Nanomaterials For Solid State Hydrogen Storage [1 ed.] 9781848001527

Citation preview

Nanomaterials for Solid State Hydrogen Storage

Fuel Cells and Hydrogen Energy Series Editor:

Narottam P. Bansal NASA Glenn Research Center Cleveland, OH 44135 [email protected]

Aims and Scope of the Series During the last couple of decades, notable developments have taken place in the science and technology of fuel cells and hydrogen energy. Most of the knowledge developed in this field is contained in individual journal articles, conference proceedings, research reports, etc. Our goal in developing this series is to organize this information and make it easily available to scientists, engineers, technologists, designers, technical managers and graduate students. The book series is focused to ensure that those who are interested in this subject can find the information quickly and easily without having to search through the whole literature. The series includes all aspects of the materials, science, engineering, manufacturing, modeling, and applications. Fuel reforming and processing; sensors for hydrogen, hydrocarbons and other gases will also be covered within the scope of this series. A number of volumes edited/authored by internationally respected researchers from various countries are planned for publication during the next few years. Titles in this series Nanomaterials for Solid State Hydrogen Storage R.A. Varin, T. Czujko, and Z. S. Wronski ISBN 978-0-387-77711-5, 2009 Modeling Solid Oxide Fuel Cells: Methods, Procedures and Techniques R. Bove and S. Ubertini, eds. ISBN 978-1-4020-6994-9, 2008

Robert A.Varin • Tomasz Czujko Zbigniew S. Wronski

Nanomaterials for Solid State Hydrogen Storage

Robert A. Varin University of Waterloo Department of Mechanical and Mechatronics Engineering 200 University Ave. W Waterloo, Ontario Canada N2L 3G1

Tomasz Czujko University of Waterloo Department of Mechanical and Mechatronics Engineering 200 University Ave. W Waterloo, Ontario Canada N2L 3G1

Zbigniew S. Wronski CANMET Energy Technology Centre Hydrogen Fuel Cells and Transportation Energy Natural Resources Canada 1 Haanel Drive Ottawa, Ontario Canada K1A 1M1

ISBN: 978-0-387-77711-5 e-ISBN: 978-0-387-77712-2 DOI: 10.1007/978-0-387-77712-2 Library of Congress Control Number: 2008929618 © Springer Science+Business Media, LLC 2009 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper springer.com

Preface

Although hydrogen as a chemical element has been known to the humankind and used in various capacities for a very long time, only in the past 15 years its importance to the world population as an energy vector has gradually emerged. A long-term reliance of humanity on the energy derived solely from fossil fuels, such as coal in the nineteenth and crude oil and natural gas in the twentieth century, has led to a number of new challenges facing all of us in the twenty-first century, such as sharp reduction in the world crude oil and eventually coal supply, global warming and following climate changes due to the release of growing amounts of greenhouse gas CO2, and poor urban air quality. Hydrogen is essentially the only viable remedy for the growing world energy problems. Hydrogen is a very attractive alternative energy vector for replacing fossil fuel-based economy. The future Hydrogen Economy offers a potential solution to satisfying the global energy requirements while reducing (and eventually eliminating) carbon dioxide and other greenhouse gas emissions and improving energy security. Hydrogen is ubiquitous, clean, efficient, and can be produced directly from sunlight and water by biological organisms and using semiconductor-based systems similar to photovoltaics. Hydrogen can also be produced indirectly via thermal processing of biomass or fossil fuels where the development of advanced technological processes combined with a CO2 sequestration is emerging. However, this rosy picture, as it usually happens in a real life, is marred by a number of obstacles which must be overcome before the Hydrogen Economy becomes a reality. One of these obstacles is safe and efficient storage of hydrogen particularly for mobile/automotive applications where hydrogen gas will be supplied to fuel cells that, in turn, will power the transport vehicles in a clean, inexpensive, safe, and efficient manner. From all possible solutions to hydrogen storage the one which relies upon storage in solid media (hydrides) is the most attractive one. The fast emerging nanoscience/nanotechnology will allow fabricating nanomaterials for solid-state hydrogen storage that can, in a long run, revolutionize hydrogen storage. This book is our modest contribution to this innovative area of hydrogen storage. Wherever possible we tried to illustrate the hydrogen storage behavior by our own results. In Chap. 1, we introduce the reader to the motivation for the transformation to the Hydrogen Economy. In a number of following sections/subsections, we v

vi

Preface

provide a comprehensive synchronic history of development of hydrides and nanomaterials including the existing fabrication methods with a special emphasis on ball (mechanical) milling in high-energy mills. Important hydride properties and experimental techniques for assessing hydrogen storage behavior are also discussed. In Chap. 2, we review hydrogen storage properties of selected simple metal and intermetallic hydrides with the most emphasis on magnesium hydride (MgH2) which now can be treated as a model hydride whose hydrogen storage properties in nanostructured form can be used as a benchmark for comparing the properties of other hydrides. Chapter 3 brings a thorough review of the properties of complex hydrides whose high volumetric and gravimetric capacities make them most attractive for the vehicular solid-state hydrogen storage in transportation. Chapter 4 provides information on carbons and nanocarbons as alternative means of hydrogen storage to solid hydrides. This includes diamond and nanodiamond, graphene, ordered graphites and nanographites, disordered and active carbons, fullerenes, carbon nanotubes, and other nanoshapes. Chapter 5 is a sort of an executive summary where we provide a critical assessment of the present state of knowledge and make predictions for the future developments. Waterloo, ON Waterloo, ON Ottawa, ON

Robert A. Varin Tomasz Czujko Zbigniew S. Wronski

Contents

1

Introduction ..............................................................................................

1

1.1 1.2

1

Motivation: The Hydrogen Economy ................................................ Brief, Synchronic History of Development of Hydrides and Nanomaterials......................................................... 1.2.1 Early Investigations of Metal–Hydrogen Systems and Hydrides .......................................................................... 1.2.2 Early Routes to Nanomaterials .............................................. 1.2.3 Historical Development of Classical Hydrogen Storage AB5 Alloys................................................................ 1.2.4 Historical Development of Interstitial Hydrides in Other Intermetallic Systems .............................................. 1.2.5 Historical Development of Nanophase AB2 Intermetallic Hydrides ........................................................... 1.2.6 New Routes to Nanomaterials: Mechanical Alloying and Mechanochemical Activation ......................................... 1.2.7 Historical Development of Lightweight Metal Hydrides and Hydride Complexes ........................................................ 1.2.8 Early Studies of Noninterstitial Transition Metal Ternary Hydrides ................................................................... 1.2.9 Toward Chemical/Complex Hydrides ................................... 1.2.10 Historical Development of Nanocarbons and Carbon Nanotubes .......................................................... 1.2.11 New Materials and Techniques.............................................. 1.3 Nanoprocessing in Solid State in High-Energy Ball Mills................ 1.3.1 Processes for the Synthesis of Nanostructured Materials ................................................................................ 1.3.2 Milling Processes and Equipment ......................................... 1.3.3 Nanoprocessing Methods and Mechanisms .......................... 1.3.3.1 Mechanical Milling ................................................ 1.3.3.2 Mechanical Alloying ..............................................

7 7 10 13 15 16 17 18 20 21 23 25 27 27 28 37 38 39 vii

viii

Contents

1.3.3.3 Mechanochemical Activation ................................. 1.3.3.4 Mechanochemical Synthesis (Mechanosynthesis) of Nanohydrides .................... 1.3.3.5 Mechanical Amorphization .................................... 1.4 Important Hydride Properties and Experimental Techniques ........... 1.4.1 Thermodynamics ................................................................... 1.4.1.1 Pressure–Composition–Temperature (PCT) Properties ..................................................... 1.4.1.2 Calculation of Activation Energy ........................... 1.4.2 PCT and Kinetic Curves Determination by Volumetric Method in a Sieverts-Type Apparatus ................................... 1.4.3 Microstructural Characterization of Ball-Milled Hydrides ................................................................................ 1.4.4 Weight Percent of a Hydride Phase and Hydrogen by DSC Method ..................................................................... References .................................................................................................. 2

40 52 55 56 56 56 60 65 71 73 74

Simple Metal and Intermetallic Hydrides .............................................

83

2.1

83 83

Mg/MgH2 ........................................................................................... 2.1.1 Crystallographic and Material Characteristics ...................... 2.1.2 Hydrogen Storage Characteristics of Commercial Mg and MgH2 ............................................... 2.1.2.1 Absorption .............................................................. 2.1.2.2 Desorption .............................................................. 2.1.3 Hydrogen Storage Characteristics of Mechanically (Ball) Milled MgH2................................................................ 2.1.3.1 Microstructural Evolution During Milling and Subsequent Cycling of Commercial MgH2 Powders .................................................................. 2.1.3.2 Hydrogen Absorption of Ball-milled Commercial MgH2 Powders ................................... 2.1.3.3 Hydrogen Desorption of Ball-milled Commercial MgH2 Powders ................................... 2.1.4 Hydrogen Storage Characteristics of MgH2 Synthesized by Reactive Mechanical (Ball) Milling of Mg ...................... 2.1.5 Aging Effects in Stored MgH2 Powders ................................ 2.1.6 Other Methods of Synthesis of Nanostructured MgH2 than Ball Milling ......................................................... 2.2 MgH2 with Catalytic Additives ......................................................... 2.2.1 Mg/MgH2–Metals and Intermetallics .................................... 2.2.1.1 Desorption in Vacuum ................................................. 2.2.1.2 Desorption at Atmospheric Pressure of Hydrogen..... 2.2.2 Mg/MgH2–Metal Oxides ....................................................... 2.2.3 Mg/MgH2–Carbon / Graphite and Carbon Nanotubes ...........

87 87 93 102 103 112 115 129 146 147 151 152 152 153 165 169

Contents

3

4

ix

2.3 Other Metal Hydrides Containing Mg............................................... 2.4 AlH3 ................................................................................................... 2.5 Other Metal and Intermetallic-based Hydrides: New Developments ........................................................................... 2.5.1 Metal Hydrides ...................................................................... 2.5.2 Rare-Earth AB5 Compounds.................................................. 2.5.3 Titanium–Iron AB Compounds ............................................. 2.5.4 Titanium and Zirconium AB2 Compounds ............................ 2.5.5 Other Novel Intermetallic Hydrides ...................................... References ..................................................................................................

170 174

Complex Hydrides ...................................................................................

195

3.1

Ternary Transition Metal Complex Hydrides ................................... 3.1.1 Mg2NiH4 ................................................................................ 3.1.2 Mg2FeH6 ................................................................................ 3.1.3 Mg2CoH5 ................................................................................ 3.2 Alanates ............................................................................................. 3.2.1 NaAlH4 .................................................................................. 3.2.2 LiAlH4.................................................................................... 3.2.3 Mg(AlH4)2 and Ca(AlH4)2...................................................... 3.3 Amides .............................................................................................. 3.4 Metal Borohydrides........................................................................... 3.5 Destabilization of High Desorption Temperature Hydrides by (Nano)Compositing ...................................................... 3.5.1 MgH2–LiAlH4 Composite System ......................................... 3.5.2 MgH2–NaAlH4 Composite System ........................................ 3.5.3 MgH2–NaBH4 Composite System ......................................... References ..................................................................................................

196 196 198 204 206 206 213 223 231 240

Carbons and Nanocarbons ......................................................................

291

4.1 4.2

291 294 294 295

Diamond and Nanodiamonds ............................................................ Graphene, Ordered Graphite, and Nanographites ............................. 4.2.1 Graphene................................................................................ 4.2.1.1 In-Plane σ and Out-of-Plane π Bonding................. 4.2.1.2 Van der Walls Interplanar and Intermolecular Interactions ............................................................ 4.2.1.3 Physisorption of Hydrogen on Carbons ................. 4.2.1.4 Chemisorption of Hydrogen on Carbons ................ 4.2.2 Graphitic Nanofibers, Whiskers, and Polyhedral Crystals .... 4.2.3 Graphite ................................................................................. 4.3 Disordered and Active Carbons......................................................... 4.3.1 Disordered Graphites and Mechanically-Activated Carbons ..................................................................................

177 179 181 182 183 183 183

253 255 265 270 281

296 297 298 299 299 301 301

x

Contents

4.3.2 Active Carbons and Chemically Activated Carbons ............. 4.3.3 Amorphous Carbon ............................................................... 4.4 Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns...................................................................... 4.4.1 Fullerenes and Hydrofullerenes ............................................. 4.4.2 Carbon Nanotubes ................................................................. 4.4.3 Carbon Nanohorns ................................................................. 4.4.4 Nanostructured Carbon Shells and Carbon Onions ............... References ..................................................................................................

305 305 308 312 314 317

Summary...................................................................................................

321

5.1 5.2 5.3

Metal/Intermetallic Hydrides ............................................................ Complex Hydrides ............................................................................. Nanocarbons and Others ...................................................................

322 323 324

Index ................................................................................................................

327

5

303 304

Chapter 1

Introduction

1.1

Motivation: The Hydrogen Economy

The energy supply to the humankind in the last two centuries was solely based on fossil fuels such as coal in the nineteenth century and crude oil and natural gas in the twentieth century. Unfortunately, this fossil fuel-based economy has led to a number of new challenges facing all of us in the twenty-first century such as global warming and following climate changes due to the release of growing amounts of greenhouse gas CO2, poor urban air quality, and reduction in the world crude oil supply, which could reach so-called Hubbert’s Peak around year 2011–2020. It is also noted that no major oil field has been discovered since 1970 [1]. Since the mid-1970s a concept of the ecologically clean Hydrogen Economy has been gaining momentum as essentially the only viable remedy for the growing world energy problems. The Hydrogen Economy offers a potential solution to satisfying the global energy requirements while reducing (and eventually eliminating) carbon dioxide and other greenhouse gas emissions and improving energy security. Hydrogen is a very attractive alternative energy vector. It is ubiquitous, clean, efficient, and can be produced directly from sunlight and water by biological organisms and using semiconductor-based systems similar to photovoltaics. Hydrogen can also be produced indirectly via thermal processing of biomass or fossil fuels where the development of advanced technological processes combined with a CO2 sequestration [2] have the potential to produce essentially unlimited quantities of hydrogen in a sustainable manner. For example, electricity produced by wind turbines or nuclear power plants during off-peak periods [3] can be used for the electrolysis of water into hydrogen [4] and the latter stored for future distribution to places of use [1]. When hydrogen burns, it releases energy as heat and produces water: 2H2 + O2→2H2O. Since no carbon is involved then using hydrogen produced from renewable or nuclear energy as energy resource would eliminate carbon monoxide and CO2 emissions and reduce greenhouse warming. However, if air is used for flame combustion of hydrogen, small amounts of NOx may be produced. In a fuel cell, hydrogen is converted directly to electricity in a similar reaction to the earlier one for burning hydrogen, and in essence just electricity and water are produced. Pure hydrogen enters the anode channel in a fuel cell and diffuses through a porous R.A. Varin et al., Nanomaterials for Solid State Hydrogen Storage, DOI: 10.1007/978-0-387-77712-2_1 © Springer Science + Business Media, LLC 2009

1

2

1

Introduction

anode toward the catalyst (Pt) where the hydrogen molecules H2 are stripped of their electrons and become positively charged ions (protons) (H2 → 2H+ + 2e−). Protons then migrate through the proton-permeable polymeric (Nafion) membrane (Proton Exchange Membrane also called Polymer Electrolyte Membrane – PEM) and the electrons generated during oxidation pass through the external circuit to the cathode, thereby creating electric current. On the cathode side, humidified air enters the cathode channel and diffuses toward the cathode-side catalyst layer. At the catalyst Pt surface, hydrogen protons recombine with electrons and oxygen molecules in air to produce water and heat according to the overall reaction ½O2 + 2H+ + 2e− → H2O + heat. This waste heat gives a PEM fuel cell an operating temperature of 60–80°C [5]. And last but not least is the fact that a hydrogen fuel-cell car can convert hydrogen energy into motion about 2–3 times as efficiently as a normal car converts gasoline energy into motion: depending on how it is designed and run, a good fuel-cell system is about 50–70% efficient in converting hydrogen to electricity, while a typical car engine’s efficiency from gasoline to output shaft averages to only about 15–17% [6]. Scott [7] succinctly summarized the inevitability of hydrogen becoming the fuel of the future by two rationales: depletion-based rationale and climate-based rationale. Driven by depletion, civilization must move from fossil fuels to sustainable energy sources. Realistically, the only way sustainable sources can be harvested to make chemical fuels is via hydrogen. Otherwise, how else can we get energy from wind, solar, or nuclear power to fuel an airplane? On the other hand, atmospheric CO2 growth is such that the concentration of CO2 in the atmosphere has increased from 280 to 370 ppm over the past 150 years. CO2 emissions can only be slowed by the extensive use of hydrogen and can only be stopped with the supremacy of sustainable-derived H2 among chemical fuels. The realization of the enormous benefits of the Hydrogen Economy has triggered over the last 15 years intense activities in the development of hydrogenrelated technologies. There are three major technological obstacles to the full implementation of the Hydrogen Economy in the next few decades. The first is the cost of safe and efficient production of hydrogen gas. At present, 48% of hydrogen is produced from methane steam reforming, 30% from oil/naphtha reforming, 18% from coal gasification, and only 3.9% from the electrolysis of water [8]. Apparently, the bulk of hydrogen production still relies upon fossil fuels, a by-product of which is CO2. However, the fossil fuel-based processes are much cheaper than the electrolysis of water. The efforts are underway to reduce the price of electrolysisderived hydrogen to $2–3 per kg. The second obstacle is the further development of PEM fuel cell (PEMFC), which is the primary cell most suitable for transportation. Most importantly, the research is focused on the extension of the usable service life, water flooding, dynamics, and reliability. The present cost of energy derived from a PEMFC is around $200 per kW and this must be reduced to around $30 per kW before PEMFC could be fully commercialized. The third obstacle is hydrogen storage for supplying PEMFC. There are three major competing technologies for hydrogen storage: compressed gas cylinders, liquid hydrogen tanks, and metal hydrides [9, 10]. Their comparison is shown in Table 1.1. A major drawback

1.1

Motivation: The Hydrogen Economy

3

Table 1.1 Comparison of three major competing technologies for hydrogen storage (based on [9, 10])

Storage system

Volumetric hydrogen capacity Drawbacks (kgH2 m−3)

Compressed hydrogen gas under 80 MPa pressure

~40

Liquid hydrogen at cryogenic tank at −252°C (21 K) Solid state hydrides

~71 80–160

Safety problem since enormous pressures are required; cost of pressurization; large pressure drop during use; hydrogen embrittlement of storage tanks Large thermal losses (open system); safety; cost of liquefaction None of the above

of compressed hydrogen storage for transportation applications is the small amount of hydrogen that may be stored in a reasonable volume (volumetric capacity/density). As can be seen in Table 1.1 compressed hydrogen gas technologies, even at such enormous pressures as ~80 MPa, also suffer from low volumetric densities not exceeding ~40 kgH2 m−3. As pointed out by Sandi [10] even at such a high pressure as 70–80 MPa, the energy content of compressed hydrogen is significantly less than that for the same volume of gasoline: 4.4 MJ L−1 (at 70 MPa) for hydrogen compared with 31.6 MJ L−1 for gasoline. Even though considered to be quite simple and inexpensive, the high pressure of 80 MPa involved in hydrogen gas cylinders raises safety concern. There is also some cost involved with compression to such high pressures. Another consideration is the large pressure drop during use. In addition, because most of the system parts exposed to hydrogen will be metallic, there is a concern of well-known hydrogen embrittlement [10]. The liquid hydrogen tank for its part offers almost twice as high storage capacity by volume as pressurized hydrogen; however, this is still less than half that required by the Department of Energy (D.O.E. FreedomCAR goal) (the requirements will be discussed later). A major drawback of liquid storage is a big cost of liquefaction, which today can add as much as 50% to the cost of H2 [6, 9, 10]. There are also safety issues associated with the handling of cryogenic liquids and the problem of evaporative loss. Solid-state hydrides that include metal/intermetallic and complex (chemical) hydrides are characterized by the highest volumetric capacities, and they do not suffer drawbacks as those experienced by compressed and liquid hydrogen. Because of the low pressures involved in metal hydride technologies and the fact that the release of hydrogen takes place via an endothermic process, this method of hydrogen storage is, however, the safest of all. Moreover, the hydrogen released from a metal hydride is of very high purity and, therefore, can be used directly to feed a PEM fuel cell. The U.S. Department of Energy (DOE) introduced a number of targets for onboard hydrogen storage systems within the frame work of its FreedomCAR program for the years 2007, 2010, and 2015 [11, 12], which are listed in Table 1.2. It is now appropriate to discuss solid state hydrogen storage in hydrides in the context of the targets shown in Table 1.2.

4

1 Introduction

Table 1.2 US DOE FreedomCAR hydrogen storage system targets [11, 12] Targeted factor Specific energy (MJ kg ) System gravimetric capacity (wt%) System volumetric capacity (kgH2 m−3) Energy density (MJ L−1) Storage system cost ($ per kgH2) System cost ($ per kg per system) Operating temperature (°C) Min/max delivery temperature (°C) Cycle life-time (absorption/desorption cycles) Flow rate (full throttle) (g s−1) Delivery pressure from tank to FC (bar) Transient response(s) (10–90% and 90–0%) Refueling rate (kg H2 min−1) −1

2007

2010

2015

– 4.5 36 – 200 – −20/50 −30/85 500 3 2.5 30 0.5

7.2 6 45 5.4 133 6 −30/50 −40/85 1,000 4 2.5 15 1.5

10.8 9 81 9.72 67 3 −40/60 −40/85 1,500 5 2.5 15 2.0

Table 1.3 Hydrogen storage properties of intermetallic compounds (from [12]) Maximum hydrogen capacity Type

Intermetallic

Hydride

wt%

Temperature for 1 atm Pdesorption (°C)

A 2B AB AB AB2 AB5/MmB5 AB2

Mg2Ni FeTi ZrNi ZrMn2 LaNi5 TiV2 (TiV0.62Mn1 5)

Mg2NiH4 FeTiH2 ZrNiH3 ZrMn2H3.6 LaNi5H6 TiV0.62Mn1 5H2 5

3.6 1.86 1.85 1.77 1.49 2.15

255 −8 292 167 12 −6

Conventional metal hydrides based on metals such as V, Nb, Pd, Li, Na, etc. have gravimetric capacities too low for any commercial consideration in mobile hydrogen storage with the exception of LiH, which has high capacity but extremely high desorption temperature [13]. A notable exception of a metal hydride is Mg, which has a relatively high gravimetric capacity and can desorb at 300°C and slightly below after nanostructuring treatment (it will be discussed later). In essence, none of the metal hydrides can meet the DOE targets [13, 14]. Similarly, hydrides based on intermetallic compounds AB (FeTi, ZrNi), AB2 (ZrMn2/TiMn2/TiCr2), AB5 (LaNi5 or MmNi5 where Mm – mischmetal), and A2B (Mg2Ni) have relatively low gravimetric storage capacities as shown in Table 1.3, which are unsuitable for transportation storage although a number of them desorb hydrogen within the temperature range targeted by DOE at the desorption pressure (Pdesorption) of 1 atm (which is more or less an operating pressure of a PEM FC). However, there exist many complex hydrides having high and very high gravimetric storage capacities some of which are shown in Table 1.4 [13–22]. Their theoretical capacity is calculated as the ratio of the atomic mass of hydrogen in

1.1

Motivation: The Hydrogen Economy

5

Table 1.4 Hydrogen storage properties of selected high-capacity hydrides [13–22] Metal– Theoretical maxi- Theoretical revers- Approx. desorphydrogen ible gravimetric mum gravimetric tion temperature system Hydride H2 capacity) (wt%) capacity (wt%) range (°C) Li–B–H Mg–B–H Fe–B–H Ca–B–H Na–B–H Li–Al–H Al–H Mg–Al–H Li–N–H Zn–B–H Ca–Al–H Mg–H Na–Al–H Mg–N–H Mg–Fe–H Na–N–H

LiBH4 Mg(BH4)2 Fe(BH4)3 Ca(BH4)2 NaBH4 LiAlH4 AlH3 Mg(AlH4)2 LiNH2(+LiH + TiCl3) Zn(BH4)2 Ca(AlH4)2 MgH2 NaAlH4 Mg(NH2)2(+LiH) Mg2FeH6 NaNH2

18.4 14.9 12.1 11.6 10.6 10.6 10.0 9.3 8.8 8.5 7.9 7.6 7.5 7.2 5.5 5.3

~13.8 ~11.2 Unknown Unknown 10.6 ~7.9 10.0 ~7.0 ~6.0 8.5(?) ~5.9 7.6 5.6 ~7.0 5.5 Unknown

~470 ~300 Unknown ~320(?) 400–600 110–260 ~150 110–160 150–280 85–140 80–180 300–400 229–247 140–250 300–400 10 °C s−1). Now, let us come back to our other historical thread: the development of hydrided materials. Large body of results was coming in the 1960s from research on apparently nanometric structures observed in fine metal powders and deposits,

10

1

Introduction

like dispersed Fe, and cathodically charged Ni layers prepared in electrochemical cells [42]. Yet, the thermodynamic information extracted from such electrochemically charged metals concerned inherently nonequilibrium processes, and was considered doubtful and less of value. In fact, an important experiment on direct synthesis of nickel hydride, NiH, at 18 kbar of H2 gas in a high-pressure cell, carried out by Baranowski in Poland’s Institute of Physical Chemistry in Warsaw in 1966 [43] was driven by the desire to achieve true equilibrium conditions between H2 gas and solid nickel hydride phase. At that time a common approach in investigations of metal–hydrogen systems was to conduct studies on well-ordered polycrystalline metals or, if possible, single crystals. Obviously, ordered metals used in experiments let investigators to construct the most transparent and instructive models for a metal-hydrogen system. Although by the end of 1960s many metals were reported to form hydrides, those were either too stable, like ZrH2, or too unstable, as NiH to attract interest in respect to reversible hydrogen sorption.

1.2.2

Early Routes to Nanomaterials

There is a strong interplay between science and technology. New science often creates new technological opportunity, and reciprocally, new technology development often triggers new opportunity to advance science. Such interplay is well illustrated in the tread in which the science of new materials for hydrogen storage has always been strongly intertwined with the parallel development of innovative materials processing techniques and technologies. Initially, those were techniques of highpressure physics and chemistry, and instrumental techniques in electrochemistry. Later development was fueled by an emerging discipline of materials science. The advent of nanomaterials, as one of a few founding blocks of future nanotechnology, came from materials science laboratories around the world. Nanotechnology is a technology that owes its name to the prefix nano, a Greek word for dwarf, as applied to objects that exhibit billionth (10−9) meter dimensions. The first reports on unusual new metallic phases, where observed structures exhibit partitioning into nanosized grains, or are simply grainless, have been coming from groups in the USA and other countries worldwide. A new process of rapid quenching of metals, with the cooling rate of 106 K s−1, first applied by Pol Duvez and his colleagues [44] in the California Institute of Technology in Pasadena in 1960, produced unexpected new fine-grained and grainless, metastable alloy phases, the first nanometals and amorphous metals. Using a quench technique capable of cooling metal melts to ambient temperatures with such extraordinary cooling rates, through just spraying and splashing of milligrams of melt alloys on a chill surface [sic!], the process of nucleation and growth, as described by Avrami 20 years earlier, was kinetically bypassed to yield a configuration of frozen liquid or amorphous metal. Duvez et al. realized this possibility by reporting complete solid solubility in Cu–Ag and Ga–Sb–Ge systems and formation of new nonequilibrium phases in Ag–Ge and

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

11

Au–Si – these being the first nanocrystalline alloys and the first amorphous alloys. Cohen and Turnbull [45] were quick to point out that amorphous alloys can exhibit glass-to-crystal transition. Indeed, when processed by low-temperature annealing they formed first metal alloys with nanometric grains, hence, the first nanoalloys. The new science of amorphous and noncrystalline solids would not have developed so rapidly if Chen and Polk had not invented useful amorphous ferromagnetic materials and this technological breakthrough had not been vigorously pursued by the big US company, the Allied Chemical Corporation (later Allied Co). Indeed, in 1972, a new chapter in metallurgical technology has been opened, when a new rapid quenching process of melt spinning, viz., casting of a stream of molten metal on a copper drum, rotating with speeds ranging from 5 to 20 m s−1, was used by Chen and Polk that worked for the Allied Chemical Corporation in USA [46] to spin the first noncrystalline ferrous (and ferromagnetic) metal ribbon, Fe80B20. Further to this a new process to form alloy ribbons with structures ranging from nanometric (then called microcrystalline) to amorphous was invented by several researchers, among them Bedell, Narashiman, and Ray working for corporate research in companies in the USA. It came as a shock to many physicists that amorphous iron alloy can be ferromagnetic in spite of absolute lack of crystalline lattice (and indeed, absence of a theory of amorphous magnetism). With the invention of a new class of soft magnetic materials it became clear that applications could be found that will utilize the outstanding combination of physical and chemical properties as being the direct consequence to the lack of structural crystalline long-range order (LRO), and the presence of short-range order (SRO); the latter being limited to nanometric distances between atoms. Among those new materials were binary Fe–B, Fe–Zr, Cu–Zr, Ni–Zr, Pd–Si, Mg–Zn, and many ternary or even quaternary amorphous systems. Many patents were filed in the USA and Japan, and by many groups worldwide, on amorphous alloys and intermetallics that were termed metallic glasses or glassy metals [47] and had been manufactured by a new rapid solidification technology. This technology became intensively developed by MIT’s Center for Materials Science and Engineering in the USA [48] and Japan’s Tohoku University, among many other materials research laboratories worldwide. From the historical point of view it is important to realize that many if not all of these amorphous alloys, i.e., metallic glasses, when carefully annealed at low temperatures, change to devitrified, nanostructured alloy phases. Those were the first nanomaterials engineered on purpose. They were produced in form of thin ribbons and wires via rapid solidification processing of melt alloys. They allowed controlled exploration of chemical, mechanical, magnetic, and other properties as arising from nanostructures. Shortly, the most prominent benefits of new nanostructures were achieved in the outstanding improvement in the energy product of new neodymium supermagnets produced by crystallization of amorphous Fe–Nd–B alloys (as reviewed in [49]). Shortly after, among research stimulated by the outstanding combination of chemical and magnetic properties in metallic glasses were the first papers on the action of hydrogen on amorphous intermetallics and magnetic metallic glasses. Absorption and diffusion of hydrogen in amorphous alloys was studied in UK by Harris [50] and by Cantor with his colleagues [51], in Germany by

12

1

Introduction

Kirchheim et al. [52, 53], and in the USA by Bowman [54], among few others. In Canada, Wronski et al. [55] studied hydrogenation of F90Zr10 amorphous intermetallic via cathodic polarization, and concluded that hydrogen can be reversibly released by low-temperature annealing at less than 100°C. The hydrogenation was accompanied by a concurrent drastic change of magnetic properties, from the paramagnetic nonhydrogenated phase to the ferromagnetic hydrogenated amorphous solid solution; during those transformations, the F90Zr10 remained amorphous in its bulk, yet Fe magnetic nanocrystals precipitated on the surface layer of the hydrogenated amorphous F90Zr10 [56]. In the USA, Johnson and colleagues in Caltech observed quite the opposite: the first example of a hydrogen-induced crystal-toglass transformation in metallic alloys resulting in the formation of amorphous hydrides [57]: Zr0.75 Rh 0.25 + H 2 → am Zr0.75 Rh 0.25 H1.14

(1.3)

Reaction (1.3) proceeded only at a relatively low temperature just above 150°C and although amorphous alloy exhibited large H/M (hydrogen-to-metal) atom ratio, the charging and discharging of amorphous phase proceeded in a range of pressures, while a crystalline alloy could have been cycled at almost constant pressure [58]. Bowman [59] explained that the strong exothermic reactions between hydrogen and most amorphous intermetallics release excess energy, which causes crystallization of amorphous phase. He also reported that many metallic glasses tend to absorb less hydrogen then their nanocrystalline counterparts; he explained this behavior by the random order of atoms in amorphous phase, which might restrict the number of interstitial sites that are favorable for hydrogen occupancy. While interest in amorphous alloys was becoming vane over the time, the paradigm shift in materials science brought about by the discovery of amorphous metals and rapid solidification cannot be underestimated. The interest of physicists, chemists, materials scientists, and engineers became refocused from well-ordered crystalline materials to disordered and nanocrystalline phases. Shortly, substantial R&D programs were initiated in the USA, Japan, and worldwide both at universities and national government laboratories. Foundations have been established for the advance of the science of nanomaterials. However, the transition from nanoscience to nanotechnology had to come from yet a concurrent innovation in tools used by scientists. This was the invention of the first scanning tunneling microscope (STM) in 1981 [60], followed by the invention of the atomic force microscope (ATM) in 1986 [61]. While STM and ATM microscopes provided powerful tools to advance new science of solids at nanometric region materials scientists were also greatly inspired by papers coming from Gleiter and his group at Saarbrücken University in Germany. Those were the papers that redefined nanocrystalline materials as such where the fraction of atoms located at grain boundaries of polycrystalline material is comparable to the fraction of atoms at the core of grains when the grain size reaches nanometers [62]. A claim was put forward that such materials are to be fundamentally different from, and often superior to, those of the conventional

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

13

polycrystals and the amorphous solids. Gleiter observed that nanometer-sized crystalline materials being polycrystals with very small crystallite sizes of about 2–10 nm in diameter are composed of randomly oriented high-angle grain boundaries. The first such nanocrystalline phases produced on purpose came from Saarbrücken about 1984 by evaporation of the material in a high-purity inert gas atmosphere followed by condensation and compaction in ultrahigh vacuum [63]. By this time chemists working in the field of metal catalysts were well aware of importance of nanometric features, such as edge sites and surface clusters on chemical catalysis. Molecular orbital calculations performed 10 years earlier predicted that the ionization potential of nanostructured metals should increase as their grain size decreases [64]. The high surface-to-volume ratio observed in nanostructured materials not only appeared important to the number of active sites in a catalyst, but also seemed to influence the defect chemistry (like oxygen and other anion defects). Siegel [65] computed that the percentage of metal atoms on the surface of grain increases from a few percent in a 100-nm particle to about 90% in a 1-nm crystallite. These simple experiments by Gleiter and equally simple calculation by Siegel were eye opening for others in research community in material science. Immediately, a new consensus reached among materials scientists was that extending our understanding of structure–property relationship in solid materials down to the nanometer regime should be attractive for development of engineering materials with an outstanding combination of properties or novel properties. Among materials that became studied were nanophases produced by mechanical alloying [66] and nanohydrides where the size effect determines hydrogen storage properties [67].

1.2.3

Historical Development of Classical Hydrogen Storage AB5 Alloys

Development of first practical hydrogen storage alloys, AB5-type intermetallics, had its beginning in a typical accidental laboratory experimentation, although it was predetermined by a vigorous development of a new discipline of materials science and engineering in late 1960s. The outstanding hydrogen sorption properties of rare-earth AB5 intermetallics were accidentally discovered in Philips Laboratories in Eindhoven, Netherlands at about 1969 in a program to develop a new permanent magnet alloy [68]. In the work on Sm–Co magnet alloy it was observed that a loss of coercivity occurs, when magnets were aged in humid air, which was related to two concurrent reactions: corrosion of SmCo5 intermetallic with release of hydrogen, and sorption of the released hydrogen by still uncorroded alloy: 2SmCo 5 + 3H 2 O → Sm 2 O3 + 10Co + 3H 2

(1.4)

SmCo 5 + 1.25H 2 ↔ SmCo5 H 2.5

(1.5)

14

1

Introduction

Zijlstra et al. [69] demonstrated that the reaction of (1.5) is reversible at several bars of pressure at ambient temperature. In the following studies on the origin of magnetic coercivity in these new magnets, Phillips laboratories came with the hydride of LaNi5, lanthanum–nickel hydride, which reversibly bonded more than six atoms of H per one formula unit (H/M ratio > 1) [70]. LaNi 5 + 3.35H 2 ↔ LaNi 5 H 6.7

(1.6)

Reversibility of hydrogen charging was very good at room temperature. This is because the hexagonal lattice of the metal host, like this of LaNi5, does not undergo major transformation as hydrogen is inserted interstitially. This was the first AB5type interstitial hydride in which hydrogen is stored between metal atoms. Those rare-earth AB5-type hydrides were quickly utilized in rechargeable nickel metal hydride batteries where electrochemical hydrogen charging and discharging take place at ambient temperature. Such electrochemical hydrogen storage is reversible, when the negative hydride electrode (anode) is combined with the positive Ni electrode (cathode) in the battery cell.. AB5 H x + xOH - ↔ AB5 + xH 2 O + xe -

(1.7)

NiOOH + H 2 O + e - ↔ Ni (OH )2 + OH -

(1.8)

As a matter of fact, the first hydrides with practical hydrogen storage capacities were realized in rechargeable nickel metal hydride batteries. For more information on electrochemical hydrogen storage in rechargeable batteries a reader can be referred to several recent reviews on this subject [71–73]. Renewed interest in solid-state hydrogen storage came with first Oil Crisis in 1970s. By early 1973 a patent was granted to US Brookhaven National Laboratory, where Reilly had produced first MmNi5 alloy for hydrogen storage [74]. MmNi5, where Mm – mischmetal – is a unrefined (cerium free) mixture of rare-earth metals, mostly La and Nd, obtained from Bastanite from California (also found in large deposits in Inner Mongolia, China), was much less expensive than La and has already been used as a deoxidizer in steel industry. Soon after, the potential for energy-related application for nickel and nickel alloys attracted the major Ni metal and alloy producer in Canada and the USA, INCO. Garry Sandrock, then with INCO, achieved optimization of MmNi5 compositions for gas hydrogen storage through careful partial substitutions for A and B in AB5 formula [75]. The compositional changes have reduced the raw materials costs to about 30% of that for the LaNi5 alloy. LaNi5 has an attractive value of the equilibrium hydrogen desorption pressure dropping in the range between 1 and 2 atm at 25°C. In the MmNi5 this pressure rises to a much less attractive 30 atm, which can imply difficult engineering design for gas hydrogen storage, and was completely impractical in MmNi5 used for the anode in a NiMH cell; obviously, battery cells operate at near-atmospheric pressure. The substitutions of MmNi5 with Ca for Mm and Al for Ni were quite effective in lowering the equilibrium pressure of hydrogen back to about 1–2 atm at

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

15

25°C, which was desired for easy hydrogen charging and discharging at ambient conditions. Approaches along the same line came in the same time from Japan [76]. The alloying of polycrystalline AB5 alloy was an excellent (and very practical) result of new materials design involving adjustment of the strength of chemical bonds between metallic elements in AB5 hexagonal lattice and its interstitial hydrogen atoms. The bond strength in the metallic host lattice was further optimized with cosubstitutions, which was effective in reducing the volume change on hydriding of the hexagonal AB5 unit [77]; the density of material can, indeed, be related to the bond lengths through an approximate formula: Ro = 0.0115Aav/a2c, where Aav is an average atomic weight for the AB5 formula unit, and a and c are the hexagonal lattice parameters in nanometers [78]. The lower dilatation of the AB5 cage on placing hydrogen in it was equivalent to better cohesiveness of atomic structure in this intermetallic, hence improved resistance to alloy decrepitation on repetitive (cyclic) charging and discharging, hence better corrosion resistance of MmNi5 alloys in an aggressive caustic battery electrolyte. Although substitution with Al came at the price of the lowered hydrogen capacity, another Mn substitution ameliorated this problem [79]. With compliance with the well-observed 10–15-year shift from new material development to its full commercialization, the Al and Mn substitutions have been combined in multielement MmNi(Co, Mn, Al)5 battery alloy and optimized for the use in anodes in NiMH commercial batteries. According to recent review a typical commercial AB5 alloy can consist of many elements, such as in hydroalloy F from GfE Metallen und Materialen GmbH: La0.64Nd0.36Ni0.95Cr0.19Mn0.41Co0.15 (in wt%) [71]. Again, one can observe how the development of materials for gas hydrogen storage for fuel cells has been intertwined with a parallel development of alloys for electrochemical hydrogen storage in rechargeable battery cells. AB5 alloys used in either hydride battery cell or in a hydrogen storage tank for fuel cell are not per se nanocrystalline but the microstructural design process, which led to their development in 1980s and 1990s, reached nanoscale phase partitioning.

1.2.4

Historical Development of Interstitial Hydrides in Other Intermetallic Systems

The first systematic research into gas hydrogen storage for practical applications began in the early 1970s at the Brookhaven National Laboratory (BNL) on New York’s Long Island. The investigations were conducted in a program that explored possibility of storing electric energy produced in off-peak hours by nuclear power stations through generation of hydrogen, and then the storage of hydrogen in metals instead of storing electric current. So, at the same time as the AB5-group hydrides were intensively investigated in the BNL, new AB-type hydrides came into view. Iron–Titanium, FeTi, hydride was discovered in 1974 by Reilly and Wiswall [80] in the same Brookhaven National Laboratory. This metal hydride formed from a 50–50 atom ratio of titanium and iron, the lowest cost and the most convenient

16

1

Introduction

hydride, offered a reversible capacity of almost 1.5 wt% H2 operating at ambient temperatures. A desirable feature of this new hydride was that the hydriding– dehydriding reaction occurs near ambient temperature and during the charging pressure ranges between 500 and 15 psi. The prototype 400-kg FeTi hydride storage unit was built by a New Jersey’s electric utility. The charge–discharge reactions were observed to be very rapid but strongly limited by the rate of heat transfer. FeTi material was not losing hydrogen storage capacity even after thousands of cycles when high-purity hydrogen was used. Yet, the alloy was very sensitive to O2 and CO poisons, and even to traces of H2O. Obviously, while meeting the expectations of stationary storage of nuclear energy in the form of solid hydride it was too heavy for storage of hydrogen for transportation energy. Anyway, the needs for cars propelled by hydrogen fuel were not in place at the end of the 1970s.

1.2.5

Historical Development of Nanophase AB2 Intermetallic Hydrides

A new venue for hydride research has been evolving from the study of multielement Ti–Ni alloy thin films for hydrogen oxidation in fuel cells conducted by Stanford Ovshinsky in the USA in the early 1980s. The new disordered multielement, multiphase AB2-type alloys for hydrogen storage exhibited a large degree of structural disorder and submicron partitioning of phases. Ovshinky has been pioneering a view [81] that structural disorder at nanoscale results in outstanding combination of hydrogen storage in crystalline grains, while the disordered grain-boundary phase provides channels for hydrogen delivery and release from nanocrystalline grains. A first, quintessential tool of the coming nanotechnology, a scanning tunneling microscope, was then used to reveal that in the TiVZrCrNi multiphase alloy the hydrogen is preferentially sorbed in the Ti-rich phase, while V-rich phase was likely to corrode preferentially to expose Ni nanoparticles [82]. This shift of view was congruent with enormous intellectual potential, which was brought about by research on amorphous and rapidly quenched phases (amorphous selenium, chalcogenide glasses, and metallic glasses) in the 1960s and 1970s. US Ovonics patents came just in time when interest has been growing in Europe, Japan, and North America in chemical properties of amorphous metals and metallic glasses prepared by rapid solidification technology. The materials challenge was to optimize hydrogen capacity and corrosion properties in multiphase nanostructures for the anode in nickel metal hydride (NiMH) cell. Shortly, the outcome of international race began toward development of a second generation of multiphase and multielement NiVZrTiCr hydrides. The nanostructured battery materials were developed applying new principles of materials science established in previous decade and reflecting on the role of disorder and nonequilibrium processing in overcoming the conflicting requirements for materials properties. In the early 1990s sealed nickel metal hydride batteries became available in Japan and elsewhere for consumer goods. Shortly afterward, the new nanopowders, which were proven in electrochemical hydrogen storage, began to be investigated for gas hydrogen storage. This came when new

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

17

drivers emerged for growth of hydrogen energy in the 1990s. The concern now was for clean environment, and mitigation of greenhouse gases, and appeared in the context of evidence for accelerating global climate change.

1.2.6

New Routes to Nanomaterials: Mechanical Alloying and Mechanochemical Activation

Like the discovery of nonequilibrium quenching of metal melts in the 1960s and 1970s led to the development of disordered multiphase AB2 hydrides in the 1980s, an interest in nonequilibrium milling of metal powders, in the 1980s, has been setting foundations for the development of the nanohydrides of the following decade. The nonequilibrium synthesis of new metal alloys by high-energy ball milling of powders was developed almost 20 years earlier by John Benjamin and his coworkers at the International Nickel Company (INCO) [83]. The development came from research on new alloys intended for gas turbines, and was an outcome of an effort to produce nickel superalloys, which were mechanically hardened through an inoculation with fine refractory metal oxides, Al2O3, Y2O3, and ThO2, in so-called oxide dispersion-strengthened (ODS) process. In his research Benjamin pointed out that mechanical alloying (MA), a new process induced by intensive milling of metal and compound powders, could make new materials with unique properties [84]. (In a historical note like this, the credit for coining the term mechanical alloying must be given to a Patent Attorney for INCO company.) However, it was not until the early 1980s that the discovery by Koch and coworkers [85] of amorphous alloys made by MA renewed interest in intermetallic compounds, which were difficult to synthesize by orthodox metallurgy. Those authors demonstrated that, during mechanical alloying of elemental metal powders of nickel and niobium in a special high-energy ball mill (SPEX™ model), the final product becomes an amorphous alloy. This was demonstrated by the complete disappearance of the sharp Bragg reflections in X-ray diffraction pattern, which are characteristic of any crystalline phase, and appearance of a broad maximum, characteristic of amorphous alloys; the latter was previously seen in metals that were rapidly quenched from melt. The amorphous structure was confirmed in the exothermic glass-to-crystal transformation peak revealed by differential scanning calorimetry (DSC). Two years before the publication by Koch, Yermakov et al. [86] reported on the amorphization of a number of Y–Co intermetallic compounds by grinding of ingots, which were prealloyed by casting. This process was later termed mechanical milling (MM) process. Further early investigations were carried out by Schwarz et al. in the US Los Alamos National Laboratories [87], as well as by groups in Germany and the Netherlands. Already, in these early stages of new research, it was well recognized that by ball milling the average size of grains in powder particulates was reduced to nanometric dimensions. Either MA or MM processes drop under more general class of solid-state amorphization reactions, SSAR. Amorphization by irradiation of solids was observed yet in the era of study of materials for nuclear reactors. In 1962, Bloch [88] amorphized U6Fe by exposing it to fluxes of nuclear fission fragments. Others observed amorphization

18

1

Introduction

of intermetallic compounds by high-voltage electrons [89]. Schwarz and Johnson [90] demonstrated the phenomenon and a new process of amorphization by diffusion between thin-film sandwiches of crystalline La and Au. In the late 1970s, Johnson and coworkers [57] in Caltech demonstrated that in addition to techniques of amorphization by rapid solidification of molten alloys and amorphization by mechanical alloying of solid alloys, the process can be conducted by repetitive exposure of an alloy to hydrogen pressure and vacuum. This provided a new solid-state processing route to amorphization by hydrogenation. To this class also belong reactions of hydrogen decrepitation of metals and alloys, the method proposed by Harris [50] for preparation of fine powders for subsequent sintering into new generation of neodymium supermagnets. Through cycling of a cast metal ingot of Fe–Nd–B in hydrogen the ingot was effectively changed into fine powder; however, nanometric hydride phases that must have formed concurrently to the decrepitation were not considered for hydrogen storage. So it took another thread in materials chemistry history tracks to bring mechanical processing for advancing research on hydrogen sorption in solid media. This thread can be attributed to studies of effects of mechanical grinding on activation of inorganic solids for chemical reactions. As a matter of fact, the approach to apply mechanical energy, in form of grinding and milling, as to activate or bring about new paths for chemical reactions was much older. In a historical review on mechanochemistry, as a branch of solid state chemistry dealing with mechanically activated chemical reactions, pioneering role is attributed to photochemist M. Carey-Lea who as early as 1892 [91] began publishing reports of systematic studies on decomposition of silver and mercury halides, then dozens of other stable compounds, under applied mechanical pressure, using so simple a method as grinding compounds in a mortar. Extracting mercury from its sulfide while rubbing it in a copper mortar was known yet in ancient Greece [92]. However, in modern times it was a large body of work on mechanochemical activation of inorganic compounds that was coming from laboratories in Russia and the USA that demonstrated the role of defects in activation of materials for chemical reactions conducted in a solid state or at the interface of solids and gases. Researchers from the US Naval Civil Engineering Laboratory engineered mechanical alloys as composites where the second phase of Fe (or Ni, Ti, or Cu) has been dispersed at nanometric manner in the Mg phase; these alloys were intended as supercorroding alloys to react rapidly with seawater to produce heat and hydrogen gas in underwater welding of naval vessels [93]. Shortly after, in years 1984–1997, Ivanov et al. [94] from Russia were the first to explore mechanical alloying to obtain magnesium alloys for hydrogen storage.

1.2.7

Historical Development of Lightweight Metal Hydrides and Hydride Complexes

Formation of magnesium and calcium hydride compounds was first recognized by German chemists yet at the end of nineteenth century [95] but it took more than a half a century before a substantial yield of Mg hydride was obtained from direct

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

19

synthesis of the elements and the first determination of values for the decomposition pressure on temperature, the enthalpy of formation, and the activation energy were obtained [96]. Since then, Mg metal with its hexagonal lattice was well researched for hydrogen storage. MgH 2 → Mg + H 2

(1.9)

Magnesium dihydride can store several times as much hydrogen per unit weight as AB-type TiFe hydride. However, the first and the most limiting problem with lightweight Mg alloys for hydrogen storage was always the very poor kinetics of hydrogenation. Indeed, metallic magnesium does not react with hydrogen at low temperatures, and reacts only very slowly at high temperatures. Because of the relative inertness with respect to hydrogen, Mg was even recommended as a structural metal suitable for contact with hydrogen up to the melting temperature [sic!]. Not surprisingly, many controversial and poor reproducing data have been published on the hydrogenation/dehydrogenation kinetics. The natural inertness of Mg to molecular hydrogen can be attributed to well-known factors: (1) thick surface oxides that cover magnesium metal do not provide sites catalytically active for dissociation of hydrogen molecule, which is known to occur on contact with a metallic surface (clean metallic surface often provides active sites for catalytic breaking of very strong H–H bond), (2) surface-adsorbed impurity gases further slow the kinetics of hydrogen absorption into the bulk grains (the process that may be related to the hydrolysis of Mg with water vapor and causing even more thick oxide/hydroxide surface barriers; Sect. 2.1.5). On the other hand hydrogen dissolved in magnesium metal is readily removed by heating it in vacuum up to 300–400°C. The stable Mg dihydride was prepared as early as 1951 by direct heating of Mg in gaseous hydrogen under high pressures; however, the reaction was slow and difficult to complete [97]. While the chemical reactivity of Mg metal to hydrogen is poor it was early realized in works by Dymova et al. [98] that its dihydride can be a strong chemical reductor, and its increased reactivity depends strongly on small grain size. So the effect of grain size was pointed on as being one of high importance. Mg2Ni alloys and magnesium–nickel A2B-type complex hydrides, Mg2NiH4 were well known, being first developed in the US Brookhaven National Research Laboratory [80], and studied in Nordic countries [99], where the interest in Mg has been stimulated by availability of hydropower required to produce Mg metal in the first place. (Even earlier, Pebler studied A2B-type hydride of Zr2Ni [100].) The new Mg2Ni alloy alleviated the problem inherent to Mg metal: very poor kinetics of hydrogenation reaction. Reilly and Wiswall [80] noticed that by the presence of intermetallic magnesium compounds, Mg2Ni and Mg2Cu, on the Mg surface this reaction can be greatly accelerated. The hydrogen release can proceed without major crystallographic transformation as the hydride is converted back to the Mg2Ni: Mg 2 NiH 4 ↔ Mg 2 Ni + 2H 2

(1.10)

20

1

Introduction

The next alkali earth in the periodic table is Ca and its ternary hydride was found to be less stable than the Mg ternary hydride. Indeed, the desired reaction for CaNi5 is: CaNi 5 + 3H 2 ↔ CaNi 5 H 6

(1.11)

and this reaction was observed to work repeatedly and reversibly at near room temperature. Yet, the following disproportionation reaction is a preferred reaction path as long as the temperature is slightly increased (to 200°C) and the diffusion of the metal atoms becomes significant in comparison to the diffusion of smaller H atoms. The reaction product observed in the disproportionation was the mixture of more stable dihydride and metallic nickel: CaNi 5 + H 2 → CaH 2 + 5Ni

(1.12)

Although in later studies most of the intermetallic compounds were found to be thermodynamically unstable and falling into the disproportionation reaction path easily their actual tendency to do so varied markedly from system to system, as later it was learnt from studies on transition metal ternary hydrides. Many of these hydrogen-storing hydrides became available by the end of 1970s, most of them under the trademark of commercial HY-STOR hydrogen storage alloys: 300 Series (Mg-base), 200 Series (Ni-base, including both CaNi5-type and MmNi5 type), and 100 Series (FeTi-based). The first and the most important alloy, for which the H storage properties clearly benefited from nanostructure and nanoprocessing, was Mg2Ni, an A2B-type intermetallic. Although yet in 1961 Dymova et al. [98] pointed to the small grain size as a primary factor in improving hydrogen sorption in Mg, and other Russians, Ivanov et al. [94], demonstrated the high-energy ball milling route to hydrogen storage in magnesium, it took another four decades and series of studies reported from the McGill University in Canada to bring again the benefits of fine grains to the attention of the community of researchers in hydrogen storage. The clear demonstration of the effect of nanostructuring came in works of Zaluskis’ [101]. A study of ball milled powders showed that when the smallest X-ray diffracting metal grains in milled particles reach the range 5–50 nm the kinetics of both absorption and desorption is improved by an order of magnitude. These were nanostructured hydrides or nanohydrides. It must, however, be pointed out that they completely ignored the effects of simultaneous decrease of particle size during milling (Sect. 2.1.3).

1.2.8

Early Studies of Noninterstitial Transition Metal Ternary Hydrides

With the beginning of the 1990s the interest in hydrides was somehow refocused. Firstly, interest evolved in other hydrides of Mg with transition metals, beyond already well-researched Mg2Ni stoichiometry. Mg and Fe do not alloy to form

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

21

Mg2Fe binary intermetallic by either ingot casting or metal powder sintering procedures; however, when sintering is done in hydrogen, a very stable Mg2FeH6, noninterstitial, ternary magnesium–iron hydride, forms [102]. The Fe and Co transition metal hydrides (in reality they are complex hydrides) with Mg were firstly synthesized in the University of Geneva in 1991 in the same group that observed formation of this hydride during sintering of powders. While presenting a formal derivative of A2B stoichiometry, the hydrides were, in essence, Fe–metal and Co–metal hydride anionic complexes [103]. The new anionic complexes of Fe and Co were stable rendering the desorption temperature too high. When desorbed, the hydrides could not change to the hydrogen-free A2B alloys because Mg2Fe and Mg2Co do not exist in the binary Mg–Fe and Mg–Co systems (e.g., iron and magnesium are immiscible in the solid state and form a large miscibility gap in the liquid state). Indeed, the following reaction (1.13) is in striking contrast to reactions (1.2–1.4) and (1.5–1.6) where the dehydrogenated phase was converted back to stable host alloy (LaNi5 or Mg2Ni), and indeed similar to the disproportionation reaction (1.12) Mg2 FeH 6 → 2MgH 2 + Fe + H 2

(1.13)

One can notice that in (1.13) the Fe3+ in the (FeH6) complex is reduced to metallic Fe, the process that requires a major rearrangement of atoms. Therefore it is not surprising that the kinetics of desorption was poor while being controlled by diffusion of the metal atoms. Although more research had been coming, particularly from groups in the University of Geneva and in Stockholm, research on Mg2FeH6 and similar transitionmetal complex hydrides was put aside for some time. This, perhaps, was unjustified as some of those ternary hydrides formed with alkali earths and transition metal complexes, viz., BaReH6 [104], exhibited interesting storage properties: desorption below 100°C and 2.7 wt% H2 capacity at very high volumetric density of 134 g H2 L−1. Although too expensive for storage in cars, such hydrides would find a niche in fuel cell power sources in portable electronics, if only the desorption temperature is lowered. However, the needs were not in place, at the time of great success of hydrides in rechargeable NiMH. A stable aluminum hydride, AlH3, has been known to exist since some time. It has been prepared by various techniques including the direct reaction of metallic Al with atomic hydrogen by Siegel [105]. AlH3 is soluble in ether when freshly synthesized but polymerizes while aged and forms insoluble precipitate whose structure depends on the degree of polymerization of –(AlH3)– monomer.

1.2.9

Toward Chemical/Complex Hydrides

With the coming of a New Millennium, and in the dramatic context of the September 11, 2001 terrorist attacks other drivers for hydride research became evident: energy security as to secure uninterrupted supply of energy for still unsustainable world

22

1

Introduction

economies, and ever-growing needs for fuels for mass transportation. Also, the rapid economic growth in China and Asia put big pressure on markets for fossil fuels. USA Congress and President issued a “Great Challenge” in respect to development of hydrogen storage for future fuel cell cars. The interest has been renewed for storage of hydrogen in chemical media other than metal and alloy powders. So far, there are about 170 hydrides listed on the website hydpark.ca.sandia.gov. Of particular interest became complex hydrides with participation of alkaline metal ions and metal complexes and in mixed ionic and covalent coordination compounds. This class of hydrides has become known as chemical or complex hydrides. Among them are nontransition metal sodium and lithium alanates and borohydrides. The two most investigated are sodium aluminum hydride (sodium alanate), NaAlH4, and sodium borohydride, NaBH4. The former, alanate, was synthesized almost two decades earlier in Russia by Dymova et al. [106]. Those complex hydrides were not thought to be suitable for solid-state hydrogen storage media as they release hydrogen irreversibly at conditions that are outside the temperature– pressure range desirable for fuel cell propelled cars. In particular, their hydriding and dehydriding characteristics have been creating significant challenges. This opinion has been changed with a breakthrough work by Bogdanovic´ and Schwickardi [107], who discovered that the addition of Ti compound (chloride) to the complex hydride NaAlH4 causes reversible hydrogen storage in the range up to 3.7 wt% H2 under moderate conditions of temperature and pressure. However, the reaction has long been known to proceed in two steps [108] NaA1H 4 ↔ 1 / 3Na 3 A1H 6 + 2 / 3H 2 → NaH + Al + 3 / 2H 2

(1.14)

and again, like in the reaction of (1.13), the completely discharged phase did not form an alloy, NaAl, from the original metals; instead alanate is reduced to metallic Al, in fact nanoaluminum, that does not easily recharge with H2. In 1997 a study conducted by Ford Corp. (USA) compared the weights and volumes of various hydrogen storage media in the context of their use for a lightweight four- to five-passenger fuel-cell car with a 500-km range per one hydrogen charge [109]. Similar studies were solicited by other major car manufacturers, and government organizations in several G7 countries. The interest was raised by the theoretical electrical conversion efficiency for an ideal hydrogen–oxygen fuel cell being an impressive 83%, and up to 60% of the hydrogen chemical energy converted to electric energy (electric current). This compared very favorably with circa 45% achieved in hydrogen internal combustion engines. Since then, search for new reversible lightweight complex hydrides was heating up. Interest aroused in other complexes of hydrogen with Al and B that have been known since 1950s. Alkali metal and magnesium tetrahydroborides were synthesized [110, 111] and investigated by Germans and Russians for thermal decomposition from the late 1950s to the early 1970s. Lithium tetrahydroboride, LiBH4, which has already been known as a strong reducing agent in organic synthesis, has received renewed attention after Züttel [112] reported the onset of hydrogen desorption at approximately 200°C promoted by SiO2 admixed to this borohydrides.

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

23

Among the alanates, the sodium alanate and the lithium alanates have been well studied. Magnesium aluminum hydride (magnesium alanate), Mg(AlH4)2, with the total 9.3 wt% H2, has been known since its synthesis by Wiberg in the 1950s; however, there have been difficulties with good reaction yield to produce this hydride until more than 50 years later Fichtner and colleagues [113] produced a sufficiently pure compound for structural investigations. Again, the reversibility of the reaction was questionable as the experiments so far demonstrated that after discharge of hydrogen the alanate yields Al phase, not unlike sodium alanate in the reaction of (1.14). Nevertheless, the intensive research on complex hydrides has established itself solidly in place in early years of the new Millennium.

1.2.10

Historical Development of Nanocarbons and Carbon Nanotubes

Concurrent stream of the development of nanomaterials for solid-state hydrogen storage comes from century-old studies of porous materials for absorption of gasses, among them porous carbon phases, better known as activated carbon. Absorption of gases in those materials follows different principles from just discussed absorption in metals. Instead of chemisorption of H2 gas into the crystalline structure of metals, it undergoes physisorption on crystalline surfaces and in the porous structure formed by crystals. The gases have also been known to be physisorbed on fine carbon fibers. Reports on growth of fine filaments of carbon have been coming since 1890, when filamentous carbon was obtained by passing cyanogens over red-hot porcelain. Hollow carbon fibers were reported in a Russian journal in 1952 [114]. An interesting view point about the discovery of carbon-nanotube-like materials can be found in Guest Editorial to the journal Carbon [115]. Fine fibers showing graphitic structures were frequently encountered as unwanted phases in both steel industry and in industrial catalysis, where they deposited on walls of metallurgical furnaces and chemical reactors. Since then development of carbon filaments was driven by their use as reinforcement phase in lightweight composite materials for space and aerospace industries. Research focused on fabrication of carbon fibers from polymerbased precursors such as rayon, polyacrylonitrile (PAN), or mesophase pitch. In the 1970s Oberlin et al. [116] published images of what can be now called single- or double-wall carbon nanotubes (SWNT, DWNT), although at this time, images obtained from transmission electron microscopes cannot clearly reveal the number of concentric hollow tubes in the fiber. In the 1980s he produced hollow carbon filaments by chemical vapor deposition (CVD) in a floating catalyst reactor where they were nucleated and grown on 10-nm metal catalyst nanoparticles. Such CVD method to produce fine hollow filaments became with time the most versatile production method for carbon nanotubes (CNT). At the time, it was mechanical property that was of primary interest, as it was realized early that well bulk carbon fibers resist crack propagation. Therefore, little attention was paid toward the

24

1

Introduction

Oberlin’s small and hollow objects, which only latter were termed carbon nanotubes, and were becoming studied around the world for other than mechanical, mostly electronic, and later hydrogen storage properties. The situation changed diametrically in 1991 when Iijima [117] published a breakthrough paper entitled “Helical microtubules of graphitic carbon.” Iijima’s rediscovery of CNTs came during the research conducted to follow on a breakthrough research on another carbon nanophase, fullerene, and spherical C60 carbon atom configuration, which was discovered 6 years earlier by Kroto et al. [118] and published in a seminal paper on C-60 buckminsterfullerene. The latter molecule consisted of 60 carbons that satisfied a mathematical theorem by Leonhard Euler requiring a spherical surface, as to be entirely built up from pentagons and hexagons, the molecule must have exactly 12 pentagons like the geodesic domes of the architect R. Buckminster Fuller, and not unlike a soccer ball. Shortly after, the now clear CNT images were observed by powerful high-resolution transmission electron microscope (HRTEM) in soot produced by the just published Kratschmer-Huffman method on the laboratory scale production of C60 [119]. CNTs can be seen as single, one atom-thick, graphitic planes, viz. graphene planes, rolled to form nanoscale cylinders, which are often closed at the end by half a sphere of the earlier discovered fullerene. These structures, while comprising of only one single-wall cylinder were termed single-walled nanotubes, SWNTs, whereas those consisting of two or more concentric graphene cylinders became known as multiwalled nanotubes (MWNTs). Depending of how the graphene sheet is rolled up, the three forms of CNTs, zigzag, armchair, and chiral, change their electronic properties from metallic, to superconducting, to insulating. In 1992 it was shown that MWNTs can be produced in DC arc discharge in He with use of two graphite electrodes [120], and in the following year the SWNTs were also produced in arc-discharge using alloy catalysts (like Fe–Co) [121–123]. Recently, unrolled single-layer graphene phase was also prepared by Novoselov et al. [124]. In 1997 Dillon and coworkers [125] reported the first experimental result of high hydrogen uptake by carbon nanotubes; their estimate was 5–10 wt% H2. This was followed by hot race to claim by many researchers ever-larger potency of carbon nanotubes to absorb and retain hydrogen. The most known became the claim by Baker et al. [126] that some nanofibers can absorb over 40–65 wt% H2. However, these experimental results and their interpretation were strongly criticized and have not yet been able to be reproduced. More careful studies, that follow, have shown that only 0.7–1.5 wt% of hydrogen gas is adsorbed in nanofibers under ambient temperature and pressures slightly above 100 bar [127]. Since then, the area of nanocarbons and carbon nanotubes for hydrogen storage has seen continuing interest from many research groups worldwide. Carbon nanotubes, as graphene and graphite, are highly ordered carbon phases. However, a separate line can be drawn for historical development of disordered carbon phases; among them is an amorphous carbon (am-C) . In it, strong bonding between carbons did not allow for completely chaotic distribution of carbon atoms in solid-state phase. Instead, amorphous carbon exhibits random distribution of three possible coordinations of carbon atoms in a planar sp2, tetrahedral sp3, and

1.2

Brief, Synchronic History of Development of Hydrides and Nanomaterials

25

even linear sp1 configurations of electronic orbitals. Some of these phases were shown to be greatly stabilized by hydrogenation. Hydrogenated amorphous carbon has been observed to contain as much as 40–60 at.% H2 in a structure consisting of more than 70% tetrahedrally bonded atoms in sp3 configuration [128]. At lower hydrogen content at 25–30 at.% H2 the sp3 fraction is high. At 70 at.% H2, the hydrogenated tetrahedral amorphous carbon, known better as diamond-like carbon (DLC), has been developed for many very useful applications [128]. These amorphous phases have only been produced as thin films in deposition from high-density plasma. Recently, a new spherical graphitic nanophase, termed carbon nanoshell, was found by Wronski and Carpenter [129] in carbon residues obtained by acid leaching of Ni from commercial carbonyl nickel powders produced in century-old CVD carbonyl nickel process in INCO refineries in Wales, UK and Canada. However, these and other new nanocarbons are still waiting to be investigated for hydrogen storage. As for the year 2007, the recent Viewpoint Paper by Chahine and Bénard [130] states clearly that these new nanocarbons and nanostructures that store hydrogen by physisorption will still require development to bring about qualitative changes; means are still waiting for looking at them from a new angle.

1.2.11

New Materials and Techniques

In the past several years, some old compounds were put into perspective from a new angle keeping still in mind that to meet hydrogen storage targets in hydrogen storage for vehicular applications such storage must be reversible in near-ambient temperatures (less than 100°C) and pressure range. One of the systems that gained great interest was lithium nitride. Li3N has been known since the 1910 study by Ruff in Germany to compound with hydrogen, and release it under strong heating. However, it was only in recent investigations by Ping Chen and coworkers in National University of Singapore [131] that the Li–N–H system was reassessed and demonstrated hydrogen absorption in as low temperature as 100°C, and desorption that is rapid, yielding as much as 9.3 wt% H2 in half an hour when temperature is raised to 255°C. What attracted wide interest in this research was that about 6.3 wt% H2 was desorbed below 200°C, and the desorption reaction follows the reverse of absorption: Li3 N + 2H 2 ↔ Li2 NH + LiH + H 2 ↔ LiNH 2 + 2LiH

(1.15)

However, the desorption proceeds under 10−5 bar vacuum, and traces of ammonia that is toxic for fuel cells were detected. While lithium nitride was an old material revisited, zeolites are well-researched materials for their catalytic and physisorption properties. These microporous inorganic frameworks were known to trap hydrogen; however, their potential for hydrogen storage was not great as they show only moderate storage capacities. On the other hand, a new microporous metal–organic frameworks (MOFs) ignited return of interest in hydrogen storage based on physisorption rather then chemisorption.

26

1

Introduction

In 2003, Rossi et al. [132] reported a MOF material with high hydrogen sorption capacity, 4.5 wt% H2 at cryogenic temperature of 78 K. The promising hydrogen sorption properties of MOF materials were discovered only 4 years from the time the first synthesis of this class of materials was performed [133]. Much more modest, 1 wt% H2 storage, yet at room temperature and 20-bar pressure was observed. This started another vigorous research in materials, in which scaffolding-like nature leads to extraordinarily high surface area of 2,500–4,000 m2 g−1, i.e., more than four times the specific surface area of zeolites. In the following research the capacity of systems such as IRMOF-6 and -8 in room temperature exceeded that of carbon nanotubes in cryogenic temperatures. Since then, the area of mesoporous hydrogen storage media, where hydrogen is sorbed in very-high porosity materials, has been experiencing great interest from the community of researchers in hydrogen storage. The intertwining of the development of nanomaterials and methods of nanoprocessing is still ongoing. While development of new MOF-type materials is an example of a quintessential bottom–up nanotechnology, an opposite route, top–down nanotechnology approach, realized through reduction in the grain and particle size has also been furthered. New techniques have become developed, which have potential to take materials research beyond metals and carbons, and beyond the limitations imposed by the choice between Scylla and Charybdis of a strong chemisorption in nanohydrides and a weak physisorption in nanocarbons. New explored materials are nanocomposites or hybrid hydrogen storage media where chemical bonds are optimized for reversible storage of hydrogen in solids. One of these new techniques having potential for preparation of such materials is the controlled reactive hydrogen ball milling/alloying (CRMM/MA) of metals, nonmetallic elements, and compounds in hydrogen-filled ball mills. This approach to preparation of new hydrides and hydride nanocomposites has been developed at the University of Waterloo and CANMET’s government laboratories in Ottawa, Canada. The milling is conducted in specialty, high-energy ball mills, where trajectories of balls are well controlled, and H2 gas can be supplied and absorbed during milling. The complex Mg2FeH6 hydride was prepared in this way by Varin et al. [134] in the direct, mechanically driven synthesis, viz. direct mechanosynthesis reaction by reacting inexpensive elemental Mg and Fe metals: 2Mg+Fe+3H 2 → Mg2 FeH 6

(1.16)

The reaction of (1.16) follows reverse path to the thermal decomposition reaction of (1.13) and proceeds at room temperature and only slight overpressure of hydrogen supply. This presents a new mechanical activation route to manufacturing of nanomaterials for hydrogen storage. This brief history of century-old investigations toward hydrogen interaction with solid materials and nanomaterials brings us to the current state of affairs when the hydrogen storage for fuel cell systems still remains to be solved. Indeed, in the first decade of the new Millennium, and at the advent of the Hydrogen Economy, fuel cell stacks for use in mass transportation, like those developed by Ballard Power Systems based in Canada, are ready for mass commercialization. Also, hydrogen

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

27

production systems, as those widely used in the chemical and fossil-fuel (oil)refining industries are ready to supply hydrogen fuel to fuel cell cars. Sources of hydrogen are numerous. More than 90% of hydrogen is today produced through thermochemical reforming of methane gas, CH4, at high-temperature (800–1,700°C) reaction with steam or oxygen [(1.17) and (1.18)] to make the syngas…not unlike the gas produced by gasification of coal in the nineteenth century (e.g., (1.1)). CH 4 + H 2 O → CO + H 2 (reforming reaction)

(1.17)

CO + H 2 O → CO2 + H 2 (gas shift reaction)

(1.18)

Alternative production of clean hydrogen is considered in the context of new nuclear reactor technologies, like Generation IV reactors, and as a complement to nuclear electric energy systems in which hydrogen is to be produced electrolytically from off-peak nuclear power. Research, development, and demonstration of production of clean hydrogen from water via innovative photochemical water splitting, and from renewable energy sources, such as wind and solar energy exhibit surge of activity. Yet, at the same end to first decade of Millennium, time-efficient and safe hydrogen storage for use in mass transportation is still unsolved. The hydrogen challenge is waiting solution. Can advances of nanotechnology and nanomaterials meet the hydrogen challenge?

1.3 1.3.1

Nanoprocessing in Solid State in High-Energy Ball Mills Processes for the Synthesis of Nanostructured Materials

Since the interest arouse in unusual chemical, physical, and mechanical properties of nanostructured materials, progress in the synthesis of such materials has only been accelerating. Already established wet chemistry methods, with use of aqueous and nonaqueous solutions, have been reexamined with respect to feasibility to conduct chemical precipitation of nanocrystalline phases under conditions that hinder growth of customary crystalline phases. These methods involve precipitation of metals from its salts with use of strong reductors as borohydrides. In general, methods have been explored that accelerate the rate of reaction and thus limit the time available for ions to diffuse and contribute to growth of the nucleating crystals. Among the advantages of the established wet chemistry methods, like sol–gel processing, are good control over microstructure and particle morphology, and no need for special equipment. Sol–gel approaches were delivering good results even when a synthesis of complex structures was required (e.g., in complex multimetal oxides). Generation of nanometer-sized phases through deposition of metals and metal oxides from gas phase was among the first dry-chemical methods [62]. Many of new processes required development of special equipment or methods as to conduct

28

1

Introduction

physical vapor deposition (PVD) or chemical vapor deposition (CVD) under thermodynamically nonequilibrium conditions. Among these old and new processes for preparation of nanocrystalline alloys and compounds are the following: • • • • • • • • • • • •

Solution-precipitation methods, e.g., those with use of strong borohydride reductors Sol–gel technologies Reverse-micelle synthesis Polymer-mediated synthesis Protein-templated methods Rapid solidification and devitrification of amorphous metals and metallic glasses Combustion-flame chemical vapor condensation processes (Kear) Induction-heating chemical vapor condensation processes DC and RF magnetron sputtering, inclusive of the method of thermalization Laser ablation methods Supercritical fluid processing Sonochemical synthesis and microwave hydrodynamic cavitation synthesis

Many if not all of these processes have already been developed and well established even before the term nanomaterials was coined; they remain further refined with focus on manufacturing materials for particular properties, electronic, etc. All of them are well covered in books and monographs, which are too many to make good recommendation to our reader. For recent progress the reader can consult Nanomaterials Handbook, where the editor, Yury Gogotsi, assembled impressive body of authors to write chapters on processing and properties of many nanomaterials, particularly well covered being carbon-based nanomaterials, metallic semiconductors, and ceramic nanomaterials [135]. All of these methods stem from a natural property of atoms to self-organize into ordered structures and crystallites as being driven by energy benefits when atoms join into structures. Short-range ordering in metal alloys may be formed during deposition of atoms or atom clusters from gas phases, and survive nonequilibrium processing of metals via solidification from rapidly quenched melts. The short-range order of atoms may even be preserved in amorphous metallic alloys. The approach to build up nanocrystals through such self-organization of atoms is well recognized in the preparation of nanomaterials and termed the bottom–up nanotechnology approach. Bottom–up processing of nanomaterials relies upon use of either liquid solvents and vapor gas phases, or solidification from molten metals and compounds. However, in this book we concentrate on nanoprocessing conducted entirely in a solid state. This is a class of methods based on processes conducted in high-energy ball mills. Such methods are based on high-energy grinding and milling of materials. They provide top–down approach toward manufacturing nanomaterials.

1.3.2

Milling Processes and Equipment

The milling, grinding, and pulverizing of materials have always been one of unit operations in chemical engineering. They are well described in chemical engineering

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

29

handbooks along with heat transfer, mass transfer, fluid flow, and thermodynamic processes. Each of these unit operations gather knowledge of physical laws and practice that is necessary for reliable engineering design in many industries, viz., mineral, ceramic, and powder metallurgy. The primary objectives of milling have always been mixing or blending, change in particle shape (morphology), and first of all: size reduction. The use of the milling and pulverizing for causing the change in fundamental mechanical, chemical, and physical properties of the materials itself was not the target of milling or grinding. Use of mechanical milling for synthesis of new alloys, compounds, and nanomaterials was not envisioned until recent decades. Equipment used for milling may be classified according to the way in which mechanical forces are applied: (1) between two solid surfaces (crushing, shearing), and (2) at one solid surface (impact). The milling in ball mills combines both crushing/shearing and impact forces combined in various proportions, depending on the equipment used. There are many different designs of ball mills, which can be used for processing of advanced materials. Among them are the following: • • • • • • •

Tumbler, jar, drum, or cannon ball mills Szegvari attritor vertical mills and other vertical stirred ball mills Planetary Fritsch and Retsch model mills Shaker (vibratory) SPEX model mills A.O.C. magnet-controlled mechanical model mill (Uni-Ball magnetomill) A.O.C. electric discharge-assisted mechanical mill ZOZ continuous-fed horizontal mill, and other horizontal high-energy bead mills

Ball or tube mills have a cylindrical or conical shell, rotating on a horizontal, vertical axis. The ball mill differs from the tube mill by being short, and having the crucible length and diameter almost identical. The typical ball mill, used as much in laboratory like in industry, has been the tumbler ball mill (in laboratory practice also known as the jar mill and in industry drum mill). The grinding balls (usually in large numbers) impact upon the powder charge when cylindrical container placed on rollers rotates along horizontal axis. The balls may roll down the inside wall surface, which produces shear forces on powder trapped between the wall and the ball, but mostly they fall freely accelerated only by gravitation force, then impacting the powders (and other balls) beneath them. To maximize the impact forces imposed on powder the criterion of critical speed may be applied, where Nc expressed in rotations per minute (RPM) is the theoretical speed at which the centrifugal force on a ball (at the height of its orbital path) becomes equal to the force on it due to gravity [136]: N c = 42.3 / D ½

(1.19)

where D is diameter of the mill in meters, and the ball diameter is kept small with respect to the mill diameter. The milling in ball mills can be described as kinetic processing when the contact of the balls with the material is the main event of kinetic energy transfer from the

30

1

Introduction

grinding media into the powder. A well-known fundamental equation describes the relation between the kinetic energy – Ekin, and the mass – m, and the velocity – v: Ekin = (1 / 2 )mv 2

(1.20)

The velocity v of the ball in free fall is gt, where g is the Earth’s gravitational (downward) acceleration of the ball, equal to 9.8 m s−2, and t is the time to cover a distance from the height of its path on the shell to the impacted powder at the bottom of the shell. Hence Ekin = (1 / 2 )m(gt)2 [kgm 2 s − 2 ] or 48.02mt 2 [Nm]

(1.21)

where the N in brackets represent force units in Newtons. Since the time-of-flight of a ball in free downward direction is greatly limited by diameter D of the shell, and the energy scales as square root of time, the kinetic energy, which can be achieved in a simple jar ball mill, is greatly limited. The contact, on impact, between the grinding (balls) and ground (powder) media is also limited. Therefore, laboratory jar and industrial drum (tumbler) mills are low-energy mills (although they can provide higher energy milling if sufficiently large diameter in order of meters and many balls are used, and the drum is operated just short of the critical speed (see 1.19). From the earlier equations it becomes clear that the maximum velocity of balls relative to milled material is the primary factor in determining the energy transferred to or deposited in the material. Larger acceleration than gravitational force, hence higher velocity and higher kinetic energy of balls can be achieved in Szegvari attritor mills. In such mills the milling is conducted in a cylinder filled with balls that are stirred by rotating a horizontal shaft (agitator). The impact of shaft causes differential velocities between the balls and the powder. This can be seen in Fig. 1.2. The collective impact of a group of balls, and the impact on powder particulates trapped between a ball and the wall, is suppressed. Attritor mills are vertical mills but the same mechanical milling principle is extended to variety of other vertical mills generically referred to as stirred ball mills. There are also horizontal stirred mills. In the latter the shell is stationary and filled with multitude of small balls or beads (diameter 0.6 cm; ¼ in. or smaller); hence they are also known as bead mills. Attritor-type bead mills stir the milled media at speeds from 100 to 1,500 RPM. These mills are available commercially in batch, continuous, and circulation types. The shear mechanical milling is a predominant mode [137] in these mills as the energy of accelerated balls is dispersed among the other balls in the mill (see Fig. 1.2) whose relative motion exerts mostly shearing forces on trapped powder particulates. An attritor-type ball mill delivers ten times more energy than comparable size drum ball mill, and some of its models can be considered as medium-energy ball mills. Although the efficiency of vertical attritor mills can be low (when the volume available for powder is small and the reactor cylinder is full with balls, and when the powder has tendency to fall by gravity to the bottom of the cylinder), the new horizontal bead mill attritors are quite efficient in preparation of fine-grained materials.

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

Fig. 1.2 Motions of balls in attritor mills; the dotted vertical line represents the axis of rotation of a vertical shaft on which a number of horizontal shafts (arms) are mounted, which strike balls and agitate their motion in a very energetic manner; powder particles trapped between the balls are subjected to high-rate deformation processes

Planetary mill

31

rotating shaft

Vibrational mill

ωp ωp

Strong shearing

a

b

Strong shearing Strong impact

Fig. 1.3 Motion of balls in a planetary (a) and a vibrational (b) mill

In planetary ball mills the force acting on balls is increased. Centrifugal force of 50 times gravitational force – m × g –can easily be produced. The motion of the shell and the balls in a planetary mill is shown in the schematic Fig. 1.3a. The size of a planetary ball mill will be smaller than the size of a drum (tumbler) ball mill of the same energy. Like attritor and horizontal ball mills they can be considered as medium- to highenergy ball mills, however, milling times needed to process submicron size and nanostructured powders may be long. The commercially available mills, popular in research laboratories, particularly in Europe are Fritsch Pulverisette™ planetary mills. The models where two or four batches of powders can be processed in the cylinders mounted on rotating plate were the workhorse instruments in many laboratories. The recent Pulverisette model 7 was designed with larger rotational speeds specially for processing of powders to nanometric-level size. The Fritsch model Pulverisette™ mills, models 4, 5, and 6, are compatible with an interesting reactor cylinder, in which the cover is equipped with pressure and temperature sensor and a small radiofrequency unit allows for wireless transfer of the sensors’ data to the receiver interfaced with computer. This GTP (temperature–pressure) system developed

32

1

Introduction

Fig. 1.4 Planetary mill vial with antenna mounted on cover equipped with temperature and pressure sensors for GTP wireless monitoring gas conditions during milling

by the Fraunhofer Institute in Dresden, Germany, allows for monitoring temperature and pressure of gases evolved (or absorbed) in the reactor vial in in-situ and real-time conditions. This vial is shown in Fig. 1.4. Since the energy transferred to the material under processing in solid state increases with the frequency of impacting as square of the ball velocity, the most energetic ball mills are such that the balls are shot at high speeds and at high frequencies. This takes place in vibratory ball mills. In those mills, known also as shaker mills, an eccentric motion is imparted to the cylindrical container (or rather armature on which it is mounted) at frequencies ranging from several impacts per minute, viz., several hertz, up to 1,800 Hz, and at small amplitudes of vibrations. The balls oscillate in three mutually perpendicular, or more, planes within a small vial (several tens of cm3) as illustrated in Fig. 1.3b. At average frequency 1,200 and the amplitude of ball vibration ca. 5 cm (diameter of the rotating vial) the balls achieve velocities ca. 5 m s−1 at the moment of impact. The kinetic energy transferred to the material can be very high, even with one 5-g ball, as one can realize substituting the square of this velocity to (1.20). Most of energy transfer in vibratory mills is conducted in the mechanical impact mode although substantial shear of powder particulates is also present, as the balls rotate on the shell and the particulates are trapped between the shell and the balls. Maurice and Courtney [138] have calculated that the shock impacts imparted on the powder in the high-energy SPEX model vibrational mill are typically 40 kbars with stainless steel balls of 6 mm in diameter vibrating in a vial 50 mm of diameter. The SPEX model mill has been used extensively in research laboratories to process small batches of nanomaterials (ca. 10 cm3 or 5 g). Although this mill has been developed for grinding hard materials, such as ZrSiO4, TiO2, SiO2, and Al2O3, in analytical spectroscopy laboratories, this has been the mill that brought the solid nanoprocessing and mechanical alloying

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

33

research to laboratories worldwide, and particularly in North America. Figure 1.5 shows the SPEX Model 8,000 Duo high-energy mill for simultaneous milling of two powder batches, as manufactured by the CertiPrep Company (Metuchen, NJ, USA). The vials, Fig. 1.5, are made of hardened ferritic steel, but vials made of zirconia, alumina, agate, and hard-metal tungsten carbide are also used to greatly limit contamination by grinding media (Fig. 1.6), which can be as high as several percent of Fe when milling in steel vials. The milling that yields nanometric structure takes place about 10 times faster in a SPEX ball mill than in a planetary ball mill. The milling time needed to produce fine powders and nanostructure depends obviously on the nature of material, but also on other milling parameters. The weight of the milling balls is one important parameter, the ratio of the weight

Fig. 1.5 Vibratory high-energy SPEX ball mill

Fig. 1.6 Tungsten carbide and agate vials for SPEX mill

34

1

Introduction

of balls to the weight of powder charge B/P, is the other. One can opt for balls having particular density, which can vary from 2.3 g cm−3 to 4.0, 5.7, 7.8, and 16.4 g cm−3 for alumina, zirconia, stainless steel, and tungsten carbide, respectively. The ball-to-powder weight ratio can vary too, from 1 to 100, and obviously, this is important to maintain the same ratio when one intends to compare results from different milling. As a first approximation one can expect the frequency of the impacts to be proportional to the number of balls used (usually two or three balls are sufficient) but also on the B/P ratio. The time needed for milling to the same fines of a powder usually decreases with the increase in the B/P ratio. Said this, we have to bring the reader’s attention to often unexpected behavior of powder during high-energy milling. For example, the charge of pure aluminum powder milled for as short as 30 min in a SPEX mill changes to a multiplicity of small aluminum balls with 2–4-mm diameter, instead grinding the coarse Al powder into fine Al powder [sic!]. Agglomeration and cold welding of metal particulates to the reactor wall and the milling balls or to form balls made of the milled powder itself are common phenomena that occur during milling of pure ductile metals in vials filled with inert gas, such as argon. On the other end, milling of graphite leads rapidly to amorphous carbons. Therefore, when formation of nanostructure is this intended task, one must design the process carefully so that neither cold-welded balls nor amorphous powders are produced. Other way to apply force to a milling ball, besides the gravitational and the centrifugal means discussed earlier, is to drive ball motion by magnets. On such a magnetic principle works the A.O.C. model Uni-Ball Mill magnetic mill developed in Australia (Wollongong, NSW, Australia). The ferritic steel balls in the reactor cylinder are attracted to the inside surface of the shell by strong Nd–Fe–B permanent magnets placed outside this shell (Figs. 1.7 and 1.8). The direction of the strong magnetic force is well defined by the lines of magnetic induction, which penetrate the shell made of nonmagnetic austenitic steel to end up in the ferritic (hence soft-magnetic) balls. The pull imparted by the external magnet on the magnetic balls inside is so strong that the centrifugal force acting on the balls becomes a secondary factor in milling. High-energy milling can be achieved at low rotations (30–200 rpm). At low rotation the mill design is greatly simplified and two cylinders rotated on the same horizontal pulley can be filled sequentially with a process gas (such as hydrogen). What is more important, the motion of the balls inside the shell is well controlled by the external magnets to the point that the fraction of the impact and the shear modes of mechanical milling can be controlled. The highenergy mechanical shear force is imparted on the powder trapped between the shell and the ball when the magnet is in the 6 o’clock position. In high-energy impact mode each ball travels from the 6 o’clock position, at the bottom of the cylinder, to the 3 o’clock position, and then falls to the bottom of the reactor. The point where each ball is detached from the wall is well determined by the position of the magnet; hence each ball imparts the same energy impact on the milled powder. Since each ball travels only quarter of the circumference of the shell it falls four times per one rotation. This results in 525 impacts per minute for a shell rotating at relatively low speed of 175 rpm.

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

35

Fig. 1.7 Motion of balls in the magnetic ball mill (courtesy of A.O.C. Scientific Engineering, Australia)

Fig. 1.8 Milling cylinder and magnet mounted for mechanical nanoprocessing in the mechanical strong impact mode (IMP2)

36

1

Introduction

At this frequency of impacts the energy transferred to the powder would be comparable with that achieved in a planetary ball mill. The milling cylinder suitable for reactive milling is shown in Fig. 1.8. High-energy ball milling is a complex process, which requires optimization of many parameters to assure repeatability of nanostructure from batch to batch. To illustrate this complexity we can list the important parameters that must be decided when conducting the process in the magnetic A.O.C. model Uni-Ball Mill: • Milling mode depends on the magnet position. By changing the magnet position one can change milling mode from low-energy shearing, through high-energy shearing, to low-energy impact, and to strong impact. Two-magnet options are also possible: (very strong shearing mode and very strong impact mode). • Number of balls used for milling usually varies from two to four balls. Maximum five balls can by used in one cylinder but the optimal number is four. • Milling speed that can be controlled in a range of 0–200 rpm depends on milling mode. More energetic modes usually require fast rotation. • Milling time that is closely related to milling mode, working distance, and type of process. Ball milling requires usually shorter time than mechanical alloying or reactive ball milling • Milling atmosphere, which is a neutral protective gas (helium or argon) during mechanical milling or hydrogen under pressure up to 0.9 MPa under reactive mechanical alloying processes. • Ball-to-powder weight ratio: This parameter depends on the mass of milled powder and number of balls and usually is in a range 10–100. The maximum mass of powder in one cylinder is 25 g what allows for milling 50 g of powder at once in two cylinders. Ball-to-powder ratio affects efficiency of milling or synthesis process and can be controlled by changing mass of milled powder or number of balls. However, there is no visible influence of the ball-to-powder weight ratio on the particle size of the synthesized MgH2 after reactive milling for 30 h. • Working distance (WD) is the distance between magnet (magnets) and cylinder. This parameter affects mainly the attractive force between the magnet and balls inside the cylinder. It is shown that increasing the WD reduces substantially the attractive force between the magnet and the 25-mm steel ball. Neodymium (NdFeB) magnets used in the A.O.C. mills may vary in attraction force. For high-energy ball milling that requires two magnets (Fig. 1.9), they are usually applied in tandem such that stronger magnets are paired with weaker ones on both sides of the Uni Ball Mill 5 (left and right) and the weaker magnet is always above (top) the stronger magnet (bottom). Ideally, magnets should have opposite polarity to double magnetic induction on balls operating in magnetic air gap. There are other advantages of employing magnetic ball mills besides the control of mechanical milling modes. Since the centrifugal force becomes a secondary factor in milling, and the reactor shell rotates at low RPM, contamination from balls and shell wear is lower than in a vibrational or a planetary mill; there is less ball wear involved and contaminations with Fe from steel become less of the problem. Also lower rotations and uniaxial movement of reactors paced on horizontal axle allow

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

37

12 Right bottom

Left bottom Force (kgf )

10 8 6 4 2

Left top

Right top

0 0

2

4

6

8

10

12

14

16

Distance (mm)

Fig. 1.9 The attractive force between the 25-mm steel ball metal vs. distance from the cylinder (working distance – WD) and four various magnets

equipping reactors with valves for filling them with reactive or inert gases. Most of the experimental results in this book that we use to demonstrate the effects of ball milling in processing of nanomaterials for hydrogen storage come from such magnetic mills employed in our laboratories. In controlled reactive mechanical milling/mechanical alloying modes (CRMM/ CRMA), discussed latter in this chapter, powders can be nanoprocessed under sequential supply of reactive process gases, such as N2, O2, etc. When the process gas is H2 and the powder is a metal, we can consider this mechanical alloying with hydrogen and term this process hydrogen alloying (HA). This has been used for a direct mechanical synthesis (mechanosynthesis) (MAS) of interstitial metal hydrides. We will discuss these processes in the following chapters.

1.3.3

Nanoprocessing Methods and Mechanisms

Five major processing methods that are employed during milling of materials in high-energy ball mills are as follows: • • • • •

Mechanical milling (MM) and mechanical disordering (MD) Mechanical alloying (MA) Mechanochemical activation synthesis (MCAS) Mechanochemical synthesis (MCS), and hydrogen alloying (HA) Mechanical amorphization (MAM)

In the following paragraphs we will briefly describe all the aforementioned methods, while paying particular attention to MA, which is at the core of all benefits brought about by mechanical processing of nanomaterials. Yet, multiplicity of phenomena of mechanochemical coupling that occurs during processing of alloys and compounds in high-energy ball mills have not been well understood, yet well reviewed before.

38

1.3.3.1

1

Introduction

Mechanical Milling

High-energy ball milling is the only nanotechnology top–down approach for the synthesis of nanoparticles. In the process of mechanical milling (MM) brute force is applied to material, whether it is a metal, a pre-alloyed intermetallic, or a solid chemical (stoichiometric) compound: the force that is sufficient to disperse the material into fine nanometric particulates or agglomerates of such material. Since only reduction in particle size is required and not substantial rearrangement of atoms takes place in milled particles the time required for MM is short. Yet, even short milling times can break thin chemically passive surface coatings (e.g., surface oxides) and expose fresh, clean chemically active metallic surface. Such milling can also introduce defects into solid compounds. It results in an increased chemical activity of milled media toward both gasses and chemical reactions in solutions and electrolytes; the process is often termed mechanochemical activation synthesis (MCAS). Longer milling of intermetallics leads to changes in the long-range order in intermetallics. Such a process termed sometimes mechanical disordering (MD) leads to formation of disordered intermetallics, alloys, and compounds. A process wherein mixture of elemental metal powders, or powders of metal and nonmetal are milled long enough to trigger alloying of elemental powders in a solid-state process is termed mechanical alloying (MA). Since substantial rearrangement of chemical species must take place during alloying the latter requires long times of milling, usually more than twice the time needed for preparation of nanopowders by mechanical milling. Even longer milling results in what we can term mechanical amorphization (MAM) of a crystalline solid. The amorphous alloys produced in this way are analogous to amorphous metal phases (metallic glasses) produced by rapid quenching of metal melts as described in “Introduction,” Sect. 1.2.2. When the milling process initiates a solid-state reaction and yields a new stoichiometric or quasi-stoichiometric chemical compound, such as carbide, nitride, silicide, etc., such a process can be termed reactive mechanical alloying or reactive mechanical milling (RMA/RMM). When mechanical modes of milling (shear, impact) are controlled we have controlled reactive mechanical alloying or controlled reactive mechanical milling (CRMA/CRMM). Reactive mechanical milling can be realized by reacting two chemical compounds A and B in a solid-state synthesis that yields a distinct chemical compound C. This is termed mechanical synthesis or mechanosynthesis. Reactive milling can also be conducted by milling metal powders in ball mills filled with a reactive gas, such as N2. Hydrogen can also be used for reactive milling and leads to reduction of oxides during milling. Hydrogen can also alloy with metals to form interstitial metal-hydrogen solutions, hydrides, or chemical hydrides. Reactive mechanical milling in hydrogen-fed ball mills, we can term hydrogen alloying (HA), is a new efficient way to the discovery and development of new hydrogen storage materials. Obviously, hydrogen alloying can be used for mechanosynthesis of chemical hydrides. In the following several paragraphs we describe briefly mechanisms that may be at play during ball milling of powders in high-energy ball mills.

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

1.3.3.2

39

Mechanical Alloying

Mechanical alloying is conducted through milling of two elemental metal powders. Longer milling, order of hours, leads to atomic-level mixing of the metals and produces an alloy consisting of these metals. Until the advent of this new, nonequilibrium and low-temperature solid-state processing method, metal alloys could only be manufactured by melt casting and metal foundry practices. There is a sequence of consecutive mechanical events together with concurrent events of chemical nature that leads to the realization of the mechanochemical process described as mechanical alloying. Mechanical alloying can be regarded as a repeated stressing, deformation, fracture, and cold welding of powder particulates, as illustrated schematically in Fig. 1.10. These mechanical processes are interlaced with enhanced diffusion of chemical species and increase in structural disorder. The first step is development of high-rate stresses in the milled powders. The key step is repeated high-rate, severe plastic deformation followed by series of chemical and mechanical processes, which result in generation of structural disorder, extended solid solubility of chemical species, and formation of metastable phases and nanostructures. The sequence of concurrent mechanical and chemical events can be written as follows: Mechanical stressing ® severe plastic deformation ® formation of a submicron lamellar microstructure ® cold interdiffusion of metal atoms between lamellae or nanograins (cold welding) ® fracture ® formation of nanostructure ® extended solid solubility ® mechanical alloying with formation of thermodynamically stable and/or metastable phases ® amorphization. A great deal of research papers, also review articles, was published on the subject of mechanical alloying. Also, there are serial conferences and international meetings on the subject, where new developments in mechanical alloying for processing of nanomaterials, including those for

repetitive stressing

welding

fracturing

Fig. 1.10 Mechanical alloying through repetitive cold welding and fracturing

40

1

Introduction

use in hydrogen storage, are reported.. Since this field of materials science is growing rapidly at the time we write this book it would be difficult to select scholarly papers published on this complex subject. Instead, we will refer the reader to an extensive review, which could be of particular help [139].

1.3.3.3

Mechanochemical Activation

Stresses, Deformations, and Equilibrium Mechanical Processing The local pressure applied to solid by two colliding balls, or a ball hitting a solid trapped between it and the shell of the cylinder can reach several GPa [140]. The mechanical stress in the material under milling develops as particles are subjected to the effect of pressure created between two milling surfaces. If this stress on ball impact is normal to the surface of solid, tensile or compressive stress occurs. In tangential movement of the milling surfaces the stresses are of shear stress type (Fig. 1.11). Actually, both stresses occur simultaneously as the material is processed. In attritor (beads) or vibrational ball mills the particles move relative to milling balls and collide either with the milling surface or with each other. The material particles and the milling balls can not only move forward but can also rotate (Fig. 1.2), so more complex stresses develop in addition to the impact stresses. In more general continuum mechanics approach to three-dimensional solid bodies, one must identify several planes on which the forces act, and to calculate the stress on each plane in the milled material, as it is shown in Fig. 1.12. One must note that if the stresses are expanded in terms of the normal-impact and tangential-shear

tension-on ball impact

compression-on ball impact

shear, on tangential ball impact

Fig. 1.11 Deformation of an ideal material under normal (tension, compression) forces and shear forces exerted by balls. The positions of balls are shown for impact and shear mode of milling

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

41

Fig. 1.12 Expansion of the ball-milling stress into normal (impact) and shear stresses

components, then the set of all components can be called a tensor σ in Euclidean space. Then, any stress state can be represented by its minimal and maximal components, known as principal components [141]. These stresses can be measured as the average stress in many milled particles, and reflected in the stress broadening in addition to the grain size broadening of X-ray diffraction peaks in milled powders (how, it will be shown latter in this section for the Williamson–Hall method, Fig. 1.19, and in Sect. 1.4.3 for the Cauchy/Gaussian method). The principal stresses can also be measured in larger single milled particles used in a stress diffractometer [142]. For example, during milling of cast Mg crash in SPEX mill, normal compressive stress, several times greater than the shear compressive stress was measured and exhibited gradual change to 0, while the shear negative (compressive) stress changed to positive (tensile). Such processing resulted in mechanical activation of Mg metal for absorption of hydrogen [142]. During milling conducted in the A.O.C. Uni-Ball Mill the occurrence of predominant stresses, shear or normal (impact), can be somehow controlled through special configuration of magnets that affects the trajectory of motion of milling balls. While action of forces and stresses during high-energy ball milling has not been well investigated it is apparent that they cause a high degree of elastic and plastic deformation. In simple terms elastic deformation is defined as the reversible deformation that occurs when a load is applied. Most materials deform in a linear elastic fashion i.e., the amount of reversible deformation is a linear function of the applied stress up to a certain stress level, i.e., the strain (tensile, compressive, or shear) divided by original length e, is proportional to the applied stress, s (force divided by cross-section area), according to the Hooke’s Law s = Ee

(1.22)

where E, Young’s modulus is the materials constant, which quantifies elastic behavior. Most nonmetallic materials, such as salts, oxides, and ceramics deform also in such a linear fashion, although in a very small range: if the applied force is further increased the compound fractures in a catastrophic manner. One can notice that the atoms in the stressed body change their relative positions (with respect to one another) more under the shear elastic stress than under tensile

42

1

Introduction

or the compressive elastic stresses. Therefore, we can focus more on the elastic shear stress. If, further to the above, one assumes proportional relationship to reflect on lowering energy by the action of mechanical force then a term proportional to the elastic shear stress, a • s, must be subtracted from the activation energy EA [143]: k = k0 exp[ −

( EA − as ) ] RT

(1.23)

where a is a proportionality constant, and k is the rate at which covalent bonds in a nonmetallic material are ruptured. Now, one can substitute the stress for the strain, and the relation between applied elastic stress and the rate k can be proposed: k = k0 exp[ −

( EA − Eae ) ] RT

(1.24)

Therefore, the rate at which chemical bonds break increases with elastic shear stressing of the material. The rupture of chemical bonds, hence fracture of material, leads to its fragmentation into particles. This reduces the average particle size in powder as fractured particles multiply into even smaller particles. Equation (1.24) points to the importance of elastic shear strains in mechanical activation of chemical bonds for particle size refinement and production of nanoparticles. Most metals initially deform elastically and then begin to deform plastically. Plastic deformation allows energy of stresses to be dissipated rather than building up to the point where the material breaks. Such dissipated energy is used for grain size refinement in each of the deformed material particles. This results in decrease of the average grain size. Therefore, the energy of balls, which was transferred to the material, is utilized for both particle and grain size refinement. Amount of energy used for each of these two processes is largely determined by the value of Young’s modulus. We will come back to the issue of grain refinement in the following paragraphs. Nonequilibrium Mechanical Processing and Excitations Although less to be seen this way, (1.24) points also to the importance of elastic shear strains in a possible mechanism of mechanical activation of materials for chemical reactions. Apparently, ruptured chemical bonds should lead to increased reactivity of the milled powder toward gases, and predominately toward hydrogenation reaction. However, for the sake of efficient mechanical activation the elastic strains should be isolated from the plastic strains. One way of doing this could be preventing the dissipation of energy injected into material by the impacting ball. Dissipation of energy in deformed materials needs time, and if there is no time sufficient for such dissipation between two stressing events (being these impact or shearing events) then the energy must be reabsorbed locally in near

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

43

adiabatic processes. This should demonstrate itself in formation of adiabatic shear bands, as caused by strain-localization phenomena that are generally attributed to a plastic instability arising when thermally activated softening exceeds work hardening in the material. Adiabatic shear bands are well known to play a key role in dynamic fracture and fragmentation, especially in hard (or work hardened) materials [144]; unfortunately, such processes have not been well described in the context of events taking place during high-energy milling, when powder particles are deformed severely at high rates of mechanical deformation. Nevertheless, it would be possible to view a picture where the energy of mechanical process is transformed locally and instantaneously into the elastic energy of either vibrational modes of thermally softened lattice of atoms or into higher oscillational modes of electrons in the atoms in such a lattice. The atoms, whose electrons achieved higher oscillation states, become ready for chemical reactions; thus, the material becomes mechanically activated for chemical reactions. The mechanism we are putting forward here is one of several possible mechanisms that may be at play during nonequilibrium processing and mechanochemical activation. The usual way to activate material for chemical reaction is through heating it to the average temperature that triggers the reaction. This constitutes thermal activation. It cannot be said firmly that milling process does not include local heating and local rises in temperature. Such a local and instantaneous rise in temperature is difficult to quantify, yet it was reported to reach 400°C in highly energetic milling of Ni–Zr powder mixtures [145]. However, the average temperature of the powder, measured with a thermocouple touching the wall of milling cylinder rarely exceeds 60°C in a SPEX mill. Theoretical analysis of ball milling process in a SPEX mill demonstrated the same ~60°C increase [140]. In thermal activation, mechanical energy transformed to kinetic energy of vibrating atoms can activate atoms for chemical reactions. Thermally activated reactions can be described in mechanistic (kinetic) terms where atoms or molecules with increased kinetic energy (hence temperature) exchange energy and interact with each other in collisions. The kinetic energy of the colliding atoms is on an average lower than the energy of chemical bond, but the local quasi-stresses at the moment of impact can be sufficient to cause extensive polarization (deformation of shape) of the molecule, or cause polarization of the chemical bond as in Fig. 1.13. For atoms in a crystal lattice, the change in the force that bonds atoms, from zero net force (point r0) to attraction (rt) or repulsion (rc), will result in increasing the population of higher oscillational states of electrons (not shown in this picture). Then, the relaxation of elastic energy with the excitation of interatomic bonds triggers formation of an unstable transition state or a short-life complex in which interatomic bonds differ from the initial molecules. Stabilization of such a complex and its dissociation along a new reaction path may lead to formation of new reaction products. Whichever be the mechanism, mechanochemical activation in high-energy ball mills is thus an overlooked method to activate solids for chemical reactions. We will discuss this in the following sections.

44

1

Introduction

V(r/r0)

rt

attraction net force

repulsion

r0

r

rC

compressive stress

tensile stress

Fig. 1.13 The elasticity of the chemical bond, gray area regions points to the polarization of chemical bond between the atoms in position 0 and r0, rc or rt

Disordering, Faulting, and Grain Refinement The relaxation of elastic energy with the excitation of interatomic bonds may be one fundamental mechanism that drives mechanochemical activation process. The other mechanism can be breaking of passive oxide layers during milling of metal powders, hence creation of fresh metal surfaces. There is, yet another important aspect of milling processes that may lead to mechanochemical activation. This is disordering and faulting that occurs in crystalline lattice near the particle surface, as well as in the particle interior. As we stated before, original source of novel properties of mechanically alloyed powders is a stressing process. If there is enough time for dissipation of energy imparted on material by ball milling, such energy is dissipated by plastic deformation processes. Such processes generate highly disordered structures within the processed particles. In metal particles, plastic deformation requires the movement of dislocations within the crystal structure, as shown, although in a simplistic manner, in Fig. 1.14. Movement of dislocations can be hindered by foreign solute atoms, precipitations of second phases, and other dislocations. Pinned dislocations can contribute further to generation of new dislocations since they form sources that emit more dislocations, as in the famous FrankRead source. As dislocations become entangled, the work hardening begins.

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

45

Fig. 1.14 One step in the movement of an edge dislocation through the crystal lattice. As a consequence of the extra-half plane of atoms regions are formed in the vicinity of dislocation where compressive and tensile stresses/strains are imposed on the neighboring atoms

The dislocations snag and pile up. Material loses its plasticity. It hardens. As the cohesive forces that keep atoms in the hardened material are overcome by the milling forces, which now cannot be plastically dissipated, the material structure becomes disjoined and rearranges into a polycrystalline structure consisting of many grains. Once polycrystalline structure is formed the Hall-Petch relation can relate the yield strength, YS, and the grain size, d YS = YS0 + Kd −1/ 2 ,

(1.25)

where YS0 and K are material constants. It must be noted that starting from grains so small that dislocations can not pack and pile up, the inverse Hall-Patch relationship can be at work and increased plasticity (superplasticity) has been reported in ceramic nanomaterials. However, until the grains achieve critical size the ball milling produces multigrain particulates. The atomic structure of a single grain is, in principle, across the grain. However, in the regions of grain boundaries, the average atomic density, interatomic spacing, and the coordination between nearest neighbor atoms deviates from that inside the grain, and differs from region to region. The presence of these two structural constituents (grains and grain boundaries) at comparable volume fractions and with typical crystal sizes of a few nanometers is crucial for the properties of nanocrystalline materials [62]. In consequence, many of the physical and mechanical properties of nanocrystalline solids, such as thermal expansion, elastic constants, fracture stress, and ductility, are widely different from those of the same material having conventional grain sizes. This is a direct consequence of the fraction of atoms in intercrystalline positions being significant. Consequently, interface structures in these materials are bound to play a major role in material properties. Once the crystal size and boundary dimensions become comparable within certain length scales new physical effects are to be expected. This grain boundary fraction is truly disordered, if not plainly amorphous. This is illustrated in Fig. 1.15. The fraction of atoms that counts into grain boundaries should increase with milling time; however, in powder particulates below some

46

1

Introduction

a grain A grain B b

c

Fig. 1.15 Enlarged view of hydrogen atoms (black circles) in the grain boundary between two grains A and B (gray circles). Note varying short-range order around atoms a and b, which belong to the grain boundary, and fixed order around atoms of type c, which belong to the grain core

critical size there is little or no grain boundaries. This may have significant consequences on chemical reactivity of milled powders because the disordered grainboundary phase is seen as the place when diffusion of chemical species is faster and chemical reactions are facilitated. Nanostructured materials are such materials where the density of grain boundaries is maximized. It must be noted that nanopowders do not necessarily consist of nanometric particles (nanoparticles). The powder particles may be actually of micrometric sizes, yet the particle interior may be divided into many nanograins, as this is depicted in Fig. 1.16. Such nanograins are the smallest diffracting domains in X-ray diffractometry, and being incoherent to X-ray diffraction they cause peak broadening in X-ray diffraction (XRD). The average grain size is determined from such broadening of peaks. We will discuss how this can be done later on. Hydrogen Entry into Grain-Refined and Disordered Powder Structures Hydrogen molecule, H2, dissociates on contact with catalytic metal surface, and then enters the material as dissociated hydrogen atoms through grain boundaries. Subsequently, the hydrogen atoms diffuse preferentially through quasi-amorphous grain boundary network (regions a and b in Fig. 1.15) before they begin to diffuse into grain interior (region c). Therefore, the specific surface area (SSA) and

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

47

H2

particle size

grain size

Fig. 1.16 Particle size and grain size in a nanostructured powder. Black points depict hydrogen molecules () in pores between particles, and hydrogen atoms () diffusing through a grain boundary network and gradually entering the grain interior; black triangles point to triple points that are preferential sites for hydrogen atom accumulation

the fraction of material that belongs to grain boundaries are equally important. The beauty of nanostructuring in ball mills comes in the observation made that both the particle size and the grain size are equally rapidly decreased with even short processing in a high-energy ball mill, as shown in Fig. 1.17. Yet, the picture in which the particle size and the grain refinement are sole reasons for mechanochemical activation of powders for reactions with hydrogen is not complete. Nanostructuring by ball milling introduces a variety of defects, vacancies, dislocations, and stacking faults besides the earlier described grains and grain boundaries. These defects raise the free energy of the system making it accessible to formation of thermodynamically metastable phases. Also, defects lower the activation energy of reactions limited by poor kinetics. One type of such defects that would be of particular interest is stacking faults. Heavy deformation via cold working, like that occurring in ball mills, introduces stacking faults on (111) planes in fcc metals, such as Ni, and in the basal (0001) or prismatic {1010} planes of the hexagonal close-packed metals, such as Mg. The former is a well-known catalyst for hydrogen storage, and the latter is a hydrogen storage metal itself, storing as much as 7 wt% H2 in a reversible process. There are few investigations toward the effect of stacking faults [147] – such as twins and deformation faults – on gas hydrogen storage in hexagonal metals and alloys.

48

1

Introduction

Commercial MgH2-ball milled 100 Particle size ECD (mm)

350

Grain size (nm)

300 250 200 150

10

1

0.1

100

0

5

50

10 15 Milling time (h)

20

25

0 0

5

10 15 Milling time (h)

20

25

Fig. 1.17 Simultaneous reduction of the particle size and the grain size in fully hydrogenated magnesium powder, MgH2 [146]

Interplanar Hydrogen in Disordered Layered Compounds Formation of stacking faults has been observed in some inorganic compounds under even mild milling conditions. Some layered inorganic compounds of interest for sorption and desorption of gas hydrogen (H2) and electrochemical hydrogen (H+) exhibit disorder in stacking of atomic planes. Among them are graphites and some nanocarbons to be discussed in Chap. 4. Stacking faults were also observed in battery-grade nickel hydroxide, Ni(OH)2, used in the positive electrode of rechargeable alkaline nickel metal hydride (NiMH) battery [148]. We will turn to nickel hydroxide powder to demonstrate the effect of ball milling on improved reversible insertion and deinsertion of hydrogen ions between hexagonal basal planes of layered nanocrystals. Generic Ni(OH)2, having a Mg(OH)2 brucite-type structure, is a compound with well-stacked hexagonal basal planes of NiO2 composition, seen in TEM image, Fig. 1.18a, and H atoms bonded to O atoms in hydroxyl, –OH, groups that hang inside interplanar space. The planes are stacked through action of relatively weak van der Walls bonds. Battery-grade Ni hydroxides were found to have large degree of disorder in stacking of planes, Fig. 1.18b, in contrast to well-stacked planes in generic hydroxide (Fig. 1.18a). The existence of such stacking fault disorder and its role in improving insertion of hydrogen ions into electrodes were recently investigated [148, 149]. Mechanochemical activation of these materials in ball mills was proposed to be an efficient method for improvement of reversible insertion of electrochemical hydrogen, viz., H+, into these structures [148]. A concept of stacking fault disorder, which was postulated on the evidence from X-ray diffraction (Fig. 1.19)

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

49

a

b Fig. 1.18 Stacking of hexagonal basal planes in Ni(OH)2. (a) Generic and (b) ball milled; sets of chevrons underline the shift of planes due to shearing forces in the milling process, after [72]. Also, shown hydroxyl – OH – groups, light grey – O, and black – H atom dumbbells extending into the interplanar space

and direct imaging in high-resolution electron microscope (Fig. 1.18b) [148], has been since confirmed in computer simulation [149] and batch-scale experiments with use of ball mills [150]. Even a short ball-milling for less than half an hour in SPEX mill results in proliferation of stacking faults, followed by fragmentation of the structure into small domains of layered nanocrystals of this compound [151]. Therefore, insertion and deinsertion of hydrogen ions is facilitated in short-layered crystallites, i.e., layered nanocrystallites, and this leads to the observed improvement in the electrode capacity, lower swelling on charge/discharge cycles, and extended durability [152]. Stacking-fault-type disorder is rarely discussed in the context of milled metals, alloys, and hydrides. We shall not discuss it more here either. However, we will come back to this type of disorder later in Chap. 4 when we will discuss disordered graphites and nanocarbons.

50

1

Introduction

a

b Fig. 1.19 (a) X-ray diffraction pattern with nonuniform line broadening, sharp lines for h − k = 3n, and broad lines for h − k ≠ 3n, which gives evidence to presence of stacking fault-type disorder in the structure of hexagonal Ni(OH)2 for electrochemical hydrogen insertion; (b) WilliamsonHall plot construction is used to determine the degree of stacking fault disorder [38]; the plot (b* = bcosq/l) vs. (d* = 2sinq/l) exhibits large scatter of b* values for the material with stacking faults(●, ) as compared with the material without faults (▲, Δ); the points for the indices hk0 (●, ▲,) are not affected by the disorder, as well as are the points for the fcc Ni reference powder (▲); l and q are the wavelength and the half of two-theta angle at which Bragg peaks are centered (adopted from [148])

Interstitial Hydrogen in Disordered Alloys Well-ordered intermetallic compounds (alloys), when processed in high-energy ball mills exhibit atomic (chemical) disordering in the early stages of ball milling [153]. Let us take as an example ordered AlRu intermetallic crystals that are of B2 type and β-CuZn structure. This structure consists of two simple cubic interpenetrating

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

51

sublattices, one mostly occupied by Cu atoms and the other having mostly Zn in lattice nodes as represented by black and white atoms in Fig. 1.20a. At high temperatures, above the critical temperature of 1,015 K, the long-range ordering disappears and both kinds of atoms become randomly distributed in the two sublattices in equal fractions. Long-range chemical order is lost and two sublattices become indistinguishable in X-ray diffraction pattern, characteristic of BCC structure. Such disordering phenomena that occur in alloys at increased temperatures have long been known, and large body of knowledge has already been accumulated in the domain of physical metallurgy and material science. However, similar disordering, induced without use of thermal activation and proceeding in room temperature, has become to be apprehended only since the advent of mechanical alloying. Indeed, high-energy ball milling of AlRu in ambient temperature for 32 h in a vibratory ball mill under high vacuum generates stable disordered solid solutions [154]. Even more intensive milling can lead to unstable solid solutions with extended solubility limits of one type of atoms in the crystalline lattice formed by other atoms. Ultimately, atoms in an alloy or a compound can be mechanically mixed to form amorphous phase, where chemical disorder is convoluted with topological disorder. We will come back to these disordered phases in the following paragraph, when we will discuss mechanochemical synthesis routes that lead to formation of nonequilibrium phases. This could have far-reaching consequences, as mechanically induced disorder can be a factor in activation of well-ordered intermetallic compounds (alloys) for reversible reactions with molecular hydrogen. Figure 1.20b shows the PCT plots, which details hydrogen sorption properties for B2-type FeTi, an important hydrogen storage alloy with capability for reversible storage at room temperature applications. The plateau of equilibrium sorption is lowered in ball milled, disordered FeTi in comparison with the not-milled counterpart. Therefore, the alloy has been

1

b sublattice

In P (MPa)

0 −1

c b

−2

a

A atom −3

B atom

0.0

a sublattice

a

b

0.4

0.8 H/FeTi

1.2

Fig. 1.20 The B2-type CsCl structure of FeTi stoichiometric, ordered, compound (left), and room-temperature hydrogen PCT properties for B2-type FeTi hydrogen storage alloys: amorphous – a, nanocrystalline − b, and crystalline − c (adopted from [155])

52

1

Introduction

mechanically activated for sorption of hydrogen at lower pressures. Inspecting the plots (Fig. 1.20b) one can also notice that overmilling the alloy to the amorphous powder is detrimental, as sorption plateau disappears, and the powder capacity for hydrogen sorption becomes null. Said this, we can let the reader to recall Fig. 1.15, where we depicted amorphous-like phase regions at grain boundaries as the pathways open for preferential diffusion of hydrogen atoms. Apparently, an alloy can benefit from some fraction of amorphous phase to improve kinetics of hydrogen absorption, but complete amorphization of crystalline lattice lowers capacity for storing hydrogen [156]. Mechanochemical activation is therefore a complex process where kinetic and thermodynamic effects must be firstly well understood, and then optimized. Summing up, we identified several reasons and phenomena for mechanochemical activation of powders for chemical reactions with hydrogen at room temperatures in high-energy ball mills. Among these discussed here were the following: ●

● ● ●

●

●

●

Breaking of oxide layers and passivation coatings, hence exposure of fresh catalytic sites on metal surfaces High levels of elastic shear and other stresses induced by milling Particle size reduction, hence increase in the specific surface area Grain size reduction and maximizing the density of grain boundaries in nanostructured particles Stacking fault disorder and fragmentation of structures into multiplicity of layered nanocrystals (viz. in oxides, hydroxides, and graphites) Atomic (chemical) disorder that leads to formation of stable and unstable (extended) solid solutions Topological disorder that leads to formation of amorphous phases

In mechanochemical activation it would be the action of several of the aforementioned parameters, and their relative importance is often difficult to separate. There may be also other phenomena and synergetic effects too, which have not been elucidated yet. Whatever are the mechanisms, mechanochemical activation works amazingly well, and presents an attractive alternative to thermal activation. It also provides easy, top– bottom method for manufacturing nanostructured hydrides and nanocarbons.

1.3.3.4

Mechanochemical Synthesis (Mechanosynthesis) of Nanohydrides

A primary method of mechanochemical synthesis of nanostructured hydrides (nanohydrides) is processing by mechanical (ball) milling. Processes of manufacturing of nanocrystalline/nanostructured hydrides by ball milling are shown in Fig. 1.21. There are three major processes all of which start from raw metallic/nonmetallic elements [157]. (1) In the first procedure, called a two-step method, either premilled pure metallic elements or a mechanically milled (MM) precast intermetallic ingot are hydrogenated under gas hydrogen pressure to form a desired metallic or intermetallic-

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

53

Raw materials Mixture of chemical compounds or hydrides

Elemental powders

Pure elements

Inert gas Inert gas

Arc melting

Preformed hydride

Mechanical alloying (MA)

H2

Bulk material

Milling of powders (MM)

Alloy powders

Reactive mechanical alloying (RMA)/ Mechano-chemical Synthesis (MCS)

Activation and hydrogenation

Mechano-chemical activation synthesis (MCAS)

By-product removed

Nanostructured hydrides

Fig. 1.21 Flow chart showing the possible methods of manufacturing nanocrystalline/nanostructured hydrides (modified from [157])

based hydride such as, for example, MgH2 or Mg2NiH4. For hydrogen alloying of a powder mixture, elemental powders are mixed in desired proportion and then mechanically alloyed (MA) by ball milling under inert gas (e.g., argon) to form intermetallic compounds. The formation of intermetallics in milled metal–metal and metal–metalloid (e.g., B, Si, C) systems is the first step of mechanosynthesis [158]. The general reactions for a binary system A and B are as follows: nA + mB → AmBn via ingot / powder

(1.26a)

AmBn + (x/2)H2 → AmBnHx hydrogenation

(1.26b)

(2) In the second method, so-called reactive mechanical milling (RMM), elemental metallic powders are milled under hydrogen atmosphere. This procedure is a one-step method. The reactions are a direct alloying of metals and hydrogen, i.e., hydrogen alloying. When sufficient amount of the hydrogen reactant is supplied this reaction yields hydrides with well-defined stoichiometry: A0 + (x/2)H20 → Ax+Hx1–

(1.27)

As seen from the change in valency of metal and hydrogen, this is a redox reaction because the metal is oxidized (loses n electrons) by the hydrogen, while the valence of hydrogen becomes, formally, −1. Magnesium dihydride, MgH2, can be manufactured in ball mills in such a manner, with the reaction yield close to 90%. The reaction of

54

1 Introduction

(1.27) is accompanied by both the particle- and the grain-reduction, as discussed in Sect. 1.3.3.3. In a variant of the second method described earlier the premixed metallic powders (or pulverized ingots) are milled under hydrogen atmosphere to directly form an intermetallic hydride. It can be also viewed as hydrogen alloying of metal powders and powder mixtures in hydrogen alloying mills. This method is called a reactive mechanical alloying (RMA) or mechanochemical synthesis (MCS). nA + mB + (x/2)H2 → Am BnHx

(1.28)

MCS-type reactions can occur while milling a simple metal hydride with a metallic element M, 2AH + 2Al+ 3H2 → 2AAlH4

(1.29a)

for instance NaH with Al [159]: NaH + Al + 3/2H2 → NaAlH4

(1.29b)

Mechanosynthesis reaction can also be conducted with use of a second hydride, instead of hydrogen gas reactant. Most often the first hydride is a simple metal hydride, while the second hydride is a complex hydride. 2NaH + NaA1H4 → Na3A1H6

(1.30)

Such reactions have been observed to occur where an alanate such as NaAlH4 is milled with alkali metal hydride, such as NaH or LiH. (3) In the third method called mechanochemical activation synthesis (MCAS), a mixture of metal compound, viz. metal chloride, is ball-milled to induce a reaction to yield a high-hydrogen capacity hydride. If an n-valent metal chloride or fluoride (AXn) is used in the reaction with alkaline metal borohydride B(BH4) then in general terms the reaction can be written as AXn + nB(BH4 ) → A(BH4)n + nBX

(1.31)

where A(BH4)n is a newly synthesized complex metal borohydride and BX is a salt (X = Cl or F). Either Li or Na borohydrides, LiBH4 or NaBH4, are used as source of hydride complex. Similar reactions are used for preparation of alumohydrides, viz. alanates, using Li or Na alanates, LiAlH4 or NaAlH4. These reactions are mutual substitution reactions where the ions in two compounds change sites in the crystalline lattice. The valence of the metals and the hydrogen do not change from the reactants to the products. This is analogous to well-known metathesis reaction, which is a common method for preparation of borohydrides in donor nonaqueous solvents (tetrahydrofuran, ethers). The MCAS advantage is that solid-state process

1.3

Nanoprocessing in Solid State in High-Energy Ball Mills

55

does not require use of an organic solvent, whereby the borohydride product does not come with the organic solvent molecules bonded in its crystalline lattice. Said this, we must observe that the MCAS process yields a mixture with alkali metal chloride or fluoride, and to purify the complex hydride product one must call upon a difference in solubility between the borohydride and the metal chloride in the same THF or ether solvents. There are also disadvantages that can be mentioned when opting for preparation of borohydrides or alanates by solid-state mechanosynthesis. Certain hydride complexes are not stable, and become stable only when borohydride or alanate molecule is solvated. Also, certain structures based on hydride complexes may be not stable under mechanical ball milling. The mixture of Mg and B milled with hydrogen yields an amorphous phase, instead of the anticipated Mg(BH4)2, with release of hydrogen gas [160]. The MCAS product can also be a mixture of an amorphous phase and a boride, such as MgB2. This brings us to the subject of synthesis of amorphous phases in mechanochemical processing.

1.3.3.5

Mechanical Amorphization

When a mixture of elemental powders is milled, the formation of the amorphous phase is due mainly to an ultimate interdiffusion of atoms that occurs at fresh surfaces and interfaces created by mechanical milling [87]. This interdiffusion is promoted by defects and chemical disorder in the crystalline structure. However, no such interdiffusion is prominent while milling intermetallics, where constituents are already alloyed. In this case the accumulation of the chemical disorder (as discussed in previous paragraphs) leads to collapse of crystalline structure if the rate of dynamic recovery of crystalline order is less than the rate of deformation. This is a nonequilibrium thermodynamic process. The phase formed under nonequilibrium milling is an unstable solid solution. In a solution consisting of two kinds of atoms that differ in size (more than 15%) their packing on the crystalline sites becomes less dense due to the disorder and defects. This constitutes a high-energy phase that has an inherent tendency to transform to more stable phase. This is a metastable amorphous phase, in which atoms abandon the crystalline sites and rearrange into close-packed topologically random structure. Yet, there is an alternative rearrangement of atoms to even lower energy structure. Such is an equilibrium crystalline compound. Addition of the third kind of atoms (in respect to atomic radii) during milling with hydrogen seems to increase ability of disordered solid to form an amorphous structure [160]. Hence, it is possible that the topological disorder brought about by elastic transformation under dynamic shear and impact stress can be stabilized by hydrogen atoms [161]. However, the alternative is to rearrange atoms into an equilibrium crystalline hydride compound. This would require nucleation of hydride grains in the matrix of unstable crystalline solid solution that was formed on milling. Since nucleation proceeds en masse, yet crystal growth rate is slow, the process yields hydride nanophase. Therefore, there are two parallel routes to lower the energy of unstable mechanically disordered solid solution: (1) formation of an intermediate amorphous phase first, followed by crystallization an

56

1

Introduction

equilibrium compound and hydride nanophases [160, 161], and (2) transformation to equilibrium crystalline compound and hydride nanophases directly. What reaction path is chosen depends on both thermodynamic and kinetic conditions, and the products in mechanochemical reactions shown in the reaction of (1.31) may come with sometimes unavoidable fraction of an amorphous phase. Summing up, we discussed four main routes toward nanoprocessing of hydrides in ball mills. These are as follows: ● ● ●

●

Mechanical milling (MM) of cast ingots followed by hydrogenation Mechanical alloying (MA) of elemental powders followed by hydrogenation Mechanochemical synthesis (MCS) by reactive milling of metal powders or intermetallics in hydrogen gas Mechanochemical activation synthesis (MCAS) by reactive milling of compounds

All of these syntheses are preceded by mechanical activation of reagents in ball mills and all of them can be considered when manufacturing of nanohydride powders is anticipated. However, solid-state reaction routes are not always well investigated, and can be complex. More often than not, the yields of reactions leading to nanohydrides are not reaching 100% and an amorphous phase, which provides lower storage capacity, is formed.

1.4

Important Hydride Properties and Experimental Techniques

1.4.1

Thermodynamics

1.4.1.1

Pressure–Composition–Temperature (PCT) Properties

Pressure–composition–temperature (PCT) curve called also pressure–composition isotherm (PCI) curve can be a source of important fundamental information related to thermodynamic properties of solid hydrides. There are several methods of determining PCT properties ranging from thermogravimetric to precise volumetric measurements obtained by using a classical Sieverts-type apparatus. The thermogravimetric methods unfortunately are extremely limited in pressure applied during test to the maximum 2 MPa, which is usually around equilibrium pressure only for low-capacity hydrides. The plateau pressure at the decomposition temperature for high-capacity hydrides is much higher and reaches from 2.5 up to 15 MPa. A typical isotherm of a reversible hydride is shown in Fig. 1.22 By measuring the changes in hydrogen pressure and corresponding changes in hydrogen concentration in metal at a given temperature, PCT curves can be constructed that are expected to give a flat plateau. Most practical hydrides do not show perfectly flat plateau or zero hysteresis. Sloping behavior is observed possibly due to different equilibrium pressure, localized defects, and surface inhomogenities [21] (will be discussed in more detail in Sect. 2.1.4).

1.4

Important Hydride Properties and Experimental Techniques

57

H2 absorption

lnP

Hysteresis = lnPa /lnPd

H2 desorption

Slope = dlnPa / dHcap

H capacity wt.%

Fig. 1.22 Schematic isothermal pressure–composition hysteresis loop

Tc

T3

lnP = ΔH/RT - ΔS/T

↵ α+β

T2

↵

lnP

T1

a

↵ lnP

α

↵ β

−ΔH/R

↵

↵

H capacity wt.%

b

1/T

Fig. 1.23 (a) Pressure–concentration–temperature plot and (b) Van’t Hoff plot

The effect of temperature on the PCT curves is shown in Fig. 1.23a. The metal initially dissolves only small amount of hydrogen ( P2 for absorption and P1< P2 for desorption. Rearranging (1.40) and (1.41) we obtain: n1 =

PV PV 1 n2 = 2 RT RT

(1.42)

Therefore, the difference between number of moles of hydrogen in the system resulting from absorption or desorption is as follows: Δ n = n1 – n2 = ΔP

V RT

(1.43)

where DP = P1 − P2. The mass of absorbed or desorbed hydrogen can be calculated using number of moles of gas and molecular mass of hydrogen: mH= 2.016Δn, which finally gives us: mH = 2.016ΔP

V RT

(1.44)

When change in hydrogen mass is known using (1.33) we can easily calculate hydrogen capacity in the investigated material. The ideal gas law should be corrected by the Van der Waals equation for the volume of gas molecules and molecular interactions at higher hydrogen gas pressures. ⎛ n2 a ⎞ ⎜⎝ p + V 2 ⎟⎠ (V − nb ) = nRT

(1.45)

where a is the measure of attraction between hydrogen particles (0.2476 L2 bar mol−2) and b is the volume excluded by a mole of hydrogen particles (0.02661 L mol−1). From our tests it seems that (1.45) should be used for pressures higher than about 5 MPa. From the scheme of Sieverts apparatus presented in Fig. 1.31, we can see that the volume that must be taken into account during calculation of quantity of desorbed or absorbed hydrogen does not consist of only a reactor volume. The volume V in (1.44) in a real system is made up of volume of reactor plus internal volume

1.4

Important Hydride Properties and Experimental Techniques

67

of connecting pipes, valves, as well as the transducer. The direct measurement of these parameters is impossible. Moreover, the transducer measures pressure changes in this part of apparatus where temperature is ambient, so for calculation using (1.44) we apply T = 297 K (ambient temperature). Nevertheless, increasing the reactor temperature to the temperature of process (usually in the range from 40 to 400°C) increases the temperature of hydrogen gas in the entire system. That is why an application of the ideal gas law would not be possible. Therefore, to evaluate the volume of gas that participates in sorption process the system volume calibration process is necessary. Calibration process consists of a few steps. At first we have to obtain advisable pressure P1 (usually about 10 atm) of argon in a calibrated volume VC and atmospheric pressure in the rest of the system (reactor and connections). Then by opening calibrated volume cut-off valve VC and by pressure reduction to the value P2 the total volume of the system for the apparatus with the relative transducer can be calculated using the formula: P1Vc = P2 (Vc + Vs )

(1.46)

where Vs is made up of volume of reactor plus internal volume of connecting pipes, valves, and the transducer. Rearranging (1.45) we obtain

Vs =

(P1 − P2 )V P2

c

(1.47)

To eliminate the temperature effect, the calibration curve must be obtained by repeating described process at various temperatures (RT to Tmax, where Tmax – maximum temperature of process). To determine PCT curve by volumetric method at first we have to know mass of analyzed powder (hydride or pure metal). The typical mass of powder used in volumetric method is in a range 50–500 mg and depends on VR (reactor volume with volume of connecting pipes, valves, and transducer). After the mass measurement, the powder is loaded into specimen holder and then it is placed in the Sieverts apparatus reactor. To prevent any oxidation and for safety reason the system must be purged a few times by argon and then evacuated. However, one must be careful how much powder is appropriate for the absorption/desorption volume of a Sieverts-type apparatus. Figure 1.32 shows the desorption experiments from two different masses of MgH2 doped with nano-Ni (n-Ni) powder. It is seen that the mass of 140 g gives about 3.0 wt% of desorbed hydrogen while the smaller mass of powder (50 mg) let more hydrogen to be desorbed. This behavior can be explained if we resort to the imaginary desorption PCT curve at 275°C in Fig. 1.33. The equilibrium plateau pressure at 275°C is higher than 0.1 MPa at which desorption is carried out and at this temperature MgH2 can desorb at atmospheric pressure. However, the kinetics of desorption will depend on the driving force (as shown in Fig. 1.26). Since a larger mass of hydride will produce

68

1

Introduction

Hydrogen desorbed [wt.%]

8.00 MgH2 + n-Ni milled for 10 min - 275⬚C

7.00 6.00

4.6wt.%H2 Mass = 50 mg

5.00 4.00 3.00

3.0wt.%H2

2.00 Mass = 140 mg

1.00 0.00 0

1000

2000 Time [s]

3000

4000

Fig. 1.32 The effect of the mass of desorbing hydride powder on the amount of hydrogen desorbed at 275°C at 0.1 MPa from MgH2 doped with 5 wt% of nanonickel (n-Ni)

lnP Very small driving force PCT at 275ⴗC Mass >100 mg Mass 50 -100 mg 1 atm (0.1MPa)

Reasonably large driving force

wt.%H2

Fig. 1.33 Schematic plot showing the effect of mass of MgH2 hydride on the driving force for desorption at 275°C at atmospheric pressure

a higher overpressure, much above 0.1 MPa, in a confined desorption volume of a Sieverts apparatus, which is closer to the plateau at 275°C, then kinetics could be slower and as a result a smaller amount of hydrogen could be desorbed as opposed to a small amount of hydride creating lower overpressure and higher driving force allowing more complete desorption. When we want to obtain a PCT curve for decomposition process the pressure of hydrogen must be set up above the expected equilibrium pressure to prevent decomposition during heating to this temperature. If we do not have any information about equilibrium pressure for investigated material we should apply maximum pressure

1.4

Important Hydride Properties and Experimental Techniques

69

that is allowable by our Sieverts apparatus. High hydrogen pressure should prevent decomposition of hydride during heating up to the temperature of the test. When temperature of the test is reached and the system is thermally stable we can start gradually decreasing the pressure and registering the amount of released hydrogen related to the equilibrated pressure. In a case of the PCT absorption curve after evacuation, the system must be heated up to a test temperature and thermally stabilized. Then by gradually increasing pressure we will observe hydrogen absorption related to a particular pressure of hydrogen. It is important that the time of visible pressure change at every step is closely connected with the kinetics of process at an applied temperature. An analysis of absorption and desorption kinetic is very similar to the PCT curve determination. Before starting the desorption test, the inner tubing of the apparatus is also evacuated and purged a few times with argon and then two times with hydrogen. Subsequently, hydrogen at a high pressure, preventing desorption, is admitted into the airtight cylindrical desorption chamber containing the hydride powder and the temperature of the chamber is gradually increased up to the desired desorption temperature. The high pressure barrier is applied to prevent any desorption during the period of temperature stabilization (15–30 min.) within the desorption chamber. Once the temperature is stabilized the hydrogen pressure is quickly reduced to 0.1 MPa (1 atm), and the hydrogen desorption process begins (pressure increases above initial 0.1 MPa). No elaborate activation procedure usually is applied to the powders. Absorption test starts with purging process as well as evacuation and then system is thermally stabilized under vacuum. Subsequently, hydrogen at desired absorption pressure is admitted into the system and by observing the pressure decreasing as a function of time, the kinetic curve is registered. The difference in pressure between the initial pressure and that at the instant of time is taken for calculation of the mass of absorbed/desorbed hydrogen from (1.44) and the wt% of absorbed/desorbed hydrogen (capacity) from (1.33). However, a question arises whether or not the hydride annealing under highpressure barrier that is applied to prevent any desorption during the period of temperature stabilization (15–30 min) can influence the sorption properties? Bouaricha et al. [174] measured PCT and kinetics of hydrogen sorption using activated samples. The samples were activated at 350 or 400°C, by exposure to hydrogen at a high pressure. This was done until the hydrogen capacity of the material reached a maximum and did not evolve anymore with time. Generally, this state was reached after 12 h. However, it is clearly visible that time applied for activation process using high-pressure annealing must be definitely longer than approximately 30 min. To be absolutely sure that thermal stabilization of measuring system under high preventing pressure of hydrogen does not affect hydrogenation/dehydrogenation properties of analyzed hydride we studied some properties, important from a decomposition point of view, for powder before and after annealing process. Magnesium hydride (type ABCR – see Chap. 2) ball milled for 100 h under HES 57 mode (Sect. 1.3.2) was heated up to various temperatures (300, 325, and 350°C) under preventing 4.5 MPa pressure of hydrogen in a Sieverts apparatus exactly in this same manner as during thermal stabilization. To eliminate any doubts the annealing time was increased up to 45 min. After annealing, the powder was cooled

70

1

Introduction

down up to ambient temperature and investigated by DSC and XRD analysis. Figure 1.34 shows the influence of annealing temperature during thermal system stabilization on DSC test results. A slight increase of both onset and peak temperature is observed, which means that tendency to the decomposition of annealed powder is even less than that for the powder directly after milling. The structural analysis by XRD method (Fig. 1.35 and Table 1.7) shows progressive decrease of the amount of γ-MgH2 (will be discussed in more detail in Sect. 2.1) in powder after annealing as compared with phase composition in a powder directly after milling. Moreover, slight increase of grain size of β-MgH2 during thermal stabilization of a Sieverts apparatus is observed. However, the crystallite (grain) size of magnesium hydride is still in the nanorange (35 nm for powder annealed up to 350°C). These observations clearly prove that the thermal stabilization of Sieverts apparatus under high pressure of hydrogen carried out under previously specified conditions can not be treated as an activation process and it causes negligible structural changes in the investigated hydride. Finally, a small but important technical detail related to the method of measuring pressure by a transducer attached to a Sieverts apparatus (T in Fig. 1.31) may be in place here. Some transducers measure pressure relative to the atmospheric pressure and display a 0 value when pressure equals atmospheric pressure. In this case, during constructing PCT curve 1 atm should be always added to each pressure reading from the transducer’s display. If a Sieverts’ system is equipped with a transducer measuring and displaying the value of 1 atm at an atmospheric pressure, this correction is redundant. Also, this correction is redundant during measuring in hydrogen absorption/desorption kinetic curve because in this case what is needed for calculation of the hydrogen quantity is just ΔP (1.44).

420

Temperature [ⴗC]

400 380 360 340

Peak Onset

DSC analysis

320 300 Milled

300

325

350

Annealing temperature [ⴗC]

Fig. 1.34 The influence of annealing temperature on onset and peak temperature of decomposition peak obtained by DSC test

1.4

Important Hydride Properties and Experimental Techniques

71

10000 b-MgH2 g-MgH2 MgO

8000

Counts

6000

Annealed 350ⴗC 4000 Annealed 325ⴗC 2000

Annealed 300ⴗC Milled

0 30

40

50 60 Degrees 2-Theta

70

80

90

Fig. 1.35 XRD pattern of MgH2 hydride after milling and annealing in Sieverts apparatus during thermal stabilization of the measuring system for 45 min at 300, 325, and 350°C under high pressure of hydrogen (4.5 MPa)

Table 1.7 Phase composition and grain size of β and γ-MgH2 after milling and annealing in Sieverts apparatus during thermal stabilization of the measuring system, at 300, 325, and 350°C under high pressure of hydrogen (4.5 MPa) Sample

Grain size of β-MgH2 (nm)

Grain size of γ-MgH2 (nm)

Phases present in powder

Milled Annealed 300°C Annealed 325°C Annealed 350°C

13 ± 2 19 ± 1 28 ± 1 30 ± 2

13 ± 0.1 19 ± 0.3 29 ± 2 –

β-MgH2, γ-MgH2, MgO β-MgH2, γ-MgH2, MgO β-MgH2, γ-MgH2, MgO β-MgH2, MgO

1.4.3

Microstructural Characterization of Ball-Milled Hydrides

Two important morphological parameters characterizing ball-milled powders are the particle and grain size of constituent phases within the powders. In our laboratory, the size measurement of the powder particles is carried out by attaching loose powder to sticky carbon tape and taking pictures under secondary electron (SE) mode in the SEM. The images are then analyzed by an image analysis software. The size of the powders is calculated as the particle equivalent circle diameter, ECD = (4A/p)1/2, where A represents the projected particle area. Usually from ~300 to 700 particles are analyzed for each batch.

72

1

Introduction

The crystalline structure of hydride powders is characterized by powder diffraction. The nanograin (crystallite) size of phases residing in the milled powders is calculated from the broadening of their respective X-ray diffraction (XRD) peaks. Since the Bragg peak broadening in an XRD pattern is due to a combination of grain refinement (nanograin/crystallite) and lattice strains, it is customary to use computing techniques by means of which one can separate these two contributions. The separation of crystallite size and strain is obtained from a Cauchy/Gaussian approximation by a linear regression plot according to the following equation [39]: d 2 (2q ) Kl ⎛ d (2q ) ⎞ = + 16e2 tan 2 q L ⎜⎝ tan q sin q ⎟⎠

(1.48)

where the term Kl/L is the slope, the parameter L is the mean dimension of the nanograin (crystallite) composing the powder particle, K is a constant (≈1) and e is the so-called maximum microstrain (calculated from the intercept), l is the wavelength, and q is the position of the analyzed peak maximum. The term d(2q) = B[1−(b2/B2)] (rad) is the instrumental broadening-corrected pure XRD peak profile breadth [39], where B and b are the breadths in radians of the same Bragg peak from the XRD scans of the experimental and reference powder, respectively. The B value is approximated as the full width at half maximum, FWHM, and calculated by the diffractometer software. The b value is approximated as FWHM from the X-ray diffraction pattern of a compound LaB6, the National Institute of Standards and Technology (NIST) standard reference material (SRM) 660 for subtracting the instrumental broadening from the experimental FWHM and finding d(2q) as given earlier. It must be noted that when FWHMs of the instrumental line profiles are obtained in this manner, the Bragg peaks for the LaB6 SRM are occasionally at different 2q angles than those of the analyzed hydride in the milled powders. The interpolated FWHM values between angles for the SRM peaks are found using a calibration curve. A typical regression plot according to (1.48) is shown in Fig. 1.36. 0.00035

d2(2q)/tan2(q)

0.0003 0.00025 0.0002

kλ/L = 0.0039

L = 39nm

λCu = 0.1541838 16e2 = 1.0E-05 e = 9.0E-04

y = 0.0039x + 1E-05 R2 = 0.99

0.00015 0.0001 0.00005 0 0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

d(2q) / [tan(q)sin(q)]

Fig. 1.36 Cauchy/Gaussian plot (1.48) for the grain size of dehydrogenation of MgH2 (ABCR) milled for 20 h and catalyzed by 5 wt% Ni

1.4

Important Hydride Properties and Experimental Techniques

73

It must be also noted that modern X-ray diffractometers are usually equipped with software that calculates grain size and lattice strain.

1.4.4

Weight Percent of a Hydride Phase and Hydrogen by DSC Method

In DSC measurements, the weight percent of a phase was calculated using the peak area of the DSC curve and its reported heat of formation. For example, weight percent of the β-MgH2 in a reactively milled powder can be estimated using the peak area of the DSC curves and the reported β-MgH2 heat of formation (−74 kJ mol−1 [175], which equals to −2,811 J g−1). The DSC curve was analyzed by the NETZSCH thermal analysis software. First, the onset and end temperature of the peak were determined. Then, the peak area was calculated using the linear approach from the onset temperature to the end temperature (Fig. 1.37) by the DSC software. The weight percent of β-MgH2 is given by: wt% of β-MgH2 = peak area (J g−1)/β-MgH2 heat of decomposition (J g−1), where heat of decomposition = −(heat of formation). From the decomposition of β-MgH2 (MgH2 → Mg + H2), the weight percent of desorbed hydrogen can be calculated by: wt% of H2 = wt% of MgH2 × (molecular weight of H2/molecular weight of MgH2).

DSC /mW/mg 2.5

[1] 332.8⬚C

exo

2.0

1.5

1.0

0.5 [1] 346.9⬚C

[1] 321.5⬚C

0

[1]

280

300

320 Temperature /ⴗC

340

360

Fig. 1.37 Schematic of calculating the weight percent of a hydride phase by DSC analysis

74

1

Introduction

References 1. J.O.M. Bockris, “Will lack of energy lead to the demise of high-technology countries in this century?” Int. J. Hydrogen Ener. 32 (2007) 153–158. 2. K. Hashimoto, H. Habazaki, M. Yamasaki, S. Meguro, T. Sasaki, H. Katagiri, T. Matsui, K. Fujimura, K. Izumiya, N. Kumagai, E. Akiyama, “Advanced materials for global carbon dioxide recycling,” Mater. Sci. Eng. A 304–306 (2001) 88–96. 3. B. Yildiz, M.S. Kazimi, “Efficiency of hydrogen production systems using alternative nuclear energy technologies,” Int. J. Hydrogen Ener. 31 (2006) 77–92. 4. J.A. Turner, M.C. Williams, K. Rajeshwar, “Hydrogen economy based on renewable energy sources,” The Electrochem. Soc. Interface 13(3) (2004) 24–30. 5. V. Ramani, H.R. Kunz, J.M. Fenton, “The polymer electrolyte fuel cell,” The Electrochem. Soc. Interface 13(3) (2004) 17–45. 6. A.B. Lovins, “Twenty hydrogen myths,” Rocky Mountain Institute, Snowmass, CO, 2005, available at: www.rmi.org. 7. D.S. Scott, “Smelling land-the hydrogen defense against climate catastrophe,” Canadian Hydrogen Association, Westmount, QC, 2007. 8. B.C.R. Ewan, R.W.K. Allen, “A figure of merit assessment of the routes to hydrogen,” Int. J. Hydrogen Ener. 30 (2007) 809–819. 9. A. Züttel A, “Hydrogen storage methods,” Naturwissenschaften 91 (2004) 157–172. 10. G. Sandi, “Hydrogen storage and its limitations,” The Electrochem. Soc. Interface 13(3) (2004) 40–44. 11. S. Satyapal, S. Petrovic, G. Thomas, C. Read, G. Ordaz, “The U.S. national hydrogen storage project,” Proceedings of the 16th World Hydrogen Energy Conference, 13–16 June, Lyon, France, (2006), pp. 1/10–9/10. 12. Basic research needs for the hydrogen economy, Second Printing, U.S. Department of Energy, Basic Research Challenges for Hydrogen Storage, Washington, DC, 2004, pp. 31–51. 13. W. Grochala, P.P. Edwards, “Thermal decomposition of the non-interstitial hydrides for the storage and production of hydrogen,” Chem. Rev. 104 (2004) 1283–1315. 14. G. Sandrock, “A panoramic overview of hydrogen storage alloys from a gas reaction point of view,” J. Alloys Compd. 293–295 (1999) 877–888. 15. L. Schlapbach, L.A. Züttel, “Hydrogen-storage materials for mobile applications,” Nature 414 (2001) 353–358. 16. D.K. Slattery, M.D. Hampton, “Complex hydrides for hydrogen storage,” In Proceedings of the 2002 U.S. D.O.E. Hydrogen Program Review NREL/CP-610–32405, Golden, CO. 17. G. Sandrock, R.C. Bowman Jr., “Gas-based hydride applications: recent progress and future needs,” J. Alloys Compd. 356–357 (2003) 794–799. 18. J.A. Ritter, A.D. Ebner, J. Wang, R. Zidan, “Implementing a hydrogen economy,” Mater. Today 6(9) (2003) 18–23. 19. A. Züttel, “Materials for hydrogen storage,” Mater. Today 6(9) (2003) 24–33. 20. F. Schüth, B. Bogdanovic´, M. Felderhoff, “Light metal hydrides for hydrogen storage,” Chem. Commun. (2004) 2249–2258. 21. D. Chandra, J.J. Reilly, R. Chellappa, “Metal hydrides for vehicular applications: the state of the art,” JOM 58 (2006) 26–32. 22. R.A. Varin, S. Li, Z. Wronski, O. Morozova, T. Khomenko, “The effect and sequential and continuous high energy impact mode on the mechano-chemical synthesis of nanostructured complex hydride Mg2FeH6,” J. Alloys Compd. 390 (2005) 282–296. 23. S. Orimo, Y. Nakamori, G. Kitahara, K. Miwa, N. Ohba, S. Towata, A. Züttel, “Dehydriding and rehydriding reactions of LiBH4,” J. Alloys Compd. 404–406 (2005) 427–430. 24. E. Jeon, Y. Whan Cho, “Mechanochemical synthesis and thermal decomposition of zinc borohydride,” J. Alloys Compd. 422 (2006) 273–275.

References

75

25. S. Srinivasan, E. Stefanakos, Y. Goswami, “New transition metal assisted complex borohydrides for hydrogen storage,” Proceedings of the 16th World Hydrogen Energy Conference, 13–16 June, Lyon, France, (2006), pp. 1/6–6/6 26. J.O. Jensen, Q. Li, R. He, C. Pan, N.J. Bjerrum, “100–200°C polymer fuel cells for use with NaAlH4,” J. Alloys Compd. 404–406 (2005) 653–656. 27. A. Maddalena, M. Petris, P. Palade, S. Sartori, G. Principi, E. Settimo, B. Molinas, S. Lo Russo, “Study of Mg-based materials to be used in a functional solid state hydrogen reservoir for vehicular applications,” Int. J. Hyd. Ener. 31 (2006) 2097–2103. 28. G. Sandrock, J. Reilly, J. Graetz, W.-M. Zhou, J. Johnson, J. Wegrzyn, “Accelerated thermal decomposition of AlH3 for hydrogen-fueled vehicles,” Appl. Phys. A 80 (2005) 687–690. 29. T. Graham, “On the absorption and dialytic separation of gases by colloid septa,” Phil. Trans. Roy. Soc. (London), 156 (1866) 399–439. 30. A. Sieverts, “Occlusion and diffusion of gases in metals,” Z. Phys. Chem. 60 (1907) 129–201. 31. A. Sieverts, G. Zapf, H. Moritz, “Solubility of H2, D2, and N2 in Fe,” Z. Phys. Chem. 183 (1938) 19–37. 32. A. Sieverts, E. Jurish, A. Metz, “Solubility of hydrogen in solid alloys of palladium with gold, silver, platinum,” Z. Anorg. Chem. 91 (1915) 1–45. 33. L. Kirschfeld, A. Sieverts, “Titanium and hydrogen,” Z. Phys. Chem. 145(3–4) (1929) 227–240. 34. M. Avrami, “Kinetics of phase change. II: Transformation-time relations for random distribution of nuclei,” J. Chem. Phys. 8 (1940) 212. 35. M. Avrami, “Kinetics of phase change. I. General theory,” J. Chem. Phys. 7 (1939) 1103. 36. M. Avrami, “Kinetics of phase change. III: Granulation, phase change and microstructures,” J. Chem. Phys. 9 (1941) 177. 37. G.G. Libowitz, H.F. Hayes, T.R.P. Gibb Jr., “The system zirconium-nickel and hydrogen,” J. Phys. Chem. 62 (1958) 76. 38. G.K. Williamson, W.H. Hall, “X-ray line broadening from filed aluminum and wolfram,” Acta Metall. 1 (1953) 22–31. 39. H.P. Klug, L. Alexander, “X-ray diffraction procedures for polycrystalline and amorphous materials,” (1st Ed. 1954; 2nd Ed. 1974), Wiley, New York. 40. H.M. Rietveld, “A profile refinement method for nuclear and magnetic structures,” J. Appl. Crystallogr. 2 (1969) 65–71. 41. G.J.C. Carpenter, “The dilational misfit of zirconium hydrides precipitated in zirconium,” J. Nucl. Mater. 48 (1973) 264–266. 42. B. Baranowski, M. Smialowski, “Charging of nickel films with hydrogen evolved electrolytically in the presence of cathodic poisons,” J. Phys. Chem. Sol. 12 (1960) 206–207. 43. B. Barnowski, “Metal-hydrogen systems at high hydrogen pressures, in hydrogen in metals II,” In G. Alefeld, J. Völkl, Editor. (1978), Springer, Berlin-Heidelberg. pp. 157–200. 44. J.W. Klement, R.H. Willens, P. Duwez, “Non-crystalline structure in solidified gold-silicon alloys,” Nature 187 (1960) 869–870. 45. M.H. Cohen, D. Turnbull, “Composition requirements for glass formation in metallic and ionic systems,” Nature 189 (1961) 131–132. 46. H.S. Chen, D.E. Polk, US Patent 3,856,513. (1974), Allied Chemical Corporation, U.S.A. 47. J.J. Gilman, “Metallic glass materials, in frontiers in materials technologies,” In M.A. Meyers, O.T. Inal, Editor. (1985), Elsevier, Amsterdam. pp. 175–199. 48. N. Grant, “Rapid solidification technology, in frontiers in materials technologies,” In M.A. Meyers, O.T. Inal, Editor. (1985), Elsevier, Amsterdam. pp. 125–172. 49. Z.S. Wronski, “Rapid solidification of iron-neodymium-boron magnetic alloys, in rapid solidification technology-an engineering guide,” In T.S. Srivatsan, Editor. (1993), Technomic Publishing, Lancaster, Basel. pp. 257–298. 50. I.R. Harris, A. McGuiness, D.G.R. Jones, J.S. Abel, “Nd-Fe-B permanent magnets: Hydrogen absorption/desorption studies (HADS) on Nd16Fe76B8 and Nd2Fe14B,” Phys. Scr. T16B (1987) 435–440.

76

1

Introduction

51. M. Kijek, A.M. Cantor, R.W. Cahn, “Diffusion in amorphous alloys,” Scripta Metall. 14 (1980) 337–1340. 52. R. Kirchheim, F. Sommer, G. Schluckebier, “Hydrogen in amorphous metals-I,” Acta Metall. 30(6) (1982) 1059–1068. 53. R. Kirchheim, “Solubility, diffusivity and trapping of hydrogen in dilute alloys, deformed and amorphous metals-II,” Acta Metallurgica, 30(6) (1982) 1069–1078. 54. R.C. Bowman Jr., A.J. Maeland, “NMR studies of diffusion in metallic glass TiCuHx,” Phys. Rev. B 24(5) (1981) 2328–2333. 55. Z.S. Wronski, A.H. Morrish, A.M. Stewart, “Hydrogen induced magnetic transformation in Fe90Zr10 glass,” Phys. Lett. 101A(5,6) (1984) 294–296. 56. Z.S. Wronski, X.Z. Zhou, A.H. Morrish, A.M. Stewart, “Magnetic microcrystals and surface layers in as-quenched and hydrogenated metallic glasses,” J. Appl. Phys. 57 (1986) 3548–3550. 57. X.L. Yeh, K. Samwer, W.L. Johnson, “Formation of an amorphous metallic hydride by reaction of hydrogen with crystalline intermetallic compounds – a new method of synthesizing metallic glasses,” Appl. Phys. Lett. 42 (1983) 242–244. 58. K. Aoki, M. Kamachi, T. Masumoto, “Thermodynamics of hydrogen absorption in amorphous Zr-Ni alloys,” J. Non-Crystalline Solids 61–62 (1984) 679–684. 59. R.C. Bowman Jr., “Preparation and properties of amorphous hydrides,” Mater. Sci. Forum 31 (1988) 197–227. 60. G. Binning, H. Rohrer, C. Gerber, E. Weibel, “Surface studies by scanning tunneling microscopy,” Phys. Rev. Lett. 49 (1982) 57–61. 61. G. Binnig, C.F. Quate, C. Gerber, “Atomic force microscope,” Phys. Rev. Lett. 56 (1986) 930–933. 62. H. Gleiter, “Nanocrystalline materials,” Prog. Mater. Sci. 33 (1989) 223–315. 63. R. Birringer, H. Gleiter, H.P. Klein, P. Marquardt, “Nanocrystalline materials- an approach to a novel solid structure with gas-like disorder?” Phys. Lett. A 102(8) (1984) 365–369. 64. R.C. Baetzold, “Molecular orbital description of catalysis by metal clusters,” J. Catal. 29 (1973) 129. 65. R.W. Siegel, “Cluster assembled nanophase materials,” Annu. Rev. Mater. Sci. 21 (1991) 559. 66. V. Chawla, S. Prakash, B.S. Sidhu, “State of the art: Applications of mechanically alloyed nanomaterials – a review,” Mater. Manuf. Proc. 22 (2007) 469–473. 67. V. Berube, G. Radtke, M. Dresselhaus, G. Chen, “Size effects on the hydrogen storage properties of nanostructured metal hydrides: a review,” Int. J. Energy Res. 31(6–7) (2007) 637–663. 68. J.J.G. Willems, K.H.J. Buschow, “From permanent magnets to rechargeable hydride electrodes,” J. Less-Comm. Metals 129 (1987) 13–30. 69. H. Zijlstra, F.F. Westendorp, “Influence of hydrogen on the magnetic properties of SmCo5,” Solid State Comm. 7 (1969) 857–859. 70. J.H.N. van Vucht, F.A. Kuijpers, H.C.A.M. Bruning, “Reversible room-temperature absorption of large quantities of hydrogen by intermetallic compounds,” Philips Res. Repts. 25 (1970) 133–140. 71. M. Beltowska-Brzezinska, A. Czerwinski, J. Kleperis, M. Kopczyk, J. Skowronski, G. Wojcik, “Electrochemical behaviour of metal hydrides,” J. Solid State Electrochem. 5 (2001) 229–249. 72. Z.S. Wronski, “Materials for rechargeable batteries and clean hydrogen energy storage,” Int. Mater. Rev. 46(1) (2001) 1–46. 73. Z.S. Wronski, “Electrochemical storage of hydrogen in nanostructured solid-state hydrides and nanocarbons,” International Conference on Processing and Fabrication of Advanced Materials- PFAM XII, Singapore (2004). Stallion Press, Singapore-London, pp. 275–287. 74. J.J. Reilly, R.H. Wiswall Jr., US Patent 3,825,418 “Alloys for the isolation of hydrogen,” (1974). Brookhaven National Laboratory, USA.

References

77

75. G.D. Sandrock, “Development of low cost nickel-rare Earth hydrides for hydrogen storage,” In Hydrogen Energy Systems, W. Seifritz, T.N. Vezeroglu, Editor. (1978), Pergamon Press, Oxford, U.K. pp. 1625–1656. 76. Y. Osumi, H. Suzuki, A. Kato, M. Nakane, Y. Miyake, “Absorption-desorption characteristics of hydrogen for mischmetal based alloys,” J. Chem. Soc. Jpn. Chem. Industrial Chem. (Nippon Kagaku Kaishhi) 11 (1978) 1472–1477. 77. H.H. van Mal, K.H.J. Buschow, F.A. Kuijpers, Hydrogen absorption and magnetic properties of LaCo5xNi5–5x compounds. J. Less-Common Metals, 32 (1973) 289–296. 78. G.D. Sandrock, “A new family of hydrogen storage alloys based on the system nickel -mischmetal-calcium,” Proceedings of the 12th Intersociety Energy Conversion Engineering Conference, Washington, DC, 28 August–2 September 1977. American Nuclear Society, pp. 951–958. 79. C.E. Lundin, F.E. Lynch, “Modification of hydriding properties of AB5 alloy type hexagonal alloys through manganese substitution,” Proceedings of the Miami International Conference Alternate Energy Sources, United States, 5–7 December 1977. University of Miami, (1978) pp. 3803–3820. 80. J.J. Reilly, R.H. Wiswall, “Formation and properties of FeTi hydride,” Inorg. Chem. 13 (1974) 218. 81. S.R. Ovshinsky, K. Sapru, K. Dec, K. Hong, US Patent 4,431,561: Hydrogen storage materials and method of making same. Energy Conversion Devices, Troy, MI (1984). 82. N. Kuriyama, D. Chartouni, M. Tsukahara, K. Takahashi, H.T. Takeshita, H. Tanaka, L. Schlapbach, T. Sakai, I. Uehara, “Scanning tunneling microscopy in situ observation of phase-selective cathodic hydrogenation of a V-Ti-Ni-based multiphase alloy electrode,” Electrochem. Solid-State Lett. 1 (1998) 37–38. 83. J.S. Benjamin, “Dispersion strengthened superalloys by mechanical alloying,” Metall. Trans. 1 (1970) 2943–2951. 84. J.S. Benjamin, “Mechanical alloying,” Sci. Amer. 234 (1976) 49–48. 85. C.C. Koch, O.B. Cavin, C.G. McKamey, J.O. Scarbrough, “Preparation of “amorphous” Ni60Nb40 by mechanical alloying,” Appl. Phys. Lett. 43 (1983) 1017–1019. 86. A. Ye, Yermakov, Ye., Ye., Yurchikov, V.A. Barinov, “Magnetic properties of amorphous powders of Y-Co alloys prepared by mechanical alloying,” Fiz Metal. Metalloved. (Phys. Met. Metall. USSR), 52(52–56) (1981) 1184–1193. 87. R.B. Schwarz, R.R. Petrich, C.H. Saw, “The synthesis of amorphous Ni-Ti alloy by mechanical alloying,” J. Non-Cryst. Solids, 76(2–3) (1985) 281–301. 88. J. Bloch, “Effect of neutron irradiation of uranium-iron alloys dilute in iron,” J. Nuclear Mater. 6 (1962) 203–212. 89. D.E. Luzzi, M. Meshii, “The crystalline to amorphous transition of intermetallic compounds under electron irradiation -a review,” Res. Mechanica 21(3) (1987) 207–247. 90. R.B. Schwarz, W.L. Johnson, “Formation of amorphous alloy by solid-state reaction of the pure polycrystalline metals,” Phys. Rev. Lett. 51(5) (1983) 415–418. 91. M. Carey-Lea, (1892)“Quicksilver from cinnabar: The first documented mechanochemical reaction,” J. Metals 34 46 Phil. Mag. cited after: Takac Laszlo, Jan 2000, pp. 12–13. 92. Theophrastus, De Lapidibus, Translation and Commentary by D.E. Eichholz, Oxford University Press, New York (1965) p. 8. 93. S.A. Black, “Development of supercorroding alloys for use as timed releases for ocean engineering applications, Report US Navy Civil Engineering Laboratory,” (1979), Civil Engineering Laboratory (Navy), Port Hueneme, CA, p. 40. 94. E. Ivanov, I. Konstanchuk, A. Stepanov, V. Boldyrev, “Magnesium mechanical alloys for hydrogen storage,” J. Less-Comm. Metals, 131 (1987) 25–29. 95. C. Winkler, “Berichte der Deutwchen Chemischen Gasellschaft,” 24 (1891) 1966. 96. F.H. Ellinger, C.E. Holley Jr., B.B. McInteer, D. Pavone, R.M. Potter, E. Staritzky, W.H. Zachariasen, “Preparation and some properties of magnesium hydride,” J. Am. Chem. Soc. 77 (1955) 2647.

78

1

Introduction

97. E. Wiberg, H. Goeltzer, R. Bauer, Z. Naturforsch, B. Teil 6 (1951) 394; cited after B. Bogdanovic´, S-T. Liao, M. Schwickardi, P. Sikorsky, B. Spliethoff, “Catalytic synthesis of magnesium hydride under mild conditions,” Angew. Chem. Int. Ed. Engl. 19(10) (1980) 818–819. 98. N.T. Dymova, Z.K. Sterlyadkina, V.G. Safranov, Zh. Neorg. Khim. (J. Inorganic Chemistry), 6 (1961) 763; cited after B. Bogdanovic´, S-T. Liao, M. Schwickardi, P. Sikorsky, B. Spliethoff, “Catalytic synthesis of magnesium hydride under mild conditions,” Angew. Chem. Int. Ed. Engl. 19(10) (1980) 818–819. 99. D.Z. Noreus, “Properties of formal low-valence transition metal-hydrogen complexes in Mg2NiH4 and Na2PdH2,” Phys. Chem. NF 163 (1989) 575–578. 100.A. Pebler, E.A. Gulbransen, “Thermochemical and structural aspects of the reaction of hydrogen with alloys and intermetallic compounds of zirconium,” Electrochem. Techn. 4 (1966) 211–215. 101. L. Zaluski, A. Zaluska, J.O. Strom-Olsen, “Nanocrystalline metal hydrides,” J. Alloys Compd 253–254 (1997) 70–79. 102. J.J. Didishiem, P. Zolliker, K. Yvon, P. Fischer, J. Schefer, M. Grubelmann, A.F. Williams, “Diamagnesium iron (II) hydride, Mg2FeH6, containing octahedral FeH64- anions,” Inorg. Chem. 23 (1984) 1953–1957. 103. P. Selvam, K. Yvon, “Synthesis of Mg2FeH6, Mg2CoH5, and Mg2NiH4 by high-pressure sintering of the elements,” Int. J. Hyd. Ener. 9 (1991) 615–617. 104. N.T. Stetson, K. Yvon, P. Fischer, “On the structure of BaReH9,” Inorg. Chem. 33 (1994) 4598–4599. 105. B. Siegel, “The reaction between aluminum and atomic hydrogen,” J. Am. Chem. Soc. 82 (1960) 7. 106. T.N. Dymova, N.G. Eliseeva, S.I. Bakum, Y.M. Dergachev, “Direct synthesis of alkalii metal aluminum hydrides in the melt,” Dokl. Akad. Nauk SSSR 215 (1974) 1369–1372. 107. B. Bogdanovic´, M. Schwickardi, “Ti-doped alkali metal aluminium hydrides as potential novel reversible hydrogen storage materials,” J. Alloys Compd 1–9 (1997) 253–254. 108. T.N. Dymova, Y.M. Dergachev, V.A. Sokolov, N.A. Grechanaya, “Dissociation pressure of NaAlH4 and Na3AlH6 ,” Dokl. Nauk SSSR 225 (1975) 591–592. 109. Ford Motor Co., Direct-hydrogen-fueled proton-exchange membrane fuel cell system for transportation applications: hydrogen vehicle safety report, D.T. Inc., ed., Arlington,VA (1997). 110. D. Goerrig, Verfahren zur herstellung von boranaten, German Patent DE1077644. (1958) Germany, http://depatisnet.dpma.de 111. V.N. Konoplev, V.M. Bakulina, “Some properties of magnesium borohydride,” Russian Chem. Bull. (Translated from Izviestiya Akad.Nauk SSSR, Ser. Khim.) 20(1) (1971) 136–138. 112. A. Züttel, P. Wenger, S. Rentsch, P. Sudan, “LiBH4 a new hydrogen storage material,” J. Power Sources 118 (2003) 1–7. 113. M. Fichtner, G. Engel, O. Fuhr, A. Gloss, O. Rubner, R. Ahlrichs, “The structure of magnesium alanate,” Inorg. Chem. 42 (2003) 7060–7066. 114. L.V. Radushkevich, V.M. Lukyanovich, “O structure ugleroda, obrazujucegosja pri termiceskom razlozenii okisi ugleroda na zelaznom kontakte” (in Russian), Zurn. Fisic. Chim. 26 (1952) 88–95. 115. M. Manthioux, V.L. Kuznetsov, Guest editorial: “Who should be given credit for the discovery of carbon nanotubes?” Carbon 44 (2006) 1621–1624. 116. A. Oberlin, M. Endo, T. Koyama, “Filamentous growth of carbon through benzene decomposition,” J. Cryst. Growth 32 (1976) 335–349. 117. S. Iijima, “Helical microtubules of graphitic carbon,” Nature, 354 (1991) 56. 118. H.W. Kroto, J.R. Heath, S.C. Obrien, R.F. Curl, R.E. Smalley, “C-60- Buckminsterfullerene,” Nature 318(6042) (1985) 167. 119. W. Kratschmer, W. Lamb, L. Fostiropoulos, D. Huffman, R. Donald, “C60: a new form of carbon,” Nature 347 (1990) 354–358.

References

79

120. T.W. Ebessen, P.M. Ajayan, “Large-scale synthesis of carbon nanotubes,” Nature 358(6383) (1992) 220–222. 121. D.S. Bethune, C.H. Kiang, M.S. Devries, G. Gorman, R. Savoy, J. Vazques, R. Beyers, “Cobalt-catalyzed growth of carbon nanotubes with single-atomic layers,” Nature 363(6430) (1993) 605–607. 122. C. Journet, D.K. Maser, P. Bernier, A. Loiseau, M. Lamy de la Chapelle, S. Lefrant, P. Deniard, R. Lee, J.E. Fischer, “Large-scale production of single-walled carbon nanotubes by electric-arc technique,” Nature 388 (1997) 756–758. 123. S. Iijima, T. Ichihashi, “Single-shell carbon nanotubes of 1 nm diameter,” Nature 363(6430) (1993) 603–605. 124. K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jing, S. Zhang, V. Dubonos, I.V. Grigorieva, A.A. Firsov, “Electric field effect in atomically thin carbon films,” Science 306(5696) (2004) 666–669. 125. A. Dillon, K. Jones, T. Bekkedahl, C. Kiang, D. Bethune, M. Heben, “Storage of hydrogen in single-walled carbon nanotubes,” Nature 386(6623) (1997) 377–379. 126. R. Baker, A. Chambers, C. Park, N. Rodriguez, R. Terry, “Hydrogen storage in graphite nanofibers,” J. Phys. Chem. B 103 (1998) 277–281. 127. T. Bose, R. Chahine, E. Poirier, “Hydrogen adsorption in carbon nanostructures,” Int. J. Hydrogen Energy 26(2001) 831–835. 128. J. Robertson, “Diamond-like amorphous carbon,” J. Mater. Sci. Eng. R Rep. 37(4) (2002) 129–281. 129. Z.S. Wronski, G.J.C. Carpenter, “Carbon nanoshells obtained from leaching carbonyl nickel metal powders,” Carbon 44(2006) 1779–1789. 130. P. Bénard, R. Chahine, “Storage of hydrogen by physisorption on carbon and nanostructured materials,” Scripta Materialia 56(2007) 803–808. 131. P. Chen, Z. Xiong, J. Luo, J. Lin, K.L. Tan, “Interaction of hydrogen with metal nitrides and imides,” Nature 420(6913) (2002) 302–304. 132. N.L. Rossi, J. Eckert, M. Eddoaudi, D.T. Vodak, J. Kim, M. O’Keeffe, O.M. Yaghi, “Hydrogen storage in microporous metal-organic frameworks,” Science 300(5622) (2003) 1127–1129. 133. M. Eddaoudi, H. Li, B. Chen, T. Reineke, M. Fehr, D. Kelly, T.L. Groy, O.M. Yaghi, “Design and synthesis of metal-carboxylate frameworks with permanent microporosity,” Topics Catal. 9 (1999) 105–111. 134. R.A. Varin, S. Li, Z. Wronski, O. Morozova, T. Khomenko, “The effect of sequential and continuous high-energy impact mode on the mechano-chemical synthesis of nanostructured complex hydride Mg2FeH6,” J. Alloys Compd. 390 (2005) 282–296. 135. Y. Gogotsi, ed., “Nanomaterials handbook,” (2006), Taylor & Francis, CRC Press, pp. 800. 136. R.H. Perry, D.W. Green, J.O. Maloney, eds., “Perry’s chemical engineers handbook,” 50 Ed. (1984), McGraw-Hill, New York, pp. 8–32. 137. R.W. Rydin, D. Maurice, T.H. Courtney, “Milling dynamics. Part 1: attritor dynamics: Results of cinematographic study”, Met. Trans. A 24A (1993) 175–185. 138. D.R. Maurice, T.H. Courtney, “The physics of mechanical alloying: A first report,” Met. Trans. A 21 (1990) 289–303. 139. C. Suryanarayana, “Mechanical alloying and milling,” Prog. Mater. Sci. 46 (2001) 1–184. 140. D.R. Maurice, T.H. Courtney, “Modeling of mechanical alloying: Part 1. Deformation, coalescence, and fragmentation mechanisms,” Metall. Met. Trans. A 25(1) (1994) 147–148. 141. G. Roy, “Orientation of principal stress components,” Exp. Tech. 18 (1994). 37–40. 142. G. Roy, J. Neima, Z.S. Wronski, R.A. Varin, “X-ray diffraction study of stress in a magnesium-hydrogen system produced by high-energy milling of powders,” Mater. Sci. Forum 539–543 (2007) 2713–2718. 143. S.N. Zhurov, “Mechanochemistry: The mechanical activation of covalent bonds,” Inter. J. Fracture Mech. 1 (1965) 311.

80

1

Introduction

144. C.G. Stelly, C.C. Dormeval, “Adiabatic shear band phenomena, In L.E. Murr, M.A. Meyers, Editors., Metallurgical applications of shock-wave and high-rate phenomena,” (1998), Marcel Dekker, New York-Basel. pp. 607–632. 145. J. Eckert, L. Schlutz, E. Hellstern, K. Urban, “Glass forming range in mechanically alloyed NiZr and the influence of the milling intensity,” J. Appl. Physics 64 (1988) 3224. 146. R.A. Varin, T. Czujko, Z.S. Wronski, “Particle size, grain size and g-MgH2 effects on the desorption properties of nanocrystalline commercial magnesium hydride processes by controlled mechanical milling,” Nanotechnology 17 (2006) 3856–3865. 147. F. Bernard, E. Gaffet, Y. Champion, N. Boudarina, A. Ustinov, “Correlation between ball milling conditions and planar effects on Cu-nanostructured powders,” J. Physics IV France 12 (2002) 6. 148. Z.S. Wronski, G.J.C. Carpenter, P. Kalal, “An integrated characterization approach for ranking nickel hydroxides designed for high-performance positive electrodes in batteries for electric vehicles,” Electrochem. Soc. Proc. 96-14 (1996) 177–188 (Exploratory R&D of Batteries for Electric and Hybrid Vehicles, A.R. Langrebe and B. Serosati, eds.) 149. C. Delmas, C. Tessier, “Stacking faults in the structure of nickel hydroxide: a rationale of its high electrochemical activity,” J. Mater. Chem. 7(8) (1997) 1439–1443. 150. Y.Y.C. Lou, X. Zhang, L. Ma, B. Xia, “Comparative study on microstructure of β-Ni(OH)2 cathode material for Ni-MH battery,” Sci. China Ser. E: Technol. Sci. 49 (3) (2006) 297–312. 151. G.J.C. Carpenter, Z.S. Wronski, “Nanocrystalline NiO and NiO-Ni(OH)2 composite powders prepared by thermal and mechanical dehydroxylation of nickel hydroxide,” Nanostructured Mater. 11(1) (1999) 67–80. 152. Z.S. Wronski, “Mechano-chemical activation of powders used in electrochemical power sources,” In 50th ISE Mtg and Symposium on Electrochemical Power Sources after 200 years, Pavia, Italy. (1999). International Electrochemistry Society, Geneva, Switzerland. 153. H. Bakker, G.F. Zhou, H. Yang, “Mechanically driven disorder and phase transformations in alloys,” Prog. Mater. Sci. 39 (1995) 159–241. 154. E. Hellstern, H.J. Fecht, Z. Fu, W.L. Johnson, “Structural and thermodynamic properties of heavily mechanically deformed Ru and AlRu,” J. Appl. Phys. 65 (1989) 305–310. 155. L. Zaluski, P. Tessier, D.H. Ryan, C.B. Donner, A. Zaluska, J.O. Strom-Olsen, “Amorphous and nanocrystalline Fe-Ti prepared by ball milling,” J. Mater. Res. 8 (1993) 3059–3068. 156. N. Eliaz, D. Eliezer, “An overview of hydrogen interaction with amorphous alloys,” Adv. Perform. Mater. 6 (1999) 5–31. 157. R.A. Varin, T. Czujko, “Overview of processing of nanocrystalline hydrogen storage intermetallics by mechanical alloying/milling,” Mater. Manuf. Proc. 17 (2002) 129–156. 158. R.A. Varin, Ch. Chiu, T. Czujko, Z. Wronski, “Feasibilty study of the direct mechanochemical synthesis of nanostructured magnesium tetrahydroaluminate (alanate) [Mg(AlH4)2] complex hydride,” Nanotechnology 16 (2005) 2261–2274. 159. J.M.B.van Colbe, M. Felderhoff, B. Bogdanovic´, F. Schüth, C. Weidenthaler, “One-step direct synthesis of a Ti-doped sodium alanate hydrogen storage material”, Chem. Commun. (2005) 4732. 160. Z. Wronski, R.A. Varin, Ch. Chiu, T. Czujko, A. Calka, “Mechanochemical synthesis of nanostructured chemical hydrides in hydrogen alloying mills,” J. Alloys Compd 434–435 (2007) 743–746. 161. Z.S. Wronski, R.A. Varin, Ch. Chiu, T. Czujko, “Mechanosynthesis of nanocrystalline MgB2 ceramic powders in hydrogen alloying mills via amorphous hydride intermediate,” Adv. Sci. Technol. 45 (2006) 309–314. 162. A. Andreasen, PhD thesis (2005) Risø National Laboratory, Roskilde, Denmark. 163. R.B. Schwarz, “Hydrogen storage in magnesium-based alloys,” MRS Bull. 24(11) (1999) 40–44. 164. D.A. Porter, K.E. Easterling, “Phase Transformations in Metals and Alloys,” 2nd ed., CRC Press, Boca Raton, (1992). 165. H. Kissinger, “Reaction kinetics in differential thermal analysis,”Analytical Chem. 29 (1957) 1702–1706.

References

81

166. N. Bazzanella, R. Checchetto, A. Miotello, “Catalytic effect on hydrogen desorption in Nb-doped microcrystalline MgH2,” Appl. Phys. Lett. 85 (2004) 5212–5214. 167. K.F. Kelton, “Analysis of crystallization kinetics,” Mater. Sci. Eng. A 226–228 (1997) 142–150. 168. M.H. Mintz, Y. Zeiri, “Hydriding kinetics of powders,” J. Alloys Compd. 216 (1994) 159–175. 169. T.R. Jensen, A. Andreasen, T. Vegge, J.W. Andreasen, K. Ståhl, A.S. Pedersen, M.M. Nielsen, A.M. Molenbroek, F. Besebbacher, “Dehydrogenation kinetics of pure and nickeldoped magnesium hydride investigated by in situ time – resolved powder X-ray diffraction,” Int. J. Hyd. Ener. 31 (2006) 2052–2062. 170. G. Barkhordarian, T. Klassen, R. Bormann, “Effect of Nb2O5 content on hydrogen reaction kinetics of Mg,” J. Alloys Compd. 364 (2004) 242–246. 171. G. Barkhordarian, T. Klassen, R. Bormann, “Kinetic investigation of the effect of milling time on the hydrogen sorption reaction of magnesium catalyzed with different Nb2O5 contents,” J. Alloys Compd. 407 (2006) 249–255. 172. G. Liang, J. Huot, S. Boily, A. Van Neste, R. Schulz, “Hydrogen storage properties of the mechanically milled MgH2 –V nanocomposite,” J. Alloys Compd. 291 (1999) 295–299. 173. J. Huot, G. Liang, S. Boily, A. Van Neste, R. Schulz, “Structural study and hydrogen sorption kinetics of ball-milled magnesium hydride,” J. Alloys Compd. 293–295 (1999) 495–500. 174. S. Bouaricha, J.P. Dodelet, D. Guay, J. Huot, S. Boily, R. Schulz, “Hydriding behavior of Mg–Al and leached Mg–Al compounds prepared by high-energy ball-milling,” J. Alloys Compd. 297 (2000) 282–293. 175. J.A. Kennelley, J.W. Varwig, H.W. Myers, “Magnesium-hydrogen relationships,” J. Am. Chem. Soc. 64 (1960) 703–704.

Chapter 2

Simple Metal and Intermetallic Hydrides

2.1 2.1.1

Mg/MgH2 Crystallographic and Material Characteristics

Reaction of hydrogen with the elemental Mg is one of the most widely researched reactions in the field of solid-state hydrogen storage. The Mg–H system is quite simple as shown in the binary phase diagram in Fig. 2.1. At moderate hydrogen pressures the only hydride phase existing in equilibrium with Mg is magnesium dihydride, MgH2, more commonly referred to as “magnesium hydride.” Table 2.1 shows the crystal structure data of the phases existing in the Mg–H system. Pure Mg has a hexagonal crystal structure and its hydride has a tetragonal lattice unit cell (rutile type). The low-pressure MgH2 is commonly designated as β-MgH2 in order to differentiate it from its high-pressure polymorph, which will be discussed later. Figure 2.2 shows the crystal structure of β-MgH2 where the positions of Mg and H atoms are clearly discerned. Precise measurements of the lattice parameters of β-MgH2 by synchrotron X-ray diffraction yielded a = 0.45180(6) nm and c = 0.30211(4) nm [2]. The powder diffraction file JCPDS 12–0697 lists a = 0.4517 nm and c = 0.30205 nm. The density of MgH2 is 1.45 g/cm3 [3]. Noritake et al. [2, 4] investigated the bonding nature of MgH2 employing the maximum entropy method and synchrotron radiation powder data. They found that the bonding nature of hydrogen in MgH2 was quite complex, consisting of the mixture of ionic and covalent bonding. As stated by the authors, there are weak but significant covalent bonds between Mg and H as well as between H and H. The weak covalency of the Mg–H bond may be advantageous on hydrogenation/ dehydrogenation performance. Modeled charge density distribution revealed that the ionic charge of Mg and H can be represented as Mg1.91+ and H0.26−, respectively. It means that Mg is ionized almost as Mg2+, while hydrogen is very weakly ionized. They also found weak but significant bonds between Mg and H as well as between H and H. Under increasing hydrogen pressure, substantial changes occur in the Mg–H system. Bastide et al. [5] investigated the behavior of MgH2 phase under high pressures up to 80 kbar and found that at ambient temperature (20°C) and 80 kbar of R.A. Varin et al., Nanomaterials for Solid State Hydrogen Storage, DOI: 10.1007/978-0-387-77712-2_2 © Springer Science + Business Media, LLC 2009

83

84

2

0

1

675

Simple Metal and Intermetallic Hydrides

Weight Percent Hydrogen 2 3 4

650⬚C

6

7 8

P = 25 MPa

L + H2 643⬚C

L

5

625 (Mg) + H2

575

566⬚C

525 MgH2

Temperature, ⬚C

(Mg)

475

425

375 0 Mg

10

20

40 50 30 Atomic Percent Hydrogen

60

70

Fig. 2.1 Mg–H binary phase diagram. Reprinted with permission of ASM International. All rights reserved. www.asminternational.org [1]

t.1 t.2 t.3 t.4 t.5 t.1 t.2 t.3 t.4 t.5

Table 2.1 Crystallographic data of the phases in the Mg–H system [1] Phase

Composition at.% H

l

Space group

Strukturbericht designation

Prototype

(Mg) MgH2

0–11 66.7

hP2 tP6

P63/mmc P42/mnm

A3 C4

Mg TiO2 (rutile)

Fig. 2.2 The crystal structure of b-MgH2 [2]

2.1

Mg/MgH2

85

pressure the β-MgH2 phase (designated α-MgH2 in the original paper [5]) transformed partially into another polymorphic phase γ-MgH2 forming a mixture (β + γ)-MgH2. X-ray diffraction (XRD) studies showed that γ-MgH2 has an orthorhombic unit cell structure of α-PbO2 type with the lattice parameters a= 0.453 nm, b= 0.544 nm, and c = 0.493 nm (this phase is included in the powder diffraction file JCPDS 35–1184). They also found another metastable phase δ-MgH2 (designated β-MgH2 in the original paper [5]), which upon release of pressure transformed into γ-MgH2. They claimed that above 350°C, γ-MgH2 was supposedly transformed into the equilibrium β-MgH2 phase. However, as will be shown later, the γ-phase decomposes above 300°C. Hydrogen absorption and desorption characteristics of two different commercial MgH2 powders were investigated in our laboratories. The first powder was purchased from Degussa-Goldschmidt, sold under the trade name “Tego Magnan.” Its average purity claimed by the supplier is ~95% (the remaining Mg) but we found that the powder was slightly more inhomogeneous than the ABCR (a German company) powder. The theoretical purity-corrected hydrogen content is ~7.2 wt.%. The powder particle size was provided by the supplier as 325 mesh. Scanning electron micrograph of the morphology of the Tego Magnan powder is shown in Fig. 2.3a. The size of the powders was calculated as the particle equivalent circle diameter (ECD) as described in Sect. 1.4.3 (~260 to 650 individual particles were usually analyzed). The size distribution is plotted in Fig. 2.3b. It is observed that the ECD distribution is well fitted by a lognormal curve (marked by a solid line in Fig. 2.3b) as confirmed by a quantitative statistical analysis. It must be pointed out that a lognormal ECD distribution has been a common characteristic of all MgH2 powders either commercial or synthesized from the elemental Mg which we have investigated so far. The evidence will be presented throughout this book. From the statistical point of view, a lognormal distribution of the powder particle sizes seems to be an indicator that two populations of particle sizes are present in a powder, one small and the other large, whose closely spaced normal distributions partially overlap giving in effect a lognormal distribution curve. Hence, the calculated mean particle size (ECD) and its standard deviation (SD) should only be considered as an estimate since according to the statistical point of view the concept of the mean value is valid only for normal (Gaussian) distribution. Nevertheless, the mean ECD is still a useful estimate for comparing powders subjected to various processing, e.g., ball milling. The mean value of the particle size of as-received Tego Magnan MgH2 is 36 µm with SD = ±16 µm (Fig. 2.3b). The grain (crystallite) size and lattice strain of the as-received Tego Magnan MgH2 were determined from the broadening of the X-ray diffraction (XRD) peaks using a Cauchy/Gaussian approximation as described in Sect. 1.4.3. The estimated value of ~67 nm (no lattice strains) [6] is at the border of nanocrystallinity if one defines it as the grain size smaller than 100 nm [7]. Our result correlates very well with the grain size of 78 nm reported very recently by Kojima et al. [8] for their commercial MgH2. Both results suggest that certain commercial varieties of MgH2 could be subjected to either ball milling or other type of postdeformation in the proprietary manufacturing process, which results in the final nearly nanosize grains (crystallites).

86

2

Simple Metal and Intermetallic Hydrides

Fig. 2.3 (a) Scanning electron micrograph of the morphology of as-received Tego Magnan MgH2 powder and (b) powder particle size distribution (equivalent circle diameter, ECD)

The second type of MgH2 powder was purchased from ABCR GmbH & Co. KG, sold under the trade name MG-5026. Its average purity claimed by the supplier is ~98% (remaining Mg). For simplicity, it will be referred to hereafter as the “ABCR powder.” Volumetric desorption tests were very reproducible, giving the average value of desorbed hydrogen equal to ~7.5 wt.% which is nearly identical to the theoretical purity-corrected capacity of 7.51 wt.%H2 at 98% purity. This testifies to

2.1

Mg/MgH2

87

a high homogeneity of the powder. Since its morphology and particle size (ECD) distribution are identical to those shown in Fig. 2.3, they are not shown here. Its mean ECD value is 41 µm with SD = ±21 µm. Indeed, particle size is very close to that of Tego Magnan. However, the grain (crystallite) size determined from the XRD peak broadening is ~300 nm (no lattice strain). This is a major microstructural difference between ABCR and Tego Magnan MgH2 powders.

2.1.2

Hydrogen Storage Characteristics of Commercial Mg and MgH2

2.1.2.1

Absorption

Hydrogenation/dehydrogenation studies of commercial Mg date almost 50 years back to the pioneering studies of Stampfer et al. [9] and Kennelley et al. [10]. The former authors reported the enthalpy of MgH2 formation being in the range of –70.8 to –72.6 kJ/mol and corresponding entropy as –127.3 to −130.6 J/molK. The latter authors reported the heat (enthalpy) of formation of MgH2 as −74.1 ± 2.9 kJ/mol. Vigeholm et al. [11–16] carried out very thorough investigations of hydrogen absorption/desorption of commercially pure Mg. They concluded that any comminuted Mg product with characteristic dimensions of less than approximately 75 µm and with impurity concentrations and compositions typical of normal Mg products will absorb hydrogen readily above 300°C at pressures exceeding the equilibrium level (i.e., plateau pressure on a corresponding pressure–composition–temperature (PCT) curve). They claimed that activation was in general either unnecessary or required only a few cycles under normal absorption conditions. The highest absorption rate was observed at an optimal temperature of around 340–350°C. Increasing the temperature above this value had slightly adverse effect on absorption kinetics. At ~263°C and 1.2 MPa hydrogen pressure, the magnesium powder absorbed ~2.3 wt.%H2 in roughly 6,000 s. At ~340°C, the same powder absorbed ~5.4 wt.%H2 in roughly 3,600 s. At the range of absorption temperatures up to ~340°C, the increase of pressure above 2 MPa increased the initial rate of kinetics but simultaneously reduced the degree of conversion of Mg into MgH2. In general, in order to obtain the degree of reaction close to 100% (i.e., Mg completely converted to a nearly single phase MgH2), the absorption temperature had to be higher than 400°C at a moderate hydrogen pressure of about 2.0–2.5 MPa. For example, in the initial absorption process at 400°C (after initial activation procedure by annealing in helium for 1 h), it was observed that the lower the hydrogen pressure, the closer the ultimate reaction approached stoichiometric hydride: 97.6% at 2.0 MPa, 95.5% at 3.2 MPa, and 81.5% only at 4.8 MPa [16]. In absorption process, the reaction of magnesium with hydrogen is a nucleation and growth mechanism where the nucleation rate is pressure dependent. They estimated the enthalpy and corresponding entropy of MgH2 formation as −70.0 kJ/mol and −126 J/mol K, respectively.

88

2

Simple Metal and Intermetallic Hydrides

More recently, very detailed thermodynamic studies of the Mg–H system were carried out by Bogdanovic´ et al. [17]. They specifically focused their attention on the possible effect of particle size on thermodynamics. Fractions with particle sizes of 25–40 µm and 70–100 µm were chosen for PCT determinations. Prior to all measurements, Mg powder fractions were activated. However, the range of selected temperatures 400–520°C was rather high. The hysteresis between absorption and desorption isotherms was very small. The reaction enthalpy and entropy for different hydrogen loadings and temperatures in the plateau region were determined using the Van’t Hoff relationship. A summary of their results is listed in Table 2.2. Thermodynamic parameters calculated for a hydrogen content of 50% (average value) are marked (calc.). It is quite clear that within the range of investigated particle sizes and temperature there is no dependence between particle size and thermodynamic parameters. In our laboratory, absorption experiments were carried out on an unmilled but activated ABCR MgH2 powder. In the preliminary stage, the powder was heated up to 350°C in a volumetric Sieverts-type apparatus (Sect. 1.4.2) under a preventing pressure of 3.5 MPa (so as not to allow desorption), held for about 30–45 min to stabilize temperature, and then subjected to the following desorption/absorption cycles: 1. Desorption at 350°C; annealing at 350°C/pre-vacuum/15 min; cooling to 325°C under pre-vacuum; absorption at 325°C/1.2 MPa/4,750 s; 2. Desorption at 350°C; annealing at 350°C/pre-vacuum/15 min; cooling to 300°C under pre-vacuum; absorption at 300°C/1.2 MPa/4,750 s; 3. Desorption at 350°C; annealing at 350°C/pre-vacuum/15 min; cooling to 275°C under pre-vacuum; absorption at 275°C/1.2 MPa/4,750 s; 4. Desorption at 350°C; annealing at 350°C/pre-vacuum/15 min; cooling to 250°C under pre-vacuum; absorption at 250°C/1.2 MPa/4,750 s. Consequently, the powder which absorbed at 250°C went through four full cycles as described above. Figure 2.4a shows typical kinetic curves of absorption at various temperatures. The absorption curves shown in Fig. 2.4a are quite similar to the ones reported by Vigeholm et al. [12] under the same 1.2 MPa hydrogen pressure as the one used in our work. In our case, about 4.5 wt.%H2 is absorbed at 325°C in ~4,500 s vis-à-vis about 4.5 wt.%H2 at 312°C absorbed in 4,000 s as reported by Vigeholm et al. At 263°C, Vigeholm et al. reported ~2.3 wt.%H2 absorbed in ~6,000 s which compares favorably with ~2 wt.%H2 absorbed at 250°C in 4,500 s in our experiments (Fig. 2.4a).

Table 2.2 Dependence of formation enthalpy and entropy of MgH2 (average values) upon the particle size of Mg powders used [17] ΔHf (calc.) ΔSf(calc.) Particle size ΔSf(J/molK) (kJ/mol) (J/mol K) Mg (µm) Temp. (°C) ΔHf (kJ/mol) 25–40 70–100

400–520 400–520

−72.6 ± 1.7 −74.7 ± 1.7

−132.2 −134.7

−72.9 −72.9

−132.3 −132.3

2.1

Mg/MgH2

89

5.00

1-250ⴗC 2-275ⴗC 3-300ⴗC 4-325ⴗC

Absorbed H2 (wt.%)

4.50

4

4.00

2

3.50

3

3.00 2.50 2.00

1

1.50 1.00 0.50 0.00 0

1000

a

2000 Time (s)

3000

4000

−3 −4

y = -64703x + 7.0755 R2 = 0.9969

ln k

−5 −6 −7 −8 −9 0.00019

b

0.0002

0.00021

0.00022

0.00023

0.00024

1/RT

Fig. 2.4 (a) Absorption kinetic curves of unmilled but activated and cycled ABCR powder and (b) estimate of apparent activation energy of absorption from the Arrhenius plot of ln k vs. 1/RT using data for all four temperatures: 250, 275, 300, and 325°C (EA ~65 kJ/mol). Coefficient of fit R2 = 0.997

Fernández and Sánchez [18] investigated the kinetics of hydrogen absorption and desorption by activated magnesium powder (several cycles of hydrogen absorption/desorption at 375°C) using a volumetric technique. They pointed out that formation/decomposition of metal hydrides comprises a number of steps taking place in series: H2 transport to the surface, H2 dissociation, H chemisorption, surface– bulk migration, H diffusion and nucleation, and growth of hydride/metal (absorption/desorption) phases. A summary of the detailed mechanisms taking place during nucleation and growth of MgH2 upon absorption and their dependence on absorption pressure (temperature) was provided by Friedlmeier and Groll [19]. Within the framework of the nucleation and growth model, they divided an isothermal absorption event into two cases depending on the applied hydrogen pressure as visualized in Fig. 2.5.

90

2

Mg

b

Mg

1

0

a

b

a

2nd Phase

a

1st Phase

1

Mg

2nd Phase

b

1st Phase

Mg

Simple Metal and Intermetallic Hydrides

0

t

t

b

Fig. 2.5 Schematic of the cross-section of an Mg particle at two different times and typical α = f(t) curves for hydriding at (a) pappl ≈ ppl and (b) pappl >> ppl (pappl – applied hydrogen pressure at T = const; ppl – plateau pressure at the same T = const; α – fraction transformed; β – β-MgH2 hydride) (adapted from [19])

Figure 2.5a shows that for an applied hydrogen pressure close to the equilibrium plateau pressure, ppl, the rate-controlling step is the chemical reaction which is proportional to the Mg: b-MgH2 interface area. The absorption curve is then sigmoidal. In this case, only a few hydride grains nucleate and they subsequently grow to relatively large sizes until a completely closed hydride layer is formed within the particle (first phase in Fig. 2.5a). The situation is similar to the solidification from the melt with a very small undercooling (a limited number of nuclei). The transformation mechanism in the first phase results in a large hydrogenated volume of a Mg particle. Normally, hydrogen diffusion takes place either through Mg phase or, alternatively, through the Mg:β-MgH2 interface. However, once the massive hydride layer is already formed, the phase transformation of Mg into β-MgH2 essentially stops because of the virtual impermeability of the β-MgH2 layer to hydrogen diffusion. It must be recalled now that the hydrogen diffusion rate in β-MgH2 is a few orders of magnitude lower than that in metallic Mg [20]. A very slow diffusion of hydrogen is now the rate-controlling reaction step (second phase in Fig. 2.5a). As shown in Fig. 2.5b, for high applied hydrogen pressures (pappl >> ppl), a large amount of β-MgH2 nuclei is formed similar to the nucleation during solidification from the melt under a large undercooling. The growth rate is much smaller as compared to that in Fig. 2.5a and only a very thin layer of β-MgH2 can be formed near the surface of each particle until the reaction stops (first phase in Fig. 2.5b). The first phase is most probably surface controlled (chemisorption). The complete chemical reaction is excluded since no sigmoidal curve is observed. The reaction in

2.1

Mg/MgH2

91

the first phase results in a small hydrogenated volume of a Mg particle, although the initial kinetics (the slope of the linear portion of the curve in Fig. 2.5b) is faster than that in Fig. 2.5b. Furthermore, as pointed out by Friedlmeier and Groll [19], the mechanisms delineated in Fig. 2.5 mean that, in practice, if one allows a sufficiently long period of time, higher hydriding capacities will be achieved for lower pressures. The higher the applied pressure, the faster the start of the hydriding reaction, but the lower the hydriding capacities achieved. All this leads to a crossing of isothermal α = f(t) curves obtained under various applied hydrogen pressure, the effect which is not observed for most hydride materials. Karty et al. [21] pointed out that the value of the reaction order η and the dependence of k on pressure and temperature in the JMAK (Johnson-Mehl-AvramiKolmogorov) equation (Sect. 1.4.1.2), and perhaps on other variables such as particle size, are what define the rate-limiting process. Table 2.3 shows the summary of the dependence of η on growth dimensionality, rate-limiting process, and nucleation behavior as reported by Karty et al. [21]. The absorption curves in Fig. 2.4a were analyzed by a linear fitting to the JMAK equation from which the reaction rate constant, k, and the reaction order,η, can be determined. The values of the reaction order η are listed in Table 2.4. At the lowest absorption temperature of 250°C, the η parameter is close to 2, and then it decreases to about 1, remaining close to this value at all the other temperatures. The different values of the reaction order suggest that different mechanisms are controlling the rates at various temperature ranges of absorption. It can be seen from Table 2.3, that the value of ~2 (1.83 in Table 2.4) suggests that the transformation of Mg to Table 2.3 Dependence of η in the JMAK equation [21] η Growth Rate limiting process dimensionality Diffusion 1 2 3 Interface transformation 1 2 3

Constant nuclei number 0.5 1 1.5 1 2 3

Table 2.4 The values of the reaction order η in the JMAK equation for the absorption experiments on the unmilled, activated, and cycled ABCR powder presented in Fig. 2.4 Absorption temperature (°C) Reaction order η 200 250 275 300 325

no absorption 1.83 0.91 0.74 0.84

Constant nucleation rate 1.5 2 2.5 2 3 4

92

2

Simple Metal and Intermetallic Hydrides

β-MgH2 upon absorption is either diffusion-rate limited occurring by two-dimensional growth at constant nucleation rate, or alternatively, it is interface-controlled transformation with one-dimensional growth at constant nucleation rate or two-dimensional growth at constant nuclei number. From the standpoint of the previous discussion of the transformation mechanisms shown in Fig. 2.5 and a nearly sigmoidal shape of the transformation curve in Fig. 2.4a, it is quite likely that at the lowest temperature of 250°C the Mg to β-MgH2 transformation during absorption is interface controlled. However, one-or two-dimensional growth is inconsistent with microscopic observations of three-dimensional hydride grain growth [19]. The η value close to 1 suggests either diffusion rate–limited transformation occurring by two-dimensional growth at constant nuclei number or, alternatively, interface-controlled transformation with one-dimensional growth at constant nuclei number. A word of caution is needed here. As pointed out by Karty et al. [21], the listing in Table 2.3 does not exhaust all possible η behavior, but it does illustrate that in general the determination of η does not provide a unique identification of the kinetics. However, the two extreme η values of 0.5 and 4 do more or less define unique processes. Karty et al. reported that the value of η for hydriding (absorption) of Mg2Cu-catalyzed Mg was equal to 0.5. In turn, η was equal to 1 for hydrogen absorption transformation of vapor-deposited Mg. The latter is in good agreement with the results shown in Table 2.4 for the temperature range higher than 250°C. Fernández and Sánchez [18] reported that a nucleation and growth (NG) mechanism as expressed by the JMAK equation with exponent values of η = 0.5–1 for absorption gave the best fit to the experimental data. The apparent activation energy of absorption in Fig. 2.4a was estimated through the Arrhenius plot of rate constant, k, with temperature (Fig. 2.4b). All the fitting procedures have already been described in Sect. 1.4.1.2. The most surprising result is a relatively low value of the apparent activation energy equal to ~65 kJ/mol. It must be pointed out that in view of the fact that the mechanisms of the transformation of Mg to β-MgH2 can depend on temperature range as clearly shown in Table 2.3, the estimated apparent activation energy of absorption reflects the occurrence of all those mechanisms. It must be mentioned that the elimination of the absorption curve at 250°C, which is characterized by the largest η = 1.83 (Table 2.3) from the estimation, does not change the final value of the apparent activation energy for absorption. Jensen et al. [22] compiled experimentally determined apparent activation energy values for hydrogenation/dehydrogenation of MgH2 from a number of published references. The apparent activation energy of the hydrogenation of cycle-activated Mg powder was reported as ~100 kJ/mol. For vapor-deposited Mg it was 138 kJ/mol. It must be pointed out that in our work absorption in Fig. 2.4a was conducted on a cycle-activated commercial MgH2 powder rather than on pure Mg. This factor could also contribute to the lowering of the apparent activation energy (Fig. 2.4b). The highest value of ~270–314 kJ/ mol was reported by Jensen et al. for deactivated (oxidized due to air exposure) Mg foil. According to Jensen et al., the major factor affecting the apparent activation energy and giving rise to a substantial scatter among values reported in the literature is the presence of MgO/Mg(OH)2 surface layers, retarding the diffusion

2.1

Mg/MgH2

93

of hydrogen into Mg (absorption) or out of MgH2/Mg (desorption). This factor will be discussed in more detail later on.

2.1.2.2

Desorption

As mentioned earlier, Kennelley et al. [10] carried out pioneering studies of dehydrogenation (desorption) of MgH2. They reported the equation for the equilibrium temperature–pressure relationship in the following form: log p (mm of Hg) = −3857/T + 9.78 where p is an equilibrium pressure and T is temperature in K for MgH2 decomposition. Later on, Stander [23] investigated the decomposition kinetics of MgH2 (92% purity) synthesized from the elemental Mg powder under 15 MPa hydrogen pressure at 400°C for 20 h. He reported that if the difference between the experimental dissociation pressures (e.g., 384 kPa) and the equilibrium dissociation pressure corresponding to T = const via PCT curve (e.g., 890 kPa at T = 368°C for MgH2 as determined by Stander) is large, the nuclei of Mg metal are formed virtually instantaneously at sites on the MgH2 particles and the rate of decomposition (kinetics) of MgH2 is controlled by movement of the cylindricalshaped Mg–MgH2 interface as given by the equation: 1 – (1 – α)0.5 = k1t

(2.1)

The apparent activation energy of decomposition estimated from the Arrhenius plot for k1 gave 120 kJ/mol. Conversely, Stander noticed that if the difference between the experimental dissociation pressures (e.g., 384 kPa) and the equilibrium (plateau) dissociation pressure corresponding to T = const (e.g., 404 kPa at T = 335°C for MgH2) is relatively small, then better fits were obtained with the model of random nucleation followed by one-dimensional growth or instantaneous nucleation followed by two-dimensional growth as given by the equation: (– ln (1 –α))0.5 = k2t

(2.2)

The Arrhenius plot for k2 yielded the apparent activation energy of 126 kJ/mol. Since a higher desorption temperature is usually related to a higher equilibrium pressure via Van’t Hoff equation, there is a tendency to attribute the kinetic effects to the difference in temperatures when desorption temperature is being changed from a relatively low value to a high value. However, one must remember that the underlying factor is still the difference between the experimental desorption pressure and the equilibrium dissociation pressure. Stander [23] concluded that slow movement of the phase boundary is the ratecontrolling step at pressures appreciably different from the equilibrium pressure. At pressures near the equilibrium pressure nucleation seems to play a role. The difference in activation energy for the two mechanisms cannot be regarded as being significant. Furthermore, as the reaction proceeds, nucleation will become less important. Eventually he concluded that nucleation is kinetically unimportant in the decomposition

94

2

Simple Metal and Intermetallic Hydrides

of MgH2. With regard to desorption, Vigeholm et al. [12] observed that the increase of desorption pressure from 0.15 MPa to over 1 MPa dramatically reduced the rate of desorption. Complete decomposition required high temperatures (in excess of 390°C to ensure completion in less than 10 min) and pressures below 100 kPa. In general, they found that during desorption the reversed nucleation took place. They also found that the reaction rates were not influenced significantly by the number of absorption/ desorption cycles within 150 cycles but the absorbed/desorbed amount of hydrogen dropped to 50–60% of the stoichiometric value up to 500 cycles. Figure 2.6a shows typical kinetic curves of first desorption carried out in a Sieverts-type apparatus at the initial hydrogen pressure of 0.1 MPa (atmospheric pressure of 1 bar) for the as-received, nonmilled, and nonactivated Tego Magnan powder. For each temperature, a fresh load of sample was desorbed. At each temperature in Fig. 2.6a, the desorption process is complete with 100% of MgH2 des-

Hydrogen desorbed [wt.%]

9

Desorption

8 7

4 3

6

1

2

5 4 1-350ⴗC 2-375ⴗC 3-400ⴗC 4-420ⴗC

3 2 1 0 0

1000

a

2000 Time [s]

3000

−3 −3.5

y = -120307x + 16.664 R2 = 0.9961

−4 ln k

−4.5 −5 −5.5 −6 −6.5 −7 0.00017

b

0.00018

0.00018 0.00019 1/RT

0.00019

0.0002

Fig. 2.6 (a) Desorption kinetic curves at various temperatures under initial hydrogen pressure of 0.1 MPa of the as-received, nonactivated, commercial MgH2 powder Tego Magnan and (b) the Arrhenius plot of the desorption rate for the estimate of the apparent activation energy, EA, using kinetics data for four temperatures: 350, 375, 400, and 420°C (EA ~120 kJ/mol). Coefficient of fit R2 = 0.996

2.1

Mg/MgH2

95

orbed (~7.6 wt.%). It must be pointed out that no desorption was observed at temperatures below 350°C after a testing time of a few hours. It is not to say that desorption would not occur after a much longer time at temperatures below 350°C. For instance, Vigeholm et al. [12] observed about 83% of hydrogen desorbed (~6.3 wt.%H2) after 15 h at ~320°C. As can be seen in Fig. 2.6a, at 350°C it takes 3600 s to completely desorb the powder (~7.6 wt.%H2 desorbed). However, an increase of temperature to 375°C and higher increases the rate of desorption enormously such that the powder is completely desorbed within 200–700 s range. The desorption kinetics observed in Fig. 2.6a is faster than that reported by Huot et al. [24] for an unmilled, commercial purity MgH2 (~95%). At 350 and 375°C, their unmilled MgH2 desorbed ~5.0 wt.%H2 in 2000 s and ~6.5 wt.%H2 in ~900 s, respectively. Figure 2.6b shows the Arrhenius plot for the estimate of the apparent activation energy of desorption, EA, using all the kinetic curves from Fig. 2.6a. The obtained value is ~120 kJ/mol with an excellent coefficient of fit R2 = 0.996. Another example of hydrogen desorption kinetic curves is presented in Fig. 2.7 for the second commercial powder ABCR. There is no desorption at 325°C but full desorption occurs relatively easily at 350°C and higher temperatures (~7.5 wt.%H2 desorbed). In this respect, the ABCR powder behaves in a manner similar to its Tego Magnan counterpart. At the first glance, the desorption rates at the same temperatures in Figs. 2.6 and 2.7 are not very different. Figure 2.7b shows the Arrhenius plot for the estimate of apparent activation of desorption, EA, using all the kinetic curves from Fig. 2.6a. The obtained value is ~168 kJ/mol with an excellent coefficient of fit R2 = 0.996. The activation energy value is much higher than the one obtained for the Tego Magnan powder. In order to shed more light on the noted discrepancy, we analyzed the reaction order parameters η in the standard JMAK equation. Table 2.5 compiles the obtained values of η for both analyzed powders. It is interesting that the highest value of ~3.6 for both powders corresponds to the lowest desorption temperature of 350°C. At higher temperatures, the η parameter substantially decreases to the average value of ~1.5. Apparently, the transformation mechanism is quite different at 350°C as compared to that at the higher temperature range. However, for both powders the values of η are remarkably close to each other in spite of the substantial differences in the apparent activation energy of desorption for Tego Magnan and ABCR powders. Unmilled, commercial Tego Magnan and ABCR powders in the as-received state were also subjected to activation by short cycling. In the first step, the powders were heated to 350°C under a hydrogen pressure of 2.7 MPa to prevent desorption during temperature stabilization. This step took about 15 min. Subsequently, the powders were subjected to activation, which consisted of three cycles and each cycle consisted of the following steps: (a) desorption at 350°C under initial atmospheric pressure of hydrogen (0.1 MPa) for approximately 60 min, (b) annealing under pre-vacuum at 350°C for 15 min, and (c) absorption at 350°C under hydrogen pressure of 2.7 MPa for 30 min. After activation, the powder was desorbed at a constant temperature in a volumetric Sieverts-type apparatus under atmospheric pressure of hydrogen. After each desorption, the same powder sample was re-absorbed at 350°C under a hydrogen pressure of 2.7–3.5 MPa for 15–30 min and this was followed by desorption at a

96

2

Hydrogen desorbed [wt.%]

8.00

Simple Metal and Intermetallic Hydrides

5

7.00 Desorption

4

6.00

3 5.00 2

4.00 3.00 2.00 1.00

1

1-325ⴗC 2-350ⴗC 3-375ⴗC 4-400ⴗC 5-420ⴗC

0.00 0

1000

a

2000

3000

4000

Time [s] −2 y = -167806x + 25.419 R2 = 0.9955

−3

ln k

−4 −5 -6 −7 −8 0.00017

b

0.000175

0.00018

0.000185

0.00019

0.000195

1/RT

Fig. 2.7 (a) Desorption kinetic curves at various temperatures under an initial hydrogen pressure of 0.1 MPa of the as-received, nonactivated commercial MgH2 powder ABCR and (b) the Arrhenius plot of the desorption rate for the estimate of the apparent activation energy, EA, using kinetics data for four temperatures: 350, 375, 400, and 420°C (EA ~168 kJ/mol). Coefficient of fit R2 = 0.996

Table 2.5 The values of the reaction order h in the JMAK equation for the desorption experiments on the unmilled and nonactivated powders presented in Fig. 2.6 and 2.7 Reaction order η Desorption temperature (°C) Tego Magnan® ABCR GmbH & Co. 350 3.60 3.65 375 1.76 1.45 400 1.39 1.73 420 1.56 1.31

2.1

Mg/MgH2

97

desired temperature. The kinetic curves of desorption are shown for Tego Magnan and ABCR powder in Figs. 2.8 and 2.9, respectively. A low rate desorption is already observed at 300°C, and a fair amount of desorption is observed at 325°C for both powders (Figs. 2.8a and 2.9a). This behavior is in a stark contrast to the lack of desorption at 325°C for a nonactivated powder (Fig. 2.7a). As mentioned previously, both nonactivated Tego Magnan and ABCR powders did not desorb hydrogen at temperatures lower than 350°C. It is quite obvious that the activation procedure allows desorption to occur at much lower temperatures. It must, however, be pointed out that after activation cycling, the maximum desorbed capacity of hydrogen, even

Hydrogen desorbed [wt.%]

7 Desorption

6

4

5

3 4

1-300ⴗC

3

2-325ⴗC 3-350ⴗC

2

2

4-375ⴗC

1 1 0

a

0

1000

2000 Time [s]

3000

−4 −4.5

y = -118468x + 16.746 R2 = 0.9964

−5

ln k

−5.5 −6 −6.5 −7 −7.5 −8 −8.5 0.00018 0.00019 0.00019 0.0002

b

0.0002 0.00021 0.00021 0.00022

1/RT

Fig. 2.8 (a) Desorption kinetic curves at various temperatures under an initial hydrogen pressure of 0.1 MPa of the as-received, activated commercial MgH2 powder Tego Magnan and (b) the Arrhenius plot of the desorption rate for the estimate of the apparent activation energy, EA, using kinetics data for four temperatures: 300, 325, 350, and 375°C (EA ~118 kJ/mol). Coefficient of fit R2 = 0.996

2 Hydrogen desorbed [wt.%]

98 7.00 Desorption

6.00 4

5.00

1 3

4.00

1-325ⴗC

2

3.00

ln k

2-350ⴗC

1

2.00

3-375ⴗC

1.00

4-400ⴗC

0.00

0

1000

a

b

Simple Metal and Intermetallic Hydrides

−3 −3.5 −4 −4.5 −5 −5.5 −6 −6.5 −7 −7.5 −8 0.00017

2000 3000 Time [s]

4000

y = -126254x + 18.056 R2 = 0.9749

0.00018

0.00019

0.0002

0.00021

1/RT

Fig. 2.9 (a) Desorption kinetic curves at various temperatures under an initial hydrogen pressure of 0.1 MPa of the as-received, activated commercial MgH2 powder ABCR and (b) the Arrhenius plot of the desorption rate for the estimate of the apparent activation energy, EA, using kinetics data for four temperatures: 325, 350, 375, and 400°C (EA ~126 kJ/mol). Coefficient of fit R2 = 0.975

at a high temperature range of 375–400°C in Figs. 2.8 and 2.9, seems to be lower than that of nonactivated MgH2 in Figs. 2.6 and 2.7. This drop in storage capacity after cycling was reported by Vigeholm et al. [13–15] and is definitely not related to the presence of residual impurity gases in the reacting hydrogen. They calculated that under the assumption that all available ppm impurities in high-purity H2 gas actually reacted with Mg, a maximum passivation of Mg per cycle would be about 0.002%, i.e., completely negligible. Most probably, the drop is related to the kinetics of the absorption stage of cycling due to applied hydrogen pressure being much higher than the equilibrium one as explained in Fig. 2.5, which will be discussed in more detail in connection to the PCT curve in Fig. 2.11. The values of the reaction order η in the JMAK equation for the desorption experiments on the unmilled and activated powders are presented in Table 2.6. On average, the values are slightly larger than those obtained for the nonactivated powders (Table 2.5), which may suggest slightly different transformation mechanisms in the nonactivated and activated powders (Table 2.3). Nonetheless, the η values are quite similar for both Tego Magnan and ABCR powders. As suggested by Dal Toè et al.

2.1

Mg/MgH2

99

Table 2.6 The values of the reaction order η in the JMAK equation for the desorption experiments on the unmilled and activated powders presented in Fig. 2.8 and 2.9 Reaction order η Desorption temperature (°C) Tego Magnan ABCR GmbH & Co. 300 3.52 – 325 2.44 2.63 350 2.04 2.44 375 1.84 1.86 400 – 1.57

[25], since MgH2 powder constitutes an agglomerate of particles with greatly varying sizes, the particles within a certain size range can decompose/compose involving slightly different rate-limiting processes and therefore different values of the reaction order η. Therefore, it seems that the use of free η values obtained from a double-logarithm fitting procedure as described in Sect. 1.4.1.2 is preferable to using fixed η values. Table 2.7 lists the estimated values of the apparent activation energy of desorption. Nonactivated and activated Tego Magnan as well as activated ABCR have almost the same average apparent activation energy of ~118–126 kJ/mol. The estimated average is quite close to that reported by Stander [23] (120 and 126 kJ/mol). Only for the nonactivated ABCR powder, the apparent activation energy of ~168 kJ/ mol is much higher than the others. Huot et al. [24] reported ~156 kJ/mol for the unmilled powder, which was estimated assuming a constant η = 3 in the JMAK equation. It was not specified whether the powder was activated or not. Jensen et al. [22] made an overview of experimentally determined apparent activation energies for desorption (dehydrogenation) of MgH2. They found that for activated MgH2 samples the activation energies of desorption were scattered within the range 120–160 kJ/mol. The samples intentionally exposed to air (called “deactivated” by the authors) exhibited very high values of the reaction rate parameter η, being on the order of 4–7, and extremely high apparent activation energies on the order of 230–300 kJ/mol. They attributed this behavior to the existence of magnesium oxide (MgO) surface layer formed on exposure to air. We also observed activation energies of this magnitude in so-called “aged” MgH2 powders, as will be discussed in Sect. 2.1.5. Friedrichs et al. [26] showed that a thin, amorphous magnesium hydroxide (Mg(OH)2) layer can form on the surface of nanocrystalline MgH2 powder after even a relatively short exposure to air. Apparently, moisture (H2O) in air can easily react with the surface of MgH2 according to the following hydrolysis reaction: MgH2 + 2H2O ⇒ Mg(OH)2 + 2H2

(2.3)

Furthermore, Varin et al. [27] reported that a long-term air exposure of nanocrystalline MgH2 for a few months leads to a massive transformation of a large fraction of MgH2 particles into the crystalline Mg(OH)2 phase. Such a reaction is especially detrimental to a nanocrystalline MgH2 synthesized by reactive ball milling as will be

100

2

Simple Metal and Intermetallic Hydrides

discussed in more detail in the following sections. Highly hydrolyzed samples of MgH2 may exhibit much higher values of activation energy, being within 230–300 kJ/mol range. Andreasen et al. [28] pointed out that the case of Mg variations in the reported apparent activation energies correlates with the presence of a MgO surface layer inhibiting diffusion of hydrogen. Thus, oxidized samples show large apparent activation energies and well-activated samples show smaller activation energies. However, the mere surface hydrolysis/oxidation layer model cannot explain these peculiar differences in the activation energy of nonactivated/activated Tego Magnan and ABCR powders in Table 2.7. We propose that it can be explained by a combined effect of hydroxide/oxide layer on the surface and, in addition, differences in grain size between both powders: ~67 and 300 nm for Tego Magnan and ABCR, respectively. A model explaining this behavior is shown in Fig. 2.10. It shows particles of Tego Magnan and ABCR powders before and after activation covered with a layer of hydroxide/oxide. It is logical to assume that the impurity layer is always broken/discontinuous along the intersections of grain boundary planes with the particle surface, which provide excellent paths for hydrogen penetration and diffusion into the bulk. Since Tego Magnan has much smaller, nearly nanosized grain, the layer on its particles will be heavily broken into small pieces and practically fully discontinuous owing to a large number of intersections of grain boundary planes with the surface (Fig. 2.10a). Activation, in principle, will not change this picture because the layer cannot be broken into even smaller pieces (Fig. 2.10b). Since hydrogen can easily penetrate and diffuse at many places, the activation energy of desorption remains nearly the same, at a relatively low level before and after thermal activation of Tego Magnan. In contrast, owing to much larger grain size, the ABCR powder has much less grain boundary plane intersections with the particle surface and the layer is semicontinuous (Fig. 2.10c), leading to more difficult penetration of hydrogen and high activation energy of desorption of nonactivated powder. After activation, the layer is heavily broken (Fig. 2.10d) allowing easy hydrogen penetration, which is associated with much lower activation energy of desorption. Figure 2.11a shows an example of PCT desorption curves for the activated commercial MgH2 Tego Magnan powder obtained at four different equilibrium temperatures. Figure 2.11b shows the Van’t Hoff plot of ln p vs. 1,000/T (where p is the pressure and T is the temperature), from which the enthalpy of decomposition of MgH2 of 71kJ/molH2can be obtained. This value is very close to all the values for the enthalpy of formation of MgH2 obtained during absorption as reported Table 2.7 Summary of apparent activation energies of desorption for unmilled Tego Magnan and ABCR powders estimated from the Arrhenius plot Apparent activation energy Coefficient of fit Kinetic curves at of desorption, EA R2 in the Arrhenius temperatures taken Powder (kJ/mol) equation for calculation (°C) Activation Tego Magnan 120 0.996 350, 375, 400, 420 No Tego Magnan 118 0.996 300, 325, 350, 375 Yes ABCR 168 0.996 350, 375, 400, 420 No ABCR 126 0.975 325, 350, 375, 400 Yes

2.1

Mg/MgH2

101

a

b

c

d

Fig. 2.10 (a) Particle of a Tego Magnan powder with the grain size of ~67 nm before activation and (b) after activation. (c) Particle of ABCR powder with the grain size of ~300 nm before activation and (d) after activation (dark envelope around particles represents Mg(OH)2/MgO)

previously in Sect. 2.1.2.1. It can be seen that the maximum hydrogen capacity of activated samples obtained in a PCT test (~5.5wt.%) is about 1.7 wt.% short of the purity-corrected hydrogen capacity of Tego Magnan powder (~7.2 wt.% at 95% purity). The same hydrogen capacity deficit is seen in Fig. 2.8a and 2.9a where the maximum amount of hydrogen desorbed from the activated MgH2, even at the highest desorption temperature, is always smaller than that desorbed from non-activated samples (Fig.2.6a and 2.7a). As mentioned previously, during activation the samples were subjected to absorption at 350° under 2.7 MPa pressure. In addition, intermediate re-absorption steps after each desorption were also conducted at the same temperature and similar pressure. One can notice in Fig. 2.11a that at 350°C the equilibrium (plateau) pressure is about 1MPa for desorption. Assuming a more or less similar plateau pressure for absorption, it means that all the re-absorption steps were conducted at the condition of applied pressure much higher than the plateau pressure (pappl >> ppl). As shown in Fig. 2.5b, under this condition a large amount of β-MgH2 nuclei is being formed, the growth rate is relatively small, and only a thin layer of β-MgH2 can be formed near the surface of each particle until the reaction stops, leading, in effect, to faster kinetics but lower hydriding capacities. Obviously, a smaller amount of absorbed hydrogen results in a smaller amount of desorbed hydrogen, as clearly seen in Fig. 2.11a. However, it must be pointed out that such a hydrogen deficit of practical maximum capacity does not affect the thermodynamic behavior of the powder during desorption, resulting in a very true enthalpy of decomposition/formation.

102

2

Simple Metal and Intermetallic Hydrides

10.00 Desorption Pressure (MPa)

2.77 MPa 1.92 MPa 1.11 MPa

1.00 0.565 MPa

400C 375C 350C 325C 0.10

0

1

2

a

3 4 5 6 H2 capacity (wt.%)

7

8

4 DHdes = 71 kJ/molH2

3.5 3

ln p

2.5 2 y = -8.5824x + 16.135 R2 = 0.9909

1.5 1 0.5 0 1.45

b

1.5

1.55 1.6 1000/T (1/K)

1.65

1.7

Fig. 2.11 (a) PCT desorption curves at various temperatures for the activated commercial MgH2 Tego Magnan powder; numbers indicate the average mid-plateau pressure. (b) The Van’t Hoff plot for finding the enthalpy and entropy of decomposition, which is equal to ~71 kJ/mol and~134 J/ mol K, respectively. Note excellent coefficient of fit R2 = 0.991 (p – pressure)

Figure 2.12 shows an example of a curve obtained in a differential scanning calorimeter (DSC), which illustrates general features of hydrogen desorption from a nonactivated MgH2 sample. A single and quite symmetrical endothermic peak characterizes hydrogen desorption. The shape of the peak for nonactivated MgH2 is quite narrow but very strong. The peak maximum is centered at 428.3°C (at heating rate 4°C/min), which is the most common range of the peak maximum temperatures for nonactivated MgH2 powders. Usually, most of the peak temperatures fall within the range 410–440°C (depending on the DSC heating rate).

2.1.3

Hydrogen Storage Characteristics of Mechanically (Ball) Milled MgH2

As discussed in Sect. 1.3, the major objective of mechanical (ball) milling is to obtain a substantial nanostructuring of hydrides (nanohydrides), which in effect can

2.1

Mg/MgH2

103

DSC / mW/mg 14

+

exo

[1] 428.3⬚C

12 10 8 6 4 2 0 200

250

300

350 400 Temperature / ⬚C

450

500

Fig. 2.12 DSC curve of a commercial, as-received and nonactivated Tego Magnan powder obtained under argon flow and a heating rate 4°C/min

improve their hydrogen storage properties. The first pioneering experiments on the synthesis of nanohydrides by ball milling were conducted in the early and mid1990s by the group led by Prof. J.O. Ström-Olsen at the McGill University in Montréal, Canada, in which a prominent role was played by L. Zaluski and A. Zaluska [29–33]. A parallel line of work on the same subject was initiated by the group led by O.N. Srivastava in India [34]. However, it must be pointed out that in technical terms the hydrides in early research were not physically ball-milled in order to induce nanocrystallinity. In reality, either intermetallic compounds were first synthesized by mechanical alloying, or pure metals (e.g., Mg) were simply ballmilled under argon and subsequently subjected to hydrogenation at the appropriate temperature and hydrogen pressure. Such a method is called in the present text “a two-step” method. In addition, those early works, especially in the group led by Prof. J.O. Ström-Olsen, overemphasized the effect of nanograins (crystallites) formed within the heavily milled powder particles on the hydrogen absorption/desorption properties and somehow marginalized the role of the reduction of particle size that occurs simultaneously with the decrease of nanograin (crystallite) size. Especially, for milled Mg, as well as for the Mg-based and other intermetallic compounds having nanograin microstructure, subsequent hydrogenation/dehydrogenation cycles at elevated temperatures should give rise to nanograin growth. Hence, it is hard to understand why improved hydrogen storage properties would, indeed, still require the nanosized grains which in all practical terms do not exist any longer. This very important issue will be discussed in more detail in the following sections.

2.1.3.1 Microstructural Evolution During Milling and Subsequent Cycling of Commercial MgH2 Powders In the context of this book, the name “commercial” MgH2 will be used to describe any MgH2 powder purchased from a supplier of chemicals, the prime examples of

104

2

Simple Metal and Intermetallic Hydrides

which are Tego Magnan from Degussa-Goldschmidt and ABCR from ABCR GmbH&Co.KG, which were already discussed in Sect. 2.1.1. The first attempts to ball-mill a commercial undoped MgH2 powder were undertaken by the group in the Hydro-Quebec Research Institute in Quebec, Canada [24, 35–36]. In contrast to the prior work in the Ström-Olsen’s group, the Hydro-Quebec authors at least pointed out that not only nanograins are being formed but also particle size reduction associated with an increase of the specific surface area (SSA) occurs simultaneously during ball milling. Nevertheless, no attempts to correlate the structural changes occurring during ball milling with hydrogen sorption properties were made. In our laboratories, we have carried out a thorough investigation of microstructural evolution occurring during ball milling of both commercial Tego Magnan and ABCR MgH2 powders. Figure 2.13 shows the evolution of XRD patterns as a function of controlled mechanical milling (CMM) of the commercial ABCR MgH2 powder under the HES57 mode in a hydrogen gas atmosphere in the magneto-mill Uni-Ball-Mill 5 (Sect. 1.3.2). It is clearly seen that, indeed, ball milling induces drastic changes in the microstructure of the milled powders. There are three important microstructural changes occurring upon milling of MgH2 that have already been described to a smaller or larger extent in a number of papers mentioned above. First, the breadth of the XRD peaks of β-MgH2 increases with increasing milling time. As discussed in Sect. 1.4.3, this is related to the formation of crystallites (nanograins) within the powder particles, which may be accompanied by the introduction of lattice strains. Table 2.8 lists nanograin size and lattice strain of β-MgH2 as a function of milling time estimated from the procedure described in Sect. 1.4.3. 30000

β-MgH2 γ-MgH2 Mg MgO s substrate

Counts

20000

s

s

s

As received 0.25 h

10000

1h 5h 10 h 20 h

0 30

40

50 60 Degrees 2-Theta

70

80

90

Fig. 2.13 Evolution of XRD patterns as a function of ball milling time of ABCR powder under the HES57 mode (high energy shearing; two magnets at five and seven o’clock positions) in hydrogen atmosphere under ~600 kPa pressure in the magneto-mill Uni-Ball-Mill 5

2.1

Mg/MgH2

105

Table 2.8 Grain size variations of b-MgH2 as a function of milling time of ABCR powder under HES57 mode from Fig. 2.13 Number of XRD Milling time (h) Grain size (nm) Strain R2 peaks 0 (as received 299 0 0.9989 4 0 (as received) 303 0 0.9936 6 0.9904 4 0.25 (15 min) 52 9.31 × 10−4 0.9954 3 1 30 3.16 × 10−3 0.9901 4 5 12 1.17 × 10−3 10 10 0 0.9999 3 0.9926 3 20 11 5.26 × 10−3

1000

Grain size (nm)

ABCR ABCR cycled after milling Tego Magnan

100

Average grain size after cycling

10

1 0

5

10 15 Milling time (h)

20

25

Fig. 2.14 Grain size of both commercial powders Tego Magnan and ABCR as a function of milling time in the magneto-mill Uni-Ball-Mill 5 under shearing and impact modes. For comparison, the level of the grain size measured after activation by a short cycling for the powders milled for the same duration is also shown (cycling scheme: heating to 325°C for ~15 min under 3.4 MPa H2 to prevent desorption-first desorption at 325°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 3.4 MPa H2 for 30 min/second desorption at 350°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 2.7 MPa H2 for 30 min)

As can be seen, the nanograin size is being reduced pretty fast with milling time such that barely after 15 min of milling it is already reduced approximately sixfold as compared to the original grain size of the as-received ABCR powder. After milling for approximately 5 h, the nanograin size is already saturated within the ~10–12 nm range and further milling for 10–20 h does not bring about any more changes in the nanograin size. It must be pointed out that the time to reach the saturation level of the β-MgH2 nanograin size depends on the milling mode, such as low-energy shearing (LES), high-energy shearing (HES), and impact, however when the milling process is carried out in the magneto-mill Uni-Ball-Mill 5 the saturation time is more or less within a few hours range regardless of the mode. The situation could be slightly different when using different type of mill, e.g., Spex or Fritsch, but there is no systematic study of a saturation behavior in these types of mills. Figure 2.14 shows

106

2

Simple Metal and Intermetallic Hydrides

a plot of grain size as a function of milling time for two commercial powders Tego Magnan and ABCR, which were ball-milled under various shearing and impact modes in the magneto-mill Uni-Ball-Mill 5. It is apparent that within the experimental scatter the grain size saturation level does not depend on the type of commercial MgH2 powder and the mode of milling. The strains of the β-MgH2 phase in Table 2.8 are minimal (1%). It seems that the strains reported by Huot et al. are rather overestimated. For clarification, it is to be pointed out that the XRD peaks of MgO observed in Fig. 2.13 arise because of the exposure of residual Mg present in the as-received MgH2 powder to air during powder handling for XRD tests. After milling, the residual Mg becomes nanostructured and in this state exhibits a very strong affinity to oxygen in air. Second, even after a short milling time of 15 min (0.25 h), the XRD peaks of an orthorhombic γ-MgH2 phase already appear on the pattern (Fig. 2.13). As pointed out by Schulz et al. [35, 36], the pressure increase due to the mechanical action of the milling balls produces the structural transformation of a tetragonal β-MgH2 into an orthorhombic γ-MgH2 which normally would occur under enormous static pressures of around 8 GPa (Sect. 2.1.1). Interestingly, the γ-MgH2 XRD intensities do not seem to increase measurably with increasing milling time beyond 1 h, which suggests that their volume fraction may saturate after a relatively short milling time under this specific milling mode. Huot et al. [24] estimated using the Rietveld refinement of an XRD pattern that the γ-MgH2 phase abundance was around 18% and did not increase with increasing milling time. They explained this apparent dilemma by two possible mechanisms. The first mechanism relies upon the occurrence of two competitive processes, one promoting the formation of the γ−phase due to mechanically driven transformation, and the other enhancing the formation of the β−phase due to thermally driven γ− β transformation. The alternative mechanism is based on the assumption that the increase of strain in the γ−phase beyond a certain level induces transformation of γ into β. It is, however, hard to decide unambiguously which mechanism is indeed true. However, as will be discussed later, our results show that the volume fraction of the γ−phase seems to be higher after milling for 100 h than that after 20 h. Third, parallel to the decrease of grain size there is always a decrease of the particle size of milled powders, as shown in Fig. 2.15. The particle size reduction occurs within a very similar time frame as does the reduction of grain size. As can be seen in Fig. 2.15a, barely after about 15 min the particle size is reduced from the initial ~40 to ~1 µm. Further prolonged milling up to 100 h may bring about very incremental particle size reduction down to ~0.6 µm (Fig. 2.15b). This behavior is of a very general nature and practically does not depend on the mode of milling and the type of MgH2 commercial powder as can be seen in Fig. 2.15. Almost identical time frame for grain and particle size variations as a function of milling time, as

2.1

Mg/MgH2

107

Particle size ECD (mm)

100 ABCR IMP68

ABCR HES57

Tego HES57

Tego IMP68

10

1

0.1 0

20

a

40

60

80

100

Milling time (h)

Particle size ECD (mm)

100 ABCR HES57

Tego HES57

Tego IMP68

10

1

0.1

b

ABCR IMP68

0

1

2

3

4

5

Milling time (h)

Fig. 2.15 Powder particle size vs. milling time for two commercial MgH2 powders Tego Magnan (Tego) and ABCR which were milled in the magneto-mill Uni-Ball-Mill 5 under shearing and impact modes. (a) Milling for up to 100 h and (b) enlarged section up to 5 h (HES57 – high-energy shearing with two magnets at five and seven o’clock positions; IMP68 – strong impact with two magnets at six and eight o’clock positions)

shown in Figs. 2.14 and 2.15, makes it rather difficult to identify unambiguously which factor is, indeed, governing hydrogen storage characteristics. This difficulty has led to a common belief that the grain size is mostly responsible for the observed enhancement of hydrogen storage properties, which is not necessarily the case as will be discussed later. Also, it must be pointed out that even after milling up to 100 h, which was the longest milling duration in the Uni-Ball-Mill 5, and regardless of the mode of milling, we never observed a contamination of MgH2 with Fe coming from the steel milling tools. The presence of Fe was not detected either by energy dispersive spectroscopy (EDS) or XRD measurements. Apparently, the content of Fe, if any, was below the detectability level of both analytical techniques. Recently, Ares et al. [37] reported that the concentration of Fe in the MgH2 milled in a steel vial increased to 4 wt.% after 700 h of milling in a Fritsch P5 planetary mill. They found

108

2

Simple Metal and Intermetallic Hydrides

that such a high concentration of Fe facilitated particle agglomeration and coldwelding. It is also interesting to note that Ares et al. found that the lattice strains of MgH2 milled in a steel vial were much lower than those observed in the material milled in a ceramic vial. In our work, a stainless steel vial was used, resulting in negligible lattice strains of MgH2 (Table 2.8). Thermal activation by desorption/absorption cycling at elevated temperatures of premilled MgH2 reverts some of the microstructural changes induced by milling. ABCR powders initially milled, as shown in Fig. 2.13, were subjected to the following cycling: heating to 325°C for ~15 min under 3.4 MPa H2 to prevent desorption/first desorption at 325°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 3.4 MPa H2 for 30 min/second desorption at 350°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 2.7 MPa H2 for 30 min. Figure 2.16 shows the XRD patterns of the premilled powders after the last absorption stage in a cycling scheme. By comparison with the XRD patterns of the milled powders in Fig. 2.13, noticeable changes occurred after cycling. First, the diffraction peaks of γ-MgH2 completely disappeared. Second, the diffraction peaks of the β-MgH2 phase have become much narrower. Table 2.9 shows the estimated grain sizes and lattice strains of β-MgH2. There has been substantial grain 40000

β-MgH2 Mg MgO s substrate

Counts

30000

20000

s

0.25 h

s

1h

10000

5h 10 h 20 h

0 30

40

50 60 Degrees 2-Theta

70

80

90

Fig. 2.16 XRD patterns of initially milled ABCR powders from Fig. 2.13 after short cycling: heating to 325°C for ~15 min under 3.4 MPa H2 to prevent desorption/first desorption at 325°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 3.4 MPa H2 for 30 min/second desorption at 350°C under 0.1 MPa H2 pressure for ~4,700 s/annealing under pre-vacuum at 350°C for 15 min/absorption at 350°C under 2.7 MPa H2 for 30 min

2.1

Mg/MgH2

109

Table 2.9 Grain size variations of premilled ABCR powder after short cycling Number of Milling Grain XRD peaks time (h) size (nm) Lattice strain R2 0 166 1.09 × 10-3 0.9865 3 0.25 47 0 0.8441 3 1 34 0 0.9171 3 5 40 0 0.9994 3 10 35 0 0.9992 3 20 43 0 0.9907 3

Particle size (ECD) (mm)

2.5 2 1.5 1 0.5 0 0

1

2

3

4

5

6

Number of cycles

Fig. 2.17 Particle size (ECD) of Tego Magnan powder milled for 100 h under the low-energy impact mode with one magnet and subsequently subjected to five cycles of desorption at 300°C under 0.1 MPa H2 pressure for ~4,700 s/absorption at 300°C under 2.5 MPa H2 for 60 min

growth, although the β-MgH2 grains can be still classified as nanosized (180°C, mixtures milled for 5 and 10 h exhibit three strong endothermic effects centered at around 271, 315 and 452°C but the one milled for 40 h shows only two endo effects at around 292 and 452°C.

3.2

Alanates

225 Mg(AlH4)2 NaCl Al b-MgH2

12000

Counts

8000

40h 4000 10h

5h 0 40

30

50 60 Degrees 2-Theta

70

80

90

Fig. 3.13 Evolution of XRD patterns as a function of milling time during MCAS of Mg(AlH4)2 from the mixture of NaAlH4 and MgCl2 DSC / mW/mg [1] 452.2⬚C

↓ exo 0.8

[2] 451.9⬚C

0.6

0.4

[3] 451.0⬚C

1. 5h 2. 10h 3. 40h

[2] 270.7⬚C [3] 291.6⬚C [1] 270.0⬚C [1] 314.8⬚C [2] 315.5⬚C

0.2 2

3 1

0 100

150

200

250

300 350 Temperature / ⬚C

400

450

500

Fig. 3.14 DSC curves of powders after milling for 5, 10 and 40 h upon heating to 500°C registered at the scan rate of 4°C/min

We analyzed the phase composition of the Mg(AlH4)2 + 2NaCl mixture during heating in a DSC apparatus by stopping the test at various temperatures and taking powder for XRD tests. Figure 3.15 shows XRD patterns of powder milled for 10 h and subsequently heated in a DSC apparatus to the selected temperatures shown in Fig. 3.15. The XRD pattern from the sample heated up to 180°C contains only Bragg

226

3 Complex Hydrides NaCl Al(Mg) Al b-MgH2 Al3Mg2

10000

8000 425ⴗC Counts

6000 350ⴗC 4000 295ⴗC 2000 180ⴗC 0 30

40

50 60 Degrees 2-Theta

70

80

90

Fig. 3.15 XRD patterns of powder milled for 10 h and then heated to 180, 295, 350 and 425°C in a DSC test in Fig. 3.14

peaks of NaCl which does not undergo any changes during heating, Al and small amount of β-MgH2. The absence of Bragg peaks of Mg(AlH4)2 shows that the thermal events seen in Fig. 3.14 around 140°C correspond to a decomposition of Mg(AlH4)2, synthesized after 10 h of milling (Fig. 3.13), into MgH2 and Al according to the reaction of (3.20a). The XRD pattern of sample heated up to 295°C contains the Bragg peaks of NaCl, the solid solution of Mg in Al designated Al(Mg), weak peaks of Al3Mg2 and very weak peaks of β-MgH2. Apparently, around the 271°C peak the majority of β-MgH2 decomposes in the endothermic reaction into free Mg and H2. Simultaneously, reactions of Mg and Al to form Al3Mg2 and Al(Mg) solid solution must have occurred. The reactions can be written as follows: β-MgH 2 + 2Al (+2NaCl) → Mg + H 2 + 2Al (+2NaCl) (225 − 340° C at 4° C/min) Mg + 2Al → 0.5Al3 Mg2 + 0.5Al(Mg) (225 − 340° C at 4° C/min)

(3.24a) (3.24b)

After DSC run up to the temperature of 350°C in Fig. 3.15, which is higher than the temperature of the second DSC peak maximum at ~ 315°C, the decomposition

3.2

Alanates

227

of the remnant MgH2 is completed and the microstructure consists of residual NaCl, large amount of Al3Mg2 intermetallic compound (high intensity XRD peaks) and large amount of Al(Mg) solid solution (high intensity XRD peaks). As such, two endothermic peaks at ~ 271 and ~ 315°C might arise either due to the decomposition of β-MgH2 and the formation of the Al3Mg2 intermetallic compound or alternatively, due to the two-step decomposition of β-MgH2. Mamatha et al. [117, 118] and Kim et al. [119] reported only a single endothermic peak with the maximum around 280–290°C for the mixture Mg(AlH4)2 + 2NaCl. Subsequently, the lattice parameters of Al(Mg) were calculated as shown in Table 3.1. It is clear that the lattice parameter of Al in the powders milled for 10 h is much larger after heating in DSC up 295, 350 and 425°C than that of the sample heated up to 180°C. The lattice expansion is consistent with the magnitude of the atomic radius of Al and Mg which is equal to 0.1432 and 0.1604 nm, respectively [121]. Fossdal et al. [114] claimed the formation of the Al(Mg) solid solution during low-temperature desorption of Mg(AlH4)2 below 180°C which is not confirmed here. Mamatha et al. [117, 118] never reported the formation of the Al(Mg) solid solution although Kim et al. [119] mentioned about increase in the lattice parameter of Al most probably due to formation of the Al(Mg) solid solution. The endo peak centered at 292°C for the mixture milled for 40 h in Fig. 3.14 is due to the decomposition of β-MgH2 and formation of Al3Mg2 and Al(Mg). It is to be pointed out that the DSC run of the 40 h milled powder at the scan rate of 4°C/min up to 350°C which is above the peak maximum at ~ 316°C in Fig. 3.14, resulted in the formation of the mixture of Al3Mg2 and Al(Mg) solid solution whose lattice parameter is also listed in Table 3.1 [120]. This clearly confirms that the formation of Al3Mg2 and Al(Mg) in the present work occurs during DSC heating from the elemental Mg and Al formed according to the reactions of (3.20a) and (3.20b). Finally, the high-temperature peak at ~ 452°C in Fig. 3.14 is due to the eutectic melting of the Al3Mg2 and Al(Mg) mixture according to the binary Mg–Al phase diagram [122]. Thermogravimetric analysis (TGA) showed [120] that the weight losses in the first and second step decomposition for the 5 h milled powder were 2.37 and 0.08 wt%, respectively. Corresponding weight losses for the 10 h milled powder were 2.14 and 0.02 wt%. Assuming the MCAS is completed (i.e., products have a 1 mol of Mg(AlH4)2 and 2 mol of NaCl), the theoretical hydrogen capacity in the mixture Table 3.1 Lattice parameters of Al and Al(Mg) solid solution in the as-milled and DSC tested samples Milling time (h)

Temperature of DSC test (°C)

10 As-milled 10 180 10 295 10 350 40 350 10 425 a Al(Mg) solid solution

Lattice parameter of Al (FCC) (nm) 0.4037 ± 0.0015 0.4039 ± 0.0015 0.4059 ± 0.0018a 0.4062 ± 0.0016a 0.4065 ± 0.0012a 0.4063 ± 0.0013a

228

3 Complex Hydrides

of Mg(AlH4)2 + 2NaCl would be ~ 3.97 wt% rather than 9.3 wt% as in a pure Mg(AlH4)2. From TGA analysis, the largest total weight loss ~ 2.45 wt%, which was observed for the 5 h milled powder, is much less than the theoretical value of ~ 3.97 wt%. A deficiency of ~ 1.5 wt% might be a combined effect of the partial decomposition of Mg(AlH4)2 during milling for 5 h and the underestimation of the weight loss in the second step (~ 270–300°C) due to the oxidation of Mg during TGA run under flowing nitrogen. The total TGA hydrogen desorption of only ~ 2.16 wt% for the 10 h milled powder is most probably due to the prior partial decomposition of Mg(AlH4)2 during milling. No weight loss observed in TGA in the 100–180°C range for the 40 h milled powder further confirms that Mg(AlH4)2 is fully decomposed after a prolonged milling time. In order to shed more light on the nature of thermal events close to 140°C in Fig. 3.14, we carried out DSC experiments at the scan rate of 4 and 20°C/min. Figure 3.16a shows the enlargement of the low temperature region up to 200°C of the DSC curve of powder ball milled for 10 h. Taking the flat DSC trace portion for the 40 h sample as the baseline (at this temperature range), one can invoke the existence of either two exothermic peaks or one endothermic peak due to the decomposition of Mg(AlH4)2. DSC traces at the scan rate of 20°C/min in Fig. 3.16b of one sample milled for 5 and one milled for 10 h, analyzed against the baseline, show the DSC humps which could be considered as endothermic events. In contrast, the other samples milled for 5 and 10 h show a diffuse trough that could be considered as an exothermic reaction. XRD analysis showed that samples heated up to 125 and 140°C still contained Mg(AlH4)2, residual NaCl, traces of β-MgH2 and the elemental Al rather than the Al(Mg) solid solution as mentioned earlier [120]. This in essence is almost the same phase content as that after ball milling for 10 h (Fig. 3.13). The presence of the traces of β-MgH2 indicates that the beginning of the decomposition of Mg(AlH4)2 starts around 125°C (at a scan rate 4°C/min). After a DSC run up to 150°C the microstructure contained only a negligible amount of Mg(AlH4)2. The absence of Mg(AlH4)2 is an indication of its complete decomposition. This analysis has clearly shown that approximately up to 180°C, Mg(AlH4)2 is already decomposed. However, the nature of this transformation as being either endothermic or exothermic could not be unambiguously established. Most probably, this situation arises owing to a very small enthalpy of decomposition of Mg(AlH4)2 to MgH2, Al and hydrogen (20a) which is cited in the literature as being equal to ~ 1.7 kJ/molH [118]. It was also predicted theoretically by Zhou [123] as being equal to 2.2 kJ/mol which agrees very well with the value obtained in [118]. It is quite possible that at such a small value of the enthalpy of decomposition of Mg(AlH4)2, which in all practical terms is almost close to zero, even small microstructural fluctuations always present in the powders synthesized by mechanochemical reactions may change the character of the decomposition reaction from endothermic to exothermic. In addition, as pointed out by Mamatha et al. [118] owing to such a small enthalpy of decomposition of Mg(AlH4)2 the thermodynamic stability of this alanate is far below that suitable for reversible hydrogen storage, roughly in the range of 20–40 kJ/mol (Sect. 1.4.1). Wang et al. [95] dehydrogenated Mg(AlH4)2 doped with 2

3.2

Alanates

229

DSC *10−2 / mW/mg 2 ↓ exo 1 0 [1] 142.9⬚C

−1 −2

baseline [1] 136.6⬚C

−3 −4

4ⴗC/min

[1] 157.2⬚C

−5 120

100

160 140 Temperature / ⬚C

180

200

DSC / mW/mg ↓ exo 0.5 0.4 0.3

[1] 162.7⬚C [3] 169.5⬚C

10h (endo)

0.2 0.1

(baseline)

20ⴗC/min

0 100

b

10h (exo)

5h (exo)

5h (endo) 120

140

160

180

200

Temperature / ⬚C

Fig. 3.16 DSC traces up to 200°C of NaAlH4 + MgCl2 mixtures ball milled for 5 and 10 h (the baseline is the trace of the mixtures milled for 40 h where Mg(AlH4)2 decomposed during milling)

mol%TiCl3 at 125°C for 16 h under 3.5 atm. hydrogen and attempted to rehydrogenate the sample at 125°C under ~ 8 MPa hydrogen pressure. No hydrogen uptake was observed. Apparently, this experiment confirms that Mg(AlH4)2 is not reversible under reasonable conditions of temperature and pressure. We tried to overcome this problem by investigating if mechanical reversibility would be possible by controlled reactive mechanical alloying (CRMA) of elemental Mg and Al powders under hydrogen in the magneto-mill Uni-Ball-Mill 5 [124]. We used three stoichiometric Mg–2Al mixtures such as (a) elemental Mg and Al powders, (b) elemental Al powder and commercial AZ91 alloy (Mg–Al–Zn alloy) and (c) powder of as-cast Mg–2Al alloy. Unfortunately, no successful synthesis of Mg(AlH4)2 has been achieved. The only nanocrystalline hydride formed up to 270 h of CRMA was β-MgH2. It has been found that there is a strong competition

230

3 Complex Hydrides

between formation of Al(Mg) solid solution and the β-MgH2 hydride occurring to a various extent up to ~ 10 h of CRMA in all three Mg–2Al mixtures. A hypothesis has been put forward that the presence of Al(Mg) solid solution inhibits the reaction of β-MgH2, Al and H2 to form Mg(AlH4)2. Furthermore, despite the fact that after prolonged milling the Al(Mg) solution eventually decomposes into secondary Al(s) (derived from solid solution), the latter retains its physico-chemical characteristics of the former solid solution which still inhibits the reaction to form Mg(AlH4)2. It is to be noted that the accelerated formation of the Al(Mg) solid solution has been observed during reactive milling of the elemental powders or pre-alloyed Mg–2Al ingots and during thermal decomposition of Mg(AlH4)2 as discussed above. This behavior has its roots in a very substantial solubility of Mg in Al vs. temperature as can be seen in the binary Mg–Al system [122]. It is to be pointed out that the major problem in the MCAS technology is the formation of a large amount of waste by product NaCl. With 2 mol of NaCl per only 1 mol of Mg(AlH4)2 (the reaction of (3.23)) the excellent theoretical hydrogen capacity ~ 9.3 wt% of pure Mg(AlH4)2 is reduced to the very inferior theoretical ~ 3.97 wt%H2 in the mixture of Mg(AlH4)2 + 2NaCl. As reported by Mamatha et al. [118] and Sterlin Leo Hudson et al. [125], it is possible to separate Mg(AlH4)2 from NaCl by the Soxhlet extraction method which is based on the suspension of the Mg(AlH4)2 + 2NaCl mixture in the diethyl ether (Et2O) solvent. However, there are a couple of serious disadvantages of this extraction method. First, the extracted Mg(AlH4)2 is usually contaminated with the solvent adduct [119]. Second, the extraction process needs several days to be completed and then the product requires some additional long-time annealing in vacuum [118]. Such a long production time may not be well suited for industrial environment. On the other hand, a solvent mediated metathesis reaction results in the formation of Mg(AlH4)2 with adducts which can substantially affect its hydrogen desorption behavior [108–111, 117–119]. A very similar alanate to Mg(AlH4)2 is its counterpart Ca(AlH4)2. However, the theoretical maximum hydrogen gravimetric capacity of Ca(AlH4)2 is 7.9 wt% and the theoretical reversible hydrogen gravimetric capacity is only 5.9 wt%. These numbers are smaller than those of Mg(AlH4)2 (Table 1.4 in Sect. 1.1). The MCAS synthesis of Ca(AlH4)2 has been investigated by Mamatha [117, 118] and very recently by Komiya et al. [126]. In this synthesis instead of MgCl2 as in the reactions of (3.23) a CaCl2 is used. The thermal dissociation of the mixture of Ca(AlH4)2 + 2NaCl obtained by the MCAS route and that of a single-phase Ca(AlH4)2 was observed upon heating from room temperature to 250°C at 4°C/min and yielded ~ 2.5 wt%H2 within ~ 80 min and ~ 5 wt%H2 within ~ 40 min. The dissociation took place in two steps (similar to the reactions of (3.20a) and (3.20b) where MgH2 is substituted by CaH2) with approximately equal amounts of hydrogen released in both steps. The expected thermolysis products, CaH2 and Al, apart from NaCl, could be identified by the XRD patterns. Addition of 2 mol%TiCl3 to the Ca(AlH4)2 + 2NaCl led to the accelerated desorption during ball milling resulting in about 2 wt%H2 desorbed within ~ 200 min of milling [117, 118]. The Ca(AlH4)2·THFx synthesized by a metathesis reaction of CaCl2 and NaAlH4, in a tetrahydrofuran (THF) solution, appeared to decompose in three steps at ~ 170, 241 and 277°C as observed by using

3.3

Amides

231

a thermal desorption mass spectroscopy (TDS) but seemed to be more stable than Mg(AlH4)2 [126]. In summary, neither Mg(AlH4)2 nor Ca(AlH4)2 seem to have a potential for further development as solid state hydrogen storage materials for mobile applications. Their storage properties seem not to be better than those of LiAlH4 or NaAlH4. Nevertheless, they still deserve more research to arrive at the final verdict.

3.3

Amides

The extensive studies of hydrogen storage properties of lithium amide were activated by the pioneering efforts of Chen et al. [127]. They showed that Li3N compound can absorb and desorb hydrogen at a very reasonable pressure range according to the following reversible reaction path Li3 N + 2H 2 ↔ Li 2 NH + LiH + H 2 ↔ LiNH 2 + 2LiH

(3.25)

where Li2NH is “lithium imide” and LiNH2 is “lithium amide.” The theoretical amount of reversible hydrogen in this reaction is ~ 10.4 wt% (2H2/(Li3N + 2H2)) or 11.6 wt% if expressed per a molecule of Li3N. In fact, as far back in the past as 1910, Dafert and Miklauz reported erroneously that the reaction of (3.25) produced Li3NH4 which in reality was a mixture of LiNH2 + 2LiH [127]. Chen et al. reported that the overall heat of the reaction of (3.25) was −161 kJ/mol. They observed two plateaus on an absorption PCT curve (under 20 bar), the first one with a very low equilibrium pressure 300°C (the reaction of (3.27c)). In contrast, the sample doped with 1 mol%TiCl3/VCl3 easily desorbed ~ 5.5 wt%H2 in the 150–250°C range at the heating rate of 5°C/min. In addition, for the TiCl3 catalyzed sample no NH3 emission was detected up to 400–450°C. This results indicates that all LiNH2 was consumed as a result of complete reaction with LiH and transformed into Li2NH (the reaction of (3.27d)). The activation energy for hydrogen desorption calculated by the Kissinger method yielded 110 kJ/mol. Short cycling at 220°C desorption/180°C absorption showed that the cycle retention of TiCl3 bearing sample was good. In further studies from this group, Isobe et al. [137] reported that Tinano, TiCl3 and TiO2nano doped 1LiNH2:1LiH composites processed by ball milling revealed a superior catalytic effect on the TDS properties, while for the Timicro and TiO2micro additives catalytic effect was similar to the sample without any additives. Pinkerton [135] investigated the kinetic behavior of decomposition of ball-milled LiNH2 by ammonia release using thermogravimetric analysis (TGA). They noticed slow but significant decomposition by NH3 release in an H2 atmosphere (the reaction of (3.27d)) even under moderate pressure and temperature conditions (0.13 MPa, 175°C). The release of NH3 continued unabated even after a full conversion of the sample to LiH. The activation energy for the decomposition reaction was determined to be about 128 kJ/mole, virtually independent of the source and purity of LiNH2, its stoichiometry, ball milling time, and TGA sample size. Yao et al. [140] investigated the hydrogen storage behavior of pure LiNH2 and 1 LiNH2:1.2 LiH mixtures (overdose of LiH) with and without additives which were ball milled up to 4 h in a SPEX 8000 mill. They found that the decomposition mainly generated NH3. The only major endothermic peak at 380°C in the DSC curve corresponded to the melting of LiNH2. They noted that NH3 was released relatively slowly from the solid-state LiNH2, but rapidly from the liquid LiNH2, due to rapid diffusion of NH3 in the liquid at a high temperature. Milling only enhanced NH3 release in the solid-state LiNH2, but not in the liquid LiNH2 and NH3 started evolving from the milled LiNH2 around 50°C. The addition of Mn, MnO2, V or V2O5 to LiNH2 did not affect LiNH2 melting temperature, but enhanced NH3 release before the melting. With regard to the desorption from the 1 LiNH2:1.2 LiH mixture, 4 h of milling was needed to avoid escape of NH3. After 4 h milling the weight loss in TGA experiments was 5 mass% and the onset hydrogen desorption temperature was lowered from 160 to 120°C. No new phases were formed due to milling. They concluded that the effect of mechanical milling on the properties of (LiNH2 + LiH) should be attributed to improved contact between LiNH2 and LiH particles, as a result of homogeneous mixing and an increase in surface area of LiNH2 and LiH particles. Such an intimate contact ensures that the released NH3 is immediately captured by LiH particles and reacts according to the reaction of

234

3 Complex Hydrides

(3.27d), generating H2. Slightly higher content of LiH in the mixture (ratio 1:1.2) enhances this behavior. The addition to Mn, V and their oxides, had little influence on the production of H2 from the (LiNH2 + LiH) mixture. According to the authors it indicates that the rate-limiting step for dehydrogenation from (LiNH2 + LiH) is the reaction between LiH and NH3 and its kinetics could be enhanced by nanosized catalytic additives. Shaw et al. [142] investigated in detail the effect of ball milling in a modified Szegvari attritor on the hydrogen storage behavior of the mixture 1 LiNH2:1.1 LiH without any additives. Ball milling for 90 min reduced the crystallite size of LiNH2 and LiH to below 10 nm and slightly higher than 10 nm, respectively. Further increases in ball milling time beyond 90 min resulted in additional decreases in the crystallite size, but the reduction rate of the crystallite size was dramatically reduced. The equivalent particle diameters were reduced to 0.12, 0.11, and 0.09 µm for the powder mixture with the ball milling time of 90, 180, and 1,440 min, respectively. Apparently, high-energy ball milling reduced most of LiNH2 and LiH particles to submicrometer sizes and increased their specific surface area by at least one order of magnitude and created nanograins inside of particles. Dehydriding behavior was also greatly enhanced such that TGA curve for the powder milled for 180 min was shifted by ~ 70°C to the lower temperature range. DSC desorption peak of (1LiNH2:1.1LiH) (the reaction of (3.27b)) was reduced from 308 to 215°C by ball milling for 90 min. This peak was further reduced to 200°C if the ball milling time was increased to 180 min. Shaw et al. also observed a substantial effect of milling on the escape of NiH3. After ball milling for 180 min, the level of the NH3 intensity in the LiNH2 and LiH mixture dropped dramatically (by about 36 times). In fact, the NH3 intensity became lower than the detection limit, indicating that the intermediate reaction of (3.27d) between LiH and NH3 has been enhanced so much through high-energy ball milling that escaping of NH3 has been prevented. They have also found that escaping of NH3 is a kinetic issue. The escaping of NH3 depends not only on the powder characteristics and thus ball milling conditions, but also on the testing condition. For a given LiNH2 and LiH powder mixture, slow heating rates (e.g., no more than 5°C/min) can prevent escaping of NH3, whereas fast heating rates (e.g., 10°C/min) can promote escaping of NH3. The activation energy of desorption of (1LiNH2:1.1LiH) according to the reaction of (3.27b) gradually decreased with increasing milling time from ~ 164 kJ/mol for an “as-received” sample to ~ 63 kJ/mol after 1,440 min (24 h) of milling. Shaw et al. pointed out that the reduced particle size and increased surface area lead to an increased reaction rate, whereas a change in the activation energy is related to a change in the reaction mechanism or in the energy state of the reactants. Therefore, it was proposed that the reduction in the activation energy observed by the authors was due to an increase in the energy state of the LiNH2 and LiH mixture induced by high energy ball milling. The structural defects and the large area of grain boundaries within mechanically activated, nanostructured particles can contribute to this high-energy state. They also found the following trends: (1) the activation energy of decomposition of LiNH2 into Li2NH and NH3 according the reaction of (3.27c) exhibits the same trend as that for the reaction of (3.27b), i.e., decreases with increasing milling time, (2) the activation energy of the decomposition

3.3

Amides

235

of LiNH2 into Li2NH and NH3 (the reaction of (3.27c) is always higher than that of decomposition of a mixture (LiNH2 + LiH) (the reaction of (3.27b)) and (3) the activation energy of the decomposition of the LiNH2 and LiH mixture with milling and then mixing is higher than that for the decomposition of LiNH2 as well as for the LiNH2 and LiH mixture with direct milling. The conclusion is that the role of LiH is not simply reacting with NH3 as defined by the reaction of (3.27d). In effect, LiH not only reacts with NH3, but also has catalytic effects on the reaction of (3.27c). Thus, the presence of LiH reduces the activation energy of the reaction of (3.27c) so that the “apparent” activation energy of the reaction of (3.27b) can be lower than the activation energy of the reaction of (3.27c) in the absence of LiH. This also explains why the LiNH2 and LiH mixture with direct milling has a lower activation energy than the mixture with milling and then mixing. Direct milling provides better mixing for the LiNH2 and LiH mixture than the mixture with milling and then mixing. Because of the better mixing of LiNH2 and LiH and the catalytic effect of LiH, the LiNH2 and LiH mixture with direct milling has a lower activation energy than the mixture with milling and then mixing. The second drawback of the reaction of (3.27b) is too high a temperature of reversible hydrogen absorption/desorption. For example, Kojima and Kawai [133] reported from PCT experiments that a single phase Li2NH could reversibly absorb about 6wt%H2 at 300°C under 9 MPa H2 pressure and desorb the same amount under vacuum (0.001 MPa) (PCT desorption plateau ~ 0.15 MPa). Their calculations using the Van’t Hoff equation gave ΔHab = −64.5 kJ/molH2 and ΔSab = −118 J/molH2K (absorption), and ΔHdes = −66.6 kJ/molH2 and ΔSdes = −120 J/molH2K (desorption). This is a larger enthalpy change than that reported in [131]. Unfortunately, even more thorough understanding of various mechanisms of the reaction of (3.27b) cannot change its unfavorable thermodynamics, specifically, because its enthalpy change, Δ H, seems to be closer to ~ 60 kJ/molH2 rather than the previously estimated ~ 45 kJ/molH2. A few approaches have been adopted to improve the storage conditions given by the reaction of (3.27b). Searching for the means to destabilize Li2NH (lithium imide) a number of authors studied a partial substitution of Li by metallic elements with larger electronegativity such as Mg [129, 132, 139]. Xiong et al. [129] observed that, for example, the Li–Mg–N–H system had a desorption peak of around ~ 166°C as compared to the peak desorption temperature of ~ 272°C for the undoped Li2NH. After hydrogenation the structure consisted of Mg(NH2)2 and LiH. This system had a plateau at around 10 bar for absorption and above that value for desorption at 180°C. Orimo et al. [132] and Nakamori et al. [139] prepared M(NH2)y where M = Li−x at.%Mg (x = 0–100 and y = 1–2). For LiNH2 with partial Mg substitution at x = 30 they observed, using thermal desorption spectroscopy, that the start of the thermal desorption was at around 100°C as compared to 280°C for LiNH2 without partial Mg substitution (a mixture of LiNH2 and LiH). It must however be pointed out that a full desorption was not completed even at ~ 300°C. These developments evolved into a search for novel hydride systems substantially differing from the original mixture (LiNH2 + LiH). Some sort of a breakthrough was achieved by Luo, and Luo and Rönnebro [143–145]. They replaced LiH with

236

3 Complex Hydrides

MgH2 in the original mixture LiNH2 + LiH and used ball milling for 2 h in a SPEX 8000 mill to prepare a mixture 2LiNH2:1.1MgH2 or 2LiNH2:1MgH2 (10% excess of MgH2 to prevent LiNH2 to decompose with the release of NH3 during cycling). The PCT experiments for absorption/desorption showed the plateau pressure at 220°C at ~ 50 and ~ 40 bar, respectively. That for desorption at 240°C was close to 70 bar. Isothermal desorption of a lightly cycled sample at 220°C in a Sieverts-type apparatus yielded ~ 4.5 wt%H2 at the pressure increasing up to 1.6 bar within about 1 h. The fresh sample (uncycled) required more than 15 h for desorption at 240°C. From the Van’t Hoff plot the decomposition enthalpy change was calculated as equal to ~ 34–39 kJ/molH2. XRD studies showed that the fresh sample contained mainly LiNH2 and MgH2 but the rehydrided sample contained LiH and Mg(NH2)2 phases. On that basis they proposed the following equation for the process 2LiNH 2 + MgH 2 → Li 2 Mg(NH)2 + 2H 2 ↔ Mg(NH 2 )2 + 2LiH

(3.28)

One should note that the reversible reaction in this equation takes place between (Mg(NH2)2 + 2LiH) and (Li2Mg(NH)2 + 2H2). Further to these original papers this new system has been more thoroughly investigated by a number of researchers [146–154]. It has been found [146] that a PCT curve for hydrogen desorption at 220°C from the mixture of Mg(NH2)2 and 2LiH exhibits two pressure-dependent segments at a given temperature and about 2/3 of the hydrogen is released at relatively higher pressures and forms a pressure plateau and the other 1/3 is desorbed at lower pressures (~ 5 wt%H2 total desorbed). The overall reaction heat measured in a differential scanning calorimeter is 44.1 kJ/ molH2, while the heat of desorption of H2 in the higher pressure plateau is about 38.9 kJ/molH2. These thermodynamic values are very favorable for PEM fuel cell application. An extension of the Van’t Hoff plot to 1 bar yields ~ 90°C but TPD measurements show that desorption starts at a temperature >100°C. The calculated activation energy of desorption Ea = 102 kJ/mol is high and sets a kinetic barrier. The authors suggested that such high activation energy may be related to the energy required for the reconstruction of reactants and dissociation of chemical bonds. Chen et al. [147] reported the results of cycling absorption at 200°C at 5 MPa hydrogen pressure and desorption in a primary vacuum (pressure 1.5) Li 2 MgN 2 H 3.2 + 1.4H 2 ↔ Mg(NH 2 )2 + 2LiH

(3.29a)

3.3

Amides

237

and for the lower pressure plateau and sloping region (for 0 < wt%H22 nm), i.e., the amount of surface area excluding the macropores. With the total area of 204 m2/g the measured H2 adsorption on activated graphite was 14 mL H2/g in 77 K [17], which comes to only 0.125 wt.% H. The interplanar distance in regular graphite, 0.3354 nm, is too small for the insertion of gas hydrogen molecules, with a diameter of 0.406 nm. Even if the bulk graphite is exfoliated to multitude of isolated graphene sheets with the surface area of 1,315 m2/g, the amount of hydrogen stored by physisorption at cryogenic temperatures should not exceed 3 wt.% H [18]. Therefore, if higher value of storage capacities were reported, they must be attributed to a concurrent chemisorption. Such chemisorption usually occurs through the formation of very strong sigma bonds, which break and release hydrogen gas only at high temperatures (as discussed in Sect. 4.2.1.4). In view of its electronic structure, carbon atoms in graphite are strongly bonded in planes, and only weak van der Waals forces hold the planes together. This leads

4.3

Disordered and Active Carbons

301

to a material that can shear easily. It is also very brittle, because the s + p bonds hold the atoms of carbon very strongly in the lattice sites, and if strained, these bonds break and the crystal fractures. In fact, the graphite stands at the opposite side of scale of material hardness to the diamond. In view of such an easy sliding on hexagonal basal planes, and susceptibility to brittle fracture, graphite presents a great potential for production of disordered carbon phases and nanophases.

4.3

Disordered and Active Carbons

4.3.1

Disordered Graphites and Mechanically-Activated Carbons

Disordered graphites and nanographitic carbons feature low crystallinity as can be inferred from X-ray diffractograms. Usually, XRD patterns exhibit only strong peak for 00l Bragg reflections from well-stacked and equidistant hexagonal basal planes. The other interferences, caused by in-plane periodic structure, are weak. A number of models were proposed to explain the peculiar shape of XRD reflections. They have attempted to interpret the diffraction peak positions and shapes through partitioning a disordered graphite into a series of stacks of N equidistant graphene layers, where N is two, three, or more layers (Fig. 4.9). Certain assumptions have to be made as to statistical distribution of interlayer distances d between parallel basal planes, as to fit XRD patterns. Some models also reflected on the fact that graphene stacks can be tilted in respect to each other by the angle fi in Fig. 4.9; c-axis

ϕi c1

[001]

B A

d1

N1

d2

B A B

N2

A B

c2

Ni

A

ϕi

ci

La La (i)

Fig. 4.9 Stacking of hexagonal basal planes in graphite (left). Mechanical activation results in a structure with proliferation of faults in the plane stacking (right). The change of stacking sequences in nearby subgrains is marked by arrows. The stacking fault disorder is corroborated from the nonuniform peak broadening or absence of hk indexed peaks in XRD pattern (see Sect. 1.3.3.3)

302

4

Carbons and Nanocarbons

consequently the basal c-axis changes direction from one stack to another [19]. This is illustrated in Fig. 4.9. The origin of the partitioning of massive graphitic structure into a mosaic of nanocrystalline stacks comes from easy shearing of planes, bonded only by weak vdW interactions. Forward shearing of hexagonal basal planes results in generation of stacking fault disorder, as seen in Fig. 4.9. The stacking fault disorder was recently postulated to explain improved insertion of H-ion into the structure of other layered compound: Hexagonal layered Ni hydroxide used in rechargeable alkaline batteries [20]. We have discussed it in Sect. 1.3.3.3. Stacking faults that originate in a mechanical shear of weak bonded graphene planes are deformation faults. Disordered graphites may also exhibit distinct growth faults created during the growth of a layered stack of planes. This view of disordered compounds based on deformed hexagonal crystal, whose planes are bonded only by week interactions, has been proposed in respect to insertion of Li+ and H+ ions in graphitic carbons and layered nickel hydroxide compounds, respectively [21, 22]. It was demonstrated that simple mechano-chemical activation of layered compounds in high-energy ball mills, such as those discussed in Sect. 1.3.2, is a simple yet unusually effective method for improvement of insertion of hydrogen ions into the structure (also seen in Fig. 4.9) for reversible electrochemical hydrogen storage [23, 24]. Another kind of faults are disclinations, and plane rotations, marked in Fig. 4.9 by angles fi and øi. The graphene layers are shown here as being planar but, in fact, these are rather wavy, curled planes. When this material has curled, twisted, and rotated graphene planes, it is called tubostratic graphite. The already weak bonding in graphene stacks become even weaker when the graphene layers are short. With weak attraction between short planes, the average interplanar distance along c-axis increases from 0.3354 nm for ideal graphite to about 0.36 in turbostratic graphite. The loosely bonded planes can rotate. This is where the term turbo (means: rotation) comes from. In turbostratic graphite, the atoms are arranged in layers, as in graphite, but stacked randomly instead of the ABABABA… sequence of graphite. The interplanar spacing, d, changes randomly, although in thin stacks it does not vary much along a short distance of c-axis (Fig. 4.9, right). Since such varying spacing cannot be resolved by X-ray diffraction, the diffractogram exhibits only (00l) peaks that are an average of these spacing. Such disordered graphites and nanographites can easily be produced by ball milling in high-energy mills, as discussed in Sect. 1.3. Hydrogen absorption in disordered graphite produced by ball-milling and shearing of layered structure of graphite was studied in the low temperatures from 35 to 110 K and at pressures up to 20 MPa. Between 5.9 and 6 wt% H was reported to be absorbed on cooling to 75 K [25]. Others reported that hydrogen absorption in mechanically activated nanostructured graphite reaches up to 7.4 wt.% H, where graphite is ball-milled in H2 at 1 MPa pressure in room temperature. Also, for the deuterided mechanically-activated graphite, prepared under 1 MPa D2 by ball milling for 80 h, two desorption peaks were observed, starting at about 600 and 950 K, indicating that there were two types of C–D type of bonding that differ in their strength [26]. A laser desorption time-of-flight mass spectroscopy experiment used

4.3

Disordered and Active Carbons

303

for characterization of a sample of graphite ball-milled with H2 for 100 and more hours suggest the formation of same hydrogenated carbon clusters, with C:H ratios 1.25 and 3.5 in addition to the majority nanostructured graphite phase [27]. Apparently, the effects of mechano-chemical activation of graphite in ball mills are not limited to promoting reversible hydrogen storage via physisorption only but also induce irreversible formation of carbon–hydrogen molecules via chemisorption. The fraction of hydrogen bonded in CHx molecules is not accessible for reversible storage under ambient conditions. The proliferation of stacking faults, edge dislocations, and other defects in disordered graphites result in many dangling bonds determined by unoccupied sp2 orbitals. If the carbon sp2 orbitals are overlapping with hydrogen s orbital, as in Figs. 4.4–4.6 (site a), the hydrogen storage would occur by chemisorption, instead of physisorption. So it is tempting to increase hydrogen storage in ambient temperatures by providing substrates with large density of empty orbitals. This can occur at the periphery of sp2-bonded graphene stacks as obtained by mechanical milling of graphite. The role of defects in changing physisorption to chemisorption can not only be limited to disordered carbons and nanocarbons but also discussed in respect to highly ordered nanocarbons, such as fullerenes and CNTs discussed in the following paragraphs. A view that the nonphysisorbed hydrogen observed in carbon nanostructures would be connected with the formation of structural defects was expressed recently by Zuttel and Orimo [28]. 4.3.2

Active Carbons and Chemically Activated Carbons

More conventional process to prepare active carbons is by chemical activation. So-called activated carbons consist of multitude of stacks of graphene planes of various sizes and degree of disorder. Such carbons are highly macro- and mesoporous materials. According to IUPAC definitions, three groups of pores are distinguished: • Macropores (above 50-nm diameter) • Mesopores (2–50-nm diameter) • Micropores (under 2-nm diameter). The microporosity is often reported in recent research papers as nanoporosity. Commercial activated carbon grades have an internal surface area of 500 up to 1,500 m2/g. Powdered activated carbon comes with particle size 1–150 µm. There are also granulated or extruded materials with granule size in the 0.5–4-mm range. Activation is often conducted by processing with steam or chemical agents. Carbons activated by steam can be prepared from raw materials such as coal, peat, or lignite, which are carbonized and reacted with high-temperature water steam, in the process where fraction of carbon atoms are gasified, leaving beside porous structure. Chemically, carbon can also be activated with phosphoric acid. So-called mesocarbon microbeads (MCMBs) were produced from coal tar pitch in the Osaka

304

4

Carbons and Nanocarbons

Gas Cherry T process; they were used in Li-intercalated rechargeable batteries. Recently, this active carbon was further developed with manufacturing so-called superactivated carbon through reaction of coke or MCMBs with potassium hydroxide. This new form of carbon materials have a open cage-like type of porosity and SSA as high as 3,220 m2/g. Reversible hydrogen adsorption, following closely the characteristic Langmuir isotherm, was reported providing the storage of 5 wt.% H at 77 K and 1.3 wt.% H2 at 296 K [29] (Table 4.1). The heat of adsorption ΔH can be determined from Clausius–Clapeyron equation: ln P =

−ΔH +C RT

(4.3)

where R is gas constant, T is absolute temperature, and C is a constant. The reported values about 6 kJ/mol H2 are still short of the desirable range above 10 kJ/mol that targets bond strength required for reversible hydrogen storage and release at ambient temperatures. This desirable range would be at 10–60 kJ/mol, the range when the energy of physical adsorption is sufficient enough to form the optimal bond strength, and storage does not require the use of cryogenic temperatures. Higher energy than 50–60 kJ/mol results in chemical absorption through the formation of strong s bonds as discussed in Sect. 4.2.1.4, and demands high desorption temperature. This is shown on the “energy line” (Fig. 4.10). The hydrogen storage capacities for disordered graphites, nanographites, and activated carbons are collected in Table 4.1. One can conclude that activated carbons are better storage materials than CNTs and most experimentally investigated carbon nanophases (like GNFs). Yet, if one applies a broader definition of nanomaterials, the activated carbon phases are, indeed, the disordered and nanostructured carbons. 4.3.3

Amorphous Carbon

When the size of sp2-bonded clusters of carbons decreases below La = 1–2 nm, nanocrystalline graphitic carbon becomes amorphous carbon (am-C). In fact, am-C begin to exhibit any mixture of sp2, sp3, and even sp1 sites. This area of research is connected with investigation of thin carbon films for electronics, particularly ultrahigh density data storage in optical discs and MEMS. Amorphous carbon with high fraction of sp3 bonding like in diamond and diamondoids is known as diamond-like carbon (DLC). The amorphous carbon requires hydrogen to stabilize its amorphous structure. Amorphous carbon thin films can contain as much as 40–60 at.% H. About 70% of carbon atoms can be sp3 bonded and most of these sp3 orbitals, if not connected to carbon, are terminated with hydrogen. Like in diamondoids we discussed in Sect. 4.1, Fig. 4.5, hydrogen stabilizes the carbon structure. Lower hydrogen content of about 20 wt.% H2 results in increased fraction of sp2 hybridized bonds in a phase known as graphite-like hydrogenated amorphous carbon (GLCH). It is relatively easily produced by physical- or chemical-vapor deposition or magnetron sputtering [38]. Raman spectroscopy demonstrated that along with hydrogen covalently bonded to carbon, and hence hardly reversible, there exists a quasi-free

4.4

Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns

physisorption

H-C 0

305

chemisorption

H-Mg

H-H 1

2

3

4

H-OH 5 x102 kJ/mol

MgH2 activated carbons carbon nanotubes graphitic nanofibers

Fig. 4.10 Current and targeted range (horizontal bar) of bond strengths for nanocarbons (red), shown against the ranges for physisorption and chemisorption, and compared with the bond strength for high-temperature MgH2, and energetic molecules of H2 and H2O

hydrogen (in a form of highly stretched or polarized H2 molecules) adsorbed by graphite-like structural fragments in otherwise amorphous structure. The amount of hydrogen released by breaking of weak C–H bonds in low-temperature annealing process (at less than 500°C) was determined to be 10 at.%. Another 12 at.% was released by breaking covalent bonds at higher temperature, 750°C. Thermallyprogrammed desorption indicated that the quasi-free hydrogen began to effuse from material by annealing at 100°C and its release peaked at 400°C [39].

4.4

4.4.1

Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns Fullerenes and Hydrofullerenes

In C60 fullerene-type carbon allotrope, there is only one structure in which all the pentagons are nonadjacent and this is icosohedral symmetry-Ih (Fig. 4.11). This structure is often referred to as backyball to reflect on its full name buckminsterfullerene (after Buckminster Fuller who popularized the geodesic dome as an architectural form). Fullerenes can be made by either combustion or pyrolysis of aromatic hydrocarbons but are mostly prepared in arc-discharge processes. They have been studied extensively, and in fact, their studies developed into a new branch of organic chemistry. In textbooks of fullerene chemistry, hydrogenation is the simplest reaction and fullerene hydrides are the simplest derivatives of fullerenes. Hydrogenation can be conducted under H2 pressure and elevated temperatures. Heating C60 at 400°C and 80 atm H2 yields red solid, C60H18 [40]. Up to 48 H atoms can be added to C60 under more forcing conditions. Catalytic hydrogenation at 280°C and 160 atm with use of Ru/C catalyst produce hydrofulerenes up to C60H50 [41]. Hydrogenation is quite

77 600–1,066 (desorp-tion) 77 77 77 298 77 300 77–300 Ambient 300 77 273 273 77 77 273 77

725

988 (+173) 70 2,800 2,800 922 (+206) 3,220 3,220

119 (85) 49 (26) 120 (+120) – – – –

77

T (K)

7 (+7)

Surface area SBET(+St) (m2/g)

1 bar 1 bar 1 bar 12 12 11 1.5 0.9 11 1 bar 3 3 3 1 bar

Ambient

1 bar

p (MPa)

0.125 0.05 0.10 1.4 12.4 5.7 1–1.8 0.5 < 0.1 1.27 0.10 0.37 4.7 1.5 1.3 5

1.5 1.5

0

H2 (wt.%)

Nijkamp et al. Nijkamp et al. Nijkamp et al. Hwang et al. Fan et al. Cheng et al. Strobel et al. Hirscher et al. Hirscher and Becher Nijkamp et al. Chahine and Benard Chahine and Benard Chahine and Benard Nijkamp et al. Kojima, et al. Kojima, et al.

Hentsche et al. Hirscher et al.

Nijkamp et al.

Reported by

[17] [17] [17] [31] [32] [33] [34] [35] [36] [37] [14] [14] [14] [37] [29] [29]

[25] [30]

[17]

References

4

CNFs carbon nanofibers; GNFs graphitic nanofibers; AC activated carbon

Graphite – synthetic Graphite – ball milled Graphite – mechanically charged with H2 in ball mill Graphite – activated 100 CNF – large CNF – median diameter CNF CNF CNF GNF – herringbone GNF – mechanically Activated GNF GNF – ACF 500 CNF – Ni70Cu30-ethylene AC – AX 21 AC – AX 21 AC – Norit SX1 AC – M30 superactivated AC – M30 superactivated

Material

Table 4.1 Several reported hydrogen storage capacities in graphitic and activated carbons and nanocarbons

306 Carbons and Nanocarbons

4.4

Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns

307

Fig. 4.11 C60 fullerene. Image credit: http://www.xdray.uu.se/cluster/images/c60

effective with use of hydrogen in statu nascendi generated from reaction of Zn with HCl conducted in toluene solution; fullerene C70H38 in mixture with some C60hydrofulerenes can be prepared by this method. Also, C60H36 is easily prepared in reduction with hydrogen from Zn/conc. HCl in toluene. The reduction is very rapid if stirring can be complete in 1 h. However, synthesis of C60H36 is often accompanied by the formation of C60H18. Another method is transfer hydrogenation conducted with the use of organic compounds that have a tendency to loose hydrogen with the formation of aromatic compounds. For example, heating C60 fullerene in 9,10-dihydroanthracene in a sealed tube in 350°C for 30 min yields C60H36 in mixture with C60H18 after 24 h of heating [41]. The reaction is reversible and hydrofullerene can be back converted to C60 fullerene with DDQ reactant dissolved in toluene. Hydrogenation of fullerenes, being a reduction process, proceeds rapidly, and this means that fullerenes are quite strong oxidizers. They are able even to reduce H2S to elemental S; the product is C60H36 with some admixture of C60H18. Oxidation back to C60 can be conducted but requires wet chemical reactions. Release of hydrogen gas in solid state reaction is difficult as the hydrogen forms strong chemical bonds with fullerenes. The hydrogen bonds to the outer surface of such buckyballs. Formation of endohedral complexes of C60 with hydrogen was also shown to be possible, both, in first-principle calculations and in experiments; this fills the buckyballs with only few percent of hydrogen to about C60H composition. Apparently, C60 and other higher fullerenes show no hydrogen storage capacity in respect to reversible hydrogen solid-state storage. Molecular crystals can also be synthesized using C60 buckyballs as building blocks. The interstitial sites in molecular crystal [C60]n can be filled with K atoms. The atoms of potassium occupy tetrahedral or octahedral sites in the cubic close packed

308

4

Carbons and Nanocarbons

lattice made of n fullerene nodes. Molecular H2 can replace K in octahedral sites resulting in the maximum H/C ratio of 1:20. However, the ratio is low and precludes any application for solid-state hydrogen storage. Fullerenes in combination with metals form metallo-organic compounds, fullerids. C60 fullerene in metallo-organic combination with scandium can bind as many as 11 hydrogen atoms giving theoretical capacity ca. 9 wt.% H [42], and in combination with titanium ca. 8 wt.% H2 [43].

4.4.2

Carbon Nanotubes

A single-walled carbon nanotube (SWNT) is a single graphene sheet rolled up in a seamless cylinder, whose diameter is of the order of few nanometers (Fig. 4.12). A double-walled carbon nanotube (DWNT) consists of rolled two graphene layers, and a multiwall carbon nanotube (MWNT) exhibits several co-axial rolls of graphene sheets, one sitting in each other and separated by about 0.35 nm. CNTs are prepared by either laser ablation from graphite target, arc discharge, or chemical vapor deposition. In either case, to grow they need the presence of Co or Ni catalyst. Typically, outer diameter of CNTs prepared by a carbon arc process ranges between 20 and 200 Å, and inner diameter ranges between 10 and 30 Å. An aspect ratio (length-to-diameter ratio) is typically of 102–103. When SWNTs are prepared, they come out as mostly capped at each end with fullerene-like half-sphere- or polyhedral cups. A CNT presents a highly-ordered, hexagonal-packed structure, not unlike the C60 buckyball and other high-ordered fullerenes. Highly ordered structure is well confirmed by direct imaging in HRTEM and in Raman spectra, the two most commonly used tools for microstructural characterization of CNTs. Analysis of Raman spectra indicates that DWNT samples come as a combination of two or more kinds of double-wall tubes; for example, one reported sample exhibits inner/outer diameters 0.72/1.48 nm and 0.91/1.61 nm, respectively [44]. SWNTs have a tendency to self-organize and form more thick nanoropes. Nanoropes consist of tenths and hundreds of SWNTs, which are intertwined and aligned. Cutting

roll

2 0 1 3

Fig. 4.12 The structural relationship between graphene sheet and single-walled carbon nanotube; arrows point to two alternative directions of rolling; circles inside rings reflect on aromatic character of carbon rings and delocalization of resonant electrons inside them

4.4

Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns

309

a rope in the plane normal to the rope alignment direction, one can see a triangular lattice of circles, as illustrated in Fig. 4.13, with the intertube spacing g of 0.3–0.4 nm, as measured from the center of the tube walls. If so, such a structure should manifest itself in X-ray diffractograms. Indeed, while SWNT material exhibits no strong XRD peaks, three broadened maxima at 2q = 1.95°, 4.1°, and 11.8° were reported for the DWNT nanorope material [44]. The presence of three peaks, although they are broadened, indicates that the well-ordered structure in bundles of DWNTs forms a 2D-triangular lattice. This intertube spacing is determined by van der Walls interactions. This is a van der Walls gap: g=d – 2r, where d is the spacing between tubes in triangular lattice of tube bundles, and r is the radius of the nanotubes (see Fig. 4.13). It is well-established that the hydrogen does not enter the interior of CNTs in any larger amounts. Instead, it prefers adsorption on the exterior wall of the tube, i.e., in the van der Walls gap. Elastic and quasielastic neutron scattering results imply that at 80 K, under H2 pressure of 110 atm, H2 molecules gradually condense in the SWNT sample while being physisorbed in the interstitial tunnels of the SWNT bundles [45]. Apparently, hydrogen is stored mainly in bundles of SWNTs, with H2 molecules physisorbed at the exterior surface of the wall, in either groove (a) or interstitial (b) channels. It is only at high pressures that hydrogen can penetrate the interior of tubes to occupy endohedral sites (c); in such a case, tube diameter must be ca. 0.7 nm [46]. Since CNTs are capped with semispherical fullerene caps at their end, the cups further obstruct hydrogen molecules to enter interior channels of CNTs. Chemical methods are devised to remove these end-caps to open nanotubes for access of gas molecules. Anticipated is linear increase in hydrogen storage capacity with larger tube diameters [7]. The low values of adsorption seems to be a direct consequence of a great fraction of the surface area that is excluded from adsorption, this being either the interior of a r

c

d g

b

Fig. 4.13 Triangular network of d-spaced nanotubes in a bundle; marked are van der Walls gap (g), and positions of H atoms on the exterior wall (a, b) and in the interior of tube (c)

310

4

Carbons and Nanocarbons

CNT or the regions of nanoropes where the van der Walls gap is too wide, 0.4 nm or more. The reported gap of about 0.32 nm [47] seems also be not optimal for efficient bonding of H2 molecule; the distance between tubes is large enough for H2 molecule to pass through without bonding to carbon. Compressing the bundle can reduce this distance to 0.26 nm, at which case chemical bonding between external wall and hydrogen can become possible [48]. Optimization of CNT arrays and ropes for storage of hydrogen was computer simulated using grand-canonical system and Monte Carlo algorithms, and such simulations suggest that self-organized triangular pseudo-lattice of tubes is unlikely to result in intermolecular potentials such that hydrogen is bonded in near-ambient temperatures, as is requested by its use in vehicular fuel cell systems [49]. A stronger intermolecular potential can be realized in van der Walls gap in the bundle of DWNTs. The DWNT bundle with wide g value can absorb two times more H2 than the close-packed SWNT bundle [44]. CNT storage medium does not consist of just straight SWCT, DWNT, or MWNTs. The tubes can be bended and deformed. There is a question how such tube deformation affects hydrogen storage. A rolled up graphene sheet that forms a CNT consists of hexagons of carbon atoms, which are sp2, hybridized like in graphite. We have already discussed that sp2 hybridization directs C–C bonds to lie in the plain of hexagonal rings formed by carbon atoms. However, graphene sheet is highly curved in a CNT. One can ponder how this high curvature of single graphene sheet affects the reactivity of carbon atoms toward hydrogen. Further to this, one could calculate an excess energy that is required to pull one C atom out of its equilibrium position in the graphene plane. Such an out-of-plane bond stretching results in the strain energy, EC–C (strain), and is composed of excess energy of four atoms: the pulled-up atom 0, and its three nearest neighbors, in the in-plane positions 1, 2, and 3 (Fig. 4.12). Then, of interest would be the binding energy EC–H (strain) of H on locally strained atom C. Apparently, such energy would require energy of breaking of in-plane bonds and moving electron into empty out-of-plane p orbital. Calculations for such a scenario, conducted by DFT computer simulations, were recently reported and the model predicted enhanced hydrogenation energy at the kink site of bent CNTs [50]. In another computer model it is found that charge transfer occurs from a low curvature region to a high curvature region of the deformed CNT bundle at 80 K and 10 MPa. It was suggested that carbon atoms develop charge polarization only on the deformed structure, like an oval-shaped CNT. The long-range electrostatic interactions of polarized charges on the deformed bundle increase the ordering of H2 molecules that interact with this surface. This causes condensation of hydrogen gas on deformed CNTs, although only in bundles and not on single SWNTs [51]. The condensed hydrogen may extract electron density from the carbon on the highly curved surface of a nanotube (hence, acts as a “hydrogen acid”) and forms covalent bond, as this is shown in Fig. 4.6, position a. Therefore, enhanced hydrogenation may benefit chemisorption instead of physisorption. Through chemisorption the hydrogen atoms will exothermally bond to the C atom on the tube wall. Again, this can be enhanced by the out-of-plain pullout, hence sp3 hybridization and out-of-wall charge polarization induced by the effect of high curvature. Bonds

4.4

Highly Ordered Fullerenes, Carbon Nanotubes, and Carbon Nanohorns

311

formed via chemisorption are much more stronger, and H2 is released only upon heating to temperatures above 400°C [52]. Unfortunately, such high temperatures are still not acceptable for vehicular hydrogen storage. The intercalation of graphite with alkali metals, particularly with Li+ ions, was investigated in reference to rechargeable Li-ion batteries; however, intercalation of carbonic materials with molecular hydrogen has not been successful [4]. Molecular H2 does not intercalate into graphites and carbonic materials but physisorbs on the carbon surfaces. A graphite intercalation compound, KC8, consists of alternating graphene planes and potassium, and lithiated graphites used in Li-ion batteries constitute the lithium analogue of this compound, LiC8. Electrochemical insertion of Li into SWNTs gives one Li+ per two atoms of C; however, the dynamic diameter of H2 molecule is much larger than Li+ ion. Intercalation of graphitic material with alkali metal causes the graphene layers to swell and thus facilitate the hydrogen insertion into graphitic nanocarbons. The early results of high H2 uptake of H2 by alkali-doped CNTs under ambient pressures and moderate temperatures [53] were not confirmed. Intercalated lithium have complex chemistry and high chemical affinity to water and carbon dioxide present in air. When re-examined, the weight gain attributed to hydrogen uptake appeared to be caused rather by absorption of water than by hydrogen. Lithiated SWNTs seems to store more than 4 wt.% H2, albeit the hydrogen desorbs only at higher temperatures above 600 K [54]. Summing up, considerable time and effort has been invested worldwide into utilizing CNTs as a hydrogen storage media; however, their use as viable hydrogen storage remains questionable. The hydrogen storage capacity of SWNTs has been studied by means of gravimetric and volumetric measurements. The result reported differs greatly (Table 4.2), to the extent not often met in reports from research. Hydrogen gas charged into CNTs at cryogenic temperatures does not bind firmly Table 4.2 2 Hydrogen storage in carbon nanotubes Material

T (K)

p (MPa)

H2 (wt.%)

Reported by

References

SWNT – SSA = 610 m2/g SWNT – 75% purity

77

0.05

~0.54

Miyamato et al.

[44]