistically designed approach. J. Membr. Sci. 219: 27–45. Lim, A.L. and Bai, R. (2003). Membrane fouling and cleaning in microfiltration of activated sludge wastewater. J. Membr. Sci. 216: 279–290. Trägårdh, G. (1989). Membrane cleaning. Desalination 71: 325–335. Wagner, J. (2001). Membrane Filtration Handbook: Practical Tips and Hints, 2e. Osmonics, Inc. Deqian, R. (1987). Cleaning and regeneration of membranes. Desalination 62: 363–371. Liikanen, R., Yli-Kuivila, J., and Laukkanen, R. (2002). Efficiency of various chemical cleanings for nanofiltration membrane fouled by conventionally treated surface water. J. Membr. Sci. 195: 265–276. Argüello, M.A., Alvarez, S., Riera, F.A., and Alvarez, R. (2003). Enzymatic cleaning of inorganic ultrafiltration membranes used for whey protein fractionation. J. Membr. Sci. 216: 121–134. Li, Q. and Elimelech, M. (2003). Organic fouling and chemical cleaning of nanofiltration membranes: measurements and mechanisms. Environ. Sci. Technol. 38: 4683–4693. Ang, W.S., Lee, S., and Elimelech, M. (2006). Chemical and physical aspects of cleaning of organic-fouled reverse osmosis membranes. J. Membr. Sci. 272: 198–210. Jin, X., Huang, X., and Hoek, E.M.V. (2009). Role of specific ion interactions in seawater RO membrane fouling by alginic acid. Environ. Sci. Technol. 43: 3580–3587. Corbatón-Báguena, M.-J., Álvarez-Blanco, S., and Vincent-Vela, M.-C. (2014). Cleaning of ultrafiltration membranes fouled with BSA by means of saline solutions. Sep. Purif. Technol. 125: 1–10. Lee, S. and Elimelech, M. (2007). Salt cleaning of organic-fouled reverse osmosis membranes. Water Res. 41: 1134–1142. Corbatón-Báguena, M.-J., Álvarez-Blanco, S., Vincent-Vela, M.-C., and Lora-García, J. (2015). Utilization of NaCl solutions to clean ultrafiltration membranes fouled by whey protein concentrates. Sep. Purif. Technol. 150: 95–101. Ang, W.S., Yip, N.Y., Tiraferri, A., and Elimelech, M. (2011). Chemical cleaning of RO membranes fouled by wastewater effluent: achieving higher efficiency with dual-step cleaning. J. Membr. Sci. 382: 100–106.

377

378

7 Fouling in Nanofiltration

344 Munoz-Aguado, M.J., Wiley, D.E., and Fane, A.G. (1996). Enzymatic and

345

346 347

348 349 350

351 352

353

354 355

356

357 358

359 360

detergent cleaning of a polysulphone ultrafiltration membrane fouled with BSA and whey. J. Membr. Sci. 117: 175–187. Khan, M., Danielsen, S., Johansen, K. et al. (2014). Enzymatic cleaning of biofouled thin-film composite reverse osmosis (RO) membrane operated in a biofilm membrane reactor. Biofouling 30: 153–167. Zeiher, E.H.K. and Yu, F.P. (2000). Membranes: biocides used for industrial membrane system sanitization. Ultrapure Water 17: 55–64. Weis, A., Bird, M.R., Nyström, M., and Wright, C. (2005). The influence of morphology, hydrophobicity and charge upon the long-term performance of ultrafiltration membranes fouled with spent sulphite liquor. Desalination 175: 73–85. Kosutic, K. and Kunst, B. (2002). RO and NF membrane fouling and cleaning and pore size distribution variations. Desalination 150: 113–120. Thor en, T. (2001). Fundamental studies on membrane filtration of coloured surface water. NTNU. Trondheim, Norway. Weis, W., Bird, M., and Nyström, M. (2002). The variation of zeta-potential with pH for UF membranes subjected ro a range of fouling and cleaning protocols. In: Fouling, Cleaning and Disinfection in Food Processing (eds. D.I. Wilson, P.J. Fryer and A.P.M. Hastings), 181–188. Cambridge: Proceedings of a conference held at Jesus College. Zhu, H. and Nyström, M. (1998). Cleaning results characterized with flux, streaming potential and FTIR measurements. Colloids Surf., A 138: 309–321. Al-Amoudi, A. and Lovitt, R.W. (2007). Fouling strategies and the cleaning system of NF membranes and factors affecting cleaning efficiency. J. Membr. Sci. 303: 4–28. Simon, A., McDonald, J.A., Khan, S.J. et al. (2013). Effects of caustic cleaning on pore size of nanofiltration membranes and their rejection of trace organic chemicals. J. Membr. Sci. 447: 153–162. Shorrock, C.-J., Bird, M.R., and Howell, J.A. (1998). Membrane cleaning: removal of irreversibly fouled yeast deposits. Trans. IChemE. 76: 1–8. Bartlett, M., Bird, M.R., and Howell, J.A. (1995). An experimental study for the development of a qualitative membrane cleaning model. J. Membr. Sci. 105: 147–157. Nyström, M. and Zhu, H. (1997). Characterization of cleaning results using combined flux and streaming potential methods. J. Membr. Sci. 131: 195–205. Ramon, G., Agnon, Y., and Dosoretz, C. (2010). Dynamics of an osmotic backwash cycle. J. Membr. Sci. 364: 157–166. Motsa, M.M., Mamba, B.B., Thwala, J.M., and Verliefde, A.R.D. (2017). Osmotic backwash of fouled FO membranes: cleaning mechanisms and membrane surface properties after cleaning. Desalination 402: 62–71. Fu, P., Ruiz, H., Lozier, J. et al. (1995). A pilot study on groundwater natural organics removal by low-pressure membranes. Desalination 102: 47–56. Niewersch, C., Rieth, C., Hailemariam, L. et al. (2020). Reverse osmosis membrane element integrity evaluation using imperfection model. Desalination 476: 114175.

References

361 Troester, M., Lipp, P., Sacher, F. et al. (2014). Laser-induced

362

363

364

365

366 367

368

369

370 371

372 373

374

375

376

breakdown-detection for reliable online monitoring of membrane integrity. J. Membr. Sci. 466: 313–321. Regula, C., Carretier, E., Wyart, Y. et al. (2014). Chemical cleaning/disinfection and ageing of organic UF membranes: a review. Water Res. 56: 325–365. Mänttäri, M., Martin, H., Nuortila-Jokinen, J., and Nyström, M. (1999). Using a spiral wound nanofiltration element for the filtration of paper mill effluents; pretreatment and fouling. Adv. Environ. Res. 3: 202–214. Simon, A., Price, W.E., and Nghiem, L.D. (2013). Influence of formulated chemical cleaning reagents on the surface properties and separation efficiency of nanofiltrationmembranes. J. Membr. Sci. 432: 73–82. Kallioinen, M., Sainio, T., Lahti, J. et al. (2016). Effect of extended exposure to alkaline cleaning chemicals on performance of polyamide (PA) nanofiltration membranes. Sep. Purif. Technol. 158: 115–123. Platt, S., Nyström, M., Capanelli, G., and Bottino, A. (2004). Stability of NF membranes under extreme acid conditions. J. Membr. Sci. 239: 91–103. Mänttäri, M., Pihlajamäki, A., Kaipainen, E., and Nyström, M. (2002). Effect of temperature and membrane pretreatment on the filtration properties of nanofiltration membranes. Desalination 145: 81–86. Gillham, C.R., Fryer, P.J., Hasting, A.P.M., and Wilson, D.I. (1999). Cleaning-in-place of whey protein fouling deposits: mechanisms controlling cleaning. Trans. IchemE 77: 127–136. Cabero, M.L., Riera, F.A., and Alvarez, R. (1999). Rinsing of ultrafiltration ceramic membranes fouled with whey proteins: effects of cleaning procedures. J. Membr. Sci. 154: 239–250. Daufin, G., Merin, U., Labbe, J.P. et al. (1991). Cleaning of inorganic membranes after whey and milk ultrafiltration. Biotechnol. Bioeng. 38: 82–89. Räsänen, E., Nyström, M., Sahlstein, J., and Tossavainen, O. (2002). Comparison of commercial membranes in nanofiltration of sweet whey. Le Lait 82: 343–356. Madaeni, S.S., Mohamamdi, T., and Moghadam, M.K. (2001). Chemical cleaning of reverse osmosis membranes. Desalination 134: 77–82. Ochando-Pulido, J.M., Victor-Ortega, M.D., and Martínez-Ferez, A. (2015). On the cleaning procedure of a hydrophilic reverse osmosis membrane fouled by secondary-treated olive mill wastewater. Chem. Eng. J. 260: 142–151. Räsänen, E., Nyström, M., Sahlstein, J., and Tossavainen, O. (2002). Purification and regeneration of diluted caustic and acidic washing solutions by membrane filtration. Desalination 149: 185–190. Gésan-Guiziou, G., Boyaval, E., and Daufin, G. (2002). Nanofiltration for the recovery of caustic cleaning-in-place solutions: robustness towards large variation of composition. Desalination 149: 127–129. Forstmeier, M., Goers, B., and Wozny, G. (2002). UF/NF treatment of rinsing waters in a liquid detergent production plant. Desalination 149: 175–177.

379

381

8 Pretreatment and Hybrid Processes Jack Gilron 1 , Marianne Nyström 2 , Jukka Tanninen 2 , and Lena Kamppinen 2 1 Ben Gurion University, Zuckerberg Institute for Water Research, Blaustein Institutes for Desert Research, Midreshet Sde Boker, 84990 Israel 2 Lappeenranta University of Technology, LUT School of Engineering Science, Yliopistonkatu 34, 53850 Lappeenranta, Finland

8.1 Introduction Stricter environmental legislation and increased competition within industry have increased the need for more efficient methods of process- and wastewater treatment. Membrane technology offers several different unit operations for water treatment. One of the unit operations utilizing membrane technology is nanofiltration (NF), which is now available and has been under study for decades. NF is used for the removal of dissolved organic and inorganic components from waters and effluents, e.g. to upgrade low-grade water for industrial and potable use or to clean industrial effluents (see Chapter 10 for water treatment and Chapter 11 for water reclamation applications). Th use of one single separation technology is often inadequate in order to meet the process objectives of industry. Therefore, a sequence of processes or a hybrid process has to be used involving two or more different separation steps. Nanofiltration feed water usually needs to be filtered prior to use because of the plugging and fouling characteristics of most waste- and process waters. Nanofiltration membranes are sensitive to high concentrations of solids. Moreover, the organic content of the feed water may easily foul the membranes, leading to shorter cleaning intervals and a shorter lifespan of the membranes. Hence, it becomes cost-efficient to remove solids and fouling organic material before nanofiltration. Nanofiltration is widely used together with other membrane technologies, such as microfiltration (MF), ultrafiltration (UF), electrodialysis (ED), and reverse osmosis (RO) and is being incorporated into emerging membrane processes such as membrane distillation (MD) and forward osmosis (FO). Nanofiltration has also been combined with other conventional separation technologies such as ion exchange (IX) and evaporation to form hybrid processes.

Nanofiltration: Principles, Applications, and New Materials, Second Edition. Edited by Andrea Iris Schäfer and Anthony G. Fane. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

382

8 Pretreatment and Hybrid Processes

In this chapter, hybrid process has been defined as a process in which different kinds of separation processes have been combined into an integrated system for enhanced processing. The aim of the chapter is to give a broad view of the feasibility of nanofiltration as a tool in processes consisting of multiple separative stages and methods. Proper pretreatment methods for nanofiltration are also reviewed. An illustration of the capabilities of hybrid processes is presented in connection with seawater desalination, olive mill wastewater (OMW), and textile effluent treatment. A hybrid process involving membrane bioreactor (MBR) and NF is discussed in Chapter 17.

8.2 Pretreatment – An Overview 8.2.1

Importance of Pretreatment in NF

Pretreatment of the feed is needed to protect the membrane and to improve its performance. Protection refers usually to the prevention of fouling, but also includes protection against mechanical and chemical damage. A high solids load can damage the membrane surface mechanically and restrict the flow in the filtration system. Furthermore, oxidizing agents, e.g. chlorine and ozone, are harmful to many membrane materials. Th pretreatment requirements and methods for NF are the same as for reverse osmosis. In most cases, any loss of membrane flux that takes place during filtration is due to fouling. Fouling can affect the lifetime of the membrane, and cleaning of the membranes causes down time. The cleaning interval depends on the feed, the filtration equipment, and the process conditions. Each membrane packaging configuration has a different degree of ease of cleaning and susceptibility to fouling. Th major foulant categories in NF are sparingly soluble inorganic salts, colloidal or particulate matter, dissolved organics, chemical reactants, and microorganisms [1]. (See Chapter 7 for more information on fouling.) Fouling can be partly controlled by optimizing the filtration process: choosing suitable membrane types and materials, module configurations, or filtration parameters (e.g. flow, pressure and temperature). However, in addition to optimized filtering conditions, an appropriate pretreatment of the feed is usually needed. The most suitable pretreatment scheme for the feed depends on the composition of the feed and the application. Th requirements for pretreatment are different with different module types and packing densities, e.g. a spiral-wound element requires more pretreatment than a backwashable hollow fiber module with larger diameter (>1 mm) capillaries fed from the inside. Pretreatment prior to NF is used to [2] • • • •

reduce suspended solids and minimize the effect of colloids; reduce the microbiological fouling potential of the feed; add chemicals (antiscalant, pH adjustment); remove oxidizing compounds in the feed if required (to protect membranes). Pretreatment methods and their uses are listed in Table 8.1.

8.3 Non-membrane Pretreatment Methods

Table 8.1 Pretreatment methods in NF and their applications [2–9].

Pretreatment

Sparingly soluble Biological Oxidizing Organic salts Colloidsa) Solids treatment agents matter

pH adjustment

X

X

Scale inhibitor

X

X

Ion exchange

X

X

Lime softening

X

Fluidized bed crystallizer (e.g. CrystallactorTM )

X

Compact accelerated precipitation softening (CAPS)

X

X

X

Media filtration

X

Oxidation-filtration

X

In-line coagulation

X

Coagulation and flocculation

X

X

Dissolved air flotation

X

X

X

X

X

X

X

X

Micro- or ultrafiltration Forward osmosis

X

Cartridge filter

X X X X X

Chlorination

X

Advanced oxidation Processes (electro-oxidation, ozonation, etc.)

X

Dechlorination

X X

Shock treatment

X

Disinfection

X

Granular active carbon filtration

X

Biological treatment Electromagnetic treatment

X

X X

X

X

X X

X

a) Colloids include silt, colloidal Fe, Al, and silica.

8.3 Non-membrane Pretreatment Methods 8.3.1

Control of Inorganic Precipitation (Scaling)

Scaling occurs when sparingly soluble multivalent salts become concentrated because of retention and then exceed their solubility limit. Scale deposition leads to significant flux loss and can also negatively impact membrane performance in other ways (as described in more detail in Chapter 8). Th risk of scaling increases when the objective is a nanofiltration process with high recovery. Th most

383

384

8 Pretreatment and Hybrid Processes

common sparingly soluble salts and thus potential foulants are calcium sulfate (CaSO4 ), calcium carbonate (CaCO3 ), calcium phosphate (in wastewater), and silica. Other possible fouling salts are magnesium hydroxide (Mg(OH)2 ), barium sulfate (BaSO4 ), calcium fluoride (CaF2 ), and strontium sulfate (SrSO4 ) usually found in groundwater samples [10, 11]. Scales can be divided into two categories: alkaline and non-alkaline scales. CaCO3 , Ca3 (PO4 )2 in its various polymorphs, and brucite [Mg(OH)2 ] form alkaline scale. Most natural waters are very rich in CaCO3 . Depending on water composition different polymorphs with different solubilities can form (calcite, aragonite, and vaterite). The alkaline scales tend to be much less soluble than non-alkaline scales and tend to have inverse solubilities with temperature. Brucite will form when the pH is increased above 10 (which is done to increase silica solubility). One traditional method for preventing calcium carbonate and calcium phosphate scale is to add acid into the feed water to reduce the concentration of the scaling anions – most commonly sulfuric acid. Th addition of sulfuric acid may, on the other hand, cause corrosion problems and precipitation of calcium sulfate. Decrease of pH may also increase the fouling potential of silica, colloidal, and organic matter in the feed water. In consequence, the use of formulated chemical inhibitors has widely replaced the use of acids for scale prevention (see Section 8.3.1.2). Th most common non-alkaline scales are calcium sulfate and silica. Of calcium sulfate’s three forms (CaSO4 , CaSO4 ⋅ 1/2H2 O and CaSO4 ⋅2H2 O–gypsum), gypsum is the dominant form at the temperatures of NF operation. Calcium sulfate is the most soluble mineral of the sparingly soluble sulfate salts. 8.3.1.1

Water Softening

Lime softening is a traditional method for the removal of hardness (Ca, Mg) from water. Th addition of lime (CaO) or slaked lime (Ca(OH)2 ) and soda ash (Na2 CO3 ) reduces the level of calcium and magnesium by precipitating magnesium as hydroxide and calcium as carbonate. Another widely used method for water softening is ion exchange, where harmful calcium and magnesium ions are replaced by sodium. Other softening approaches are compact accelerated precipitation softening (CAPS), pellet softening reactor, ion exchange softening in pretreatment section of high efficiency reverse osmosis (HERO), and advanced precipitation softening (APS) [11]. Partial softening reduces the need for antiscalants and increases recovery. CAPS is a partial softening process that integrates filtration media with the softening process. Th softening vessel contains a reaction suspension of water and 1–3 wt% calcium carbonate solids in which microfiltration media are submerged. Raw water is fed into the reactor along with caustic soda to generate supersaturation with respect to calcium carbonate. Calcium carbonate precipitates in the slurry (crystal growth on the suspended particles) and on the filter cake that forms on the microfilter (nucleation and crystal growth on the pore wall of the cake). The potential of the CAPS process to remove heavy metals and silica and coprecipitate hydrophobic foulants has also been demonstrated [6–8, 12]. Pellet softening uses fluidized bed reactors as the contact media when feeding raw

8.3 Non-membrane Pretreatment Methods

water and caustic to precipitate calcium carbonate [13]. HERO uses a weak acid and strong acid cation exchanger along with acidification to drive off the alkalinity to allow operation at high pH (>10) to operate at high silica concentrations (>1000 mg/l) [14]. In a desaturation unit (DU) sparingly soluble salts are forced to precipitate on seed crystals. This stage is usually placed before the last NF stage, where the concentration of the salts in the concentrate is the highest. A high total organic carbon (TOC) content may interfere with the precipitation of some salts, e.g. barium sulfate. The mechanism of the interference is that certain organic compounds are adsorbed on the seed crystals (crystal poisoning) and have a threshold effect on crystallization [3]. A study of APS on RO concentrates showed that caustic could be used to destabilize the antiscalant on calcium carbonate and to allow precipitation of excess gypsum on seed crystals [15]. 8.3.1.2

Antiscalants

Antiscalants are chemical inhibitors for scale prevention. Both polymeric and non-polymeric antiscalants exist. An antiscalant can in sub-stoichiometrical amounts prevent the precipitation of salts that have exceeded their solubility limit (threshold effect). It may also interfere with the normal crystal growth, which leads to an irregular crystal structure with poor scale forming ability (crystal distortion effect) or it may cause repulsion between crystals by making them comparably charged (dispersancy) [16–21]. Th first scale inhibitor to be widely used was sodium hexametaphosphate (SHMP). Although still used, SHMP is losing ground to more effective antiscalants. Phosphonates are a widely used group of antiscalants. Phosphonates are organophosphorus compounds, such as diethylenetriamine-penta(methyl phosphonic) acid (DTPMPA), 1-hydroxyethylidene-1,1-diphosphonic acid (HEDP), 2-phosphonobutane-1,2,4-tricarboxylic acid (PBTC), and aminotrimethylenephosphonic acid (AMP). Numerous polymers, used as antiscalants, are also commercially available, many of them anionic polyelectrolytes. The most common polymers used are polyacrylic acid, polyacrylamide, poly-maleic acid, poly-carboxylic acid, and polysulfonate. The performance of a polymeric antiscalant depends on its functional groups, surface charge density, and molar mass. The effectiveness is also affected by the operating temperature and pH. A novel class of antiscalants, dendrimeric polymers, which are highly branched polymer blends with 3-D structure, has also emerged [22, 23]. These antiscalants are, however, not yet in widespread commercial use though they show promise for silica control. Limitations in the performance of inhibitors usually lead to a need to use a combination of at least two different types of antiscalants for optimum results. For example, HEDP inhibits markedly calcium carbonate and barium sulfate formation, but does not exhibit any marked inhibitory action on calcium sulfate [16, 24]. Some antiscalants have low tolerance for calcium ions and this can cause excess membrane fouling in the form of calcium deposits, such as calcium phosphonate or calcium phosphate [17, 18]. Many forms of particulate matter adsorb antiscalant molecules. Therefore, the concentration of functional

385

386

8 Pretreatment and Hybrid Processes

antiscalant in water rich in suspended solids can decrease remarkably and precipitation can occur [25]. 8.3.1.3

Silica Scaling and Fouling Prevention

High silica brines can cause severe and irreversible membrane fouling. A widely approved operation practice in the absence of effective silica antiscalants is to keep the silica content in the concentrate below 120–155 mg/l, depending on the pH and temperature. The chemistry of silica precipitation is complex and may consist of colloidal polymerized silica, magnesium silicates, or alumino-silicates. Calcium and magnesium are known to promote silica polymerization [26, 27]. A decrease in pH reduces the stability of naturally occurring silicate particles by reducing their negative charge. Thus, silicate particles alone or together with metal oxides or organics may aggravate colloidal fouling problems [28]. Calcium carbonate crystals, for example, can act as nuclei for silica and magnesium silicate precipitation. Th presence of trivalent cations such as Fe3+ and Al3+ is also highly significant. The risk of silica deposits can be minimized by removing calcium species and polyvalent metals and metal oxides from the feed as well as any other suspended matter [16, 25]. It is possible to remove a major portion of silica that is present in most groundwater during lime softening by the addition of AlCl3 or ZnCl2 . Conventional filtration of aluminum and silicon precipitants is, however, very difficult. In CAPS coprecipitation of silica with aluminum or zinc chloride in calcium carbonate slurry and filtering through a cake does not measurably increase the filtration time [8]. Polymeric dispersants and antiscalants are used for the prevention of silica fouling. Modern antiscalants based on polyoxazole [22] and dendrimers [22, 23] allow operation at silica levels of 250–300 mg/l [29]. 8.3.1.4

Removal of Iron, Manganese, and Zinc

Iron, manganese, and zinc also form insoluble precipitates. Some antiscalants have the ability to stabilize these ions in water [18]. Ion exchange or oxidizing prefilters, such as greensand filters, can be used for the removal of iron and manganese. At low pH, however, greensand becomes inactive or if the feed contains more than 2 mg/l dissolved oxygen, the operation of the ion exchanger can fail due to the precipitation of iron and manganese oxides. Th feed for an ion exchanger should be pretreated and free from suspended iron and manganese [30]. In conventional groundwater treatment the water is often aerated in order to precipitate iron and then it is filtered through a rapid sand filter and a cartridge filter [31]. 8.3.1.5

Magnetic Treatment of Water

One of the more controversial methods for scale prevention is magnetic or electromagnetic treatment of water [4]. This method does not remove scale forming ions, but alters them in such a way that they lose their ability to form scale. The mechanism of the prevention is not well known, but it is assumed that heavy metals, such as iron and zinc, are activated by the magnetic field and function as threshold inhibitors. Some studies show effectiveness [32] and some do not [33]. Th electromagnetic field (EMF) helps induce an electric charge on particulate

8.3 Non-membrane Pretreatment Methods

foulants and repulsion takes place. As such, it can be effective on suspended particles and colloids as well. A complete review of the methods for preventing scaling on pressure-driven membranes (both NF and RO) can be found in a review by Gilron [11]. 8.3.2

Removal of Colloids and Solids

Particulate fouling from colloids, suspended solids, and microbial cells is a persistent problem in pressure-driven membrane systems and for NF in particular. In addition to potentially blocking the spiral retentate channels, colloid enhanced concentration polarization can lead to enhanced osmotic pressure [34] and increased potential for supersaturation and scaling on the NF membrane [35]. Its complete removal by pretreatment is not always easy [36]. A wide variety of colloids are found in natural waters: colloidal silica and sulfur, precipitated iron and aluminum compounds, corrosion products, silt, clay, biological debris, and even high molar mass organic substances such as humic substances [28]. Because colloidal matter is often rather disperse and the particles may be charged and repelled by the filter media, coagulants or flocculants are added as filtration aids. Th coagulants used include alum, ferric chloride, ferric sulfate, and/or polyelectrolyte and are often added in-line to the feed stream prior to prefiltration on media or membranes [2, 37]. The dosage of a coagulant varies according to the feed and therefore, frequent readjustments and on-site tests are required for the optimum chemical dosage. Waters with low solids content need only be prefiltered before nanofiltration. With proper design it is possible to remove almost all suspended solids by prefiltration. Feed waters with high suspended solids, high turbidity, and/or a high silt density index (SDI) usually need coagulation followed by sedimentation before they are filtered. The settling time and dosage are determined by on-site tests [2]. A wide range of prefilters, such as media, bag, and cartridge filters are available. Media filters can be divided into single media, dual media, and multimedia filters and the filter materials include, among others, sand, anthracite coal, finely crushed garnet, or pumice. In addition to sieving, the filter material can be functional. Manganese greensand is used in oxidizing filters, calcite in neutralizing filters and diatomaceous earth, or activated carbon in adsorptive filters. The e are many design options: a media filter can be rapid or slow, a down-flow, or an up-flow system [21]. Of course, MF and UF can also be used for particulate removal in pretreatment before NF (see Sections 8.4.2 and 8.4.3). A newer pretreatment method involves using dissolved air flotation (DAF) to remove suspended solid and some organic substances [9]. 8.3.3

Removal of Organic Substances

Adsorption of organic substances on the NF membrane surface causes flux decline. In the most severe cases the flux loss can be irreversible. Adsorption by high molar mass compounds is most likely when the compounds are hydrophobic or positively charged. Divalent cations and low pH often increase

387

388

8 Pretreatment and Hybrid Processes

the adsorbability of natural organic matter (NOM) [2, 28]. The reason for the fouling tendency of humic acid, which is a major part of NOM, is related to its binding tendency to multivalent ions [38–40]. Multivalent ions, such as calcium and iron, may form insoluble complexes or gels with NOM and precipitate on the membrane surface. Complexation of NOM with calcium increases with pH because of the increased negative charge of NOM and in this way NOM can be precipitated. [39, 41]. A high pH is preferable for prevention of organic fouling, because both the membrane and the organic compound will then be negatively charged, if charged at all. Organics in the form of emulsion, oil, and grease should be removed before NF. An emulsion may form an organic film on the membrane and oils and greases are readily adsorbed onto the membrane surface. Coagulation and activated carbon adsorption are useful methods in the removal of oils and greases [2]. Organics occurring in natural waters are mainly humic substances. If the TOC content is very high humic substances can be removed by coagulation, ultrafiltration, or adsorption on activated carbon. However, filtration through granular activated carbon is known to run the risk of releasing bacteria. Th high inner surface of the carbon pores and the adsorbed organic nutrients promote the biological activity of the filter. When carbon filters are run at sufficiently low velocities (2–10 m/h) and with sufficiently high beds (2–3 m), all the microbial activity takes place in the upper region of the filter bed and the filtered water is almost free from bacteria and nutrients [2]. Organic material can also be effectively eliminated by biological pretreatment. In order to enhance the elimination by bacteria, the organic material can be degraded by, e.g. ozonation prior to biological treatment [42]. Biological treatment will be discussed in Section 8.3.5. Oxidative processes (ozonation, chlorination, electrochemical oxidation) can reduce the fouling potential of organic materials as well as making them more amenable to biological treatment. NF membranes (NF270) had increased average flux (49 vs. 32 l/m2 h) when treating a dyehouse effluent that was previously treated by electrochemical oxidation [43]. In the case of treating pharmaceutical residues, placing oxidative processes upstream of NF reduces the organic load that NF must retain to only those molecules that are recalcitrant [44]. These can then be recycled to the oxidative process. 8.3.4

Biological Fouling Prevention

All raw waters contain microorganisms and most of them can be regarded as particular or colloidal matter, depending on their size, and can be removed by prefiltration. However, under favorable conditions microorganisms have the ability to reproduce and form a biofilm. Dissolved organic and inorganic nutrients in the feed promote reproduction, so microbial growth prevention is a major objective of pretreatment processes [1, 2, 45, 46]. Chlorination has been used for disinfection for many years. The effect of chlorine depends on the concentration, time of exposure, and pH. Chlorine is added at the intake and left to react for 20–30 minutes. Th feed must be dechlorinated before membrane filtration in order to prevent membrane oxidation,

8.3 Non-membrane Pretreatment Methods

although membranes are now coming on the market that are chlorine tolerant (HYDRACORE10-LD, Deltapore HF NF), and there are other chlorine-resistant NF membrane materials being developed (see Chapters 2, 25, 26). An activated carbon bed is very effective for dechlorination. Residual free chlorine can also be removed by using chemical reducing agents, such as sodium metabisulfite (SMBS) [2, 45]. Ultraviolet (UV) irradiation is also being increasingly employed to destroy chlorine residuals [47]. Ozone is a stronger oxidizing agent than chlorine, but it decomposes readily. A sufficient ozone level must be maintained to kill all microorganisms [2]. De-ozonation must be performed before membrane filtration and UV has been successfully used for this purpose. Ozonation can alter the structure and charge of organic molecules, thus producing ionized groups, which may adsorb on the membrane surfaces. Ozonation of NOM is known to split the carbon–carbon double bonds and form aldehydes, ketones, and carboxylic acids [3]. The efore, ozone and other oxidative agents, such as chlorine, peracetic acid, peroxide, and potassium permanganate (KMnO4 ) are effective disinfectants, but they also decrease the biostability of the feed water. For example, humic acids are very complex large molecules and cannot as such be used as nutrients either for aerobic or for anaerobic bacteria. Once humic compounds have been oxidized to smaller pieces they are converted into assimilable organic carbon and become good food for bacteria [2, 45]. Consequently, disinfection by oxidation may indirectly cause biological fouling and increase fouling by NOM. If oxidation is followed by, e.g. biological treatment, the fouling problems caused by NOM may be eliminated. UV irradiation at 254 nm has a germicidal effect. No chemical addition is needed in UV treatment, but this method can only be used for relatively clear waters, because colloids and organic matter reduce the effect of the irradiation [2]. Bioactivity can also be controlled with biocides. Th biocide used must not harm the membrane material and therefore, formulated non-oxidizing biocides have gained ground in integrated membrane systems [48]. Experience, however, shows that long-term use of non-oxidizing biocides often causes microbial resistance. It is advisable to apply these biocides frequently as a ‘shock’ dose depending upon the rate of re-growth. Th effect is maintained if two different biocides are used contemporaneously [45]. Other commonly used biocides are monochloramines and SMBS. Chloramines, however, can have an adverse effect on NF flux, and bisulfite is more efficient against aerobic bacteria than against anaerobic bacteria [1, 2]. 8.3.5

Biological Pretreatment

During biological treatment the growth of microorganisms that use organic compounds as a carbon substrate is maintained and encouraged by favorable environmental conditions. The microorganisms that metabolize the substrate are subsequently separated from the water, leaving a relatively clean effluent. Biological processes are primarily designed for the removal of dissolved and suspended organic matter in aerobic or anaerobic conditions. Biological treatment

389

8 Pretreatment and Hybrid Processes

Desal-5 DK, 30 minutes

Desal-5 DK, VRF 5

120

100

Permeate flux (l/m2 h)

390

80

60

40

20

0 pH 5 (Original)

pH 7

Bio, pH 7

Bio + UF, pH 7

Figure 8.1 The effect of biological pretreatment (aerobic at T = 45–55  C) on the flux in nanofiltration of pressure groundwood mill circuit water in the pulp and paper industry. Desal-5 DK membranes were used at 10 bar with a flow velocity of 6 m/s. Source: Manttari et al. [49]. Reproduced with the permission of World Filtration congress.

is also capable of removing suspended solids, nitrogen, phosphorus, and heavy metals, but it is susceptible to toxic chemicals [5]. Gravity settlement and depth filtration are used for separation of the biomass from the final effluent. Th former method is the most commonly used. Sedimentation is, however, often disturbed by gas bubbles lifting sludge to the surface, or by the existence of less dense, bulky flocs that do not settle. Microfiltration and ultrafiltration membranes when coupled to biological processes (MBRs) are most often used as a replacement for sedimentation [5]. Biological pretreatment prior to NF can improve the nanofiltration flux, prevent fouling, and enhance the cleaning procedure. An example of the effect of biological pretreatment on the flux in nanofiltration (thermophilic, T = 45–55  C and aerobic) is shown in Figure 8.1. Pressure groundwood (PGW) mill water was filtered using nanofiltration with or without pretreatment. The flux was measured 30 minutes after beginning the NF and after 8–10 hours, when the volume reduction factor (VRF) was 5. Th use of biological pretreatment reduced fouling at higher concentrations, hence increasing the flux. Th change of pH from 5 to 7 was crucial in preventing instantaneous fouling, but in the long run the biological treatment became important as it degraded hydrophobic foulants [49]. Biological treatment is slow compared to chemical treatment. Therefore, biological treatment (not referring to MBRs) as a pretreatment method for NF is less utilized than other pretreatment methods. However, hybrid processes where NF is used as a post-treatment after an MBR have been successfully used, e.g. for purification of dumpsite leachate [5, 50]. For details see Chapters 15 and 16.

8.4 Pretreatment Methods Using Filter Media

8.4 Pretreatment Methods Using Filter Media In many hybrid processes filtration is used as a pretreatment method before nanofiltration. It is possible to use conventional filtration or micro- or ultrafiltration. In some cases, other types of pretreatment methods can be substituted with more economically and environmentally feasible filtration technologies. 8.4.1

Conventional Filtration

Conventional filtration processes are in many cases feasible for nanofiltration feed pretreatment. Micro- and ultrafiltration can usually offer better selectivity, less floor space, and more flexible use. In some cases, however, it is easier to add NF to an existing filtration process, such as sand filtration. In addition to large sand filters, bag filters are utilized in membrane process applications. A quite common procedure seems to be to add a 5–10 m prefiltering system in front of a membrane process to avoid excessive plugging of the membrane elements and fouling of the membranes. Sometimes, cartridge filters are added for safety measure only – to protect the membrane elements from accidental overdose of solids. The e are quite a few applications that use conventional filtration prior to NF. One such application taking advantage of both bag- and sand filters is reported by Reiss et al. [51], according to whom the treatment of secondary treated municipal wastewater can be carried out by a hybrid process consisting of slow sand filtration and NF. In the process, the wastewater is subjected to slow sand filtration, after which the effluent is run through a 5 m prefilter before entering the NF process. Th purpose of the slow sand filtration is the removal of biological and organic substances that might otherwise foul the NF membranes, which would be the case, for example, with the rapid sand filtration method. Slow sand filtration, thus leads to lower cleaning costs for the nanofilter [51]. In the case of groundwater treatment, river bank filtration can also be used as a prefiltration step before the NF process [31, 52]. In combination with proper post-treatment, such as pH adjustment, the process chain of bank filtration – aeration – green sand filtration – NF can provide biologically and chemically stable drinking water free from trace organic pollutants [52]. Use of rapid sand filters in pretreating surface water prior to nanofiltration increases steady state NF fluxes from 39 to 48 l/m2 h with 5 bar applied pressure with high NF rejection of fulvics, humics, and aromatic proteins [53]. The largest application for NF pretreatment by sand filtration is in the purification process of the surface water of Méry-sur-Oise, France. A dual-layer filtration system with additional flocculation onto the filter is placed before a huge NF process in the purification of the river water [54]. 8.4.2

Microfiltration (MF)

Th primary use of MF in combined processes of MF and NF is the pretreatment of the NF feed water. MF removes solid particles and some colloidal substances, reducing turbidity and clarifying the water. The MF permeate can then be filtered with NF units with reduced concentration polarization and fouling as a result of

391

392

8 Pretreatment and Hybrid Processes

the reduction of colloidal enhanced concentration polarization. In many cases, this kind of prefiltration is essential because NF processes usually utilize spiral modules that clog quite easily. The use of MF as NF pretreatment is not as common as UF and does not remove dissolved organic material. Th following are examples of how MF is used as pretreatment for NF processing streams from the laundry-, textile, and sugar industry with a comparison to UF pretreatment in the same industries. 8.4.2.1

Laundry Wastewater

Commercial laundry wastewater has a high biological oxygen demand (BOD), chemical oxygen demand (COD), and a high concentration of detergents and surfactants. In order to recycle water in the wash cycles, the contaminants have to be removed. Two-pass combinations of MF/NF [55] or UF/NF [56] systems have been utilized to remove oil, lint, grease, detergents, surfactants, and other impurities from the wash water to produce high quality water for reuse in the laundry wash cycles. Th first membrane unit in both cases removes oil, lint, grease, and particulates. The spiral-wound based MF (JX)/NF (DK5) system reaches 82% recovery with 76% removal of total dissolved solids (TDS), while the tubular ceramic UF/NF system reaches 75–80% recovery with 97.5% removal of COD and removal of heavy metals to below detectable limits. 8.4.2.2

Textile Industry

In related applications MF has been an effective pretreatment for NF in treating dyehouse effluents. Certain combinations of cleaned effluent could be used in dyeing with a similarity of result reaching 86% that of freshwater [57]. Previously a comparison of the processes of coagulation, MF (0.45 m), and UF (100 kg/mol molecular weight cutoff [MWCO]) as NF pretreatment for indigo dyeing effluent showed that MF/NF gave the best performance in terms of color removal and flux [58]. UF can be used both to enhance COD retention in NF-based wastewater treatment processes and to improve NF flux. The COD retention of dye-containing wastewater is 99% with NF90. By adding a pretreating UF stage the COD in the permeate was reduced by 40% and flux increased by 50% [59]. In the recovery of caustic from mercerization process used in treating cotton textiles, the application of MF or UF prevents the fouling of the base stable MP-34 NF spirals by the suspended solids in the effluent and reduces the COD load. Th NF step then reduces the COD by more than 97% leaving less than 22 mg/l COD in the effluent [60]. The recovered caustic is then concentrated by evaporation to the required concentration [60]. A later study [61] showed that NF caustic recovery could be done without MF pretreatment, when open channel modules without feed spacers were used. A tubular NF could be used but it would be at the price of higher energy consumptions for the high cross flow velocities (at least 0.79 m/s) needed in tubular configuration. 8.4.2.3

Sweetener and Sugar Industry Applications

MF can also be combined with NF in the sweeteners and sugar industry. After enzymatic treatment, 95% dextrose can be produced from starch. Th remaining

8.4 Pretreatment Methods Using Filter Media

5% includes di- and trisaccharides. Th conventional process for producing high purity dextrose (i.e. greater than 99% purity) requires expensive and time-consuming crystallization. In order to produce pure dextrose, a membrane process consisting of MF and NF has been proposed. Th MF permeate is fed to the NF unit, which fractionates dextrose from impurities, producing over 99% pure dextrose syrup as an NF permeate [62]. 8.4.3

Ultrafiltration (UF)

UF membranes can also be used as a pretreatment for NF. In some cases, UF is preferred to MF due to its tighter structure, which offers better separation efficiency. With UF it is possible to partially purify the NF feed from organic or inorganic material that might otherwise damage or foul the NF membranes or clog the modules. Usually, a UF/NF combination can be used when the treated solution contains both salts (multivalent and/or monovalent) and organic matter. After the solution has been purified of the larger organic molecules using UF, NF can be used for the removal of organic traces and multivalent salts. A recent study showed that it was preferred as pretreatment of surface water for producing drinking water by NF based on flux and organic retention [63]. It is also possible to fractionate organic materials by UF in order to concentrate or purify the product by NF (e.g. by fractionating salts and organic matter). This kind of pretreatment, which partially removes organic matter from the feed, also enhances the flux of NF in some applications. At the moment, UF/NF combinations are quite common in industrial applications where the solutions contain high amounts of organic material as shown below. 8.4.3.1

Product Purification and Production

A good example of this kind of fractionation using both UF and NF can be found in the dairy industry. Mikkonen et al. [64] have studied the effect of NF on lactose crystallization. NF is a potential method for the partial demineralization and preconcentration of whey permeate, which has been prefiltered and deproteinated by UF. The process used in the study is shown in Figure 8.2. Processing whey permeate by NF increased the efficiency of lactose crystallization by 5–12%. Furthermore, both the amount and the ash content of the mother liquor resulting from the lactose crystallization decreased. 8.4.3.2

Fouling Prevention and Retention Enhancement

NF pretreatment with UF in the meat industry has been reported by Redondo [10]. The poultry and meat processing industry in southern Europe needs to treat and recirculate up to about 2000 m3 /d effluent water with a TDS content of 2.2–2.5%. Some of the treated effluent is needed as a process solution with some 1.0–1.5% salt concentration with low sulfate concentration. Hardness should not be completely removed because a small amount of calcium is desired. Reverse osmosis rejects too much of the multivalent salts to be considered a viable separation technology in this case. Th treatment of the whole effluent with loose NF membranes has been proposed as a way of solving the problem (Figure 8.3). However, due to the

393

394

8 Pretreatment and Hybrid Processes

Cheese whey

Ultrafiltration

Permeate

Retentate

Nanofiltration

Retentate

Permeate

Concentration

Crystallization

Wash water

Separation

Mother liquor

Drying

Lactose

Figure 8.2 Block diagram of the process used for lactose production. Source: By courtesy of Mikkonen et al. 2001 [64].

high fouling potential of the effluent, some pretreatment is needed. This can be performed by conventional filtration, MF, or UF. Both MF and UF are effective in removing bacteria, cysts, and other microbiological contents. UF can also remove many viruses. In addition, operational costs for membrane technology were calculated to be more than 30% lower than for conventional pretreatment processes. Therefore, MF and UF pretreatment should be preferred to conventional pretreatment (e.g. flocculation, settling media filtration, cartridge filtration) [10]. NF permeate and brine can then be treated further to achieve the desired quality. UF can also be used as a pretreatment to NF in the leather industry for the recovery of chromium from exhausted tanning baths [65]. The conventional method for chromium recovery is based on the precipitation of chromium salt with NaOH followed by the dissolution of Cr(OH)3 in sulfuric acid [66]. The quality of the recovered solution is, however, not always optimal [67]. A membrane process can be installed to improve the quality of the recycled chromium. The process would consist of UF and NF stages. Th original feed from the tanning bath is too concentrated to be efficiently treated with NF alone. UF rejects solid components and fat substances well, letting most of the chromium pass to the permeate side. The UF permeate is then fed to the NF

8.5 Nanofiltration as a Pretreatment

stage, in which the chromium is almost totally retained (99.9% with the tested membrane), leading to a very low chromium concentration in the NF permeate. After further concentration of the NF concentrate, the recovered chromium solution can be used in tanning and retanning processes [65]. In the sugar industry, UF/NF processes are feasible, e.g. for producing sugar from beets. The juice that is separated from the macerated beets can be treated with UF and further with NF, evaporation, and crystallization, to produce white sugar [68].

8.5 Nanofiltration as a Pretreatment In many processes dealing with waste- or process waters NF is not used as an “end of pipe technology.” This is due to the fact that the NF permeate is rarely clean enough and has to be treated further to meet process objectives. However, due to its fractionating properties, NF is well suited as a pretreatment stage for several other separation technologies. Th particular properties of NF that lend it to serving as a pretreatment are as follows: I. Removal of divalent ions that can lead to mineral scaling in high recovery desalination II. Removal of low molecular weight organics that can foul polymeric membranes used in desalination and greater resistance to organic and colloidal fouling III. Higher water permeabilities (LP ) and relatively lower monovalent salt rejection (𝜎 < 1) that allows application of lower pressures when operating NF as a pretreatment as illustrated by the equation: Jv = LP (
P  𝜎
) 8.5.1

(8.1)

Pretreatment Before Reverse Osmosis (RO)

Since the rejection patterns of NF and RO are partly different, the two membrane technologies can be combined for more efficient separation performance. Owing to the looser structure of NF membranes, NF can generally be utilized as a pretreatment stage prior to RO in order to purify water from substances that might damage the membranes during the RO stage. Usually NF can be used to remove multivalent salts and organic matter so that RO can be carried out without major scaling and fouling. NF particularly excels in rejecting sulfate, which contributes to the scale forming minerals gypsum, barite, and strontium sulfate. Th role of NF as RO pretreatment is further elaborated in Section 8.6 on NF in hybrid seawater desalination processes. Another reason for the use of NF as a pretreatment step is its ability to fractionate substances from each other. Nanofiltered permeate or concentrate streams containing the product are usually quite dilute. RO can be utilized for the concentration and further purification of the product stream, as was seen in Figure 8.3.

395

396

8 Pretreatment and Hybrid Processes Low SO4 content (< 100 mg/l)

Feed: 7.86 m3/h TDS 23.5 g/l

Process effluent

Screening

UF/MF

TDS 42 g/l

TDS 50 g/L

Disposal

Process reuse 1

NF

RO

TDS 15 g/l

Process reuse 2 TDS 0.6 g/l

Figure 8.3 Flow sheet of wastewater treatment process at a meat and poultry pilot plant. Source: By courtesy of Redondo 2001 [10].

8.5.2

Pretreatment Before Electrodialysis (ED)

Industrial effluents are usually too complex and concentrated for efficient separation with ED. NF as a pretreatment stage prior to ED can remove most of the contaminants, such as multivalent metals, which could inhibit the electrical separation process. Alternatively, NF can be applied as a post-treatment unit for the ED concentrate for selective separation or alternatively for purification of the dilute stream. Th emphasis of the few hybrid processes consisting of NF and ED is currently on wastewater treatment in the fields of the metal-, dairy-, and pulp and paper industry [69–71]. One example of the feasibility of an NF/ED process in the pulp and paper industry, reported by de Pinho et al., is the result of efforts to reduce water consumption. A kraft mill with an average production of a thousand tons of pulp per day faces the disposal of approximately 6000 m3 /day of the first alkaline bleaching streams, the E1 effluent. The composition of the E1 effluent is very complex and conventional biological and physicochemical treatments are very often not technically or economically feasible [71, 72]. A block diagram of the NF/ED hybrid process used in de Pinho et al.’s study is shown in Figure 8.4. The pilot plant was installed in a kraft pulp mill of Portucel (Setúbal, Portugal). Th effluent was fed directly from the E1 effluent process pipe line to the pilot plant. Pretreatment, which consisted of pH adjustment by HCl, flocculation/coagulation, and separation with a sand filter and a 5 m cartridge filter, was carried out to remove fibers and colloids from the effluent. The feed from the pretreatment stage was fed to the NF stage, which consisted of a SEPAREM spiral wound module with an effective area of 1.7 m2 . The tests were made at a pressure of 15 bar in 30–50  C and the NF permeate was fed to the ED unit with Selemion AMV and CMV membranes from Asahi Glass [73]. Rosa et al. [74] found that NF was feasible for removing low molar mass organics together with partial demineralization. NF alone could remove 95% color, 90% organics, and 30% salts from the E1 effluent. The multivalent salts were almost totally removed. Th quality of water necessary for recycling required the coupling of NF with other separation techniques. The processing of the NF permeate by ED was sufficient to decrease the NaCl content to 60 ppm. Optimization of the NF/ED sequence for E1 effluent treatment makes it possible

8.5 Nanofiltration as a Pretreatment

E1 Effluent

Pretreatment

Level 1

Level 2 NF

Polyelectrolyte

Process water

NF Concentrated effluent

HCl

ED Brine

Figure 8.4 Block diagram of an NF/ED hybrid process. Source: By courtesy of de Pinho and Depinho 1995 [73].

to produce water in which the multivalent ions are practically absent and the NaCl concentration is low [73]. 8.5.3

Pretreatment Before Ion Exchange (IX)

Th most important applications using both IX and NF include water softening and demineralization. Th separation mechanisms of both processes take electroneutrality into consideration. In many studies, NF is being compared to IX as an alternative treatment process. In some applications IX has been replaced by NF. The costs for the two processes are quite similar, so the choice of whether to install a membrane plant is based not on the economics of the process but on the quality of the wastewater produced as a by-product of water softening [75]. However, since the two processes handle the same problem in different ways, it is natural to speculate what a combination of the two as a hybrid process could do. Th problem with IX is that it has to be customized to a particular process. In some cases, this is not possible without NF as a pretreatment to remove some unwanted substances from the water. Also, when treating highly contaminated waters, the IX capacity is quickly reached and the process must be stopped for regeneration. With an NF step removing multivalent ions prior to IX the wastewater can be partially purified to lessen the burden of the ion exchange stage. Moreover, the regeneration process itself creates an effluent, which can be treated with NF. Th benefits of pre-demineralization with NF include [76, 77]: • • • •

need for a smaller IX demineralization plant; lower water and chemical consumption; reduced volume of regeneration effluent; higher capital cost of the combined system recouped through savings in operating costs.

Th combination of NF and IX has proved its efficiency in the dairy industry in Australia, where the process has been producing 90–95% de-ashed whey. The

397

398

8 Pretreatment and Hybrid Processes

process will give whey processors the ability to tailor different products to their customers’ needs [77]. In the metal industry, NF is used to concentrate dilute effluents that cannot be treated by IX directly. For example, in 1997 5 billion gallons of runoff accumulated in the pit and threatened the planned expansion of the Mexicana de Cananea open-pit copper mine. Th runoff contained rainwater, sulfuric acid used to leach copper from crushed rock, and copper worth approximately US$20 million. The copper concentration was too low for the liquid IX system of the mine to extract it, and installing a new system would have been too expensive. Th copper was concentrated twofold by an NF process (provided by GE Water), thus enabling the use of the existing IX system to recover the copper [78]. Further examples of NF operating with IX in metal recovery can be found in Chapter 19. 8.5.4

Pretreatment Before Evaporation

Evaporation is an efficient way to concentrate many solutions but the energy consumption during evaporation is extremely high (order of 25 kWh/m3 for brine concentrator mechanical vapor compression [MVC]), making it a very expensive process in most cases. Introducing an NF stage as a pre-concentration step prior to evaporation could prove economically feasible. When using the NF permeate as the feed water for distillation, more efficient processing is enabled in desalination due to the low multivalent salt concentration of the permeate. This is elaborated in Section 8.6 on the role of NF in seawater desalination. A good example of how an evaporation process can be enhanced using NF has been reported by Turek and Gonet [79] studying the production of sodium chloride from coal-mine brines in Poland. Th se brines are the most concentrated coal mine waters and can cause severe ecological problems. Brine concentration by evaporation in a 12 stage expansion installation or vapor compression unit entails very high energy consumption. This limits the use of coal-mine brines for industrial salt production. Th application of lower energy evaporation processes, such as multistage flash (MSF) and multiple effect distillation (MED), is limited by the high concentration of calcium and sulfate ions in the coal-mine brines. Using charged NF membranes with an adequate pore size makes it possible to decrease the concentration of divalent ions in the permeate without practically any changes in the concentration of sodium chloride [79]. What is especially important in this particular application is that NF rejects calcium and sulfate almost totally. In some cases, due to the Donnan effect, monovalent salts may even be concentrated in the permeate (negative rejection) if the multivalent salt content is sufficient [80].

8.6 NF in Hybrids Related to Seawater Desalination Th concepts and principles that make NF a good pretreatment or that make it useful due to its selectivity spectrum (removal of small molar mass organics and divalents) are well illustrated by a series of hybrid processes that incorporate NF in the desalination of seawater. Th s was reviewed by Zhou et al. [81] who

8.6 NF in Hybrids Related to Seawater Desalination

showed various schemes for incorporating NF in desalination processes. Th se were presented in Table 8.2 of Zhou et al. [81], and adapted here from their review to illustrate the concepts. In each of these schemes, one or more of the NF’s three features previously highlighted at the beginning of Section 8.5 are exploited. Using two-pass NF (second NF unit treating permeate of first NF unit) and then recycling the second pass NF concentrate to the first pass NF feed will potentially lower energy consumption by using membranes with much higher permeability allowing lower pressures for each stage [82]. However, the energy savings relative to RO are only found in cases where there is no energy recovery in the RO-only process. Placing NF in front of RO will result in lower organic foulants, lower salt load, and less scaling minerals to the RO step. As a result, higher permeability RO membranes (brackish water reverse osmosis [BWRO] instead of seawater reverse osmosis [SWRO]) can be used on the second pass, and they can reach higher in-stage recoveries, or alternatively operate at lower pressures while still providing high quality permeate (100 mg/l) [85]. NF is even more necessary for the hot-seawater desalination process, which requires the near complete removal of the scale-forming components in the seawater before proceeding to the RO or thermal stage. Membranes screened for this kind of pretreatment showed that NF90 had the best divalent cation rejection but lower flux, while NF270 and NF99HF had moderate cation rejection (60–80%) and all had excellent sulfate rejection [108]. In addition, the reduced consumption of chemicals makes the NF pretreatment process more friendly to the marine environment [88, 89], though today there are SWRO processes that work without chemicals [109]. A side benefit of the NF process in tandem with RO is that two separate concentrated salt streams are obtained (see Figure 8.5) that allow recovery of both sodium chloride and divalent sulfate minerals [87]. Adding an NF stage before the thermal processes (MSF and MED) leads to a significant improvement in the seawater desalination processes by lowering the total energy consumption by 25–30% as a result of raising the top brine temperature (TBT) to 125–130  C and thus increasing the gained output ratio (GOR). In the NF-MSF process, one study shows an increase of GOR from 8 to 16 [90]. In another study the NF step resulted in increased CO2 deaeration resulting in a 60% reduction in the CO2 heat transfer resistance in the MSF system [91]. The process involving NF and MSF together is shown in Figure 8.6a [88, 92, 110]. Because NF removes scaling ions (hardness and sulfate) it is possible to go for high recovery seawater desalination schemes. One such NF-RO scheme involves a thermal process (MSF, MED, MD) applied to the RO concentrate ([110]; see Figure 8.6b). Salt production is possible by applying a conventional or membrane crystallizer [95, 111] to the final brine. Th Drioli group has done a complete energy, exergy, and economic analysis of six different hybrid pressure-driven processes for seawater desalination involving NF [95]. The processes with thermal membrane crystallizer were the most economic if salts recovered from NF and RO concentrate could be sold [95]. Where NF is designed to be a pretreatment and not just as a means of removing the scaling ions and reducing salt load, then it should be robust enough that it can be used without an additional membrane pretreatment. Th s would be best

399

Table 8.2 Summary of hybrids incorporating NF in seawater desalination.

Strategy

Hybrid scheme

Dual pass pressure-driven NF–NF membrane

NF–RO

RO–NF NF as pretreatment to NF–MSF thermal process exploitation of non-scaling thermal process to recover NF–RO–MSF salt by-product

Switch feed to exploit NF properties

NF property exploiteda I = divalent retention II = ow molecular weight (LMW) organic retention III- high water permeability Advantage I, III

Lower energy

NF flux (l/m2 h)

Overall recovery (%)

Electric consumption (kWh/m3 ) References

20 (first)–42 (second)

37.0

3.35

[82, 83]

64 (first)–73 (second)

22.5

3.4

[84]

3.8

I, III

Lower energy (seawater 39 000)

33.7–36.4

I, II

Salt by-products from evaporative lagoons

80 (NF), 50 (overall)

III

Lower energy

I

Higher recovery, Higher TBT (lower thermal energy), CO2 deaeration Lower thermal energy and incorporation with power generation

I

[81] 40.6

65 (NF)

40.6

65 (NF)

I

Salable NaCl reduces product water costs

78.0

NF–RO–MD

92.7

See dual pass

Reduced scaling Salt production

FO–NF

I, II, III

Lower energy

HIX–NF

I, II, III

Lower energy

II, III

Enable waste-thermal energy use for regeneration

I

Lower energy

Polish/post-treatment draw FO-Thermal–NF solute recovery for B removal NF–NF–ED

Source: Adapted from Zhou et al. 2015 [81].

0.94 (NF)

[88–92]

[93]

54 (overall)

NF–RO–MED– Crystallizer

NF–RO–MD–MCr See dual pass

[85, 86] [87]

[94] 1.54–1.61a

25 (MgSO4 ), 10 (Na2 SO4 ) 12

[95, 96]

[97–99]

90 (NF)

[100–105] [106] 0.245 (B step)

[107]

Seawater intake

100 m3/d 2% NaCl 1% Ca/MgSO4

Monovalents NF

RO

50 m3/d

80 m3/d 2% NaCl 30 m3/d 5% NaCl 20 m3/d 5% Ca/MgSO4

Lagoon 2 Solar crystallization

Drinking water

NaCl For chloralkali industry

Concentration Lagoon 1a Solar crystallization

CaSO4•2H2O (s)

Mg2+,SO42–, Cl–

Lagoon 1b Solar crystallization

MgSO4•7H2O

Figure 8.5 Flow sheet for producing salable salts from seawater desalination in a two-pass NF-RO membrane system. Source: Eriksson et al. 2005 [87]. Reproduced with permission of Elsevier.

402

8 Pretreatment and Hybrid Processes

Pretreatment NF unit NF product

Seawater intake NF reject Seawater out

Steam MSF unit

Cooling seawater

Condensate

MSF product

(a)

Blow down Pretreatment NF unit

NF product

Seawater intake

NF reject Blended product

RO unit RO product

MSF Product

Steam

RO reject Seawater out

MSF unit

Cooling seawater

Condensate (b)

Figure 8.6 Schemes involving NF–MSF hybrids (Figure 3 of [110]). (a) An NF–MSF dihybrid. (b) Schematic of a trihybrid NF–RO–MSF process. Source: Hamed 2005 [110]. Reproduced with permission of Elsevier.

effected if the membrane were in hollow fiber configuration and allowed backwash or air sparging [112]. One of the proposed processes for desalination is to use forward osmosis and a draw solution that can be regenerated by a pressure driven membrane process. Th forward osmosis membranes are less susceptible to fouling and scaling and

8.7 NF as Post-treatment and Polishing Technology Disposal NaCl (seawater) Reject (NaCl) Step 3

Step 1 Anion exchanger in sulfate form. Exchanges chloride ions to produce sulfate ions in the effluent Na2SO4

Regeneration of anion exchanger. Anion exchanger is restored back to sulfate form while a reject stream of sodium chloride solution is produced NF membrane Reject stream rich in Na2SO4

Current cycle operation

Permeate (desalinated water)

Next cycle operation

Step 2 Nanofiltration of sodium sulfate solution

Figure 8.7 IX–NF process for seawater desalination. Source: Sarkar and SenGupta 2008 [100]. Reproduced with permission of Elsevier.

the clean simple salt draw solution is regenerated without the hazard of scaling and fouling. Draw solutions amenable to regeneration by NF include divalent salts, EDTA, sucrose, and hydroacids [97]. By using a divalent salt as the draw solution (e.g. MgSO4 ), it is possible to exploit NF rejection of divalent cations and higher permeability to reduce the energy consumption relative to the process of FO/RO by 25% [113]. If necessary, a two-pass NF scheme can be used to achieve adequate rejection of the draw solute, with recycle of the second pass NF concentrate back to the first pass [98]. As a polishing step in seawater desalination, NF can be used to remove residual surfactant from FO draw solute after it is regenerated by heating to its Krafft point [106] separating a surfactant-rich from an aqueous-rich phase. In another process, ED helps to remove boron by polishing the second pass NF permeate [107]. Another novel process that exploits NF selectivities uses anion exchange resins to treat seawater and replace chloride with sulfate, which is highly rejected by NF [100–103]. The NF then can be used to produce high quality water (see Figure 8.7), though methods must be found for preventing scaling by calcium sulfate in the NF stage.

8.7 NF as Post-treatment and Polishing Technology 8.7.1

Purification

Although NF is generally used as pretreatment for other separation technologies (in desalination and water treatment, for example), it can also be used for

403

404

8 Pretreatment and Hybrid Processes

post-treatment. In some cases, NF is sufficient to purify a solution of unwanted substances. This is the case with, for example, the post-treatment of MF or UF permeates, shown earlier in this chapter. In other cases, pre-concentration of NF feed water enables smaller NF units and thus more economical processing. If the end product can contain some impurities, such as monovalent salts and traces of organic contaminants, NF is usually sufficient as the polishing step in the separation. Usually the question is whether to use NF or RO membranes for the final treatment. In some instances, NF is preferred to RO if some of the salt content is wanted in the permeate. Th s is the case, for example, in the treatment of textile industry wastewater for reuse, reported by Marcucci et al. [114]. Generally this can be seen as a matter of quantity vs. quality of the permeate. In NF the flux is higher, while in RO the purity of the permeate is superior. In order to achieve an adequate purification of the permeate in finishing NF processes, it is quite common to have several consecutive separation stages. When treating process waters, the NF permeate and/or concentrate usually needs to be treated further in order to reach the desired product quality. These polishing processes include evaporation, crystallization, and drying, just to mention a few. Hybrid processes that were developed quite early were reported by Rautenbach et al., which combined NF with RO [115] or applied NF to ED concentrate and an inline crystallizer [116] in treating landfill leachates. In applications that use NF with conventional separation technologies for drinking water production, NF is usually situated near the end of a long line of processing. Polishing purification methods after NF may include oxidation and/or UV disinfection. An example from the food industry presented by Mavrov et al. [117] shows that when treating a low-contaminated process water (milk and meat processing), UV disinfection alone is sufficient to purify the NF permeate of most of the remaining microorganisms. Th treated water can then be reused as boiler makeup water or warm cleaning water. However, when treating industrial effluents, NF can be utilized in some applications as a true ‘end of pipe’ technology. In one study, Mutlu et al. [118] reported the decolourization of a baker’s yeast plant wastewater by NF that was pretreated with MF. The effluent contained an average of 4 g/l of COD and was successfully treated, removing 94% of COD and 89% of color. In addition to the food industry, NF has also been proved a satisfactory finishing separation technology in other fields of industry, such as the textile industry, as reported, e.g. by Marcucci et al. [114]. Th treatment of effluents containing large amounts of recalcitrants by conventional biological treatment requires high residence times. To overcome this, a hybrid process of an activated sludge bioreactor and NF (known as Biomembrat-Plus process) has been proposed to treat dumpsite leachate. The rate of biodegradation is increased by passing the NF concentrate to the bioreactor [50]. This process has been described in more detail in Chapter 17.

®

8.7.2

Fractionation

In some cases, the aim of NF is to fractionate the feed stream into permeate and concentrate streams, each consisting of a different product. In this way there are

8.7 NF as Post-treatment and Polishing Technology

no effluents, only product streams that can be re-used in other processes or used for the production of the final product. An example of the fractionating properties of NF is shown below: a case from the metal industry presented by Osmonics (now Suez Water Technologies). In this case, the treatment of acidic rinse waters, NF is feasible for the post-treatment of RO, since the initial feed is too dilute for direct NF fractionation and must be pre-concentrated by RO whereas the osmotic pressure of the sulfuric acid is too high to allow further concentration of the copper by RO alone. A copper rod refinery generates 17 000 l/h of a 2% sulfuric acid rinse stream containing 1.2 g/l soluble copper. A membrane system was installed to reduce the volume of the waste stream by concentrating and clarifying the acid for process reuse and recovering and concentrating the soluble copper [119] (Figure 8.8). Th two RO stages are loaded with a total of 84 Desal-3 RO elements, while the NF stage is loaded with 12 Desal-5 NF elements [119, 120]. Th membrane system has two stages. The first RO stage concentrates the acid from 2% to 10%. Th second RO stage further purifies the permeate from the first RO stage for reuse as process water. The concentrate of the second RO stage is fed back to the process. Th concentrated sulfuric acid and the copper stream from the first RO stage are fed to the NF unit. Sulfuric acid permeates the NF membranes while copper sulfate is retained [119]. Th NF permeate is fed back to the pickling bath, as the concentrate is concentrated further in an evaporator. The total system, with a feed flow of 17 000 l/h, reclaims approximately 333 kg of copper sulfate per day, 14 800 l/h of waste water, and 1800 l/h of 10% sulfuric acid at pH 0.9. Th s results in a total annual cost

Figure 8.8 A membrane process to treat acidic wastewater created by a copper rod refinery. The process includes two spiral-wound RO stages and one spiral-wound NF stage. Source: van der Merwe 1998 [119]. Reproduced with permission of Elsevier.

405

406

8 Pretreatment and Hybrid Processes

saving of more than US$ 560 000 and a payback time of less than one year on the capital costs of the system [119]. As the example shows, it is sometimes wise to concentrate the NF feed water. Th process would have worked without RO but with several drawbacks. Th process would have needed a larger multistage NF system, which in turn would have needed larger evaporation and IX units for the treatment of dilute copper solutions. Th combination of RO and NF seemed to be the most economical choice in this case. At that time the projected lifetime was only 9–15 months [119], but newer NF membranes show better acid resistance [121, 122]. See Chapter 19 for more elaboration on this point. 8.7.2.1

Antioxidant Recovery from Olive Mill Wastewater

Many of the options for NF pretreatment as well as its potential to perform fractionation are illustrated in its role in isolating and recovering antioxidants from the wastewater remaining from olive oil production. Th s is the water remaining from washing, and rinsing free the oil from the crushed olive pulp during olive oil production. OMW contains both high COD, high TOC, and high levels of polyphenols. Th polyphenols have potential as antioxidants in nutraceuticals and as starting materials for pharmaceuticals [123, 124]. Figure 8.9 shows the different antioxidants that can be prepared from hydrolysis of the oleupurein under acidic conditions. Two of the main polyphenols of interest as antioxidants are hydroxyltyrosol and tyrosol. NF membranes can be used to retain some of the polyphenols [125, 126], or a more open NF can be used to pass polyphenols while retaining other organic compounds [127, 128]. However, other components of the OMW

Figure 8.9 Various polyphenols generated by hydrolysis of oleuropein contained in the olives.

8.8 Conclusions

Olive mill wastewater

Hydroxytyrosol

NF

RO

Olive pomace, leaves

Centrifuge Solvent extraction Hydroxytyrosol

RO Supercritical extraction Figure 8.10 Flow sheet for isolating polyphenols from olive pomace and olive mill wastewater. Letters identifying the streams are explained in the patent text. Source: Figure 2 from Da Ponte et al. [128].

can cause severe fouling of NF membranes. Therefore, there is need for pretreatment steps to reduce fouling, which can include acidification/flocculation, physical separation (sieving and centrifugation), and either photocatalysis [129] or microfiltration [127] or ultrafiltration [130]. In order to separate the polyphenols from the other compounds contributing to COD, there is a need to separate the polyphenols from the other COD components. This can be partially done with ultrafiltration [126] or nanofiltration [127]. Figure 8.10 shows how using open NF can allow fractionating the polyphenols as a parallel process to supercritical extraction. Th remaining membrane concentrates can be sent to anaerobic digestion or as growth media for microalgae to produce biofuels [129]. Further steps in concentrating and refining polyphenols can be osmotic distillation [127], reverse osmosis [128], or resin columns.

8.8 Conclusions Th use of NF has greatly expanded in the last 10–15 years and this has been accompanied by a greater understanding of transport of ionic species [131] and organic species through the NF membrane. For the last 15 years research and development in NF has been intensive. Th growing number of NF-based hybrid processes in different fields of industry shows the ever-growing potential of NF as a versatile separation technology in water- and wastewater treatment. Another reason for the increase in research of hybrid processes is the current trend toward designing integrated systems in order to create processes that are more feasible both economically and environmentally. As long as the development of more resistant and stable NF membranes continues, the future of NF promises innovative applications as pre- and post-treatment

407

408

8 Pretreatment and Hybrid Processes

in several fields of industry. Currently, there are dozens of different NF membranes to choose from, with different characteristics and uses for different aqueous media. Th studies and applications of hybrid processes described in this chapter are, in fact, just the tip of an iceberg, considering the number of applications that currently utilize NF. In numerous applications NF alone can meet the separation efficiency needed for the process. In many other hybrid processes, NF has been coupled with IX, ED, MD, FO, supercritical extraction, UF and RO, tailored differently for each process. Th new separation technology has come a long way since its introduction in the middle of the 1980s. With stricter environmental regulations, industry is trying to reduce the contaminant levels of industrial effluents and to create more efficient treatment processes for their process waters. As knowledge of NF and its possibilities increases, more industrial-scale applications for the treatment of process- and wastewaters will emerge.

Acknowledgements Th Outokumpu Foundation is acknowledged for financial support.

Abbreviations AMP APS BOD CAPS COD DAF DTPMPA DU ED EMF FO GOR HEDP HERO IX MCr MD MED MF MVC MBR MSF

aminotrimethylenephosphonic acid advanced precipitation softening biological oxygen demand compact accelerated precipitation softening chemical oxygen demand dissolved air flotation diethylenetriamine-penta(methyl phosphonic) acid desaturation unit electrodialysis electromagnetic field forward osmosis gained output ratio 1-hydroxyethylidene-1,1-diphosphonic acid high efficiency reverse osmosis ion exchange membrane crystallizer membrane distillation multiple effect distillation microfiltration mechanical vapor compression membrane bioreactor multistage flash

References

NF NOM OMW PBTC PGW RO SDI SHMP SMBS TBT TDS TOC UF UV VRF

nanofiltration natural organic matter olive mill wastewater 2-phosphonobutane-1,2,4-tricarboxylic acid pressure groundwood reverse osmosis silt density index sodium hexametaphosphate sodium metabisulfite top brine temperature total dissolved solids total organic carbon ultrafiltration ultraviolet volume reduction factor

References 1 Speth, T.F., Gusses, A.M., and Summers, R.S. (2000). Evaluation of nanofil-

tration pretreatments for flux loss control. Desalination 130: 31–44.

2 Redondo, J. and Lomax, I. (2001). Y2K generation FILMTEC* RO mem-

3

4 5 6 7

8

9

10 11

branes combined with new pretreatment techniques to treat raw water with high fouling potential: summary of experience. Desalination 136: 287–306. Bremere, I., Kennedy, M., Michel, P. et al. (1999). Controlling scaling in membrane filtration systems using a desupersaturation unit. Desalination 124: 51–62. Vedavyasan, C.V. (2001). Potential use of magnetic fields – a perspective. Desalination 134: 105–108. Stephenson, T., Judd, S.J., Jefferson, B., and Brindle, K. (2001). Membrane Bioreactors for wastewater treatment. London: IWA Publishing. Gilron, J., Chaikin, D., and Daltrophe, N. (2000). Demonstration of CAPS pretreatment of surface water for RO. Desalination 127: 271–282. Kedem, O. and Zalmon, G. (1997). Compact accelerated precipitation softening (CAPS) as a pretreatment for membrane desalination I. Softening by NaOH. Desalination 113: 65–71. Masarwa, A., Meyerstein, D., Daltrophe, N., and Kedem, O. (1997). Compact accelerated precipitation softening (CAPS) as pretreatment for membrane desalination II. Lime softening with concomitant removal of silica and heavy metals. Desalination 113: 73–84. Al-Zoubi, H., Al-Thyabat, S., and Al-Khatib, L. (2009). A hybrid flotation-membrane process for wastewater treatment: an overview. Desalin. Water Treat. 7: 60–70. Redondo, J. (2001). Combined systems tackle marginal resources. Desalination and Water Reuse Quarterly 11: 37–45. Gilron, J. (2013). Scaling. In: Encyclopedia of Membrane Science and Technology (eds. E.M.V. Hoek and V.V. Tarabara). Wiley.

409

410

8 Pretreatment and Hybrid Processes

12 Gilron, J., Daltrophe, N., Waissman, M., and Oren, Y. (2005). Comparison

13

14

15

16 17 18 19

20 21 22

23

24 25 26

27

between Compact Accelerated Precipitation Softening (CAPS) and conventional pretreatment in operation of brackish water reverse osmosis (BWRO). Ind. Eng. Chem. Res. 44: 5465–5471. Tran, A.T.K., Zhang, Y., Jullok, N. et al. (2012). RO concentrate treatment by a hybrid system consisting of a pellet reactor and electrodialysis. Chem. Eng. Sci. 79: 228–238. Fritz, C.H. and Tiwari, C.K. (2002), An economical new zero liquid discharge approach for power plants. Power-Gen International, Orlando, FL (10–12 December 2002). McCool, B.C., Rahardianto, A., and Cohen, Y. (2012). Antiscalant removal in accelerated desupersaturation of RO concentrate via chemically-enhanced seeded precipitation (CESP). Water Res. 46: 4261–4271. Darton, E.G. (1997). Scale inhibition techniques used in membrane systems. Desalination 113: 227–229. Patel, S. and Finan, M.A. (1999). New antifoulants for deposit control in MSF and MED plants. Desalination 124: 63–74. Gill, J.S. (1999). A novel inhibitor for scale control in water desalination. Desalination 124: 43–50. Patel, S. and Milligan, J. (2008). Magnesium hydroxide deposit control for reverse osmosis systems. In: CORROSION 2008, 14. New Orleans, LA: NACE International. Amjad, Z. (1995). Mineral Scale Formation and Inhibition. New York: Plenum Press. Amjad, Z. (1997). RO systems – current fouling problems and solutions. Desalination and Water Reuse Quarterly 6: 55–60. Neofotistou, E. and Demadis, K.D. (2004). Use of antiscalants for mitigation of silica (SiO2 ) fouling and deposition: fundamentals and applications in desalination systems. Desalination 167: 257–272. M. Truby (2012). Scale Formation and Dendrimer Chemistry. Presented at Kingdom of Saudia Arabia Water and Power Forum. Jeddah, Saudi Arabia (2–4 December 2012). http://ksawpf.com/wp-content/uploads/2012/12/ OMYA.pdf (accessed August 2016). Amjad, Z. (1985). Applications of antiscalants to control calcium-sulfate scaling in reverse-osmosis systems. Desalination 54: 263–276. Darton, E.G. (1999). RO plant experiences with high silica waters in the Canary Islands. Desalination 124: 33–41. Sheikholeslami, R., Al-Mutaz, I.S., Tan, S., and Tan, S.D. (2002). Some aspects of silica polymerization and fouling and its pretreatment by sodsodium aluminate, lime and soda ash. Desalination 150: 85–92. A. Kempter, T. Gaedt, V. Boyko, et al. (2013). Controlling silica in water treatment applications. Proceedings IDA World Congress on Desalination and Water Reuse 2013, International Desalination Association, Tianjin, China, IDAWC/TIAN13–103 (21–24 October 2013). http:// www.watersolutions.basf.com/ev/internet/watersolutions/en/function/ conversions:/publish/content/watersolutions/desalination/controlling_silica_ in_water_treatment_applications_tian13-103.pdf.

References

28 Yiantsios, S.G. and Karabelas, A.J. (1998). The effect of colloid stability on

membrane fouling. Desalination 118: 143–152.

29 Hater, W., zum Kolk, C., Dupoiron, C. et al. (2011). Silica scaling on reverse

30

31

32

33 34

35 36 37

38 39 40

41

42

43

44

osmosis membranes – investigation and new test methods. Desalin. Water Treat. 31: 326–330. Tho pson, M.A. and Azar, B. (1998). Evaluation of conventional and membrane processes for softening a North Carolina groundwater. Desalination 118: 229–237. van Paassen, J.A.M., Kruithof, J.C., Bakker, S.M., and Kegel, F.S. (1998). Integrated multi-objective membrane systems for surface water treatment: pre-treatment of nanofiltration by riverbank filtration and conventional ground water treatment. Desalination 118: 239–248. Gryta, M. (2011). The influence of magnetic water treatment on CaCO3 scale formation in membrane distillation process. Sep. Purif. Technol. 80: 293–299. Baker, J.S., Judd, S.J., and Parsons, S.A. (1997). Antiscale magnetic pretreatment of reverse osmosis feedwater. Desalination 110: 151–166. Hoek, E. and Elimelech, M. (2003). Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes. Environ. Sci. Technol. 37: 5581–5588. Tho pson, J., Lin, N., Lyster, E. et al. (2012). RO membrane mineral scaling in the presence of a biofilm. J. Membr. Sci. 415–416: 181–191. Boerlage, S.F.E., Kennedy, M.D., Aniye, M.P. et al. (1998). Monitoring particulate fouling in membrane systems. Desalination 118: 131–142. Tabatabai, S.A.A., Kennedy, M.D., Amy, G.L., and Schippers, J.C. (2009). Optimizing inline coagulation to reduce chemical consumption in MF/UF systems. Desalin. Water Treat. 6: 94–101. Nystrom, M., Kaipia, L., and Luque, S. (1995). Fouling and retention of nanofiltration membranes. J. Membr. Sci. 98: 249–262. Nystrom, M., Ruohomaki, K., and Kaipia, L. (1996). Humic acid as a fouling agent in filtration. Desalination 106: 79–87. Manttari, M., Puro, L., Nuortila-Jokinen, J., and Nystrom, M. (2000). Fouling effects of polysaccharides and humic acid in nanofiltration. J. Membr. Sci. 165: 1–17. Schafer, A.I., Fane, A.G., and Waite, T.D. (1998). Nanofiltration of natural organic matter: removal, fouling and the influence of multivalent ions. Desalination 118: 109–122. Vaisanen, P., Huuhilo, L., Puro, L. et al. (1999). Effect of pretreatments on membrane filtration in the pa-permaking process. In: Th Yearbook of Cactus – Technology Program, 23–34. Jyvaskyla, Finland: VTT Energy. Tunçal, T. and Kaygusuz, T. (2014). Treatment of dyehouse effluents using sequential combinations of electrochemical oxidation, membrane separation, and activated sludge. Environ. Prog. Sustain. Energy 33: 472–481. Ganiyu, S.O., Van Hullebusch, E.D., Cretin, M. et al. (2015). Coupling of membrane filtration and advanced oxidation processes for removal of pharmaceutical residues: a critical review. Sep. Purif. Technol. 156: 891–914.

411

412

8 Pretreatment and Hybrid Processes

45 Baker, J.S. and Dudley, L.Y. (1998). Biofouling in membrane systems – a

review. Desalination 118: 81–89.

46 Vrouwenvelder, J.S., Beyer, F., Dahmani, K. et al. (2010). Phosphate limita-

tion to control biofouling. Water Res. 44: 3454–3466.

47 Aquafine Corporation. Chlorines, Chloramines destruction. https://www

48 49

50 51

52

53

54

55

56

57

58

59

60

.aquafineuv.com/Chlorines-Chloramines-Destruction (accessed August 2016). Bertheas, U., Majamaa, K., Arzu, A., and Pahnke, R. (2009). Use of DBNPA to control biofouling in RO systems. Desalin. Water Treat. 3: 175–178. Manttari, M., Kaipainen, E., and Nyström, M. (2000). Flux enhancement in nanofiltration of pulp and paper effluents. In: World Filtration Congress 8 (3–7 April 2000), Brighton, 1003–1006. Elsevier. Rautenbach, R. and Mellis, R. (1994). Waste-water treatment by a combination of bioreactor and nanofiltration. Desalination 95: 171–188. Reiss, C.R., Robert, C., Langenderfer, N., and Taylor, J.S. (1997). Integrated membrane systems for treatment of surface waters. In: Annual Conference of American Water Works Association, Atlanta, GA, 381–399. Merkel, T.H., Speth, T.F., Wang, J., and Summers, R.S. (1998). The performance of nanofiltration in the treatment of bank-filtered and conventionally-treated surface water. In: Water Quality Technology Conference, Denver, CO, 270–281. Wu, Y., Xia, S., Dong, B. et al. (2013). Study on surface water treatment by hybrid sand filtration and nanofiltration. Desalin. Water Treat. 51: 5327–5336. Ventresque, C. and Bablon, G. (1997). The integrated nanofiltration system of the Mery-sur-Oise surface water treatment plant (37 mgd). Desalination 113: 263–266. (1996). Laundry Waste 130, Osmonics Nanofiltration Application Bulletin. http://www.roplant.or.kr/_admin/_board/_inc/filedown.asp? FileName=Industrial_Catalog.pdfhttp://www.roplant.or.kr/_admin/_board/_ inc/filedown.asp?FileName=Industrial_Catalog.pdf (accessed August 2016). Envirochemie (2006). Ideal treatment of laundry Wastewater with Ultrafiltration Installations. WRP https://envirochemie.com/cms/upload/downloadsen/fachbeitraege/Treatment_of_laundry_wastewater.pdf (accessed August 2016). Masmoudi, G., Trabelsi, R., Ellouze, E., and Amar, R.B. (2014). New treatment at source approach using combination of microfiltration and nanofiltration for dyeing effluents reuse. Int. J. Environ. Sci. Technol. 11: 1007–1016. Unlu, M., Yukseler, H., and Yetis, U. (2009). Indigo dyeing wastewater reclamation by membrane-based filtration and coagulation processes. Desalination 240: 178–185. Gozálvez-Zafrilla, J.M., Sanz-Escribano, D., Lora-García, J., and León Hidalgo, M.C. (2008). Nanofiltration of secondary effluent for wastewater reuse in the textile industry. Desalination 222: 272–279. Tunç, M.S., Y𝚤lmaz, L., Yeti¸s, Ü., and Çulfaz-Emecen, P.Z. (2014). Purification and concentration of caustic mercerization wastewater by membrane

References

61

62

63

64 65

66

67

68

69 70

71 72

73

74

75

processes and evaporation for reuse. Sep. Sci. Technol. (Philadelphia) 49: 1968–1977. Varol, C., Uzal, N., Dilek, F.B. et al. (2015). Recovery of caustic from mercerizing wastewaters of a denim textile mill. Desalin. Water Treat. 53: 3418–3426. Sievers, L.J., Binder, T.P., and Hadden, D.K. (1999). Nanofiltration process for making dextrose. US Patent 5,869,297. Archer Daniels Midland, filed 10 June 1992 and issued 9 February 1999. Sun, F.Y., Dong, W.Y., and Guo, X. (2012). Comparison of pre-treatment methods for nanofiltration (NF) in drinking water treatment related with organic removal and membrane fouling mitigation. Adv. Mater. Res.: 335–339. Mikkonen, H., Helakorpi, P., Myllykoski, L., and Keiski, R.L. (2001). Effect of nanofiltration on lactose crystallisation. Milchwissenschaft 56: 307–310. Cassano, A., Molinari, R., Romano, M., and Drioli, E. (2001). Treatment of aqueous effluents of the leather industry by membrane processes – a review. J. Membr. Sci. 181: 111–126. Cassano, A., Drioli, E., Molinari, R., and Bertolutti, C. (1997). Quality improvement of recycled chromium in the tanning operation by membrane processes. Desalination 108: 193–203. Esmaeili, A., Mesdaghinia, A., and Vazirinejad, R. (2005). Chromium(III) removal and recovery from tannery wastewater by precipitation process. Am. J. Appl. Sci. 2: 1471–1473. Donovan, M., Jansen, R.P., Hlavacek, M., et al. (2002). Sugar beet membrane filtration process. US Patent 6,406,547B1. Tate and Lyle Industries Ltd, filed 18 July 2000 and issued 18 June 2002. Adiga, M. (1997). Treatment of plating wastewater for removal of metals. US Patent 6,162,361, filed 14 February 1997 and issued 19 December 2000. Gernigon, G., Schuck, P., Jeantet, R., and Burling, H. (2011). Whey processing: demineralization. In: Encyclopedia of Dairy Sciences, 2e, 738–743. Elsevier. de Pinho, M.N., Geraldes, V.M., Rosa, M.J. et al. (1996). Water recovery from bleached pulp effluents. TAPPI J. 79: 117–124. Afonso, M.D., Geraldes, V., Rosa, M.J., and De Pinho, M.N. (1992). Nanofiltration removal of chlorinated organic compounds from alkaline bleaching effluents in a pulp and paper plant. Water Res. 26: 1639–1643. Geraldes, V. and Depinho, M.N. (1995). Process water recovery from pulp bleaching effluents by an NF/ED hybrid process. J. Membr. Sci. 102: 209–221. Rosa, M.J. and Depinho, M.N. (1995). The role of ultrafiltration and nanofiltration on the minimization of the environmental-impact of bleached pulp effluents. J. Membr. Sci. 102: 155–161. Canepa, P., Garombo, C., Szpyrkowicz, L., and ZilioGrandi, F. (1996). Comparison between ion exchange and nanofiltration for softening of industrial water. Filtr Separat. 33: 131–135.

413

414

8 Pretreatment and Hybrid Processes

76 Okawa, T., Shimada, M., Ushida, Y. et al. (2015). Demineralisation of whey

77

78 79 80 81

82

83 84

85

86

87

88

89 90

91

92

by a combination of nanofiltration and anion-exchange treatment: a preliminary study. Int. J. Dairy Technol. 68: 478–485. Hutson, T. (1997). Nanofiltration and ion-exchange for the demineralisation of whey. In: Second International Whey Conference, 88–92. Chicago, IL: International Dairy Federation. Tanninen, J., Mänttäri, M., and Nyström, M. (2006). Nanofiltration of concentrated acidic copper sulphate solutions. Desalination 189: 92–96. Turek, M. and Gonet, M. (1997). Nanofiltration in the utilization of coal-mine brines. Desalination 108: 171–177. Rautenbach, R. and Groeschl, A. (1990). Separation potential of nanofiltration membranes. Desalination 77: 73–84. Zhou, D., Zhu, L., Fu, Y. et al. (2015). Development of lower cost seawater desalination processes using nanofiltration technologies – a review. Desalination 376: 109–116. Vuong, D. (2006). Two stage nanofiltration seawater desalination system. US Patent number 7,144,511. City of Long Beach, filed 2 May 2002 and issued: 5 December 2006 Adham, S., Cheng, C.Y., and Vuong, D. (2003). Long Beach’s dual-stage NF beats single-stage SWRO. Desalin. Water Reuse (Quarterly) 13: 18–21. Liu, J., Xie, L., Wang, Z., and Yuan, J. (2014). Dual-stage nanofiltration seawater desalination: water quality, scaling and energy consumption. Desalin. Water Treat. 52: 134–144. Altaee, A. and Sharif, A.O. (2011). Alternative design to dual stage NF seawater desalination using high rejection brackish water membranes. Desalination 273: 391–397. Song, Y., Su, B., Gao, X., and Gao, C. (2013). Investigation on high NF permeate recovery and scaling potential prediction in NF-SWRO integrated membrane operation. Desalination 330: 61–69. Eriksson, P., Kyburz, M., and Pergande, W. (2005). Desalination and the environment NF membrane characteristics and evaluation for sea water processing applications. Desalination 184: 281–294. Hassan, A.M., Al-Sofi, M.A.K., Al-Amoudi, A.S. et al. (1998). A new approach to membrane and thermal seawater desalination processes using nanofiltration membranes (Part 1). Desalination 118: 35–51. Al-Sofi, M.A.K., Hassan, A.M., Hamed, O.A. et al. (2000). Optimization of hybridized seawater desalination process. Desalination 131: 147–156. Mabrouk, A.N.A. and Fath, H.E.-b.S. (2013). Techno-economic analysis of hybrid high performance MSF desalination plant with NF membrane. Desalin. Water Treat. 51: 844–856. Al-Rawajfeh, A.E. (2016). Nanofiltration pretreatment as CO2 deaerator of desalination feed: CO2 release reduction in MSF distillers. Desalination 380: 12–17. A.M.U.S. Hassan (2003). Process for desalination of saline water, especially water, having increased product yield and quality. Saline Water Conversion Corporation. US patent 6508936. Saline Water Conversion Corporation (SWCC), filed 1 October 1997 and issued: 21 January 2003.

References

93 Hamed, O.A., Hassan, A.M., Al-Shail, K., and Farooque, M.A. (2009). Perfor-

94

95

96

97 98

99

100

101

102

103

104

105

106

107

mance analysis of a trihybrid NF/RO/MSF desalination plant. Desalin. Water Treat. 1: 215–222. Turek, M. and Chora˛zewska, ̇ M. (2009). Nanofiltration process for seawater desalination–salt production integrated system. Desalin. Water Treat. 7: 178–181. Macedonio, F., Curcio, E., and Drioli, E. (2007). Integrated membrane systems for seawater desalination: energetic and exergetic analysis, economic evaluation, experimental study. Desalination 203: 260–276. Drioli, E., Curcio, E., Di Profio, G. et al. (2006). Integrating membrane contactors technology and pressure-driven membrane operations for seawater desalination: energy, exergy and costs analysis. Chem. Eng. Res. Des. 84: 209–220. Luo, H., Wang, Q., Zhang, T.C. et al. (2014). A review on the recovery methods of draw solutes in forward osmosis. J. Water Process Eng. 4: 212–223. Tan, C.H. and Ng, H.Y. (2010). A novel hybrid forward osmosis – nanofiltration (FO-NF) process for seawater desalination: draw solution selection and system configuration. Desalin. Water Treat. 13: 356–361. Zhao, S., Zou, L., and Mulcahy, D. (2012). Brackish water desalination by a hybrid forward osmosis-nanofiltration system using divalent draw solute. Desalination 284: 175–181. Sarkar, S. and SenGupta, A.K. (2008). A new hybrid ion exchange-nanofiltration (HIX-NF) separation process for energy-efficient desalination: process concept and laboratory evaluation. J. Membr. Sci. 324: 76–84. Hilal, N., Kochkodan, V., Al Abdulgader, H. et al. (2015). A combined ion exchange–nanofiltration process for water desalination: I. sulphate–chloride ion-exchange in saline solutions. Desalination 363: 44–50. Hilal, N., Kochkodan, V., Al Abdulgader, H., and Johnson, D. (2015). A combined ion exchange–nanofiltration process for water desalination: II. Membrane selection. Desalination 363: 51–57. Hilal, N., Kochkodan, V., Al Abdulgader, H. et al. (2015). A combined ion exchange–nanofiltration process for water desalination: III. Pilot scale studies. Desalination 363: 58–63. SenGupta, A.K. and Sarkar, S. (2011). Brackish and sea water desalination by a hybrid ion exchange/nanofiltration (HIX-NF) process. US Patent 7,901,577. Inventors, filed 13 February 2008 and issued 8 March 2011. Sarkar, S. and SenGupta, A.K. (2009). A hybrid ion exchange-nanofiltration (HIX-NF) process for energy efficient desalination of brackish/seawater. Water Sci. Technol. Water Supply 9: 369–377. Hoyer, M., Haseneder, R., and Repke, J.-U. (2016). Development of a hybrid water treatment process using forward osmosis with thermal regeneration of a surfactant draw solution. Desalin. Water Treat.: 1–14. Turek, M., Dydo, P., and Bandura-Zalska, B. (2009). Boron removal from dual-staged seawater nanofiltration permeate by electrodialysis. Desalin. Water Treat. 10: 60–63.

415

416

8 Pretreatment and Hybrid Processes

108 Llenas, L., Ribera, G., Martínez-Lladó, X. et al. (2013). Selection of nanofil-

109

110 111 112 113 114

115

116

117

118

119 120 121

122

123 124

tration membranes as pretreatment for scaling prevention in SWRO using real seawater. Desalin. Water Treat. 51: 930–935. AquaSwiss RO. Green sustainable water production. http://www.aquaswiss .eu/wp-content/themes/aquaswiss/images/brochure_final%20RO%20Green %202018.pdf. . Hamed, O.A. (2005). Overview of hybrid desalination systems – current status and future prospects. Desalination 186: 207–214. Drioli, E., Di Profio, G., and Curcio, E. (2012). Progress in membrane crystallization. Curr. Opin. Chem. Eng. 1: 178–182. Futselaar, H., Schonewille, H., and van der Meer, W. (2003). Direct capillary nanofiltration for surface water. Desalination 157: 135–136. Altaee, A., Ismail, A.F., Sharif, A. et al. (2016). Two-stage FO-BWRO/NF treatment of saline waters. Desalin. Water Treat. 57: 4842–4852. Marcucci, M., Nosenzo, G., Capannelli, G. et al. (2001). Treatment and reuse of textile effluents based on new ultrafiltration and other membrane technologies. Desalination 138: 75–82. Rautenbach, R. and Linn, T. (1996). High-pressure reverse osmosis and nanofiltration, a ‘zero discharge’ process combination for the treatment of waste water with severe fouling/scaling potential. Desalination 105: 63–70. Rautenbach, R., Kopp, W., and Herion, C. (1989). Electrodialysis-contact sludge reactor and reverse osmosis-phase separator two examples of a simple process combination for increasing the water recovery rate of membrane processes. Desalination 72: 339–349. Mavrov, V., Chmiel, H., and Belieres, E. (2001). Spent process water desalination and organic removal by membranes for water reuse in the food industry. Desalination 138: 65–74. Mutlu, S.H., Yetis, U., Gurkan, T., and Yilmaz, L. (2002). Decolorization of wastewater of a baker’s yeast plant by membrane processes. Water Res. 36: 609–616. van der Merwe, I.W. (1998). Application of nanofiltration in metal recovery. J. South Afr. Inst. Min. Metall.: 339–342. Acid Waste Treatment, Nanofiltration Application Bulletin 126, in Osmonics-Desal Industrial/commercial catalogue, 1998 Tanninen, J., Platt, S., Weis, A., and Nyström, M. (2004). Long-term acid resistance and selectivity of NF membranes in very acidic conditions. J. Membr. Sci. 240: 11–18. Yun, T., Chung, J.W., and Kwak, S.-Y. (2018). Recovery of sulfuric acid aqueous solution from copper-refining sulfuric acid wastewater using nanofiltration membrane process. J. Environ. Manag. 223: 652–657. Visoli, F., Romani, A., Mulinacci, N. et al. (1999). Antioxidant and other biological activities of olive mill waste. J. Agric. Food Chem. 47: 3397–3401. Vaya, J. and Aviram, M. (2001). Nutritional antioxidants mechanisms of action, analyses of activities and medical applications. Curr Med Chem Immunol Endocr Metab Agents 1: 99–117.

References

125 Russo, C. (2007). A new membrane process for the selective fractiona-

126 127

128

129

130

131

tion and total recovery of polyphenols, water and organic substances from vegetation waters. J. Membr. Sci. 288: 239. Paraskeva, C.A., Papadakis, V.G., Tsarouchi, E. et al. (2007). Membrane processing for olive mill wastewater fractionation. Desalination 213: 218–229. Garcia-Castello, E., Cassano, A., Criscuoli, A. et al. (2010). Recovery and concentration of polyphenols from olive mill wastewaters by integrated membrane system. Water Res. 44: 3883–3892. Da Ponte, M.L., Dos Santos, J.L.C., Matias, A.A.F., et al. (2011), Method of obtaining a natural hydroxtyrosol-rich concentrate from olive tree residues and subproducts using clean technologies. US Patent 8,066,881 B2, filed 28 January 2008 and issued 29 November 2011. Cicci, A., Stoller, M., and Bravi, M. (2013). Microalgal biomass production by using ultra- and nanofiltration membrane fractions of olive mill wastewater. Water Res. 47: 4710–4718. Coskun, T., Debik, E., and Demir, N.M. (2010). Treatment of olive mill wastewaters by nanofiltration and reverse osmosis membranes. Desalination 259: 65–70. Wang, J., Dlamini, D.S., Mishra, A.K. et al. (2014). A critical review of transport through osmotic membranes. J. Membr. Sci. 454: 516–537.

417

419

Part II Applications

421

9 Water Treatment Erich Wittmann 1 , Edvard Sivertsen 2 , and Thor Thorsen 2 1 30 rue Madeleine Vionnet, F-93300 Aubervilliers, France 2

SINTEF, S.P. Andersens veg 3, N-7034 Trondheim, Norway

9.1 Introduction A number of factors have been contributing to the growing interest in using membrane processes for the production of drinking water. The most important one is the trend to more and more stringent drinking water quality regulations that, in some cases, can only be met, in an economically viable way, by membrane processes. Relatively high investment and operation costs of membrane plants compared to conventional plants have been, for many years, an obstacle to quick implementation of membrane processes in the drinking water sector, where the added-value level of the product (water) is very low. However, this gap is becoming smaller and smaller because of strong efforts in Research and Development, standardization, and expansion in membrane, module, and membrane system production capacities, which allow to reduce equipment costs and to optimize the performance of the membrane processes. A further explanation for the increasing use of membrane processes in drinking water treatment is the fact that, in some cases, membrane treatment can be cheaper than conventional treatment because the undesirable components can be removed with only one treatment step, whereas a conventional treatment would require several different steps. This is particularly true for nanofiltration, which, in one step, removes dissolved organic and inorganic compounds.

9.2 Overview of Nanofiltration Applications in Drinking Water In water treatment, the separation capacities of nanofiltration allow the elimination of small dissolved organic molecules (micropollutants, disinfection by-product precursors, etc.), together with more or less significant, partial transmission of salts through the membrane [1–3]. Nanofiltration: Principles, Applications, and New Materials, Second Edition. Edited by Andrea Iris Schäfer and Anthony G. Fane. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

422

9 Water Treatment

Reverse osmosis has separation capacities similar to nanofiltration with higher rejection rates and can thus sometimes be a competing process. However, reverse osmosis is generally, because of the higher required pressures (lower permeability), more expensive in both investment and operation costs. The power consumption is significantly lower for nanofiltration compared to reverse osmosis. Furthermore, partial passage of calcium and bicarbonate through nanofiltration can be an advantage because drinking water distributed through a network should be saturated for calcium carbonate in order to avoid corrosion (calcium carbonate stabilization). Nanofiltration started to be applied for drinking water production in the late 1980s for the treatment of hard and colored water in Florida. Today, many different existing applications or potential future applications combine one or several of the following removal capacities of nanofiltration: • Dissolved mineral components – Hardness and alkalinity (softening) [4–8] – Sulfate [9, 10] – Nitrate [11–13] – Fluoride [11, 14, 15] – Other inorganic micropollutants and metals (As, Se, Ni, Cr, Cd, Fe, Mn, U, etc.) [13, 14, 16] • Dissolved organic components – Natural organic matter (NOM), organic color, and algal toxins [4, 5, 17–24] – Biodegradable dissolved organic molecules [6–8] ⚬ Biodegradable dissolved organic carbon (BDOC) ⚬ Assimilable organic carbon (AOC) – Dissolved organic molecules reacting with disinfection chemicals [6–8, 25–28] ⚬ Disinfection by-product (DBP) formation potential ⚬ Chlorine demand – Organic micropollutants ⚬ Pesticides [6–8, 29] ⚬ Emerging organic micropollutants (endocrine disruptors, pharmaceuticals, etc.) [25, 30–32] – Taste and odors [6–8] Th feedwater used can be any nonbrackish groundwater or surface water. For treatment of brackish water, nanofiltration is often not the suitable process because Cl and Na+ are among the ions with rather low rejection rates for many nanofiltration membranes. Nevertheless, very tight nanofiltration membranes (e.g. Dupont Filmtec NF90) allow in some cases to achieve a reasonably good reduction in chloride or in salinity [10, 15, 33]. Thus, these membranes, which, in terms of salt rejection, are in a gray zone between nanofiltration and reverse osmosis, can be used for the treatment of brackish water. Recently, nanofiltration has also received increasing attention in retrofitting of existing facilities, especially when targeting enhanced natural organic removal [34–36].

9.3 Plant Design

9.3 Plant Design 9.3.1

Membrane Selection

Th membrane choice is essentially determined by the feedwater composition and the desired permeate quality. Th process performance does not only depend on the membrane performance but also on the system design (especially system recovery) including pre- and post-treatment. This may imply an iterative procedure between membrane selection, design, and product quality calculations until the desired final water quality is attained. In general, the membrane manufacturer provides a computer program that does most of the calculation work (Dupont Filmtec: ROSA and WAVE, Hydranautics: IMSdesign, etc.). Some of them can be downloaded from the manufacturer’s internet site. For some water components, the membrane rejection capacity is sometimes not known or can be very dependent on the concentration of other components. In these cases, a pilot study (feasibility trial or an experimental membrane screening) may be required. 9.3.1.1

Feedwater Characterization

It is essential to have a maximum of parameters to be analyzed. Th feed quality together with the treatment objectives are among the most important factors that determine pretreatment requirements, membrane choice, design of the membrane treatment unit, and operation conditions. Specific ion concentrations are required for the evaluation of the scaling potential of the concentrate and the corrosion potential of the treated water. For parameters that vary over the time (e.g. temperature and pesticide concentrations), the variation range has to be determined. Some parameters will be changed by a pretreatment, and their values at the inlet of the membrane treatment unit need to be established by calculation or a conservative estimation. Th most important parameters are temperature, pH, total dissolved solids (TDS), alkalinity, sulfate, chloride, fluoride, nitrate, orthophosphate, calcium, ammonium, barium, strontium, iron, manganese, aluminum, silica, hydrogen sulfide, color, total organic carbon (TOC), dissolved organic carbon (DOC) and its biodegradable fraction (BDOC), free chlorine, total bacterial count, turbidity, and silt density index (SDI). A complete ionic balance is very useful for the verification of the ion concentrations. The feedwater characterization needs to be completed by the analysis of any other parameters for which a quality limit is given by the drinking water standard, if these components are suspected to be present in the water in a significant amount. 9.3.1.2

Membrane Material and Module Type

The e are four major categories of nanofiltration membranes that are used in drinking water production: • Flat sheet composite membranes in spiral wound modules • Flat sheet asymmetric membranes in spiral wound modules

423

Table 9.1 Typical commercial nanofiltration membrane modules.

Commercial namea)

Chemical familyb)

Membrane structure and modulec)

Salt rejectiond) (%)

Module flow rate (m3 /d)

Chlorine tolerancee)

Test conditions solution

Pressure (bar)

Recovery (%)

15

MUNI-NF-400

PA

CS

98

43.5

80

20–80

50–80

20–50

90

>90

>90

>90

>90

>90

>90

>95

50–80

>95

50–80

>95

>80

50–90

50–90

GAC Ozonation

>80

50–80

>80

>80

>80

>80

>80

50–80

>90

>80

>80

80

80

Na+ , respectively. When each ion rejection in groundwater NF treatment was compared to that in individual salt solutions, all monovalent ion rejections were significantly lower (over 45%) due, in part, to differences in water characteristics (e.g. increase in ionic strength); the increase in salt concentration decreases the electrostatic repulsion (due to double layer compression)

10.3 Groundwater Remediation

while multivalent anions with higher charge densities are strongly rejected by the membrane and drive monovalent anions passage through the membrane to fulfill the electroneutrality requirements [62]. In this regard, the slight decrease (97–95.2%) in arsenate rejection as HAsO4 2 (i.e. [As(V)]) can be attributed to the presence of more permeable ions such as monovalent ions; also reverse sulfate effect consistent with the As(V) rejection was enhanced by the presence of Cl , but reduced in the presence of SO4 2 due to mutual interactions between anions [60]. Th proper pre/post-treatments during NF application can vary depending on the characteristics of the process streams. A pilot plant investigation demonstrated that the use of aeration and sand filtration or MF/UF membranes reduced the content of key foulants to the subsequent NF and LPRO that has proved to be capable of removing pesticides [63]. It was also found that although a lower fouling potential could be obtained by MF/UF, sand filtration was better at removing manganese and dissolved organic matter. This suggested that the combination of sand filtration and membrane process can be a good option for pesticide removal with no chemical addition and minimum membrane maintenance. In the case of product water with low total dissolved solids (TDS), post-treatment system such as mixing with groundwater or mineralization is required to produce water with satisfactory composition [64]. The above studies suggest that one needs to account for site-specific conditions in order to establish the most appropriate system configuration and treatment trains that integrate nanofiltration. 10.3.2.3

Process Limitations

Nanofiltration provides selective ion rejection to separate divalent ions from monovalent ions. NF rejection for anions, for example, is in the increasing order of NO3  < Cl < SO4 2 . However, it is important to consider whether NF can be effective in rejecting nitrate ions in the presence of other inorganic anions as expected under field conditions. For example, in a pilot study of the removal of nitrate from groundwater, Amouha et al. [55] observed that nitrate rejection by NF90 was 60% similar to the level reported by Mahvi et al. [65]; nitrate rejection under field conditions, as reported in the above studies, was significantly lower than the range of 85–95% reported for bench-scale NF experiments under controlled laboratory conditions [55]. Although further studies are required to fully characterize nitrate rejection, NF is admittedly less efficient for nitrate removal. Given that NF membranes provide high rejection of sulfate ions, membrane mineral scaling (e.g. associated with sulfate salts of calcium) is also of concern as with RO membranes as discussed in Section 10.4. In addition, NF as well as RO membrane processes are confronted with the challenge of management of residual brine streams (see Chapter 11 on concentrate treatment). 10.3.3

Conclusions

Th capacity of NF for selective ion removal is critical to ascertain the proper integration of nanofiltration membrane for targeted contaminant removal in groundwater remediation applications. The selective separations performance of NF membranes (e.g. higher rejection of divalent relative to monovalent ions)

477

478

10 Water Reclamation, Remediation, and Cleaner Production with Nanofiltration

is well known and reasonably well described by the extended Nernst–Planck equation. In some cases, in order to take advantage of NF membrane separation performance, specific pre/post-treatment may be necessary for groundwater remediation to remove a range of ionic contaminants.

10.4 Agricultural Drainage Water 10.4.1 Project Drivers – Reduction of Agricultural Drainage Water Salinity Th use of reclaimed water for agricultural irrigation has a long history. Currently, a significant portion of the reclaimed water in the United States is targeted for agriculture (Figure 10.1). As an example, the salinity of agricultural drainage water in the San Joaquin Valley (SJV) in central California, one of the most productive agricultural areas in the United States, is in the range of about 3000–30 000 mg/l. However, the salinity of such water in some regions is of major concern, particularly when the salinity tolerance of certain crops is exceeded and arable land may have to be retired as its productivity is diminished. The efore, it is critical to reduce the salinity of agricultural drainage water, which would help (i) reduce reliance on imported irrigation water, and (ii) minimize the volume of environmentally contaminated drainage water. To this end, the application of low-pressure RO and nanofiltration membrane technologies has drawn serious attention for agricultural drainage water reuse due, in part, to a new generation of high-performance membrane materials and scalability. 10.4.2

Advantages of Low-Pressure RO/NF

Low pressure RO and nanofiltration membranes can operate at remarkably low pressures with excellent product water flux and reasonably high levels of salt rejection, which is critical to achieve a desired water quality for agriculture irrigation. Low-pressure RO/NF membrane are suitable for reducing the salinity level to 750 mg/l TDS, which is suitable for irrigation. The scalability and feasibility of membrane technologies for agricultural water treatment and reuse have been demonstrated since 1971 at the historic pilot facility at Firebaugh, San Joaquin Valley in California [66, 67]. However, membrane process optimization (e.g. type of membrane, level of pretreatment, control of mineral scale, etc.) is critical for agricultural water treatment, which is a complex mixture of dissolved and suspended organic and inorganic components as well as a wide variety of microorganisms. 10.4.3

Process Feature and Fundamentals

Th potential use of NF for agricultural drainage water treatment was demonstrated via a pilot-scale plant in the Buena Vista Water Storage District in Buttonwillow, California. Feed water composition for this area is provided in Table 10.13 as well as the raw water qualities in the pilot-scale system in Buena Vista. In

10.4 Agricultural Drainage Water

Table 10.13 The raw water qualities in the pilot-scale system in Buena Vista. Parameters

Concentration

Total dissolved solids (mg/l)

5250

Total organic carbon (mg/l)

2.77

Hardness (mg/l)

1630

Turbidity (NTU)

0.8

Na (mg/l)

1150

Ca2+ (mg/l)

555

+

Mg2+ (mg/l)

60.7

Cl (mg/l)

2010

SO4 2 (mg/l)

1020

HCO3  (mg/l)

291

Source: Lee et al. 2003 [68]. Reproduced with permission of Elsevier.

order to treat a complex mixture of agricultural drainage water, membrane selection was initially performed based on laboratory-scale performance evaluation (salt rejection and permeate flux) and a subsequent screening analysis of biofouling potential (bacterial count). Laboratory-scale membrane performance tests using NaCl and CaCl2 solutions, over the range of salt concentrations expected in the field at feed pressures from 100 to 200 psi, demonstrated a higher salt rejection and permeate flux for solutions of the divalent calcium ion than for the monovalent sodium ion, which is consistent with published studies for multivalent electrolytes [69, 70]. Laboratory experiments also confirmed that membrane selection involves a trade-off between solute rejection and permeate flux. For example, all three membranes tested had Ca2+ rejection (≳94%); however, the low fouling composite (LFC) membrane (Hydranautics), which had the highest Ca2+ rejection (≳99%), demonstrated permeability that was 30–60% lower than the TFC HR and TFC ULP membranes (Koch Membrane Systems) evaluated for desalting agricultural drainage water [68]. Pilot-scale tests using selected low-pressure loose RO membranes (TFC-ULP and TFC-HR) were operated for a period of six weeks to evaluate the “true” recovery for agricultural drainage water in the pilot plant. As shown in the process flow diagram in Figure 10.7, the plant consisted of a multimedia filtration system for pretreatment and a two-stage portable RO unit with six pressure vessels each containing three spiral-wound membrane elements, arranged in a 2 : 2 : 1 : 1 array. The plant was designed to handle a feed water flow rate up to 27 gpm operated at a feed pressure of 145–235 psi. Th plant was operated with ESPA-1 membranes (Oceanside, CA) in both first and second stages, which were replaced by TFC-HR (first stage) and TFC-ULP (second stage) membranes after operating for 530 and 720 hours, respectively. Clearly, agricultural drainage water with higher TDS levels would require a higher rejection. The data provided in Figure 10.8 shows that salt rejection up to 95% was achieved with membrane replacement,

479

10 Water Reclamation, Remediation, and Cleaner Production with Nanofiltration

Well #1 (North)

Well #2 (North) Fiber train #1 Alum

Feed filter

Cartridge filter

Boost pump

pH

P

Legend Pressure gauge

F

Flow meter

EC Conductivity meter pH pH meter

EC

Acid

Concentrate The Drain The Drain Sump Pump

Fiber train #2

Scale inhibitor

F

Baclpressure Concentrate

Needle valve

Concentrate P P

RO pressure vessel (TYP)

F1

Concentrate

F3

F2

F4

Permeate EC

1st stage

F

2nd stage

Figure 10.7 Process flow diagram of pilot plant. Source: Lee et al. 2003 [68]. Reproduced with permission of Elsevier. 100

0.300

90

0.250

80

% Rejection

0.200

% Recovery

70

Normalized flux

0.150

60 0.100

50 New 1st stage membranes

40 30

0

120

240

New 2nd stage membranes

360 480 600 720 840 Elapsed operating hours

0.050

Normalized flux (GFD/psi)

Precent conductivity rejection and precent recovery

480

0.000 960 1080 1020

Figure 10.8 Performance of the membrane pilot plant. Percent rejection was based on the measured conductivity. Source: Lee et al. 2003 [68]. Reproduced with permission of Elsevier.

which enabled the entire system to attain a relatively stable normalized flux (an average of 0.09 gal/ft2 /d/psi [GFD/psi]; equivalent to 2.2  105 l/m2 h Pa) and water recovery of 50%. Th drawback of NF is that it allows passage of monovalent cations while retaining divalent ions to a greater extent, thereby possibly worsening the sodium absorption ratio (SAR) [71]. In order to cope with the above issue, it was proposed to utilize a selective NF membrane within an integrated MF-NF-RO pilot-scale membrane system [8] for agricultural irrigation to address specific water quality requirements (e.g. maintaining an adequate nutrient content,

10.4 Agricultural Drainage Water

thereby lowering excessive sodium concentration to avoid increased soil and plant salinity). In this system, the final product water includes the NF retentate and RO permeate streams. The NF elements enrich the final product water with respect to divalent ions (e.g. Ca2+ , Mg2+ ) by rejecting divalent ions to a greater extent than monovalent ions while allowing sodium ions to pass to the RO elements for further treatment (e.g. rejection of Na+ ). In the above approach, recycled water is produced with useful levels of nutrients and other ions, and lower SAR to minimize the impact on soil salinity. An integrated NF-RO membrane system has been also reported for treating micro-filtered, biologically treated sewage effluent (BTSE) with high SAR and sodium and chloride concentrations [72]. It is interesting that NF alone did not remove sodium and chloride sufficiently and NF and RO alone did not produce water with the required ratio of SAR. However, the quality of product water that combines equal proportions of NF permeate and RO permeate from a two-stage NF-RO membrane system was suitable for irrigation with respect to the abovementioned risk factors. It is worth noting that agriculture irrigation water requires a minimum Mg2+ concentration to reduce the need for fertilizers, especially for soil that is low in minerals [73] In this regard, the generated brine from the selective NF membrane rich in Mg2+ was demonstrated to be suitable for supplementing magnesium ions to soft waters (e.g. desalinated water for agricultural irrigation and drinking water) in arid areas [74]. It should be also noted that a similar process concept (hybrid NF/RO filtration scheme) was reported (Section 10.3.2.2.) for the removal of nitrate from groundwater for groundwater treatment, whereby the RO permeate was mixed with the NF residual stream to generate product water of low nitrate concentration, yet with a balanced composition with respect to the required composition for drinking water. 10.4.4

Process Limitations and Progress

High recovery permeate production is desired to reduce the challenge of concentrate (brine) management, including safe disposal and reuse. In certain locations, the achievable recovery may be limited as in SJV’s case where brackish groundwater is often near or above saturation with respect to various sparingly water-soluble mineral salts, including gypsum (CaSO4 ⋅2H2 O), calcium carbonate (CaCO3 ), and silica (SiO2 ). Mineral scaling of the membrane surface can result in permeate flux decline, shortening of useful membrane life, and consequently, increased water production costs [15, 17–21]. For example, in order to avoid mineral scaling in a previous pilot-scale operation [68], recovery was limited to 50%. In order to assess the limitations of membrane treatment for water of high mineral scaling potential, a systematic analysis and field testing were reported in [75], specifically addressing the brackish water issue in SJV [68]. Brackish groundwater composition in SJV is known to vary temporally and spatially [75]. As shown in Table 10.14 and Table 10.15, water quality data analysis revealed that gypsum and calcite saturation indices (defined as SIi = IAPi /K SP ,i where IAPi is the ion activity product of the constituent ions and K SP ,i is the solubility product) were in the range of 0.12–0.13 and 3.0–9.5, respectively, for five representative field water sources from the SJV [75]. The gypsum and calcite saturation indices were in the

481

482

10 Water Reclamation, Remediation, and Cleaner Production with Nanofiltration

Table 10.14 Summary of average water quality and saturation indices from the DWR database (2003–2004). Name

Location

TDS (mg/l)

𝝅 0 a) (kPa)

pH

SIC a)

SIG a)

OSA

Central area

7999

7.7

408

3.9

0.75

LNW

Southern area, Lost Hills

11 944

7.5

834

2.3

0.99

CNR

Southern area, Kern Lake Bed

6987

7.7

315

5.1

0.76

VGD

Southern area, Lemoore

23 480

7.9

1000

5.3

0.84

a) 𝜋 0 , SIC , and SIg were calculated at 20  C. Source: McCool et al. 2010 [75]. Reproduced with permission of Elsevier.

Table 10.15 Selected data demonstrating variability of water quality in San Joaquin Valley. 14 July 9 September 12 November 12 January 22 March 10 May 2003 2003 2003 2004 2004 2004 Average

Date

TDS (mg/l)

3828

9344

11 000

9576

8284

7999

[Ca2+ ] (mg/l) 283

224

385

430

454

358

356

[SO4 2 ], (mg/l)

3990

2340

5550

6460

5630

4890

4810

Total alkalinity (mg/l as CaCO3 )

271

239

160

173

146

217

201

pH

7.6

7.1

7.9

7.9

7.9

7.9

7.7

𝜋 0 a) (kPa)

371

231

466

534

443

407

409

3.2

0.86

4.3

4.7

4.5

5.8

3.9

SIG a)

0.57

0.41

0.83

0.93

0.98

0.76

0.75

SIC a)

5864

a) 𝜋 0 , SIC , and SIg were calculated at 20  C. Source: McCool et al. 2010 [75]. Reproduced with permission of Elsevier.

range of 0.41–0.98 and 0.86–5.8, respectively, for one of the field water sources from the SJV. At each site the achievable product water recovery (i.e. maximum recovery or recovery limit), which is limited by mineral salt scaling, was estimated based on the knowledge of feed water chemistry while accounting for concentration polarization and potential pretreatment conditions (e.g. pH adjustment, antiscalants). For example, as product water recovery Y (defined as Y = QP /QF where QP and QF are the permeate and feed flow rates, respectively) increases, the concentration of the retentate (or brine), C R , increases by a concentration factor (relative to the feed concentration) CF (i.e. CF = C R /C F ) given as [75] 1  Y (1  Rs ) (10.6) 1Y Th product recovery limits for different feed source waters can be estimated using Eq. (10.6) by establishing the CF at which the retentate concentration CF =

10.4 Agricultural Drainage Water

reaches the mineral salt scaling thresholds. Analysis of the recovery limit imposed by mineral salt scaling (e.g. gypsum, calcite, silica) revealed that the feasible water recoveries can vary from about 46% to 69% for different source waters from SJV. Th limiting scalant depends on source water characteristics and pretreatment strategies (e.g. acid pretreatment, antiscalant). For example, calcite is the limiting scalant at natural pH 7.5; however, gypsum is the limiting scalant at acidic pH when acid pretreatment is applied to the feed source water. Acid pretreatment may be counterproductive for water of high gypsum scaling potential and low to medium bicarbonate content. For such a water source, the positive effect of bicarbonate ions on retarding gypsum precipitation may be diminished due to the protonation of bicarbonate ions at reduced pH [76]. Th range of the recovery limit for different source water from SJV implies that site-specific process optimization would be required for the implementation of low-pressure RO/NF membrane technologies for agricultural drainage water reuse given the geographic and temporal water quality variabilities. Site-specific process optimization has been established by using a novel approach based on direct observation of mineral scaling and flux decline measurements, utilizing an automated ex situ membrane monitoring system (MEMS). MEMS operates in a stand-alone single-pass desalting mode enabling direct observation of particulate deposition and mineral scaling on the membrane surface. Th s enables a step-by-step field diagnostic and RO plant membrane scaling monitoring to establish site-specific operating conditions that are applicable to brackish and drainage water treatment and desalting in SJV. For example, analysis based on direct MEMS observations revealed that suspended particulates (even for feed water of turbidity 95%) can be obtained by different NF membranes [84].

551

12 Nanofiltration in the Chemical Processing Industry

Th sulfate retention for NF of raw solution brine can be easily determined from simple membrane characterization experiments and a fitted average “pore” radius [84]. Furthermore, the chloride retention, which is close to zero for this application, can also be predicted from the membrane-average “pore” radius obtained from membrane characterization experiments [84]. For both components, the retention reduces when a membrane with a bigger “pore” radius (more open membrane) is used. It has been shown that for processing of nearly saturated brines, Donnan exclusion is negligible and sodium chloride retention is mainly governed by the salting out effect caused by sodium sulfate activity differences between retentate and concentrate [89, 106]. NF can thus (partly) replace the brine purification step in the salt production plant. To achieve this, it is essential that not only sulfate concentration in the raw brine is decreased but also calcium, strontium, and magnesium concentrations as well. Th removal of these ions by NF is helped by the presence of minor amounts of the so-called positive retention enhancing compound [84]. Calcium retentions in raw brine can be increased substantially, as can be seen by comparing calcium retentions for processing raw brine without any addition of positive retention enhancing components and raw brine to which 600 mg/kg (mg/l) DrewsperseTM 747A (a deposit inhibitor from Solenis containing 4.52% polymaleic acid, 1.74% amino tri-methylenephosphonic acid, and 0.22% maleic acid) has been added, as illustrated in Figure 12.7 for Desal DK. For this membrane, strontium and magnesium retentions in excess of 90% were obtained, while sulfate retention was in excess of 96% [85] for raw brine containing Drewsperse 747A. For NF-270, slightly lower divalent cation retentions but slightly higher sulfate retention was obtained. Samhaber et al. evaluated the potential of NF for the concentration of sodium sulfate in mother liquor from the salt crystallization process in pilot trials at Austrian Salt Works in Ebensee [86]. The intention of this application is to 100

Raw brine

90 Calcium retention (%)

552

Raw brine with 600 ppm Drewsperse 747A

80 70 60 50 40 30 20 10 0 1

1.1 Concentration factor CF (–)

Figure 12.7 The effect of addition of Drewsperse 747A on calcium retention as a function of concentration factor for Desal 5DK during lab-scale processing of raw brine at room temperature and 31 bar operating pressure.

12.2 Inorganic Chemical Industry

concentrate and recycle the sulfate to the brine purification as purification chemical for the removal of calcium and to produce sodium chloride salt crystals from the purified permeate lean in sulfate. In these trials, the mother liquor was diluted by 10–15% water or raw brine before feeding it to the NF membranes to decrease the sulfate concentration and to avoid crystallization of sodium sulfate in the NF membranes [86]. The pH of the mother liquor was kept below 10, the temperature was reduced to 30  C, and the operating pressure was set at 30 bar. Th obtained sulfate retentions were between 98% and 99%, while negative chloride (ranging from 0% to 5%) and bromide (ranging from 10% to 15%) retentions were reported. Based on the performance, a membrane lifetime in excess of 18 months was predicted. Th membrane that was used for these trials was Desal 5DK [87, 88]. In their process, excess sulfate was removed from the plant via a purge waste stream, supersaturation was avoided, and sodium sulfate crystallization was not used. A disadvantage of the NF option proposed by Samhaber et al. is the need for dilution of the mother liquor. Furthermore, the absence of sodium sulfate crystallization for removal of excess of sulfate in the plant leads to relatively high purge streams. Th presence of sodium sulfate crystallization consequently has clear benefits but is usually quite energy intensive as well. Th high-energy-consuming sodium sulfate removal crystallization process fed with salt crystallization mother liquor (saturated in both sodium sulfate and sodium chloride) can be replaced by a NF step to create a retentate supersaturated in sodium sulfate, which is subsequently sent to a crystallizer to produce anhydrous sodium sulfate crystals. Crystallization of the sodium sulfate in the NF membrane modules is avoided by the presence of small amounts of crystal growth inhibitor [89, 90]. In this process, a permeate lean in sodium sulfate and practically saturated in sodium chloride is produced. Very high (see Figure 12.8) and stable sulfate and calcium retentions were reported for pilot tests using NF-270 and Desal 5DK membranes [89], while stable and negative bromide and chloride retentions were reported, with bromide retentions 10% lower than chloride retentions (in absolute terms). The modynamic considerations as discussed in Refs. [84, 89, 101] can explain the negative retentions observed. Furthermore, it was stated that NF-270 is preferred over other membranes evaluated for this application [89]. The chloride retention appeared to be a function of the sulfate concentration difference between retentate and permeate (as for the NF of raw brine and depleted chlor/alkali brine). More information about this application can be found in Refs. [84, 89, 90]. NF is used in the salt production plant of AkzoNobel Industrial Chemicals in Hengelo (the Netherlands) since 2012, removing sulfate from brine [96]. Th application of NF installed in a partnership with Chemetics (a Jacobs Company) leads to 1.5–2% efficiency increase (less salt loss) and is reported to save 2.75 million m3 of natural gas per year.

553

12 Nanofiltration in the Chemical Processing Industry

100 80 Retetention (%)

554

60 40 20 0 Sulfate

Calcium

Potassium

Chloride

Bromide

–20 –40

Ion

Figure 12.8 NF-270 ion retentions for processing of salt crystallization mother liquor saturated in sodium chloride and sodium sulfate. Source: Adapted from Bargeman et al. [89].

12.2.3 Pollution Treatment in the Inorganic Chemical Industry, MLD and ZLD Th majority of pollution is leaving the inorganic chemical industry as liquid waste streams. Some of the major chemicals released and transferred by inorganic facilities are sulfuric acid, phosphoric acid, nitric acid, hydrochloric acid, ammonia, chlorine, and all kinds of salt streams. The acids are often contaminated with heavy metal compounds. The salt streams are often contaminated with organics or a mixture of different salts, which require innovative NF solution to separate them for reuse. These inorganic pollutions can be separated and concentrated by membranes, in order to recycle the permeate water and reuse the concentrated products. Increasingly stringent government regulations, water scarcity, and a trend to reuse products are requesting innovative solutions from the industry in the field of minimal liquid discharge (MLD) and zero liquid discharge (ZLD). The technologies that are being used in the final treatment stages like evaporation and crystallization require significant amounts of capital and operational cost. Th last stage before these high-cost unit operations is very often reverse osmosis alone or in combination with NF as a pretreatment- or salt separation purpose. 12.2.3.1

Treatment of Medium Salinity Waste Streams

Since 2016, DOW Water & Processes is reacting to these innovative developments and introduced ultrahigh-pressure reverse osmosis (UHPRO) elements that are operating in a 120 bar RO system. Th s way, the medium-salinity feed streams with a TDS below or at 6–8% can be concentrated to a high-salinity stream with 11–20% TDS. Th maximum TDS concentration is dependent on the nature and osmotic pressures of the salts dissolved in the feed. As an example, at 120 bar, the maximum achievable concentration is 11% for NaCl and 20% for Na2 SO4 . The UHPRO concentrate may be fed directly to the crystallizer; in that

12.2 Inorganic Chemical Industry

case, it may minimize or even eliminate evaporators that are of an enormous economic penalty. A NF upfront of the UHPRO is an innovative step forward in order to separate the salts from a medium-salinity feed stream into two feed streams with different compositions: – NF concentrate: Multivalent anions such as SO4 2 , HPO4 2 , PO4 3 , and CO3 2 as well as residual organics with molecular sizes of >150–300 Da are concentrated. – NF permeate: Monovalent anions such as Cl , NO3  , HCO3  , and HSO4  pass the NF membrane into the NF permeate. Th NF permeate is concentrated by UHPRO to the maximum possible concentration at 120 bar without any risk of scaling because scale building divalent cations are removed by NF. Depending on the original feed composition, the UHPRO concentrate is a nearly clean NaCl solution at 10%, which can be reused for internal processes. The UHPRO permeate has a TDS of about 1000 mg/l and can be reused. Th NF concentrate stream containing mainly multivalent salts has a potential for reuse (see Figure 12.9). 12.2.3.2

Removal of Ammonia Pollution with NF and RO

Wastewater from processes or evaporator condensate containing ammonia is neutralized by adding sulfuric acid. After neutralization, (NH4 )2 SO4 is formed and concentrated by high-pressure NF or RO up to 18%. In the case of using nitric acid as a neutralization agent, NH4 NO3 is formed and concentrated by UHPRO up to 16%. In this case, NF cannot be used because a monovalent anion passes a NF membrane. In both cases, the concentrates are products with a potential for reuse. Th approach is a two-stage system, where (NH4 )2 SO4 is concentrated up to 8% in an 80 bar stage followed by a 120 bar stage to concentrate further up to 18%. After further crystallization of (NH4 )2 SO4 , the salt can be sold as a fertilizer. Th s is an excellent example of membrane technology converting pollution into a valuable product [1]. 12.2.3.3

Acid Waste Treatment

Acid-stable (or acid-tolerant) polymer NF thin film spiral wound membranes and elements have been available since early 2000. Such elements are stable in 30% sulfuric acids up to 70  C [2]. Acid-stable NF membranes are in use to treat acidic waste streams. The objectives are as follows: 1. To purify the acid in order to reuse it 2. To recover metals for reuse Th fact that acid-stable NF membranes (e.g. DURACID from GE Water (today SUEZ) or AMS Technologies has nearly 0% rejection for sulfuric acid and nearly 100% rejection for CuSO4 can be readily explained by looking at the dissociation steps of sulfuric acid: H2 SO4 HSO4  SO4 2

555

Industrial WW Equalization basin

BWRO

SWRO

Lime softening (optional)

A

Sand filter (optional)

UF (Particle removal)

IX Softening, hardness and COD reduction *

Optional

To discharge or reuse

6–8% TDS

Crystallizer

Evaporator B 0.5–2% TDS Permeate to reuse

3–4% TDS

6 – 8% TDS to conventional downstream treatment NF

Permeate to reuse

C

Maximum 6% NaCl for reuse UHP RO: ultra high pressure reverse osmosis * IX: alternatively also between RO/NF stages

TM

DOW

20–25% TDS

UHPRO

H2O

Direct to crystallizer 10–20% TDS

Permeate to For reuse, reuse e.g. maximum 10% NaCl

Solids UHP RO replaces evaporator partially ( ) or completely ( )

Figure 12.9 ZLD/MLD process with UHPRO and integrated NF process demonstrating three options for a medium-salinity stream (TDS 6–8%). (a) Discharge or reuse, (b) direct to conventional downstream treatment, and (c) direct to UHPRO (optional with NF pre-treatment/salt separation). Source: Data from DowDuPont technical Information.

12.2 Inorganic Chemical Industry

Th first dissociation step is mainly creating monovalent bisulfate ion (HSO4  ). Th bisulfate ion is a monovalent anion and passes through the NF membrane with almost the same ease as, for instance, a chloride ion. Th next dissociation step produces a sulfate ion, which is present in lower concentration and is rejected together with the metallic cation. Th NF membrane at low pH is positively charged (pH is below the isoelectric point); this further enhances the rejection of divalent cations. Hydrochloric acid passes through the NF membrane with the same ease as sodium chloride in concentrated solutions. Metals dissolved in a hydrochloric acid solution will be rejected at rates corresponding to their molecular weight (MW), size, and charge. Th rejection of metallic ions is not complete. The partial rejection can be improved markedly by addition of a sulfate salt, such as sodium sulfate, or a small quantity of sulfuric acid to the feed solution [3]. Examples of applications of NF in acid steams: 1) Sulfuric acid/sulfates: Mexicana de Cananea used NF to treat a low pH mining water (PLS, pregnant leach solution) feed of 900 m3 /h of pH 1.6, containing approximately 2% H2 SO4 and 900 mg/l Cu2+ . The acidic permeate was reused and CuSO4 in the NF concentrate was recovered. The installed membrane area is 42,000 m2 , the membrane type was originally DK (GE Water/Osmonics), the pressure is 20–40 bar, and the recovery is 50%. Th specific flux of the NF membrane is 10.5 l/m2 h (Info Osmonics) [4]. DK is not acid stable, and the lifetime under such conditions is less than one year. Today, acid-resistant NF membranes are available, e.g. DURACID from GE (SUEZ) or acid-stable elements from DDP (DowDuPont). 2) Surface treatment of metals: Al foil production industry (BEKROMAL) [4]. Acid-stable NF membranes (DURACID from GE Water) were used to separate Al3+ from 10% H3 PO4 . The Al3+ concentration in the pickling bath is kept constant, and the lifetime of the pickling bath is increased significantly. VP GmbH in Switzerland reported that they use a DURACID membrane in the Al anodising industry to separate Al from 3–10% sulfuric acid. Th acid is an eluate from an ion-exchange resin process. The objective is to reuse the acidic eluate rather than to neutralize and dump it. Th Al-rich concentrate can potentially be used as a precipitation chemical in other industrial processes. 3) Phosphoric acid is the second most produced acid in the world after sulfuric acid. NF is used to purify phosphoric acid from heavy metals. GE Water reported about large NF systems (3000 m2 membrane area) to purify a phosphoric acid solution in the fertilizer industry [5]. An acid-stable NF membrane (DURACID) was used for purification of phosphoric acid. Th aim is to reject heavy metals, which limit the use of phosphoric acid in food application. Permeate flux was 3 l/m2 h at 69 bar (1000 psi) and 94% acid permeation and 99.2% of cationic impurities rejection was obtained [6]. 4) Nitric acid: Amafilter Germany reports about a process to remove lead from nitric acid solutions. According to that paper, the lead concentration rose up to 70 g/l, and 80–90% of the nitric acid is reused for the process [7]. Osmonics

557

558

12 Nanofiltration in the Chemical Processing Industry

reported the rejection of uranylnitrate from nitric acid in the nuclear industry. UO8 NO3 is rejected by 86%, while the 2 M acid passed the membrane completely [8]. 5) Boric acid: Osmonics (today GE Water/SUEZ) reported that radioactive nuclides are rejected from boric acid. Rejections of Co58 and Co60 were > 85% and that of Sb124 was >99%, while boric acid passed the membrane with 97% transmission [8].

12.3 Organic Chemical Industry 12.3.1

Characterization of the Industry

Th organic chemical industry produces organic chemicals (those containing carbon) used as either chemical intermediates or end products. Th industrial organic chemicals industry uses basic materials derived from petroleum and natural gas and from recovered coal tar condensation by coke production. We can classify the organic chemical industry into four main groups: • • • •

Cyclic organic chemicals, Industrial organic chemicals, Gum and wood chemicals, Petrochemical industry

12.3.1.1

Product Characterization

Th important classes of products within cyclic organic chemicals are (i) derivatives of benzene, toluene, naphthalene, anthracene, and other cyclic chemical products; (ii) synthetic organic dyes; (iii) synthetic organic pigments; and (iv) cyclic (coal tar) crudes, such as light oils and light oil products according to an EPA report [40]. Important classes of industrial organic chemicals include (i) noncyclic organic chemicals such as acetic, adipic, chloroacetic, oxalic acids, and their metallic salts, formaldehyde, methylamine, etc.; (ii) solvents such as amyl, butyl, and ethyl alcohols, methanol acetates, ethers, glycols, chlorinated solvents, etc.; (iii) polyhydric alcohols such as ethylene glycol (EG), sorbitol, synthetic glycerine, etc., synthetic perfumes, and flavoring materials such as methyl salicylate, saccharin, citral, synthetic vanillin, etc.; (iv) rubber processing chemicals such as antioxidants, etc.; (v) plasticizers, both cyclic and acyclic, such as esters of phosphoric acids, phthalic anhydride, etc.; (vi) synthetic tanning agents such as sulfonic acid condensates; and (vii) esters and amines of polyhydric alcohols and fatty and other acids. Th important classes within gum and wood chemicals are hardwood and softwood distillation products, wood and gum naval stores, charcoal, natural dyestuffs, and natural tanning materials. 12.3.1.2 Pollution Prevention Opportunities for NF in the Organic Chemical Industry

Th best way to reduce pollution is to prevent it in the first place. Processintegrated membrane systems can purify and concentrate organics before

12.3 Organic Chemical Industry

Table 12.2 Chemicals released in chemical industry. Acids (sulfuric, phosphoric, and hydrochloric)

High-application potential for NF

Alcohols, e.g. methanol, glycols, and IPA

High-application potential for NF

Solvents, e.g. toluene

NF solvent stable

they recieve the “pollution” status. It is critical to emphasize that each pollution is process specific. The chemical industry has historically released more chemicals than any other industry and membrane technology can improve the environmental performance within this industry, see Table 12.2. 12.3.2 Potential and Actual Applications for NF in the Organic Chemical Industry 12.3.2.1

UF/NF Treatment of Rinsing Waters

In a production plant for liquid detergents, UF and NF membranes have been tested in an on-site pilot plant to recover process water as well as to reduce the high organic load of the rinsing waters as the chemical oxygen demand (COD) load of up to 300 kg/m3 is the main expense factor for wastewater disposal cost. Membrane screening and pilot plant experiments have been performed. COD reductions in wastewater of up to 96% have been achieved resulting in less environmental impact and lower wastewater disposal costs [9]. A challenge for NF membranes in chemical industry is their application in organic solvents. More on this subject can be found in Chapter 20. Some examples of overviews have been published on possible applications of NF for organic solvents, such as alcohols, alkanes, esters, ethers, and ketones, and in the purification of aqueous streams containing high concentrations of organic compounds [10, 11]. 12.3.2.2

NF for Professional Motion Picture Imaging

Th s example demonstrates the NF rejection of heavy metals in the chemical photographic processing industry, despite the demise of this industry. Kodak has tested NF for the recovery of chemicals and wash water used in photographic processing. The goal was to recycle at least 80% of the water by using NF. During the trials, recycling ratios between 80% and 95% were achieved [12]. Weizman membrane type MPSW-11 was used and the NF installation contained nine elements of 5.6 m2 , resulting in a total membrane area of 50.4 m2 . The inlet pressure to the NF installation was 20–25 bar and the average permeate production was 3.5 m3 /h, which means an average flux of 70 l/m2 h. Th results of the tests are shown in Table 12.3. The pH of the concentrate was pH 3.5 and that of the permeate pH 6. From the results shown in Table 12.3, it is apparent that most of the chemicals in the wash water were retained with the exception of chloride, which had a very large negative retention of 80%. Th silver retention was very high, resulting in a loss of only 2 mg/l of silver. Th concentrated solution, however, must be treated to recover the silver. In

559

560

12 Nanofiltration in the Chemical Processing Industry

Table 12.3 Results of tests with photoprocessing wash water. Concentrate (mg/l)

Permeate (mg/l)

Retention (%)

COD

10 000

500

95.0

BOD5

4 400

Nd

–

Thi sulfate

15 000

400

97.3

Sulfate

9 100

400

Chloride

250

450

Hydrogenphosphate

360

40

88.9

Total sulfur

13 000

700

94.6

Magnesium

30

0.3

99.0

Silver

300

2

99.3

Calcium

400

5

98.75

Phosphorous

200

20

90.0

Organics

Trace

No trace

—

95.6

80.0

addition, experiments have indicated that where NF is used with separated wash water, it is possible to reuse the concentrate after some chemical adjustments have been made. Kodak made an economic evaluation for this process and the results are summarized in Table 12.4. Th total amount of recycled wash water from the NF plant would be 7000 m3 /yr. Th costs per cubic meter of water would then be €2.86 and the costs of fresh water would be €2.67 m3 . This means that recycling of the water alone gives insufficient incentive to install a NF plant. The recovery of the silver, however, was an additional advantage. Th NF technique is very efficient in eliminating most of the pollutants from the wash water, and it enables the recycling of up to 80% of wash water. It is especially efficient in silver recovery. Th trials were not conducted long enough to determine the lifetime of the membranes. This could have a significant impact on the final cost data of using NF in film processing. Both the reuse of the recovered concentrate as a replenisher as well as recycling of the wash water would bring a faster return on investment than recycling water alone. 12.3.2.3

NF of Catalysts in Solvents

Homogeneous Catalyst Separation Th separation of reaction products from cata-

lysts is a major problem existing in many forms of homogeneous catalysis, including phase transfer catalysis and transition-metal-catalyzed reactions. Catalytic organic synthesis has become a major focus for developing cleaner processes. Phase transfer catalysts avoid the use of aprotic solvents for reactions involving a water-soluble nucleophilic agent and an organic soluble electrophilic reagent. Similarly, transition metal catalysts can provide faster reactions than using stoichiometric reagents. A major drawback of such catalysis is the extensive and usually destructive postreaction work-up that is needed to remove the catalysts from

12.3 Organic Chemical Industry

Table 12.4 Economic data. Annual costs (k€)

Investment

46.65 k€

5.80

Maintenance

3% of investment

1.40

Total fixed costs

7.20

Power consumption

26 000 kW/yr

3.00

Labor

120 h/yr

1.10

Membrane replacement (2 yr)

4.5 modules/yr

5.50

Filter replacement

36 filters/yr

0.15

Cleaning

25 kg/yr

0.15

Water consumption

1100 m3 /yr

2.45

Waste

1100 m3 /yr

0.45

Total variable costs

12.80

Total

20.00

the reaction medium. Presently, hardly any industrially employed separations are aimed at the recovery of the catalyst in an active form, but rather at obtaining a pure product while recovering the metal in a form that may be recycled to a catalyst manufacturer. This leads to poor economics as the ligands forming the active catalysts, which are often more expensive than the metals, are lost. Because of the nature of the catalysis and reaction environment, only solventstable NF membranes can be applied. Usually, the homogeneous catalysts are relatively large, MW > 450 Da, and the reaction products are substantially smaller, so that a separation of the catalyst used and the reaction products is feasible with NF, see also Chapter 20. Phase Transfer Catalysis For phase transfer catalysis, usually a model reaction is used: the conversion of bromoheptane (MW 179 Da) to iodoheptane (MW 226 Da) with toluene as a solvent, using an aqueous solution of potassium chloride, and catalyzed by teraoctylammonium bromide (TOABr, MW 546 Da) [13–15]. Koch MPF-50 (MWCO 700 Da), MPF-60 (MWCO 400 Da), and Desal-5 (MWCO 350 Da) membranes showed a low rejection of the catalyst, 48–86%. Th catalyst was almost completely retained (>99%) by Grace STARMEM 122 (MWCO 220 Da), STARMEM 120 (MWCO 200 Da), and STARMEM 240 (MWCO 400 Da) membranes. Although the Desal-5 and the MPF-60 membranes have the same nominal MWCO as STARMEM 240, the actual rejection of TOABr is not at the same level [14]. The reason for this must probably be found either in the different methods that are used to establish the MWCO of a membrane or in the structure or the polymeric materials used in these membranes. Retention of the product iodoheptane of 8–15% for STARMEM 122 (today MET/EVONIK) was measured in a set of reaction filtration cycles. The reaction mixture consisted of 0.5 M bromoheptane and 0.05 M TOABr, the temperature

561

562

12 Nanofiltration in the Chemical Processing Industry

was 20  C, and the pressure for the NF filtration was 30 bar. The flux obtained decreased from 9.1 to 7.0 l/m2 h because of a combination of precipitated catalyst, osmotic pressure effects, and concentration polarization. Washing the membrane with reaction mixture could easily restore the flux. The catalyst showed no loss of activity during those three cycles [14]. Transition Metal Catalysis Th

Heck coupling of iodobenzene (MW 204 Da) with styrene (MW 104 Da) to produce trans-stilbene (MW 180 Da) is usually selected as a model for transition metal catalysis. Th catalyst used is bis-(acetato)bis(triphenylphosphine)palladium(II) (MW 749 Da) complex and the reaction solvents were tetrahydrofuran (THF), methyl-tert-butyl ether (MtBE), and a 50 : 50 mixture of ethyl acetate and acetone [16]. The membrane used was STARMEM 122. Although the flux in THF was much higher than in the other systems, 45.8 l/m2 h at 30 bar and 20  C, compared to 11.3 l/m2 h in MtBE and 32.3 l/m2 h in ethyl acetate/acetone, the reaction rate was the slowest. Th reaction in ethyl acetate/acetone was much faster than in the other solvents. During a number of reaction filtration cycles, it was noted that the reaction rate was falling below 20% of the initial value by the fourth catalyst recycle. Also, the retention of Pd decreased during these cycles from 99% to 90% [15]. Apparently, smaller Pd species are being formed that are small enough to pass the membrane. Other organometallic catalysts studied, which are used commercially, were the Jacobsen catalyst, a manganese complex with MW 622 Da, Wilkinson catalyst, a rhodium complex with MW 925 Da, and Pd-BINAP ((R)-(+)-2,2′ bis(diphenyl-phosphino)-1,1′′ -binaphtyl), a complex with MW 849 Da, in ethyl acetate, THF, and dichloromethane (DCM) as solvents [17]. The membranes tested were Desal-5, MPF-50, and STARMEM 120, 122, and 240. Good catalyst rejection, >95%, coupled with good solvent fluxes, >50 l/m2 h at 20 bar, were obtained in the majority of the systems tested. Th membranes from Grace show higher selectivities, 95.4–99.6%, than the Koch Selro and Desal types, 81.4–94.2%, although the STARMEM 240 membrane is unstable in DCM and THF, and the types 120 and 122 show no flux with DCM as a solvent. The Grace membranes are made of polyimide [18] and such a membrane is also used in the MAX-DEWAX process in the Beaumont refinery of Mobil. Also for selective hydrogenations, coupled NF reactions can be used. Th continuous enantioselective hydrogenation of dimethyl itaconate (MW 158 Da) with Ru-BINAP (MW 929 Da) and of methyl 2-aceamidocrylate (MW 143 Da) with Rh-EtDUPHOS (1,2-bis(2,5-dimetyl phospholano)benzene, MW 723 Da) is reported in Ref. [19]. The reagents were dissolved in methanol. The retentions for the catalysts were 98% and 97%, respectively, using a MPF-60 membrane. Th fluxes, measured with pure methanol, were in the order of 1.2 kg/m2 h at a pressure of 10 bar and 30  C. Th Rh-EtDUPHOS catalyst showed a slow deactivation, possibly because of oxidation of the phosphine ligand. Several companies, such as Union Carbide, Hoechst, and DSM, have obtained patents on the application of NF for recovering homogeneous catalysts, especially rhodium complexes for hydroformylation reactions [20–23], involving solvent-stable polymeric and inorganic NF membranes.

12.3 Organic Chemical Industry

12.3.2.4

NF and Aqueous Mixtures of Alcohols

Low molecular weight alcohols, primarily methanol, and ethanol are extensively used as solvents and cleaners in many processes in the chemical (and other) industries. In the course of their use, they are contaminated and, depending on the actual application, may need to be purified for further use. This step often involves costly on-site distillation or purification at a central processing plant. In most cases, low molecular weight alcohols can be economically reclaimed with NF membrane systems more effectively than with a distillation process. For example, a NF system was constructed to recover contaminated 90% methanol. The methanol waste stream is coming from a membrane manufacturing process. The methanol permeate is recycled to the process. Th system was in operation in the membrane production department of Desalination Systems Inc., Escondido, California (today GE/SUEZ). Th design and operating parameters are element DK 4040 C, three elements per housing, feed pressure 8.3 bar (120 psi), and a capacity 5.5 m3 /day (Osmonics Application Bulletin 101). Other publications describe NF studies of larger organic microsolutes in methanol solutions [24]. Glycol-based solutions are commonly used by a number of industries. Glycol-based solutions are inexpensive, relatively easy to manufacture and modify, and have a number of uses. EG has the chemical formula CH2 (OH)CH2 (OH) (MW 62 g/mol) and can pass through a NF membrane. 12.3.2.5

NF of Aqueous Mixtures of Glycols

Th common use for EG is as a heat exchange fluid because it has excellent thermal conductivity properties and EG/water solutions have high boiling points and low freezing points. EG and EG derivatives are also used during manufacture of resins, pharmaceuticals, food surfactants, lubricants, inks, solvents, polymers (plasticizers), and other products. EG is poisonous if ingested in sufficient quantity. NF membranes are used to purify glycols and reject impurities such as hardness, sulfates, color, organic molecules with MW > 300 g/mol, and other contaminants. Glycols are passing the NF membrane and larger pollutants are rejected. The NF permeate can be further concentrated by UHPRO membranes, and the maximum concentration is limited by the osmotic pressure of the concentrated glycol molecules. Examples are given below of NF systems in deicing and antifreeze processes and in the recycling of glycols in the organic chemical industry. Deicing fluid is sprayed on airplanes during temperatures below freezing point. Deicing fluid typically contains 30–50% EG, surfactants, and other chemical additives. Some airports deice their planes at each gate with a mobile deicing unit and other airports deice their planes at one central location immediately before planes take off. RO-NF systems are used to reuse deicing chemicals and some companies are specialized for this application [25]. Several systems are installed at airports in Minneapolis, MN; Dallas, Texas;

Ethylene Glycol as Deicing Fluid

563

564

12 Nanofiltration in the Chemical Processing Industry

Chicago, IL; and others [26]. A RO-NF-evaporator unit purifies and concentrates the glycol up to 50% for reuse. Propylene Glycol, Cleaning, and Recycling by NF Another application to recycle gly-

col is in the automobile industry. Used automobile antifreeze (propylene glycol) is considered a hazardous waste. High waste disposal costs make it economical to reclaim the contaminated antifreeze for reuse. Propylene glycol reclamation can be achieved with the installation of a RO-NF membrane system. Propylene glycol (e.g. 33% concentration) is passed through the membrane while multivalent salts, oils, suspended solids, and color are removed. Th specific flux is 6–10 l/m2 h at approximately 30 bar. NF can clean polypropylene glycol solutions either for direct reuse or for downstream concentration by high-pressure RO, UHPRO up to 120 bar, or evaporators. As a pretreatment to evaporators, the NF system provides an excellent quality feed that results in more efficient evaporator performance and less frequent cleaning.

12.3.2.6

Dyes and Inks

Dyes NF is becoming widely accepted in the dyestuff industry, in particular for reactive, acid, and direct dyes, as a way of producing high-quality products, maximizing yields and capacity, and saving materials. Reactive dyes are soluble anionic dyes in the molecular weight area of 600 up to 3000 Da. NF is used for both desalting and concentration of dyes and inks [27]. The dye, retained by the membrane, is desalted and concentrated, increasing its strength by reducing the content of inorganics. The first patents obtained for this application were from Ciba Geigy in 1982 [28, 29]. Plant capacity can be increased using NF before drying in the production of powder dyes. Th feed concentration to the dryer can be increased from the usual level of 7–10% total solids (TS) up to 25% solids. By running the spray dryers continuously and using NF to concentrate the dye before spraying, it is envisaged that 170% increase in production volume will be achievable. Spray drying is more efficient because the granulation of the dye takes place without the production of dust. In the manufacture of liquid dyes, almost complete desalting, which helps to improve the stability of the product, enhances the solubility of the dye. Diafiltration is used to allow a high level of desalting to be achieved. The consequence of this is that some dye passage through the membrane occurs, although this is considerably less than that from a filter press. By using NF, the dye production is increased by about 8%. With a level of 98–99+% removal of the salts, about 0.01–0.25% of the dye passes the membrane, depending on the molecular weight and type of the dye (see Figures 12.10 and 12.11) [30]. Th NF membranes used have typically a MWCO of 200–300 Da. The desalting and concentration of the dye is usually carried out in two steps: a diafiltration process desalts the raw dye slurry, containing 6–8% dye and 8–10% NaCl. When the NaCl content of the dye has reached its desired level, 99% can be achieved. It has been found that color removal decreased with increasing salt concentration. Especially in low salt

565

566

12 Nanofiltration in the Chemical Processing Industry

Figure 12.12 Typical NF plant for desalting of dyes. Source: Supplied by PCI Memtech.

concentrations, a gel layer is formed by the rejected dye on the membrane surface, operating as a resistance to the permeation of dyes [31]. Further discussion of NF applied to the dye/textile industry is given in Chapter 14. Inks In France, a NF installation has been in operation since February 1997 for

the decolorization of effluents containing fountain pen inks [32]. In the experimental stage, two membranes were tested: Koch MPT20 and MPT31, selected according to their MWCO of 600 and 400 Da, respectively, and their resistance to solvents and acid solutions. The membranes were tested in batch mode, up to a volumetric concentration of a factor 10. Th pressure was 25 bar and the temperature was 30  C. Th MPT20 membrane showed the highest flux rate, from 80 to 45 l/m2 h during the concentration, but the retention decreased from 98.8% to 96.5% in three weeks. The MPT31 membrane showed a lower flux, 55–35 l/m2 h during the 10-fold concentration, but the retention remained stable at 99.95%. Th industrial plant is very small, only 0.9 m2 , with filter batches of 2000 l for 50 hours, followed by a cleaning cycle of 1 hour. The average flux is 40 l/m2 h at 30  C, the cross-flow velocity is 2.5 m/s, and the pressure is 26–28 bar, resulting in an installed electrical power capacity of 4.1 kW. Th retention observed is 99.7–99.95%, and the volumetric concentration factor is 15–40. Th membrane lifetime is two years.

Installations Most of the installed NF plants in dye and ink are found in Southeast Asia, although some plants are installed in Europe and the United States. Th total membrane area for this application 15 years ago was estimated >100 plants with estimated 8000–10,000 m2 [100]. Some examples of NF installations are found in the dye production of CBW Chemie in Bitterfeld-Wolfen in Germany with a capacity of 1000 ton/year [33], at Dystar, also in Germany, where a NF plant of 100 m2 , with 15 spiral wound Nadir NF PS 10 membranes, is installed [34] and 2

12.3 Organic Chemical Industry

NF plants at Avecia/Stahl, in St. Claire du Rhone, France, of 270 m2 with tubular membranes and of 800 m2 with spiral wound membranes, for the desalination and concentration of leather dyes [35]. In China, a NF installation is in operation since 1993 [36]. In this plant, the membranes were exchanged after a service of three years. In Newcastle, USA, a large NF plant with 400 m2 with spiral wound membranes is in operation at Avecia/CEL for the desalination and concentration of ink-jet products [35]. 12.3.2.7

Increasing the Recovery of Pyridine Dicarboxylic Acid (PDC)

Pyridine-2,3-dicarboxylic acid (PDC, MW 167 Da) can be used as a raw material in the preparation of pharmaceuticals, agrochemicals, and colorants. One of the production processes for PDC is the oxidation of 2,3-lutidine with KMnO4 in an aqueous solution containing KOH. The MnO2 formed during the oxidation is filtered off and washed. After the crystallization of the PDC by adding HCl to obtain a pH = 1, the crystals formed are centrifuged and washed. The remaining mother liquor and wash water from the PDC filtration contain not only salts and by-products but also substantial amounts of PDC product, which is difficult to work-up. The wash water of the MnO2 filtration contains feedstock, PDC (about 5 wt%, which is the maximum solubility at high pH), and by-products, but no salt and is recycled to the oxidation step or can be combined with the mother liquor and wash water from the PDC filtration. Recirculation of the mother liquor and wash water of the PDC filtration causes an undesired accumulation of salt and by-products. Because of the fact that the mother liquor and the wash water have to be discharged, the overall recovery yield of PDC in this process is low: 65–70%. NF is used to separate the PDC from the salts and by-products [37]. The process scheme is shown in Figure 12.13. Both the mother liquor and the wash water from the PDC filtration contain about 1 wt% PDC. Th by-products in this combined stream are 0.015 wt% picolinic acid, 0.12 wt% 3-methylpicolinic acid, 0.04 wt% nicotinic acid, 0.54 wt% 2-methylnicotinic acid, and about 20 wt% KCl. The pH of the feed to the NF unit is raised to pH 8.5 by means of adding 25% KOH solution. Th membranes used were Stork Friesland (now Pentair X-Flow) WFN 0505 membranes, which show low, even negative, retention for salts at high pH values and high salt concentrations [38, 39]. The applied pressure was 30 bar, and the feed temperature was 40  C. Th local retention for PDC remained high: 91.6% at the start and 85.6% after a sixfold concentration. The retentions of the by-products and the salt also decreased with increasing concentration, even to negative values. About 85–90% of the KCl and the by-products were removed by using NF. Because of the low or negative retentions for the by-products and KCl, no accumulation of these compounds was observed. Th concentrate can therefore be recycled to the crystallization step and the permeate can be discharged without any further treatment. Th recovery of PDC with NF is 80–85%. Th overall result is an increase in product yield from 65% to 70% to about 90%, resulting in a payback period of less than one year for the NF process. In the United States, a NF installation is in operation since 1997 for PDC production with a total capacity of 600 t PDC/year. The amount of feed to the NF plant is 7200 m3 /yr, and the membrane area of the NF installation is about 30 m2 .

567

568

12 Nanofiltration in the Chemical Processing Industry

Lutidine KMnO4 KOH Water Oxidation Water Filtration

MnO2

WW, 5% PDC, pH 14 Permeate

HCl Precipitation pH 1

Retentate 5–6% PDC

Nanofiltration ηPDC = 80–85%

Water ML, 1% PDC Filtration

pH8–9 WW, 1% PDC

PDC η = 90%

KOH

Figure 12.13 PDC process with NF [37]. Source: DeltaMem, Switzerland; previously the membrane business of Sulzer.

12.3.2.8

Increasing the Yield of Organic Salts

Iminodiacetic acid (IDA), a chemical intermediate and a chelating reagent with a molecular weight of 133 g/mol, is produced as an organic salt solution. Th organic salt is converted to the desired acid product through the addition of a strong acid, leading to neutralization of the salt, and crystallization of the acid. Th presence of salt in the solution restricts the crystallization [93], and the remaining mother liquor from the crystallization step therefore contains significant amounts of IDA and salt. Application of NF for desalination and recovery of IDA from the mother liquor can increase the overall IDA yield and improve the sustainability of the process [92]. Out of three membranes evaluated for the process, NF-270 showed the highest (>90%) IDA retention over a pH range between 3 and 10. Desal 5-DL showed a lower retention between 80% and 90% over a pH range between 4 and 11 (at high pH, the IDA retention of Desal 5-DL approached the retention of NF-270) and reduced retention at pH < 4, whereas the more open membrane Nanomax50 showed a retention below 40% and reducing at lower pH [92]. For all membranes, a negative NaCl retention was found for operation at pH > 7 and a (limited) positive NaCl retention was found for pH < 6. For an industrial solution containing approximately 46 g/l IDA and 113 g/l NaCl, and operating at pH = 5.2, approximately 96% IDA recovery was reported with approximately 60% removal of NaCl, using a combination of concentration and diafiltration mode operation [92]. This combination of concentration and diafiltration leads to optimal flux, IDA recovery, salt removal, and water usage.

12.4 Pharmaceutical and Biotechnology Industry

For glyphosate (N-(phosphonomethyl) glycine), a widely used herbicide for which IDA is used as an intermediate, a similar use of NF is reported [94, 95]. In this case, glyphosate liquor, obtained after glyphosate (MW = 169 g/mol) crystallization and neutralization of the mother liquor using a caustic solution, is desalinated using NF. The thus-obtained herbicide product contains less salt, which avoids or reduces soil salination during product use. The glyphosate liquor typically contains 1–2% glyphosate, organic components in high concentration, 1–3% Na2 HPO3 , and 10–20% NaCl and has a pH of 13.5. Before NF operation, the pH needs to be reduced to 11 for membrane stability reasons. Desal-5 DK appears to be superior over Desal-5 DL and NTR7450 [94]. During batch concentration using Desal 5DK, high glyphosate retention of around 95% is obtained at low concentration factor, reducing to 90% at higher concentration factor. However, the membrane flux is relatively low starting at around 7 l/m2 h at 20 bar trans membrane pressure and a temperature of 20  C and reducing to 1.5 l/m2 h at higher concentration factor. The low flux is probably due to relatively high osmotic pressure difference and crystallization on the membrane surface [94]. The use of diafiltration operation improves the flux (although not quantitatively specified) and leads to less glyphosate loss at higher NaCl removal [95] as expected.

12.4 Pharmaceutical and Biotechnology Industry Th pharmaceutical and biotechnology industry manufactures products for human and veterinary uses. This section describes some industrial standard processes within the pharmaceutical and biotechnology industry. We try to demonstrate the role of NF membrane technology in those processes. However, most of the interesting NF applications in pharmaceutical and biotechnology industries are covered by secrecy agreements. In the pharmaceutical and biotechnology industry, there are three main stages in the process of the development of a product: • Research and development • Conversion of organic and natural substances into bulk pharmaceutical substances through fermentation, extraction, and/or chemical synthesis • Formulation of the final pharmaceutical product 12.4.1

Research and Development of Bench Test Technology

Based on data from a 1995 study by the Center for the Study of Drug Development at Tuffs University, a pharmaceutical research and development (R&D) facility discovering and developing new medical agents evaluated approximately 5000–10,000 components. However, only about five proceeded to limited human clinical testing. Product developments in general, therefore, must be faster, cheaper, and more efficient. Membrane technology bench tests for purification, separation, and concentration objectives can be performed quickly and are reliable. Systems can be made for smallest volumes available in biotech and pharma industries.

569

570

12 Nanofiltration in the Chemical Processing Industry

Figure 12.14 Lab bench test systems for flat sheet polymer cross-flow membrane. Stirred cell and pump-driven cross-flow cell for smallest volumes. (pictures: DeltaMem, Switzerland; previously the membrane business of Sulzer). Stirred cell. Source: Picture from Evonik technical manual.

12.4.1.1

Membrane Lab Bench Testing for Smallest Volumes

Th development of a membrane-based process begins at the lab bench test. Separation, purification, and concentration applications are carried out utilizing cross-flow cell systems. The membrane cells can be stirred to obtain required cross-flow. Feed pressure can be obtained by the use of pressurized gas or by a feed/recycling pump (Figure 12.14). Smallest volume requirements allow early testing. Simplicity in design and use permits quick screening of candidate polymer membranes. The conditions can be easily modified for temperature, pressure, and viscosity. Th selected polymer membrane must be confirmed by pilot tests, where an industrial scale module must confirm the results under industrial conditions.

12.4.1.2

Pilot/Prototype Applications

Once the screening lab bench test has identified the appropriate membrane chemistry to achieve the desired goal, it is directly scaled up to customized pilot test units. Scale up sometimes includes an intermediate prototype unit pilot plant (see Figure 12.15). This process creates a blueprint for the final production phase. Th smallest available test modules for chemical and biotechnology applications are spiral wound 1812 elements in a 2′′ diameter and 12′′ length dimension. Th area of an 1812 element is approximately 0.4 m2 depending on the feed spacer used. There is a complete line of different 1812 membrane products available with different element constructions (e.g. different feed spacers) to handle a wide range of solids and viscosity levels. Similar small elements are available as UF tubular test modules or as single or multifiber modules (Figure 12.15).

12.4 Pharmaceutical and Biotechnology Industry

Figure 12.15 Pilot plant for spiral wound elements up to 70 bar and 80  C, maximum 18 m2 membrane area. Photo: GE Water, technical manual.

12.4.2

General Industrial Process Description

Most pharmaceutical substances are manufactured utilizing “batch” processes and are manufactured by (i) chemical synthesis, (ii) fermentation, (iii) isolation/recovery from natural sources, or (iv) combination of those processes. Examples of different drugs produced by each of these processes are as follows: • Chemical synthesis: Antibiotics, hormones, vitamins, antihistamines, and vaccines • Fermentation: Polysaccharides, antibiotics, steroids, vitamins, clavulanic acids, and amino acids • Natural product extraction: Enzymes and digestive aids, insulin, and vaccines. 12.4.3 NF Applications in the Pharmaceutical and Biotechnology Industry Th s industry offers a great potential for NF membrane separation technologies in wastewater treatment and purification processes, and they are the least published with most of the interesting applications subject to secrecy agreements. Some of the OEM companies involved have worked for more than 40 years in pharmaceutical and biotechnology NF membrane applications and this provides indirect evidence of the significance of the role of NF. Dominant NF applications are antibiotics recovery from fermentation processes, biological active principles purification and concentration, pyrogen removal from biological liquids, enzyme recovery, protein or polysaccharide concentration, and purification. Proven pharmaceutical applications include the production of cephalosporin-C, clavulanic acid, erythromycin, penicillin G, vitamin B12, 6-aminopenicillanic acid (6-APA), 7-aminodeacetocephalosporanic acid (7-ADCA), and others [41]. Other examples of NF systems installed in recent years are purification

571

572

12 Nanofiltration in the Chemical Processing Industry

and concentration of erythromycin, chondroitin, acarbose, and glutathione. Membranes used are from SUEZ (exGE) and DDP (DowDuPont). Some of the plants are very large; Hydro Air Research (HAR) in Italy confirmed the delivery of the world’s largest ceramic plant for fermentation broth purification, followed by a purification/concentration system with >20 000 m2 NF membrane. Th expected installed NF membrane area in pharmaceuticals is ≫100.000 m2 [100]. 12.4.3.1

Antibiotics Production

NF is nowadays becoming one of the key processes in antibiotics production. In antibiotic production lines, NF processes can be used for several purposes and in different process steps. Among others, typical applications in downstream processing are the concentration of the fermentation broth after broth clarification, the preconcentration of the clarified broth before a resin step, and the concentration of the elute from resin columns before crystallization. With reference to the first two applications mentioned, a thin film spiral wound NF system is working with optimal performance after a ceramic microfiltration (MF) or ultrafiltration (UF) system, which is used to filter the fermentation broth. NF can achieve high-activity concentration without stressing the product due to thermal shock. NF is performed at a low temperature, which is usually fundamental to avoid product degradation. NF is also able to concentrate the product, while removing inorganic salts present in the solution by diafiltration, in order to obtain a higher purity of the final product. Organic synthesis frequently results in reaction mixtures containing both organic compounds and inorganic salts. Th se mixtures can be formed when mineral acids or bases are involved as catalysts or neutralization agents during the reaction. Th subsequent purification is usually straightforward because mineral salts precipitate in organic solvents and organic molecules are extracted by solvents and/or salted out from water. Th task is more difficult for mixtures with water as the solvent, when the interesting organic molecule is hydrophilic, indicating that the compound is completely (or at least very highly) soluble in aqueous media. In this case, the separation of organic/inorganic mixtures is long and tedious. Because industrial organic synthesis usually operates at high concentration levels, highly concentrated reaction mixtures are formed. As most of the published work on NF is related to diluted aqueous solutions (450

Salt retention

99.5%

NA

NA

NA

80 l/m2 h for DR80 and RB2 60 l/m2 h for CR

PVDF membrane [42]

Congo red (CR) and reactive back-5 (RB5)

∼97–99.9%

97–98.4%

NA

NA

4.8 ± 0.3 l/m2 h at 60 psi

Membranes

Dyes

PVDF hollow fiber membrane (ZW-1) [35]

Disperse orange 30 (DO30) and Disperse red 73 (DR73)

NA: data not available.

Flux

14.3 Membrane-Based Technologies for Textile Wastewater Treatment

Despite the superior performance of MBR compared to conventional biological treatment systems, recalcitrant COD and color components are still present in the treated effluent, which prevents its direct reuse for industrial process. Large variation in water composition also complicates the use of MBR in textile wastewater treatment, since it leads to varying degrees of surface fouling, which is difficult to anticipate. Hence, MBR technology is always combined with other advanced treatment technologies to achieve greater efficiency [47]. 14.3.3

Nanofiltration (NF)

Th growing popularity of nanofiltration over the last decades as an effective yet simplified textile effluent treatment technology can be attributed to the several benefits it offers in terms of dye and other auxiliary chemical removal along with partial removal of ions. One study revealed that NF was an efficient process for reducing conductivity, COD, and color in industrial effluent, to assure constant good water quality demanded by textile finishing processes [48]. Commercial NF90 (200–400 Da) and DK (150–300 Da) membranes for instance were reported to be able to achieve almost 100% color removal [49]. Desal-5 and NE-70 membranes were studied with synthetic dye wastewater and showed promising TOC rejection of 88.80–98.80% and 95.40–99.10%, respectively [50]. Subsequently, they were found to be effective in recovering water from real dye bath wastewater. Other commercial NF membranes that were used for dye removal and have shown promising results include TFC-SR2 [51], NFT-50 [52] and NF 270, TS-80, XN 45, MPF 34, and MPF 36 [33]. Th performance of NF membranes depends not only on the membrane physical and chemical properties but also on operating conditions. A comprehensive review of all these influences on membrane performance (water flux and solute rejection) as well as on antifouling properties will be provided in Sections 14.4 and 14.5. Unlike other types of pressure-driven membranes, the separation mechanisms of NF membranes are governed by both Donnan exclusion and sieving effect. Understanding the membrane separation mechanism is of importance, particularly for the treatment of textile effluent, which usually contains different classes of dyes and dissolved ions. 14.3.4

Reverse Osmosis (RO)

Reverse osmosis is very effective in achieving almost complete elimination of ions from textile effluents, producing permeate of highest quality for reuse [3]. However, it needs very high energy to operate owing to the high concentration of dissolved ions at the membrane surface. Th s causes severe concentration polarization and significantly increases osmotic pressure. The performance of RO membrane has been previously compared with NF at the same operating pressure to treat biologically treated wastewater (activated sludge) from a real textile industry treatment plant and the results showed that NF membrane was better than RO membrane in terms of water flux without compromising organic rejection [53].

633

634

14 Nanofiltration of Textile Dye Effluent

In another study, cross flow filtration tests of textile effluent were carried out using BW30 RO membrane and NF90 membrane over a wide range of concentration ratios and under different hydrodynamic conditions. The treated process streams, in both cases, satisfied the reclamation criteria and could be recycled to textile processes such as washing and dyeing [54]. However, the volume of the RO reject was very high compared to that with NF. Another interesting study further elaborated the difference between RO and NF membranes in dye removal and reported that the rejection of RO against methyl orange was 99.99% while NF achieved slightly lower rejection at 99.00%. With respect to salt removal, RO membranes always provided higher rejection than NF membranes, particular in the effluent that contained monovalent salts.

14.4 Advances in Nanofiltration Fabrication and Modification Nowadays, most of the commercial NF membranes are made by interfacial polymerization (IP) in which an extremely thin polyamide layer (several hundreds of nanometers in thickness) is formed over a microporous polymeric substrate. Th s type of membrane structure, generally known as thin film composite (TFC) membrane in the market, has been found applicable in a wide range of industrial processes that involve water and wastewater treatment processes. Although NF membranes can be industrially implemented without any major problems, continuous improvement in the separation characteristics of membranes is still needed in order to enhance competitive advantages. Herein, this section highlights the development of NF membranes over the past decade, particularly for the dye and salt removal process. 14.4.1

TFC Flat Sheet Membranes

One of the synthesis conditions that could determine the properties of TFC NF membrane is the choice of reactive monomers. In the work of Liu et al. [55], two types of new monomers, i.e. 2,5-bis(4-amino-2-trifluoromethyl-phenoxy) benzenesulfonic acid (6FAPBS) and 4,4′ -bis (4- amino-2-trifluoromethylphenoxy) biphenyl-4,4′ -disulfonic acid (6FBABDS), were synthesized, aiming to improve the inherent property of a typical polyamide layer and increase water permeation of TFC membranes. The active skin layer of the TFC NF membranes was prepared through the IP technique using amine solutions containing the novel 6FAPBS/6FBABDS and piperazine (PIP) as well as trimesoyl chloride (TMC) organic solution on the microporous polyphenylsulfone support membrane. With increase of 6FAPBS or 6FBABDS content in the amine mixture, it was reported that the water contact angle of the typical TFC membrane surface declined significantly from >70 to 90%) against three selected dyes (MW: 627, 992, and 1632 g/mol), in addition to >80% NaCl and 90% Na2 SO4 removal. Instead of using the IP process, Yu et al. [61] developed negatively charged NF membranes by dip coating sodium carboxymethyl cellulose (CMCNa) on the outside of polypropylene microporous hollow fiber membranes followed by cross-linking with FeCl3 . The fabricated composite hollow fiber membranes exhibited excellent dye rejections (with more than 99.7% removal of Methyl Blue [MW: 319.85 g/mol] and Congo Red [MW: 696.66 g/mol]) throughout the studied period due to the combined effect of steric hindrance and Donnan exclusion. As this membrane showed very poor rejection against NaCl (98%) against Eriochrome Black T (MW: 461.4 g/mol), Eriochrome Blue Black B ((MW: 416.4 g/mol), and Alizarine Red (MW: 360.3 g/mol) but also exhibited a very high water flux (75.2–80.5 l/m2 h) when tested at 0.4 MPa and 20  C. Although the authors attributed the dye rejection results to the membrane pore size that was smaller than the size of dyes, the high degree of water flux permeability (even greater than TFC membranes) was not explained in detail. 14.4.3

Positively and Negatively Charged Membranes

At present, most of the NF membranes sold in the market are negatively charged at neutral and alkaline environment and their applications are primarily oriented to the retention of small molecules and/or multivalent anions. Th development of positively charged NF membranes therefore could complement the existing short supply of such membrane property in the current market. Sun et al. [64] recommended an approach to develop positively charged NF membranes by introducing branched PEI and isophthaloyl chloride (IPC) on the surface of polyamide–imide (PAI) dual–layer hollow fiber membranes. As

14.4 Advances in Nanofiltration Fabrication and Modification

a polyelectrolyte, PEI is able to functionalize PAI membranes by cross-linking the carbonyl groups of imide rings in the polymer chains with amines to form amide groups. Because of this interaction, the NF membrane with positively charged selective layer could be produced, and it showed almost complete removal (99.8%) of Safranin O (positively charged dye). Th authors attributed the promising result to the charge repulsion effect between Na+ and the positively charged membrane. Although using branched PEI of higher MW tended to further increase membrane dye rejection, the membrane water flux was compromised. As evidenced from the field emission scanning electron microscopy images, higher MW branched PEI could lead to the formation of a thicker selective layer that increased mass transfer resistance of water molecules and thus reduced water permeability. Cheng et al. [65] modified the surface of a base membrane using PEI followed by cross-linking with butanediol-di-glycidyl-ether. Th fabrication process uses standard organic solvents and avoids the need for hazardous concentrated sulfuric acid, which significantly benefits the scale up potential of any future commercial manufacturing process. Besides showing superior water flux compared to commercial Desal-DK and Nanomax-50 membrane, the developed membrane was reported to have an excellent balance of water flux and dye rejection when it was employed to remove Methylene Blue (MW: 319.85 g/mol), a positively charged organic dye, from a simulated low concentration effluent. Th membrane performed extremely well and achieved 98% rejection at 5 bar with a flux of about 17 l/m2 h. By co-depositing branched PEI and low-cost catechol on the surface of polyacrylonitrile (PAN) microporous membrane, Xu et al. [66] developed a novel positively charged NF membrane for dye and salt removal. Streaming potential measurements confirmed that this membrane was indeed positively charged in the pH range below 9. Because of this, it showed higher rejection against Crystal Violet (MW: 407 g/mol, charge number: +1) in comparison to Orange G (MW: 452 g/mol, charge number: 2). More importantly, the novel membranes displayed stable long-term separation performance toward MgCl2 removal, and the dye-fouled co-deposited membranes could be facilely regenerated and reused with a simple static immersion operation. UV-initiated grafting has also drawn interest to prepare positively charged NF membrane owing to its easy operation and low cost. Moreover, the chemical bond between the substrate membrane and the active layer might increase the stability of the NF membranes at high-pressure operation. Zhong et al. [67] fabricated novel positively charged NF membranes using sulfonated polyphenylenesulfone (sPPSU) support via UV-induced grafting. Two types of monomers, [2-(methacryloyloxy)ethyl]trimethyl ammonium chloride (monomer A) and diallyldimethylammonium chloride (monomer B), were used during NF membrane fabrication. After UV-grafting positively charged vinyl monomers on the sPPSU substrate, the pore sizes of the resultant membranes were reduced to between 0.5 and 2 nm in their selective layers. Compared to the substrate surface, the UV-modified NF membranes exhibited higher hydrophilicity and could achieve pure water permeability of 9–14 l/m2 h bar and remove almost completely Safranin O (99.98%) due to their positive surface charges.

637

638

14 Nanofiltration of Textile Dye Effluent

Liu et al. [68] developed a positively charged loose NF membrane via UV-induced graft polymerization of diallyl dimethyl ammonium chloride (DADMAC) onto a polysulfone (PSF) support membrane. Th optimized NF membrane was achieved by irradiating the membrane coated with 50 wt% DADMAC solution for seven minutes at an irradiation distance of 22 cm. Th occurrence of N and Cl element, increasing skin layer thickness, and roughness variation all verified the successful immobilization of poly(DADMAC) layer on the membrane surface. At the optimized synthesis conditions, the NF membrane showed good hydrophilicity and pore size of 8.6 nm. In terms of separation characteristics, the membrane rejection to different valence states followed the order of AlCl3 > CaCl2 > MgCl2 > NaCl > LiCl > MgSO4 > Na2 SO4 . The rejections to divalent and multivalent cations in particular were all higher than 90%. Furthermore, the optimized membrane could achieve almost complete rejection against Methylene Blue and Congo Red. In view of the different charge properties of textile dyes, Akbari et al. [69] developed NF membranes with surface charge identical with that of dye molecules using UV irradiation technique. Negatively charged and positively charged NF membranes were prepared using monomers of sodium p-styrene sulfonate (NaSS) and [2-(acryloyloxy)-ethyl]trimethylammonium chloride (AC), respectively. Upon surface modification, it was reported that the membranes could be successfully used for the treatment of dye effluents, containing either anionic dyes or cationic dyes. It was shown that the flux of dye solutions was quite stable irrespective of the concentration, indicating that fouling and/or osmotic pressure were limited. In order to enhance water permeability, a great deal of study has been conducted through selection and modification of the microporous support membrane. A vast variety of polymers have been successfully used as support layer for TFC membrane fabrication over the years. A thermally stable support layer in particular is highly desired in textile industry, which originally discharged hot effluent. In view of this, Han et al. [70] fabricated new NF membranes using poly(phthalazinone ether sulfone ketone) (PPESK) as the support substrate layer. Th y fabricated two types of NF (negatively charged and positively charged) membranes through IP and phase inversion methods, namely, PIP/PPESK and positively charged quaternized PPESK (QA PPESK). QA PPESK membrane showed outstanding performance compared to PIP/PPESK in terms of water flux and dye rejection in the long-term operation at 60  C. Th excellent performance of the QA PPESK membrane may be explained by the hydrophilic properties of the positively charged surface that make it less susceptible to fouling. Another thermally stable substrate made of poly(phthalazinone ether amide) (PPEA) was also attempted for making the TFC NF membrane [71]. When tested at 80  C for up to 250 h using a synthetic solution containing 2000 mg/l Acid Chrome Blue K (MW: 586.4 g/mol) and 2000 mg/l NaCl, the consistent dye rejection (99.3–99.9%) achieved could testify the thermal stability of the membrane. Th s membrane exhibited a higher flux rate of 270 l/m2 h at 1.0 MPa at 60  C and demonstrated flux stability for eight operating/cleaning cycles, resulting in just 8.5% decline in the end. The minor declination of the water flux meanwhile was caused by the deposition of dye molecules on membrane surface.

14.5 Factors Affecting NF Performance

Additionally, Wei et al. [72] focused on the application of a TFC NF hollow fiber membrane synthesized by IP on a support membrane made of PSF and PES for dye desalination and concentration. Th fabricated negatively charged NF membranes yielded an MWCO of 520 g/mol and a pure water flux of 47.5 l/m2 h at 0.4 MPa. Rejections of different salts followed the order of MgSO4 > Na2 SO4 > MgCl2 > NaCl at pH 6.8, with up to 96.2% rejection for MgSO4 . With respect to dye separation, the membrane demonstrated 99.9% removal of reactive Brilliant Blue X-BR (MW: 681.39 g/mol) and Acid Red B (MW: 520.4 g/mol) owing to its negative charge properties.

14.5 Factors Affecting NF Performance Research works always indicated the high potential of NF membranes for reuse of water and chemicals from the textile effluent. In order to achieve a good balance of water flux and solute rejection, one needs to understand the factors that can influence the performance of NF during treatment process and further optimize the conditions. It is generally agreed that factors such as membrane characteristics, feed properties, and operational process conditions are the key components influencing NF performances. Th advancements in NF membrane fabrication and modification reviewed in Section 14.4 show that much has been done over the past decades to improve NF membrane performance. Th s section focuses on the influences of feed properties, membrane charge, and hydrodynamic conditions on the filtration performance of NF membranes. 14.5.1

Influence of Feed Properties

Generally, NF membranes are able to achieve very high removal rate against dissolved dyes if the characteristics (e.g. MW, concentration, and charge number) of the dyes are known. Lopes et al. [73], for instance, reported that commercial NF membrane – MPS 31 (from Weizmann) with smaller MWCO – could achieve a greater dye removal rate compared to two other commercial membranes (NF45, Dow FilmTec and DK 1073, Osmonics) as more dyes were retained by the membrane with smaller pore size. Tang and Chen [74] studied the influence of dye concentration on membrane performance and found that the rejection of Reactive Black 5 (MW: 991 g/mol) in commercial TFC-SR2 membrane (from Fluid System) remained almost constant with increasing dye concentrations from 92 to 1583 mg/l at a feed pressure of 5 bar. Similarly, Akbari et al. [75] found that the dye rejection was insignificantly affected by increasing dye concentration (from 2000 to 6000 mg/l) at an operating pressure of 10 bar. Using self-made NF membranes, Ismail and Lau [62] also observed that rejection rates against Reactive Black 5 and Reactive Orange 16 (MW: 616 g/mol) were only slightly affected (with rejection recorded at 99.54–99.96% and 95.38–99.47%, respectively) with increasing individual dye concentration from 50 to 250 mg/l. The promising dye removal rates as shown in these studies could be mainly attributed to the use of NF membranes

639

640

14 Nanofiltration of Textile Dye Effluent

with much smaller pore size compared to the size of dye components. Ong et al. [60] also observed that dye rejections remained high at >98% regardless of dye concentration, but membrane water fluxes were negatively affected with increasing dye concentration from 100 to 1000 mg/l. This observation might be explained by higher fouling as well as a higher osmotic pressure at the feed side that reduces the net driving force across the membrane. As dyes are prevented from passing through the membrane structure, they will accumulate on the membrane top surface. Thus, the higher the concentration of dyes in the feed the greater the amount of dyes depositing on the membrane surface. As a consequence, dye removal rate will increase for a long period of operation owing to the formation of a secondary barrier layer formed by retained dyes on top of the membrane skin layer. Besides dyes, salt at high concentration is normally found in the textile effluent as it is required during manufacturing process for the dye molecules to fully penetrate the cloth and provide the necessary dyeing action. It is generally known that salt rejection of NF membrane tends to decrease with increasing salt concentration. Sodium chloride, for example, can be ionized completely into Na+ and Cl in acid, alkali, or pure water and the concentration of ions increases as a function of salt concentration. Based on the principle of Donnan equilibrium, repulsive force from the negatively charged membrane would be negatively affected with increasing electrolyte concentration. A lower repulsive force indicates that more Cl anions are allowed to pass through the membrane and affects the rejection rate. Moreover, increasing salt concentration could build up concentration polarization on the membrane surface and reduce electrostatic repulsion of the membrane, leading to lower water permeability [76]. Previous work has shown that increasing salt concentration could promote the penetration of dye molecules through the membrane and affect the removal rate owing to reduced Donnan exclusion effect [77]. Nevertheless, Jiraratananon et al. [78] experienced no change in dye removal rate after NaCl was added into the dyeing solution. According to them, dye retention is mainly governed by steric exclusion effect rather than Donnan exclusion effect. Th s explanation seems more reasonable because the pore radius of NF membrane is usually much smaller than the effective hydrodynamic radius of the dye, making it unlikely to permeate through NF. Th increasing ionic strength of the feed solution tends to result in high osmotic pressure and concentration polarization, which negatively affects permeate flux [79]. In general, the rejection of divalent cations is always better than that of monovalent cations due to the ionic charge whereas among the divalent cations, for example, the rejection of magnesium is higher than that of calcium due to its lower hydration energy [80, 81]. Additionally, ionic strength effect could play a role by screening the electrokinetic effects occurring under dilute conditions. As discussed before, two factors, i.e. steric effect and electrostatic effect, are usually used to explain the separation mechanism of NF membranes. At low salt concentration, there is no significant influence of electrostatic repulsion between dye and membrane surface that affects dye retention [82]. However, increase of dye concentration in the feed increases dye retention owing to the reduced membrane pore size as a result of cake layer formation [57, 83]. By increasing the salt

14.5 Factors Affecting NF Performance

concentration, for example, from 500 to 10 000 mg/l, the sieving effect is almost not affected whereas Donnan effect becomes less effective [83]. Increasing feed water pH tends to increase the zeta potential (negative charge) of the membrane, leading to increased electrostatic repulsion, which eventually improves salt rejection. In higher pH of the feed water, MWCO influence is less important. For example, the membranes NF-90 and NF-200, which have different MWCO values, showed same rejections for negatively charged organic acid, which is due to electrostatic repulsion [84]. 14.5.2

Influence of Membrane Properties

By comparing monovalent salts (e.g. NaCl) with divalent salts (e.g. Na2 SO4 ), it is found that the divalent anion SO4 2 , which has a higher valence and bigger size than Cl , is usually rejected better by negatively surface charged NF membrane [85]. Using uncharged NF membranes, Vrijenhoek and Waypa [86] found that the salt removal rate was mainly governed by the size exclusion effect. The rejection of single salt was in the order of CaCl2 > Na2 SO4 > NaCl. Although the atomic weight of Ca2+ is smaller than that of SO4 2 , its rejection is slightly higher than that of SO4 2 . This indicates that cations are able to attract more water molecules around the ions, leading to a larger hydrated radius being formed. For the separation of multiple salt solutions containing NaCl and Na2 SO4 , Vrijenhoek and Waypa [86] found that size exclusion is the dominant mechanism for the ions’ retention where the observed order of rejection is SO4 2 > Na+ > Cl , which is in reverse order of ionic diffusion coefficients. Contradictory results are seen in the separation efficiency of NF membrane against salts as Peeters et al. [87] reported that the salt rejections in the neutral NF membrane were in the order Na2 SO4 > CaCl2 > NaCl. They attributed the rejection trend to both size exclusion and Donnan exclusion mechanisms. It must be mentioned that contradictory findings on the rejection order of salts could be affected by the differences in salt concentration as well as process conditions employed in a separate work. Further investigation using positively and negatively charged NF membrane demonstrated the rejection order of CaCl2 > NaCl>Na2 SO4 and Na2 SO4 > NaCl>CaCl2 , respectively [87]. Positively charged NF membrane is able to repel cations, particularly divalent cations, Ca2+ , and attract anions, particularly divalent anions, SO4 2 . Negatively charged NF meanwhile prefer2+ entially rejects SO2 4 but permeates Ca . As alkaline condition is necessary to create covalent fixing between dye and fabric during textile process, the use of NF membranes with negative surface charge in neutral/alkaline environment is more preferable because the membrane possesses higher electrostatic repulsion against Na2 SO4 and NaCl. Th se two types of salts are commonly used in textile industries. Several studies have investigated in detail the electrostatic interaction/repulsion between different dye molecules and membrane surface [33, 88, 89]. In general, NF membrane exhibits an isoelectric point (IEP) of 4–6 and because of this the membrane surface will have positive charge in acidic environment but negative charge in alkaline condition. When cationic dye molecules were present in acidic pH there was strong electrostatic attraction and vice versa in alkaline pH. Thus, electrostatic interaction is the important factor

641

642

14 Nanofiltration of Textile Dye Effluent

Electrostatic interaction

(–) ve dye molecules

Active layer

Membrane porous substrate

Electrostatic interaction (+) ve dye molecules

pH < IEP

IEP

pH > IEP

Figure 14.2 Mechanism of electrostatic interaction between dye and membrane surface. Source: Chidambaram et al. [33]. Reprinted with permission from Elsevier.

for the dye/salt separation. Figure 14.2 shows the mechanism of electrostatic interaction/repulsion of dye molecules and membrane surface. 14.5.3

Influence of Hydrodynamic Conditions

In addition to the feed properties, the efficiencies of dye/salt separations using NF membranes are also dependent on the operating conditions such as pressure, temperature, and cross flow velocity. Bes-Pia et al. [90] investigated the relationship between the permeate flux and feed pressure on NF90 (Dow FilmTec) and DK5 membranes (Osmonics) and found that by increasing the feed pressure from 10 to 20 bar, the permeate flux increased significantly. Since NF is one kind of the pressure-driven membrane processes, increasing operating pressure would improve membrane permeate flux accordingly. Ong et al. [85] however reported unexpected flux behavior of NF90 membrane with increasing operating pressure during the treatment of synthetic dyeing solution. Compared to the NF270 membrane (Dow FilmTec) that showed remarkable flux increment with increasing pressure from 10 to 18 bar, NF90 in this case only displayed minor flux increase with pressure. The insignificant flux increase of NF90 can be explained by the fact that the membrane is more susceptible to compaction when it is operated at high pressure. With respect to color removal, Ong et al. [85] reported that both NF90 and NF270 membranes showed almost complete rejection of color regardless of the operating pressure studied. Th excellent decolorization can be strongly

14.5 Factors Affecting NF Performance

attributed to the smaller pore size of NF membranes compared to the size of RB5. Similarly, salt rejections of both membranes were almost unchanged by varying the feed operating pressure. Applying the NF270 membrane to the treatment of industrial textile effluent, Aouni et al. [48] reported that COD and color removal (both >90%) were almost unchanged by increasing pressure from 5 to 15 bar. Th conductivity removal rate meanwhile varied between 52% and 68%. In contrast, Pal et al. [91] and Yu et al. [83] reported a lower value of membrane flux at high operating pressure. With the use of high-resolution optical microscopy, Pal et al. [91] found that this was a direct consequence of dye particles moving inside the membrane and depositing on the membrane surface after applying high pressure on the NF membranes. Appreciable increases in penetration inside the membrane surface and deposition over the membrane surface were experienced with increasing convective forces, causing flux to deteriorate over time. Yu et al. [83] also observed that permeate flux of NF membranes decreased when increasing feed pressure of the dye solution from 0.75 to 2.5 MPa. They attributed the findings to the increase in concentration polarization and dye adsorption on the membrane surface. Undeniably, increasing pressure promotes water molecule permeation through the membrane, but it also results in solute particles moving toward the membrane surface, making the deposited layer more compact in the long run. Thus, operating pressure should be carefully optimized in order to lessen the impacts of dye penetration/deposition on the water flux. Operating temperature of feed solution is another aspect that should be taken into consideration during NF membrane separation process. High water temperature requires lower operating pressure to achieve the desired flux compared to the process that operates at lower water temperature. Work conducted by Chen et al. [92] clearly demonstrated that permeate flux increased with an increase in the operating temperature following a significant decrease in water viscosity. In addition to the decreased solution viscosity, the increase in free volume of membrane at elevated feed temperature due to the increase in polymer chain mobility could also contribute to the membrane flux enhancement during the treatment of dyeing solution [60]. Han [93] also reported the improvement in NF membrane water flux with increasing feed temperature. Salt rejection and color rejection meanwhile were insignificantly affected within the temperature range of 15–95  C. Th findings were in good agreement with the work of Ong et al. [60] in which rejections of RB5, Reactive Blue 19 (MW: 627 g/mol), and Reactive Yellow 81 (MW: 1632 g/mol) remained almost unchanged at different temperatures ranging from 25 to 70  C. Employing high temperature for the NF process however is not recommended as it will reduce the membrane lifespan due to changes in surface properties. Also, it is not economically wise to run the NF process at high operating temperature. To maintain a stable filtration rate and minimize fouling propensity, the cross flow membrane filtration process is widely adopted in industrial applications. Compared to dead-end filtration, cross flow filtration mode offers significant advantages in reducing the need for frequent membrane cleaning. However, for the Reynolds number, which falls in either transition region or laminar region, the cross flow filtration process is not effective enough to prevent fouling as

643

644

14 Nanofiltration of Textile Dye Effluent

solid deposition is likely to happen on membrane surface at low flow velocity. Thus, Reynolds number of the NF process should be set at the turbulent region to achieve desirable performance [74]. In terms of color removal, Koyuncu [77] found that higher cross flow velocities (i.e. 1.11 and 0.41 m/s) were more effective to reduce the concentration polarization phenomenon on the membrane surface compared to the lower cross flow velocity of 0.11 m/s. As a result, higher membrane permeate flux and greater rejection were experienced. Enhanced water flux and dye removal of NF membranes were also reported in the work of Yu et al. [83] when cross flow velocity was increased from 0.1 to 1.1 m/s. It was elucidated that cross flow velocity plays an important role in limiting dye adsorption and concentration polarization. Nevertheless, it is quite difficult to define an optimized cross flow velocity for the NF treatment process as the value is strongly dependent on module and system design. Although there have been paradoxical reports in the literature regarding the performance of NF membranes in both acid and alkaline solutions correlating with membrane properties, most of the NF membranes do not tolerate extreme pH conditions (e.g. too low pH or too high pH values) for long-term operation. Capar et al. [94] evaluated the effect of pH on the NF membrane performance in treating acid dye bath wastewater and reported that COD rejection as high as 97% could be achieved by the membrane process with feed pH neutralization compared to only 55–77% shown by the membrane without feed pH neutralization. Color, meanwhile, was completely removed in these two options. Th y therefore suggested that pH neutralization is necessary for the recovery of acid dye bath wastewater. Th y also emphasized that pH neutralization is required when the reuse of water is of concern. Depending on the feed properties and the surface chemistry of the NF membrane, mild pH adjustment on the feed solution could be performed in order to achieve higher water flux and lower degree of dye adsorption [49].

14.6 Fouling Control Approaches Numerous studies have reported the viability of NF process in producing water of sufficient quality for reuse/recycle, but the significant flux decline that resulted from fouling is still a major concern to many when NF is considered for textile effluent reclamation. Given the fact that textile factories utilize different recipes of chemical compounds in their daily production, the characteristics of the wastewater generated are very likely to fluctuate over a wide range of values. Th s has made the identification of components (foulants) that contribute to membrane fouling very challenging. Th approaches that can be employed to reduce membrane fouling propensity are divided into several categories in which the first category consists of the use of NF membrane with improved surface characteristics and optimization of process conditions. The second category involves membrane remediation through an appropriate cleaning process and the last category is about the integration of NF process with other treatment methods.

14.6 Fouling Control Approaches

Of all these approaches, development of a less fouling-sensitive NF membrane without compromising water permeability and solute rejection is the most sustainable solution to minimize the fouling tendency during a treatment process. To improve membrane antifouling property, hydrophilic and/or charged functional groups can be introduced onto the membrane surface via coating, blending, or grafting. Upon surface modification, the resultant membrane is capable of withstanding the varying chemical composition of textile effluents and exhibiting less fouling sensitivity through decreased dye deposition/adsorption. Such information can be found in Section 14.4 in which the development of NF membrane materials for color/salt removal process is discussed. The influences of feed properties and process conditions on the NF performance meanwhile can be found in Section 14.5. Th s section will focus on the efficiency of using different cleaning methods/agents in retrieving the performance of NF membrane fouled by dye compounds. The e are a wide variety of cleaning mixtures and protocols suggested by both membrane manufacturers and researchers, but none of them are universally applicable for all types of membrane processes. Rather, the cleaning protocol that is adopted is strongly dependent on many factors. These include membrane properties, module design, and treatment process as well as feed characteristics. Th cost should also be taken into consideration for cleaning chemicals and energy consumption during the cleaning process. Chemical cleaning by far is the most widely used method in the NF membrane process as physical cleaning process (e.g. backwashing) is not suitable for densely structured membranes. Furthermore, chemical cleaning process requires relatively less energy compared to the physical cleaning method and is more effective in retrieving membrane water flux. Table 14.5 shows various chemical cleaning strategies employed by researchers in recovering the performances of NF membranes after being used for textile wastewater/synthetic dyeing solution treatment. As shown, acidic and alkaline solution are commonly used in NF cleaning process. Typically, acid solution (e.g. HCl) is used for the removal of inorganic polluting agents (scaling) while alkaline solution (e.g. NaOH) is used for the organic foulants removal. A thorough study on cleaning agents and their concentration is required because some chemicals might cause damage on the properties of membranes and reduce their lifetime. It appears from the literature that not much work has been published on this specific area and most of the findings were obtained from laboratory-scale filtration process that was conducted in very limited time. Th success of chemical cleaning methods indeed depends on many other factors. These include properties of cleaning solution (e.g. pH, temperature, and concentration), duration of cleaning and its frequency, as well as process conditions (e.g. pressure and cross flow velocity) [95]. It must be noted that there is no perfect solution (chemical cleaning procedure) that can completely avoid the degradation of membranes after they undergo numerous cycles of cleaning processes. Even so, the impacts of cleaning can be minimized using well-chosen cleaning agents through a trial-and-error approach in laboratory-scale/pilot-scale studies. A proper cleaning procedure is very critical to ensure the stable performance of NF membrane in the long run and further extend its lifespan.

645

646

14 Nanofiltration of Textile Dye Effluent

Table 14.5 Different chemical cleaning methods for NF membranes of textile effluent/dyeing solution treatment. Properties of NF membrane

Feed properties

Chemical agents/ cleaning protocol

Outcomes

Self-made hollow fiber NF membrane with pure water flux of 6 l/m2 h bar [60]

Reactive blue 19, 500 mg/l NaCl, 500 mg/l Na2 SO4 , and 50 mg/l surfactants

Cleaning with NaOH solution (0.1 wt%) followed by HCl (0.1 wt%) for 30 min

Higher flux recovery without affecting solute rejection (at least 88%)

Self-made positively charged membrane with chitosan coated on the surface of PSF substrate [116]

Textile industry effluent (Shandong, China)

Cleaning with 1 wt% NaOH solution at 1 MPa for 2 h

Restored 91.6% flux and < 1.5% decrease in salt rejection

NF90 and NF270 [117]

1000 mg/l RB5, 500 mg/l PVA, 5000 mg/l Na2 SO4 , and 3000 mg/l NaCl

Cleaning with 0.1 wt% NaOH solution for 8 h at 30  C

90% of water flux of NF270 recovered Only 32% water flux recovered for NF90

Self-made PES/SPEEK membranes [62]

1000 ppm RB5, 500 ppm PVA, 250 ppm NaCl, and 750 ppm Na2 SO4

Cleaning with 0.2 wt% HCl solution for 15 min followed by 0.5 wt% NaOH solution for another 15 min

Up to 86% of the membrane initial water flux retrieved Removed the dye deposition/cake

NE70 membrane (from Saehan) [50]

Effluent collected from Max Textile, Singapore

Cleaned with 2% citric acid and DI for 1 h and rinsed with 0.5 wt% EDTA solution for 1 h

93% of pure water flux recovered

NFT-50 membrane (from Alfa-Laval) [117]

Carpetmanufacturing factory effluent

Clean-in-place using HNO3 solution (pH 3) followed by NaOH solution (pH 9–10) for 30 min each

97% flux recovered Fouling was greatly reversible

MPF36 membrane (from Koch) with MWCO of 1000 [118]

Real industrial effluent that might contain either acid dye or reactive dye

Cleaning with 0.2 wt% HNO3 solution and 0.5 wt% NaOH solution for 45 min each

65–90% and 79–100% of water flux recovered by acidic and alkaline cleaning

14.7 Integrated Process Involving Nanofiltration Membrane fouling and NF reject volume are the two major obstacles that limit the extensive application of NF technology in the treatment of textile wastewater. It has been shown that process engineering if carried out professionally could provide an alternative pathway toward enhanced separation without

14.7 Integrated Process Involving Nanofiltration

significantly affecting operating/maintenance cost [96]. Direct NF process is a feasible solution only for the lightly contaminated effluents discharged by rinsing operations [97, 98]. For the treatment of complex textile wastewater streams, an integrated approach involving multiple treatment stages is certainly necessary. Generally, pretreatment prior to an NF process involves removing relatively large molecules that are able to cause pore narrowing/plugging on NF membrane. Significant reduction in the COD and BOD levels of the effluent prior to NF process could ensure better stability of water permeability and reduced cleaning frequency. As NF process only achieves partial demineralization, its permeate requires post-treatment if it is intended for reuse. Many different integrated treatment processes have been investigated in the past for color and salt removal. By integrating the NF process with pretreatment, it is found that the NF surface fouling could be reduced and potentially higher water recoveries can be obtained. Suksaroj et al. [99] reported that coagulation/flocculation process prior to NF process showed great potential to limit membrane fouling during water recovery. On the other hand, the utilization of microporous membrane (either MF or UF) prior to the NF filtration process has been attracting great attention among membrane scientists in textile wastewater treatment [100–104]. Besides reducing the degree of flux decline, other significant advantages offered by the membrane pretreatment process include recovery of unspent auxiliary chemicals (e.g. sizing agent – PVA), high molecular-size insoluble dyes (e.g. indigo, disperse), and caustic compounds [105, 106]. Further pretreatment using either activated sludge or MBR technology prior to UF and NF membrane process will improve the overall lifespan of membrane-based separation processes [7]. Th hybrid system using UF/MF membrane technology can be flexibly adjusted with respect to the system operating conditions to accommodate changes in the feed stream. Membrane–membrane hybrid systems are recognized as treatment processes of high efficiency, which can be competitive in comparison to the traditional methods of wastewater treatment. Preliminary removal of fine suspended solids and colloids from textile effluents is fundamental to prevent severe fouling and module damage of downstream NF, which in turn guarantees a good and constant performance of the NF system. In general, the fouling phenomenon in NF could be the result of cake layer formation and/or pore blocking. Without suitable pretreatment, the situation is expected to become worse due to cake compression with a decrease in cake porosity against operation time. Alcaina-Miranda et al. [101] highlighted the effectiveness of UF as a pretreatment stage in controlling NF fouling. Th y reported that the permeate flux of two commercial NF membranes remained practically constant during the studied period. Th small percentage difference between the initial clean water flux and the wastewater flux in NF could be explained by the fact that particles that caused fouling were efficiently removed in the UF stage. Th summarized important integrated process for textile wastewater treatment is presented in Table 14.6. On the other hand, with the use of MF followed by NF, Marcucci et al. [110] observed that significant flux decline was only recorded in the MF and NF system after 60 h of operation. Th MF flux was reported to decrease from 500 to

647

648

14 Nanofiltration of Textile Dye Effluent

Table 14.6 Integrated nanofiltration processes for efficient water and brine recovery from dye wastewater.

Integrated approaches

Reason for the integrated processes/dye waste contaminants

Outcome

Coagulation/flocculationNF [99]

Higher COD (500–1500 mg/l) and BOD level (200–400 mg/l)

Higher flux and good quality permeate (19 l/m2 h at 18.5 bar, 080

Excess biomass

Membrane bioreactor

Good

Excess biomass

>90

>80

Concentrate

Fair

Fair

>80 >85

Coagulation/flocculation Poor

Fair

Fair

—

Chemical precipitation

Poor

Fair

Poor —

Adsorption

Poor

Fair

Good >80 70–90 —

—

Oxidation

Poor

Fair

Fair

—

30–90 —

Stripping

Poor

Fair

Fair

—

90

40–60