230 11 15MB
English Pages 1242 Year 2007
Table of Contents • Part One Biosensor and Biochip Technologies
•
Introduction to Biosensor and Biochip Technologies Full Text: HTML PDF (46K)
•
1 Overview of Biosensor and Bioarray Technologies Abstract | Full Text: HTML PDF (171K)
•
2 Overview of Modern Analytical Needs Abstract | Full Text: HTML PDF (176K)
•
3 Historical Perspective of Biosensor and Biochip Development Abstract | Full Text: HTML PDF (839K) • Part Two Biological and Molecular Recognition Systems
•
Biological and Molecular Recognition Systems Full Text: HTML PDF (36K)
•
4 Protein Recognition in Biology Abstract | Full Text: HTML PDF (1195K)
•
5 Enzymology Abstract | Full Text: HTML PDF (311K)
•
6 Molecular Antibody Technologies for Biosensors and Bioanalytics Abstract | Full Text: HTML PDF (230K)
•
7 Phage-Displayed Epitopes as Bioreceptors for Biosensors Abstract | Full Text: HTML PDF (279K)
•
8 Luciferase Reporter Bacteriophages Abstract | Full Text: HTML PDF (146K)
•
9 Natural Luminescent Whole-Cell Bioreporters Abstract | Full Text: HTML PDF (116K)
•
10 Recombinant Bacterial Reporter Systems Abstract | Full Text: HTML PDF (120K)
•
11 Recombinant Whole-Cell Bioreporter Systems Based on Beetle Luciferases
Abstract | Full Text: HTML PDF (154K)
•
12 Recombinant Aequorin-Based Systems for Biomarker Analysis Abstract | Full Text: HTML PDF (246K)
•
13 Yeast-Based Biosensors and Their Incorporation of Mammalian Protein Receptors for HighThroughput Screening Abstract | Full Text: HTML PDF (579K)
•
14 Molecularly Imprinted Polymers as Recognition Elements in Sensors Abstract | Full Text: HTML PDF (324K)
•
15 Aptameric Biosensors Abstract | Full Text: HTML PDF (236K) • Part Three The Biology–Materials Interface
•
The Biology–Materials Interface: Interfacial Science and Receptor Integration Full Text: HTML PDF (36K)
•
16 Immobilization of Biomolecules by Electropolymerized Films Abstract | Full Text: HTML PDF (173K)
•
17 Electrochemical Polymerization for Preparation of Electrochemical Sensors Abstract | Full Text: HTML PDF (105K)
•
18 Smart Hydrogel Materials Abstract | Full Text: HTML PDF (268K)
•
19 Scanning Electrochemical Microscopy for Biomolecular Immobilization and Imaging Abstract | Full Text: HTML PDF (450K)
•
20 Modeling of Biosensor Interfaces Abstract | Full Text: HTML PDF (622K)
•
21 Ion Channel Biosensors Abstract | Full Text: HTML PDF (342K) • Part Four Transducer Technologies for Biosensors
•
Transducer Technologies for Biosensors and Bioarray Technologies Full Text: HTML PDF (57K)
•
22 Electrochemical Techniques in Biosensors Abstract | Full Text: HTML PDF (341K)
•
23 Conductometric Enzyme Biosensors Abstract | Full Text: HTML PDF (169K)
•
24 Chemical and Biological Field-Effect Sensors for Liquids—A Status Report Abstract | Full Text: HTML PDF (290K)
•
25 Overview of Optical Biosensing Techniques Abstract | Full Text: HTML PDF (380K)
•
26 Localized Surface Plasmon Resonance (LSPR) Spectroscopy in Biosensing Abstract | Full Text: HTML PDF (1171K)
•
27 Picoscopes, New Label-Free Biosensors Abstract | Full Text: HTML PDF (428K)
•
28 Chemiluminescent Optical Fiber Immunosensor Abstract | Full Text: HTML PDF (147K)
•
29 Bioluminescent Whole-Cell Optical Fiber Sensors Abstract | Full Text: HTML PDF (401K)
•
30 Phagocyte Luminescent Sensor Abstract | Full Text: HTML PDF (396K)
•
31 Applications of the Electrogenerated Luminescent Reactions in Biosensor and Biochip Developments Abstract | Full Text: HTML PDF (457K)
•
32 Dual Polarization Interferometry: A Real-Time Optical Technique for Measuring (Bio)molecular Orientation, Structure and Function at the Solid/Liquid Interface Abstract | Full Text: HTML PDF (249K)
•
33 Grating-Based Optical Biosensors Abstract | Full Text: HTML PDF (424K)
•
34 Holographic Sensors Abstract | Full Text: HTML PDF (207K)
•
35 Introduction to Acoustic Technologies Abstract | Full Text: HTML PDF (252K)
•
36 Love Wave Biosensors Abstract | Full Text: HTML PDF (109K)
•
37 Magnetic Acoustic Resonator Sensor (MARS)
Abstract | Full Text: HTML PDF (251K)
•
38 Thermal Biosensor and Microbiosensor Techniques Abstract | Full Text: HTML PDF (279K)
•
39 Microcalorimetry and Related Techniques Abstract | Full Text: HTML PDF (246K)
•
40 Magnetic Biosensor Techniques Abstract | Full Text: HTML PDF (462K) • Part Five Miniaturized, Micro and Particle Systems
•
Miniaturized, Microengineered, and Particle Systems Full Text: HTML PDF (40K)
•
41 Introduction to Microfluidic Techniques Abstract | Full Text: HTML PDF (297K)
•
42 Practical Aspects of Microfluidic Devices: Moving Fluids and Building Devices Abstract | Full Text: HTML PDF (564K)
•
43 Polymer-Based Microsystem Techniques Abstract | Full Text: HTML PDF (249K)
•
44 Microelectrochemical Systems Abstract | Full Text: HTML PDF (128K)
•
45 Micro- and Nanoelectromechanical Sensors Abstract | Full Text: HTML PDF (413K)
•
46 Nanobiolithography of Biochips Abstract | Full Text: HTML PDF (339K)
•
47 Nanosphere Lithography-Based Chemical Nanopatterns for Biosensor Design Abstract | Full Text: HTML PDF (268K)
•
48 Quantum Dots: Their Use in Biomedical Research and Clinical Diagnostics Abstract | Full Text: HTML PDF (91K)
•
49 Manipulation and Detection of Magnetic Nanoparticles for Diagnostic Applications Abstract | Full Text: HTML PDF (154K)
•
50 The Detection and Characterization of Ions, DNA, and Proteins Using Nanometer-Scale Pores Abstract | Full Text: HTML PDF (411K)
•
51 Conducting Polymer Nanowire-Based Biosensors Abstract | Full Text: HTML PDF (323K)
•
52 Biosensors Based on Single-Walled Carbon Nanotube Near-Infrared Fluorescence Abstract | Full Text: HTML PDF (377K) • Part Six Array Technologies
•
Array Technologies Full Text: HTML PDF (40K)
•
53 Nucleic Acid Arrays Abstract | Full Text: HTML PDF (222K)
•
54 Protein Chips and Detection Tools Abstract | Full Text: HTML PDF (263K)
•
55 Surface-Enhanced Laser Desorption/Ionization (SELDI) Technology Abstract | Full Text: HTML PDF (238K)
•
56 Fiber-Optic Array Biosensors Abstract | Full Text: HTML PDF (577K)
•
57 Surface Plasmon Resonance Array Devices Abstract | Full Text: HTML PDF (170K)
•
58 Label-Free Gene and Protein Sensors Based on Electrochemical and Local Plasmon Resonance Devices Abstract | Full Text: HTML PDF (293K)
•
59 An Electrochemical Biochip Sensor for the Detection of Pollutants Abstract | Full Text: HTML PDF (249K)
•
60 Microcantilever Array Devices Abstract | Full Text: HTML PDF (211K)
•
61 Biosniffers (Gas-Phase Biosensors) as Artificial Olfaction Abstract | Full Text: HTML PDF (348K) • Part Seven Data Analysis, Conditioning, and Presentation
•
Data Analysis, Conditioning and Presentation Full Text: HTML PDF (35K) • Part Seven Data Analysis, Conditioning and Presentation
•
62 Design of Data Algorithms for Blood Glucose Biosensors Abstract | Full Text: HTML PDF (138K)
•
63 Microarray Analysis Software and its Applications Abstract | Full Text: HTML PDF (620K)
•
64 Data Validation and Interpretation Abstract | Full Text: HTML PDF (147K)
•
65 Introduction to Bayesian Methods for Biosensor Design Abstract | Full Text: HTML PDF (153K) • Part Eight Biosensor Applications
•
Areas and Examples of Biosensor Applications Full Text: HTML PDF (40K)
•
66 Genetic and Other DNA-Based Biosensor Applications Abstract | Full Text: HTML PDF (329K)
•
67 Examples of Biosensors for the Measurement of Trace Medical Analytes Abstract | Full Text: HTML PDF (154K)
•
68 Biosensors for Monitoring Metabolites in Clinical Medicine Abstract | Full Text: HTML PDF (87K)
•
69 Need for Biosensors in Infectious Disease Epidemiology Abstract | Full Text: HTML PDF (101K)
•
70 Biosensors for Neurological Disease Abstract | Full Text: HTML PDF (140K)
•
71 Utility of Biosensors in the Pharmaceutical Industry Abstract | Full Text: HTML PDF (159K)
•
72 Glucose Measurement Within Diabetes via “Traditional” Electrochemical Biosensors Abstract | Full Text: HTML PDF (491K)
•
73 Field-Operable Biosensors for Tropical Dispatch Abstract | Full Text: HTML PDF (189K)
•
74 Lateral-Flow Immunochromatographic Assays Abstract | Full Text: HTML PDF (310K)
•
75 Chip-Based Biosensors for Environmental Monitoring Abstract | Full Text: HTML PDF (83K)
•
76 Environmental Biochemical Oxygen Demand and Related Measurement Abstract | Full Text: HTML PDF (145K)
•
77 Optical Biosensor for the Determination of Trace Pollutants in the Environment Abstract | Full Text: HTML PDF (162K)
•
78 Food and Beverage Applications of Biosensor Technologies Abstract | Full Text: HTML PDF (104K)
•
79 Agriculture, Horticulture, and Related Applications Abstract | Full Text: HTML PDF (119K)
•
80 From Earth to Space: Biosensing at the International Space Station Abstract | Full Text: HTML PDF (420K)
•
81 Life Detection within Planetary Exploration: Context for Biosensor and Related Bioanalytical Technologies Abstract | Full Text: HTML PDF (239K) • Part Nine Business and Regulatory Issues
•
Commercialization, Business and Regulatory Issues Full Text: HTML PDF (38K)
•
82 Biacore—Creating the Business of Label-Free Protein-Interaction Analysis Abstract | Full Text: HTML PDF (273K)
•
83 Commercialization of DNA Arrays—Affymetrix a Case Study Abstract | Full Text: HTML PDF (82K)
•
84 RAPTOR: Development of a Fiber-Optic Biosensor Abstract | Full Text: HTML PDF (273K)
•
85 Regulatory and Validation Issues for Biosensors and Related Bioanalytical Technologies Abstract | Full Text: HTML PDF (126K) • Part Ten Implications, Trends and Perspectives
•
The Future Full Text: HTML PDF (44K)
1 Overview of Biosensor and Bioarray Technologies Christopher R. Lowe Institute of Biotechnology, University of Cambridge, Cambridge, UK
1 INTRODUCTION
Biosensors and related bioarray techniques represent the end product of a rapidly growing field, which combines fundamental biological, chemical, and physical sciences with engineering and computer science to satisfy needs in a broad range of application areas. Not surprisingly, therefore, the term biosensor has different connotations depending on what field the user comes from: for example, to the biologist, a biosensor is “a device, which translates biological variables such as electric potentials, movement, or chemical concentrations into electrical signals”. To the chemist, a more apt definition might be “a device that uses specific biochemical reactions mediated by isolated enzymes, immunosystems, tissues, organelles, or whole cells to detect chemical compounds, usually by electrical, thermal, or optical signals”. The physicist might define a biosensor as “a device, which detects, records, and transmits information regarding a physiological change or process”. For the purpose of this review, the author defines a biosensor as “an analytical device, which converts the concentration of the target substance, the analyte, into an electrical signal through a combination of a biological or biologically derived recognition system either integrated within or intimately associated with a suitable physico-chemical transducer”.1
The term biosensor first appeared in the scientific literature in the late 1970s and the field was reviewed several times in the early 1980s.1–3 However, the first “biosensor” was generally recognized as being introduced by Clark in 19564 and subsequently exemplified by Clark and Lyons in 1962 by sandwiching soluble glucose oxidase (GOx) between an outer dialysis membrane and the gas permeable membrane of an amperometric oxygen (O2 ) electrode.5 The reduction in the concentration of dissolved oxygen was detected by the electrode and shown to be proportional to the concentration of glucose in the sample. Later, Updike and Hicks6 used a second O2 electrode to correct for oxygen variations in the sample. It was soon realized that enzyme electrodes for a variety of other clinically important analytes could be created by coupling relevant enzymes to appropriate electrode systems.7 Rechnitz8 described a selective electrode for arginine in 1977 by immobilizing living microorganisms on the surface of an NH3 gas-sensing electrode and used the term bioselective sensor. This term was subsequently shortened to “biosensor” and has remained the popular choice for any analytical device, which combines a recognition system of biological origin and a physico-chemical transducer. The biological or biologically derived element is capable of recognizing the presence, activity, or concentration of a specific target analyte in a complex mixture of other components. The recognition
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
BIOSENSOR AND BIOCHIP TECHNOLOGIES
element may comprise one of three different types: affinity biosensors are based on ligand-receptor interactions such as those involving antibodies, nucleic acids, aptamers, peptides, protein, or cell receptors, while the second principal class involves binding and catalysis, and involves enzymes, abzymes, microorganisms, organelles, plant or animal cells or tissue slices, and the third involves biomimetic receptors based on various synthetic binding and/or catalytic systems. The interaction or subsequent reaction of the recognition element with the analyte in the sample matrix results in a measurable change in a solution property, such as depletion of a reactant or formation of a product, immediately proximal to the transducer. The latter converts the change in solution property into a quantifiable and processible electrical signal. The transducer is a device, usually electronic, electroacoustic, electro-optical, electromagnetic, electrothermal, or electromechanical, that converts one type of energy (electricity, sound, light, magnetism, heat, or mechanical) into another (usually electrical) for various purposes including measurement or information transfer.
2 THE BIORECOGNITION SYSTEM
The fundamental and crucial distinguishing feature of a biosensor is the recognition system for the target analyte, since this determines both the selectivity and sensitivity of the device. Generally speaking, choice of these biological materials depends on the nature of the target analyte, its concentration in the sample matrix, the presence or absence of interfering substances, and whether the measurement is discrete “one-shot”, real-time, or continuous.
2.1
Biocatalytic Systems
Biosensors that use single or multiple enzyme sequences as the recognition and response elements are the most extensively investigated area where the target analyte is amenable to enzymatic modification and its concentration is typically in the range 10−1 –10−7 M. The high specificity and high turnover rates of enzymatic reactions suggests ways to create tailor-made sensitive and
specific enzyme-based biosensors for their respective substrates.9–12 The enzyme catalyst is integrated either on, adjacent to, or into the materials comprising the transducer in order to ensure that the biocatalytic transformation is selectively transduced into the electrical signal. From an analytical perspective, the most important classes of enzymes are the oxido–reductases, which catalyze the oxidation of substrates using molecular oxygen, NAD+ , or pyrroloquinoline quinine (PQQ), and the hydrolases, which catalyze the hydrolysis of substrates.2,3 The oxidation of glucose catalyzed by GOx is an example of the archetypal enzyme biosensor: Glucose + O2 → Gluconic acid + H2 O2 H2 O2 → 2H+ + O2 + 2e−
(1) (2)
with the steady-state current produced by oxidation of the product hydrogen peroxide on a platinum electrode poised positive (∼ + 0.7 V) relative to an Ag/AgCl reference electrode.13 Good precision and accuracy were obtained with 100-µl blood samples, but since then a wide range of amperometric enzyme electrodes, differing in electrode configuration, material, membrane composition, or immobilization approach have been described.12 Intense activity during the 1980s focused on the development of mediator based “second-generation” glucose biosensors,14,15 the use of modified electrodes for enhancing the sensor performance16,17 , and the introduction of commercial strips for self-monitoring of blood glucose.18,19 During the late 1980s and early 1990s, attention was focused on promoting direct electrical communication between the redox center of GOx and the electrode surface20–22 and the development of minimally-invasive subcutaneously implantable devices.23–25 Enzymes isolated from extremophilic organisms such as thermophilic, hyperthermophilic, psychrophilic, halophilic, acidophilic, and alkaliphilic organisms have extended the range of conventional enzyme-based biosensors26,27 by improving key operational parameters. Furthermore, enzymes have been cloned, overexpressed and engineered by site-directed and other mutagenesis techniques to improve sensor performance such as lifetime, thermostability, pH tolerance, and kinetic properties. For example, the PQQ-dependent glucose
OVERVIEW OF BIOSENSOR AND BIOARRAY TECHNOLOGIES
dehydrogenase (GDH) from Klebsiella pneumoniae has been engineered to enhance thermostability by a single amino acid replacement,28 luciferase stability has been improved by multiple replacements29 and the Km value for a cyanidase from Pseudomonas stutzeri has been made more suitable for its analytical application.30 Organelles (mitochondria, microsomes, and chloroplasts), whole cells (bacteria) and tissue slices (plant and animal sources) have been used extensively as biocatalytic packages in biosensors for a large range of analytes of clinical, environmental, or defence interest.31 Such packages include all the numerous enzymes and other cofactors required to convert substrates into products in an evolutionary optimized environment.32,33 However, a major drawback with the use of such systems is their multiple content of enzymes, which results in decreased substrate specificity and relatively slow response times. Nevertheless, whole-cell systems also have advantages since by altering the external conditions, different substrates can be measured with the same biocatalytic system, and the use of inhibitors, activators and stabilizing agents can be used to optimize the selectivity and lifetimes of cell- and tissue-based biosensors.33 Whole-cell sensors are highly appropriate for the monitoring of genotoxic substances34 and obviate the expense and time required to perform bioassays. Wholecell assays based on the use of geneticallyengineered bioluminescent bacteria have become very popular since they are more rapid and costeffective than traditional methodologies.35
if durable antibodies with affinities >10−10 M are used.39 Furthermore, a high affinity constant coupled with a labile immobilized antibody, makes regeneration of the transducer surface difficult and limits practical application to single-use devices.40 Natural receptor systems have also been investigated as recognition-response elements in biosensors. However, while ion-channel devices have been proposed,41 the stability of the lipid membrane, the incorporation of the receptor into these structures, and the combination of the modified membrane with the transducer initially proved problematic and conspired to limit their introduction in practice. However, the use of a stable membrane from a thermophilic bacterium containing gramicidin ion channels coupled to a microgold electrode (10 bonds) but not when the chains are short (80 mM). The enzyme concentration is 1 mg cm−3 . (b) Plot of equation (22) for the ferrocene acetic acid modified enzyme. Three different enzyme concentrations were used in the analysis (0.45, 1.13, and 4.5 mg cm−3 ). [Reprinted with permission Bartlett et al.28 copyright 1987, Royal Society of Chemistry.]
8
THE BIOLOGY – MATERIALS INTERFACE
exhibited clean and well-defined quasi-steadystate electrochemical responses at a glassy carbon detector electrode in dilute oxygen-free aqueous phosphate buffer solution. It is also clear from the latter diagram that the redox mediator thermodynamics (expressed as a half wave potential value) depends markedly on mediator type (typically E1/2 (SCE) = 0.30–0.33 V, ferrocene carboxylic acid; 0.13–0.18 V, ferrocene acetic acid; 0.09–0.11 V) and that for a given substrate concentration and enzyme concentration the optimal reactivity was exhibited by ferrocene acetic acid–modified glucose oxidase. Bartlett and coworkers28 estimated values for kC and KM for each of the modified enzymes from an analysis using equation (22). Their results are presented in Table 1. Note that the catalytic rate constant kC is considerably greater for the ferrocene acetic acid–modified enzyme and this explains the trend in current magnitude observed in the voltammograms. The analysis using equation (22) was replicated over a range of enzyme concentrations and the linearity predicted from the equation confirmed for all modified enzyme systems studied. The largest error was observed for low glucose concentrations, but the linearity exhibited by a typical plot (such as that outlined in Figure 5b) is very good and is supportive of the theory. We now move to the situation where a membrane is used to enclose a thin layer of solution next to the detector electrode and reexamine the reaction-diffusion expression outlined in
equation (6) but now substitute a new boundary condition at the enzyme solution/boundary membrane interface at χ = 1. Therefore, equation (9) is replaced by a zero flux condition, which reads du =0 (23) χ = 1, dχ 1 Therefore, through integration of equation (6) subject to equations (8) and (23) we obtain the following expression for the normalized concentration of oxidized enzyme adjacent to the electrode surface: u(χ) =
κγ −1/2 √ √ cosh[ γ χ] −1/2 κγ + tanh γ √ √ (24) − tanh γ sinh[ γ χ]
The concentration of reduced enzyme is given by v(χ) = 1 − u(χ). Specifically at χ = 0, u = u0 and we write u0 =
κγ −1/2 √ κγ −1/2 + tanh γ
(25)
and can immediately write the expression for the normalized reaction flux by noting √ tanh γ = 1 − u0 = √ κγ −1/2 + tanh γ =
1
1 + κγ
−1/2
coth
√
(26)
γ
Table 1. Experimental data (Ref. 28) obtained for electro-enzyme direct reaction
Enzyme modifier Glucose oxidase E1/2 (V) (vs saturated calomel reference electrode (SCE)) Number of ferrocene per enzyme kC (s−1 ) KM (mM−1 ) kU = kC /KM (dm3 mol−1 s−1 ) (a)
Ferrocene carboxylic acid
Ferrocene acetic acid
Ferrocene butanoic acid
0.3–0.33
0.13–0.18
0.09–0.11
2
13
22
29
800 20 40 × 103
5 1 5 × 103
1100 5 220 × 103
50 2 25 × 103
(a)
ca −0.4
Data obtained in author’s laboratory from analysis of cyclic voltammetry data of GOx adsorbed on carbon electrodes modified with a mesh of single walled carbon nanotubes in phosphate buffer pH 7.
MODELING OF BIOSENSOR INTERFACES
Again, we can deconstruct the latter expression by taking suitable limiting approximations. For instance when√the reaction/diffusion parameter γ √ is small, tanh γ ∼ = γ and equation (26) reduces to: √ γ 1 ∼ (27) = √ = −1/2 κγ + γ 1 + κγ −1 Alternatively when γ is large, tanh equation (26) reduces to: ∼ =
√ ∼ γ = 1 and
1 1 + κγ −1/2
(28)
which is the same as equation (15) obtained for the membrane free case. Again, looking at equation (27), when γ ≪ 1 we get two limiting cases depending on whether κγ −1 ≪ 1 or κγ −1 ≫ 1. First, if κγ −1 ≪ 1, then κ ≪ γ . We recall that the former parameter compares the flux of reduced enzyme oxidation at the electrode to the diffusive flux of reduced enzyme ′ to the site of reoxidation (κ = kET /kD = fET /fD ); whereas the latter parameter compares the flux for the homogeneous enzyme/substrate kinetics to the transit time for enzyme diffusion across the solution layer (γ = k/kD = fES /fD ). Therefore, the product κγ −1 = (fET /fD )(fD /fES ) = fET /fES compares the flux of reduced enzyme oxidation at the detector electrode surface to the flux arising from the bimolecular homogeneous enzyme/substrate reaction within the diffusion layer. Consequently, when κ ≪ γ , fET ≪ fES and the regeneration of oxidized enzyme is slow and rate determining. Under such circumstances, the normalized flux reduces to ∼ = 1 or ′ f = kET e
(16)
which again is case I, met previously for the membrane free situation. The current flow depends
9
only on the concentration of enzyme, may exhibit a potential dependence, and will be independent of substrate concentration. Conversely when κγ −1 ≫ 1, κ ≫ γ and fET ≫ fES . Here, oxidized enzyme regeneration is fast and the Michaelis–Menten enzyme/substrate kinetics is slow and rate determining. Here the normalized flux reduces to ∼ = 1 = κ −1 γ and the net reaction flux is given by κγ −1 f = ke δ =
kC e sδ KM + s
We label this situation case IV. In this case, the flux depends on substrate concentration according to the Michaelis–Menten rate law but also depends on the thickness of the solution layer trapped behind the membrane. When s ≪ KM , we have k f ∼ = KC e δs = kU e δs and we have case IVA M corresponding to unsaturated enzyme kinetics. In contrast, when KM ≪ s then f ∼ = kC e δ, we have case IVB, and we have saturated enzyme kinetics. More generally, inversion of equation (29) yields the following: 1 1 1 1 · + = f (kC /KM )e δ s kC e δ
(30)
and a plot of f−1 versus s −1 is linear with a slope given by 1/kU e δ and intercept 1/kC e δ. The situation for large γ yields equation (15), previously obtained for the semi-infinite situation. Again we have two limiting cases depending on the magnitude of the product κγ −1/2 . Firstly, when κγ −1/2 ≪ 1, ∼ = 1 and we regain case I corresponding to rate-determining electrode kinetics of enzyme regeneration. In contrast, when κγ −1/2 ≫ 1, the normalized flux reduces to ∼ = κ −1 γ 1/2 , which we have labeled case III. We can summarize the kinetic results obtained to date in Table 2. We can also geometrically represent the analysis in terms of a kinetic case diagram. This
Table 2. Summary of pertinent rate-limiting expressions for direct electro-enzyme reaction
Kinetic case
(29)
Normalized substrate flux
Heterogeneous enzyme oxidation kinetics I Reduced enzyme diffusive transport II
Modified Michaelis–Menten III
Bounded modified Michaelis–Menten IV
∼ =1 ∼ = κ1 γ 1/2 ∼ = κ γ ∼ = κ
Substrate flux ′ f = kET e k D e f = f = KkC Ds + s e M
kC e sδ f = K M +s
10
THE BIOLOGY – MATERIALS INTERFACE Membrane free direct enzyme case −1/2
kg
log k
ψ≅k
−1
III ψ ≅ k−1g1/2
II
ψ ≅ 1 −1/2 1 + kg
log g
k=1 ψ≅
=1
1 1+ k
I ψ≅1
g=1 1/2
ψ=
(a)
coth g kg−1/2 + coth g1/2
Membrane bounded direct enzyme case log k kg−1/2 = 1
ψ ≅ gk−1
−1 1/2 III ψ ≅ k g
ψ≅
1 −1/2 1 + kg
IV log g
k=1 I ψ ≅ 1 −1 1 + kg
ψ≅1
kg−1 = 1 ψ=
(b)
g=1 tanh g1/2 kg−1/2 + tanh g1/2
Figure 6. Kinetic case diagram (plot of log κ vs log γ ) for (a) membrane free direct reaction and diffusion of electroenzyme and (b) membrane bounded direct reaction and diffusion of electro-enzyme. Note that κ compares the rate of enzyme regeneration at the electrode with that of enzyme diffusion through the solution, whereas γ compares the rate of enzyme diffusion to that of homogeneous reaction between enzyme and substrate. In both cases approximate limiting expressions for the normalized flux and the expressions delineating the boundaries between specific cases are presented.
is presented in Figure 6. The natural axes defining the case diagram are log κ and log γ . The membrane free direct enzyme case is outlined in Figure 6(a), whereas the membrane bound situation is presented in Figure 6(b). Three kinetic sub-cases (I, II, and III) are relevant for the semi-infinite membrane free situation. The bounded membrane situation is well described by the sub-cases labeled I, III, and IV. Common to both are cases I and III. Case II is found only in the semi-infinite membrane free case, whereas case IV is specific to the membrane bound situation. In Figure 6(a) the II/III case boundary lies at γ = 1, the I/II boundary is at κ = 1, and the
I/III boundary is set at κγ −1/2 = 1. In Figure 6(b) we note that the I/III boundary is again defined by the line κγ −1/2 = 1, whereas the I/IV boundary is defined by the line κγ −1 = 1, and the III/IV boundary is γ = 1. We can directly compare the amperometric response obtained for the membrane free situation with that obtained for the membrane bound situation by taking the ratio of the normalized fluxes: √ √ MF coth γ (κγ −1/2 + tanh γ ) = √ √ MB tanh γ (κγ −1/2 + coth γ ) √ 1 + κγ −1/2 coth γ (31) = √ 1 + κγ −1/2 tanh γ This ratio is illustrated schematically in Figure 7 for various values of the parameters κ and γ . We note that for all values of κ, MF /MB → 1 as γ → ∞. Furthermore, the ratio MF /MB increases significantly as γ decreases. The rate of increase in the latter ratio with decreasing γ value is more marked for lower values of the parameter κ. In short the steady-state amperometric response expected for a sensor that does not have a bounding membrane coating is significantly larger than that recorded for a sensor containing a membrane when the homogeneous enzyme/substrate reaction kinetics is slower compared to the rate of diffusive movement of enzyme across the diffusion layer and when the rate of oxidized enzyme regeneration at the electrode surface is much smaller than that of enzyme diffusion.
3 AMPEROMETRIC ENZYME ELECTRODES USING SELF-ASSEMBLED MONOLAYER THIN FILMS
The immobilization of redox enzymes onto the surface of self-assembled monolayers has recently been the subject of considerable interest.29–36 Indeed in a recent review by Wilner and Katz3 it is stated that “Integration of redox enzymes with an electrode support and formation of an electrical contact between the biocatalyst and the electrode is the fundamental subject of bioelectronics and optobioelectronics.” In this section we develop a simple kinetic model to describe the operation of an amperometric
MODELING OF BIOSENSOR INTERFACES ΨMF ΨMF ΨMB
∼ =
1 + k g−1
ΨMB
=
11
1 + k g−1/2 coth g1/2 1 + k g−1/2 tanh g1/2
1+k
1000 ΨMF/ΨMB
k = 0.01 k = 0.1 k=1 k = 10 k = 100
100
ΨMF
10
ΨMB
∼ =1
1
0.1 0.01
0.1
1 g
10
100
Figure 7. The ratio of the amperometric response obtained for the membrane free configuration to that corresponding to the membrane bound situation as a function of the normalized parameter γ that compares the rate of enzyme diffusion in the solution region next to the detector electrode, with the rate describing the reaction between enzyme and substrate in the solution. The flux ratio is presented also for various values of the parameter κ that compares the rate of the reduced enzyme reaction at the detector electrode surface to that of reduced enzyme diffusion to the electrode surface.
enzyme electrode where the enzyme is immobilized within a self-assembled monolayer and where the electronic communication between the enzyme and the electrode is effected by a mobile electronic relay mediator species. This analysis will build on previous work reported by Bourdillon and coworkers,37 Tatsuma and Watanabe,38 Bartlett et al.39 and most recently by Gooding et al.36 and Lyons.40 The simplified model adopted in the present paper is schematically presented in Figure 8. The redox enzyme is assumed to be located within a thin nonconductive layer of thickness L deposited onto the surface of a conducting support electrode. This layer may be an alkane thiol film formed through adsorptive self-assembly from solution. We assume that the enzyme is homogeneously distributed in a plane located at x = L, at the interface between the thiol film and the adjacent solution. We consider the following reaction scheme: k1
kC
S + Eox ⇀ ↽ ES → P + Ered k−1
k
Ered + A → Eox + B k
′
B→ A
Enzyme layer
a0
Diffusion layer
B
S
a∞
P
A
aL
s∞
B B A
sL bL
A
L
d
Bulk solution
Figure 8. Schematic representation of immobilized enzyme electrode using a soluble redox mediator. A denotes the oxidized mediator and B is the reduced mediator. Concentration polarization of substrate S and mediator within the enzyme layer is neglected but a Nernst diffusion layer treatment for substrate and mediator transport in the solution is adopted.
where the oxidized form of the enzyme Eox reacts with the substrate S to form (through an enzyme/substrate complex as intermediate)
12
THE BIOLOGY – MATERIALS INTERFACE
product P and reduced enzyme Ered . The oxidized form of the mediator species A partitions from solution into the enzyme layer and reacts with the reduced enzyme to regenerate the active oxidized enzyme and produce the reduced form of the mediator B. A can subsequently be electrochemically regenerated at the support electrode through the reaction of B, which we assume to occur with kinetic facility. Therefore, the surface concentrations of reduced and oxidized mediator species b0 and a0 are related through the Nernst equation. Of course when measuring the steady-state amperometric response of electrodes of this type in practice, it is usual to hold the electrode at a sufficiently oxidizing potential that the mediator is entirely in its oxidized form at the electrode surface. Under such conditions we can set b0 = 0. In the following analysis this assumption is made. We also assume that the substrate S is electrochemically inert and does not react directly at the support electrode surface. It only interacts with the redox enzyme. We assume that the value of the mediator concentration in the bulk solution is a ∞ and is present only in its oxidized form. Therefore, the concentration of the reduced form of the mediator B is 0 outside the diffusion layer, which is assumed to have a thickness δ. The observed current i is related to the net flux f corresponding to the reaction of mediator B at the electrode surface through the Faraday relationship and is given by i db = f = DB (32) nF A dx 0 where n denotes the number of electrons transferred, F is the Faraday constant, and A represents the geometric area of the electrode, and DB denotes the diffusion coefficient of reduced mediator species in the enzyme layer, which differs from the corresponding diffusion coefficient in the bulk solution. This experimentally measurable flux f must be related to the substrate reaction flux fS within the enzyme layer. We correspondingly show, following an argument initially proposed by Bartlett et al.39 that the required relationship between the fluxes is given by: f = η fS
(33)
where we determine that 1/2 ≤ η ≤ 1. The η parameter reflects the fact that the flux of substrate reacting within the film is not necessarily the same as the flux of reduced mediator detected at the electrode. This is because some reduced mediator can diffuse through the thiol layer away from the electrode and be lost into the adjacent solution. We see that the precise value of η depends on how effectively species B moves away from the electrode. Since the immobilization layer is thin we can neglect diffusive depletion of substrate and mediator within the latter, and so we need only consider the Michaelis–Menten kinetics between the oxidized enzyme Eox and the substrate S and the bimolecular kinetics between the oxidized mediator A and the reduced enzyme Ered . Under steadystate conditions both of these processes are in balance, and we can write fS =
kC eox L κS sL = kLered κA aL κS sL + KM
(34)
where κS , κA denote the partition coefficients for substrate and mediator species respectively and aL , sL denote the oxidized mediator and substrate concentrations at the outer edge of the monolayer at x = L. As previously noted L denotes the thickness of the immobilization layer and eox , ered denote the concentrations of the oxidized and reduced forms of the enzyme. Note also that kC and KM represent the catalytic rate constant and Michaelis constant that are significant parameters, which define the enzyme kinetics, and k denotes the bimolecular rate constant quantifying the kinetics of the reaction between oxidized mediator and reduced enzyme. We let e = eox + ered denote the total enzyme concentration contained within the monolayer. From equation (34) we can show that eox =
(κS sL + KM )kκA aL e kC κS sL + (κS sL + KM )kM κA aL
(35)
The substrate reaction flux within the enzyme layer is then given by kC κS sL Leox κS sL + KM kC κS sL LkκA aL e = kC κS sL + (κS sL + KM )kM κA aL
fS =
(36)
MODELING OF BIOSENSOR INTERFACES
We now need to relate the substrate and mediator concentrations at the outer edge of the monolayer at x = L to the corresponding quantities in the bulk solution. This is done using the Nernst diffusion layer approximation where we note that fS = DS′
= DA′
s ∞ − sL δ
a ∞ − aL δ
= kDS (s ∞ − sL )
= kDA (a ∞ − aL )
DS′ δ
kDA =
DA′ δ
(37)
(38)
We can readily show that
fS fS = kDS 1 − sL = s − kDS fDS
fS fS ∞ = kDA 1 − aL = a − kDA fDA ∞
(39)
where fDS and fDA represent the mass transport controlled fluxes of substrate and oxidized mediator in the solution with fDS = kDS s ∞ and fDA = kDA a ∞ . Now we note from equation (39) that if fS fS fDS ≪ 1 and fDA ≪ 1, then we can neglect the concentration polarization of S and A in the solution and we can set sL ∼ = s ∞ and aL ∼ = a ∞ . This approximation is often used in the literature. If the expressions presented in equation (39) are substituted into equation (36), we can, after quite an amount of algebra, obtain the following cubic equation in substrate flux: AfS3 − BfS2 + CfS − D = 0
(40)
where we can show that A=
kκA κS a ∞ s ∞ fDS fDA
B = kκA κS a ∞ s ∞
1 1 kC e L + + fDS fDA fDS fDA
kC κ S s ∞ kKM κA a ∞ + fDA fDS
C = kC κS s ∞ + kκA a ∞ (KM + κS s ∞ )
1 1 ∞ ∞ + kkC e LκS κA a s + fDS fDA
D = kkC e LκS κA a ∞ s ∞
where DS′ and DA′ denote the diffusion coefficients of substrate and mediator in the solution phase and we have introduced the mass transport rate constants kDS =
+
13
(41)
In a recent paper dealing with monolayer enzyme electrodes fabricated using self-assembled monolayers of alkanethiols, Gooding and coworkers36 have derived a similar cubic expression for the substrate flux (equation 7 of Ref. 36) and proceeded to solve the cubic equation with a negative discriminant to obtain an expression for fS and hence the observed current i. Indeed they fit their experimental batch amperometry curves to the analytical solution of the cubic equation. This analysis procedure of course is valid, but it must be stated that the direct analytical solution of the cubic equation does not confer any real physical insight into the nature of the system. Instead it is preferable to adopt a different strategy as shown subsequently. Substituting equation (39) into equation (36) results in the following expression for the substrate reaction flux kC κS sL LkκA aL e kC κS sL + (κS sL + KM )kM κA aL fS fS ∞ ∞ 1− 1− kC ke LκS κA s a fDS fDA = fS fS ∞ ∞ kC κ S s + kκA a 1− f 1− f DS
DA fS ∞ K M + κS s 1− f
fS =
DS
(42) This rather ungainly expression may be simplified further if we invert both sides to obtain 1 fS −1 1 1− = fS ke LκA a ∞ fDA fS −1 1 1 1 − + + (kC /KM )e LκS s ∞ fDS kC e L (43) The latter expression is now physically transparent in that the first term is related to the bimolecular reaction between reduced enzyme and oxidized mediator, the second term relates to the
14
THE BIOLOGY – MATERIALS INTERFACE
unsaturated kinetics between substrate and oxidized enzyme, and the third term relates to saturated enzyme kinetics involving decomposition of the enzyme/substrate complex to form product and reduced enzyme. The first two terms on the righthand side (rhs) of equation (43) are modified by transport factors involving substrate and mediator diffusion in the solution. For instance if the substrate reaction flux in the layer becomes close to the limit imposed by either substrate transport in the solution or mediator transport in the solution then both the substrate reaction rate and the mediator reaction rate will be diminished by the transport f f terms TS = 1 − f S or TA = 1 − f S . The effecDS DA tive concentration of substrate and mediator at the layer/solution interface will be considerably lesser than their bulk values s ∞ and a ∞ . We note from equation (43) that the substrate reaction flux may be limited by three possible processes: bimolecular kinetics between oxidized mediator and reduced enzyme, unsaturated kinetics between oxidized enzyme and substrate, or saturated enzyme kinetics involving decomposition of ES intermediate. In the analysis so far we have implicitly assumed that the reaction between the oxidized mediator A and the reduced enzyme Ered can be described in terms of a simple bimolecular expression. Karube and coworkers41,42 have stated that this assumption may not necessarily be valid. It is possible that the mediator/enzyme reaction may also be described in terms of Michaelis–Menten kinetics. In such a situation the mediator may bind to the enzyme and form a complex that will subsequently decompose. In this case, equation (4) must be replaced by: fS =
kLered κA aL kC eox L κS sL = ′ κS sL + KM κA aL + KM
analysis just described and show that the inverse substrate reaction flux is now given by: 1 1 = ′ fS (k/KM )e LκA a ∞
1−
1 + (kC /KM )e LκS s ∞ +
fS fDA
−1
fS 1− fDS
1 1 + kC e L ke L
−1 (45)
The first term on the rhs of equation (45) corresponds to rate-determining unsaturated mediator/ enzyme kinetics modified by a mediator transport term. The unsaturated rate constant is given by kU′ = k′ . The second term corresponds to rateKM determining unsaturated enzyme/substrate kinetkC . The third ics, with a rate constant kU = K M term corresponds to saturated enzyme kinetics involving rate-determining decomposition of the enzyme/substrate complex, and the fourth and final term corresponds to the saturated kinetics involving decomposition of the mediator/enzyme complex. It is clear that the development of a general expression for the inverse substrate flux provides much more information pertaining to the reaction mechanism than solving the cubic equation. We now follow the procedure adopted in the previous section and develop a kinetic case diagram for an immobilized enzyme monolayer electrode system. In this analysis we neglect concentration polarization of mediator and substrate in solution. We assume that both the mediator/enzyme reaction and the substrate/enzyme reaction are described by Michaelis–Menten kinetics. Under such circumstances the reaction flux is given by:
(44)
′ where KM denotes the Michaelis constant for the mediator species. We note that when the mediator concentration in the layer is smaller ′ than the Michaelis constant κA aL ≪ KM , the expression outlined in equation (44) reduces to that previously presented in equation (4), but in this case the bimolecular rate constant takes the ′ form k/KM . For the more complicated situation where the mediator/enzyme reaction exhibits Michaelis–Menten kinetics, we can repeat the
kC ke LκS κA s ∞ a ∞ ′ (κS s ∞ +KM )kκA a ∞ +(κA a ∞ + KM )kC κS s ∞ (46) We now introduce a normalized substrate flux given by: fS =
S =
fS fS,max
=
fS kC e L
(47)
where fS,max denotes the maximum enzyme turnover rate. We also introduce saturation parameters
MODELING OF BIOSENSOR INTERFACES
α and β for substrate and mediator as follows: α=
κS s ∞ KM
β=
κA a ∞ ′ KM
(48)
We finally introduce a kinetic competition parameter γ as: ′ )κA a ∞ fME kU′ κA a ∞ e L (k/KM = = ∞ ∞ (kC /KM )κS s kU κ S s e L fSE (49) Therefore, γ compares the mediator/enzyme reaction flux to the substrate/enzyme reaction flux. When γ ≪ 1 then fME ≪ fSE and the net flux is limited by the kinetics of the bimolecular reaction between mediator and enzyme. In contrast when γ ≫ 1 then fME ≫ fSE and the net flux is limited by the kinetics of the reaction between substrate and enzyme. The substrate saturation parameter α compares the value of the substrate concentration in the layer κS s ∞ to the Michaelis constant KM for substrate. When α ≪ 1, κS s ∞ ≪ KM and we have unsaturated enzyme kinetics. In contrast, when α ≫ 1, κS s ∞ ≫ KM and saturated enzyme kinetics pertain. The mediator saturation parameter β compares the oxidized mediator concentration within the layer, κA a ∞ , to the Michaelis constant ′ ′ for the mediator KM , . When β ≪ 1, κA a ∞ ≪ KM and unsaturated mediator kinetics pertain. This is the situation usually considered in the literature. On the other hand when the mediator concentration ′ within the layer is large, κA a ∞ ≫ KM , saturated mediator kinetics will apply and β ≫ 1. If equations (47–49) are substituted into equation (46), one obtains the following expression for the normalized substrate flux in the layer:
γ =
S =
αγ γ (1 + α) + 1 + β
(50)
This normalized expression is the most general and is valid for all mediator and substrate concentrations. We can now simplify the analysis and assume firstly that the concentration of mediator within the film is low. Under these conditions β ≪ 1 and equation (50) reduces to: S ∼ =
αγ 1 + γ (1 + α)
(51)
15
We now can simplify the latter expression further depending on the value of the substrate saturation parameter α. When the substrate concentration in the layer is small, α ≪ 1 and equation (51) reduces to: αγ S ∼ (52) = 1+γ This expression is valid for the situation where the mediator and enzyme kinetics are unsaturated. We can simplify still further by examining suitable limiting values of the competition parameter γ . Firstly when γ ≪ 1 we recall that the substrate reaction flux is limited by reaction between oxidized mediator and reduced enzyme. Here the normalized flux is given by: S ∼ =αγ
(53)
We label this case IA. Transforming into an expression for the flux we get: k fS ∼ = ′ e LκA a ∞ = kU′ e LκA a ∞ KM
(54)
Therefore, when the reaction between oxidized mediator and reduced enzyme is rate determining, the flux in the film should exhibit a first-order dependence on the bulk concentration of oxidized mediator, provided the concentration of mediator is not too large. The flux should also be independent of the bulk substrate concentration and exhibit a first-order dependence both on enzyme loading and layer thickness. Secondly, when γ ≫ 1 the substrate reaction flux in the layer is limited by the unsaturated reaction kinetics between oxidized enzyme and substrate. In this case the normalized flux takes the form: S ∼ =α
(55)
We label this situation case IB. Retransforming into the usual variables we get kC fS ∼ e LκS s ∞ = kU e LκS s ∞ = KM
(56)
Therefore, we note that the reaction flux should be first order with respect to bulk substrate concentration, independent of mediator concentration, and first order with respect to enzyme loading and
16
THE BIOLOGY – MATERIALS INTERFACE
layer thickness. Therefore, case IA and IB pertain when α ≪ 1 and when β ≪ 1. We now turn to the situation where α ≫ 1 and β ≪ 1. Reexamination of equation (51) indicates that the approximate expression for the normalized flux is now given by: S ∼ =
αγ 1+α γ
(57)
Again we get two limiting cases depending on the value of the product αγ . Firstly when αγ ≪ 1 equation (57) reduces to S ∼ =αγ
(53)
which was obtained previously, and case IA is obtained again. Secondly, when αγ ≫ 1 equation (57) reduces to: S ∼ =1
(58)
We label this case II. Here the reaction flux is given by: fS ∼ (59) = kC e L and corresponds to rate-determining saturated enzyme kinetics. Here the flux is independent both of bulk substrate and mediator concentrations but depends linearly on enzyme loading and layer thickness. We have identified three cases (IA, IB, and II) when the mediator concentration in the layer is low. We now turn to the situation when the opposite pertains. In this case equation (50) reduces to αγ S = γ (1 + α) + β
(60)
Again we can simplify by taking the small α and large α limits. Firstly, when α ≪ 1 we get S =
αγ γ +β
(61)
We now compare the magnitudes of the normalized parameters γ and β. When γ ≪ β, equation (61) reduces to S =
αγ β
(62)
We label this situation case III. Transforming to the usual expression for the flux we obtain fS ∼ = ke L
(63)
Therefore, case III corresponds to the case of saturated mediator kinetics where the decomposition of the mediator/enzyme complex to form reduced mediator and oxidized enzyme is rate determining. Here the flux is independent of bulk mediator concentration and bulk substrate concentration, but depends in a first-order manner on enzyme concentration and layer thickness. On the other hand, when γ ≫ β, we get S ∼ =α
(55)
which again is case IB. Therefore, case IB pertains also when β ≫ 1 and so holds for the entire range of the β parameter. Turning again to equation (60), which holds for the case where the mediator concentration in the layer is high, and considering the case α ≫ 1, we obtain S =
αγ α γ +β
(64)
Again we can get two possible limits by comparing the magnitudes of αγ and β. Firstly when αγ ≪ β we again get case III. S =
αγ β
(62)
Whereas in contrast when αγ ≫ β we get S ∼ =1
(58)
which is case II. Therefore, case III is valid when the mediator saturation factor β is large and is valid for the entire range of bulk substrate concentrations or α values. Its region of validity is determined by the conditions γ ≪ β and αγ ≪ β. Case II corresponding to saturated enzyme kinetics is valid for the entire range of β values, and for large values of α, and subject to the restraints that αγ > β and αγ > 1. Therefore, we have identified three cases (IB, II, III) for the situation where the mediator concentration in the layer is high and all reduced enzyme is bound by mediator. These various mechanistic and kinetic possibilities are presented in terms of a kinetic case
MODELING OF BIOSENSOR INTERFACES
17
log b
log b
III
IB
III
b=g
IB
b >> 1
log g
II
II
b = ag
log g
log a
ag = 1
log a
II
IB IALog
log g II
b Cu2+ > Cd2+ > Co2+ > Pb2+ > Sr2+ ; reactivation of the inhibited enzyme with Ethylenediaminetetraacetic acid (EDTA) was shown to be probable. Conductometric biosensors applied to analyzing total solution toxicity at parathion-methyl photodegradation were presented in Refs 64 and 65. The results obtained were compared with the data from traditional, highly sensitive method of high performance liquid chromatography (HPLC) and from the Lumistox device (Germany) for toxicity determination. The solution’s toxicity was shown to increase dramatically as pesticide photodegradation began, the toxicity remained once the parathion-methyl dissociation had completed. However, the authors do not oppose the biosensor method, but consider it as an additional, fast way for early screening of numerous samples. The comparative analysis of the operational characteristics of enzyme biosensors for penicillin determination based on conductometric planar electrodes, on the one hand, and pH-sensitive fieldeffect transistors, on the other, demonstrated47 similarity in their analytical parameters: both have short response time and high operational stability. However, planar conductometric electrodes are preferable from the technological aspect, because they are cheaper and easier to manufacture, and therefore promising for practical use. The possibility of selecting the necessary dynamic range of the transducer’s operation by varying the medium’s buffer capacity was also shown.
7 CONCLUSIONS
An application of the conductometric measurement method to continuous recording in the course of enzyme processes is thoroughly examined and analyzed regarding both standard conductometers and
conductometric enzyme biosensors. As compared with conventional methods of biochemical analysis, the method considered is universal, features higher accuracy, and lower labor costs. Conductometric biosensors also have advantages over other types of transducers. First, they can be produced through inexpensive thin-film standard technology. This, along with using an optimized method of immobilization of biological material, results in considerable decrease in both primary cost of devices and the total price of analyses. For integral microbiosensors it is easy to perform a differential measurement mode, thus compensating for external effects and considerably increasing measurement accuracy. The data is convincing evidence of the great potential of conductometric biosensors. However, it is still rather a novel trend in the field of biosensors, which is why the development of commercial devices has a promising future.
REFERENCES 1. P. N. Bartlett and R. G. Whitaker, Strategies for the development of amperometric enzyme electrodes. Biosensors, 1987/88, 3, 359–379. 2. I. Karube and K. Sode, Microbial Sensors for Process and Environmental Control , in Bioinstrumentation and Biosensors, D. L. Wise (ed), Marcel Dekker, New York, 1991, pp. 149–160. 3. S. V. Dzyadevych, A. P. Soldatkin, A. V. El’skaya, C. Martelet, and N. Jaffrezic-Renault, Enzyme biosensors based on ion-selective field-effect transistors. Analytica Chimica Acta, 2005, 568, 248–258. DOI: 10.1016/j.aca.2005.11.057. 4. E. Hall, Biosensors, Open University Press, Cambridge, 1991. 5. C. Tran Minh, Biosensors, Chapman & Hall, London, 1993. 6. E. Kress-Rogers, Handbook of Biosensors and Electronic Noses: Medicine, Food, and Environment, CRC Press, New York, 1997. 7. B. S. Hoffheins, R. J. Lauf, and M. W. Siegel, Intelligent thick-film gas sensor. Hybrid Circuits, 1987, 14, 8–12. 8. J. F. Alder, P. R. Fielden, and A. J. Clark, Simultaneous conductivity and permittivity detector with a single cell for liquid chromatography. Analytical Chemistry, 1984, 56, 985–988. 9. E. N. Dorokhova, G. V. Prokhorova, Analytical Chemistry. Physical-Chemical Methods of Analysis, Vysshaya shkola, Moscow, 1991. 10. W. Gopel, T. A. Jones, M. Kleitz, J. Lundstrom, and T. Seiyama, Conductometry, in Sensors. A Comprehensive Survey, W. Gopel, J. Hesse, and J. N. Zemel (eds), VCH Verlagsgesellschaft, Weinheim, 1991, Vol. 2, Part I, pp. 314–337.
CONDUCTOMETRIC ENZYME BIOSENSORS 11. D. B. Kell, The Principles and Potential of Electrical Admittance Spectroscopy: An Introduction, in Biosensors: Fundamentals and Applications, A. P. F. Turner, I. Karube, and G. S. Wilson (eds), Oxford University Press, Oxford, 1987, pp. 427–468. 12. K. Camman, Working with Ion-Selective Electrodes, Springer-Verlag, New York, 1979. 13. K. J. Vetter, Electrochemical Kinetics, Academic Press, New York, 1967. 14. E. T. McAdams and J. Jossinet, Electrode-electrolyte impedance and polarisation. Innovation et Technologie en Biologie et Medecine, 1991, 12, 11–200. 15. L. I. Antropov, Theoretical Electrochemistry, Mir Publisher, Moscow, 1977. 16. L. D. Watson, P. Maynard, D. C. Cullen, R. S. Sethi, J. Brettle, and C. R. Lowe, A microelectronic conductometric biosensor. Biosensors, 1987/88, 3, 101–115. 17. P. T. Kissinger, W. R. Heineman, Laboratory Techniques in Electroanalytical Chemistry, Marcel Dekker, New York, 1984. 18. S. V. Dzyadevich, A. A. Shul’ga, S. V. Patskovsky, V. N. Arkhipova, A. P. Soldatkin, and V. I. Strikha, Thinfilm conductometric transducers for enzyme biosensors. Russian Journal of Electrochemistry, 1994, 30, 887–891. 19. W. Olthuis, A. Volanschi, J. G. Bomer, and P. Bergveld, A new probe for measuring electrolytic conductance. Sensors and Actuators, B, 1993, 13 – 14, 230–233. 20. U. Weimar and W. Gopel, A.c. measurements on tin oxide sensors to improve selectivities and sensitivities. Sensors and Actuators, B, 1995, 26 – 27, 13–18. 21. N. F. Sheppard, R. C. Tucker, and C. Wu, Electrical conductivity measurements using microfabricated interdigitated electrodes. Analytical Chemistry, 1993, 65, 1199–1202. 22. W. Y. Lee, S. R. Kim, T. H. Kim, K. S. Lee, M. C. Shin, and J. K. Park, Sol-gel-derived thick-film conductometric biosensor for urea determination in serum. Analytica Chimica Acta, 2000, 404, 195–203. 23. P. Jacobs, J. Suls, and W. Sansen, Performance of a planar differential-conductivity sensor for urea. Sensors and Actuators, B, 1994, 20, 193–198. 24. R. Hintsche, B. Moller, I. Dransfeld, U. Wollenberger, F. Scheller, and B. Hoffmann, Chip biosensors on thin-film metal electrodes. Sensors and Actuators B, 1991, 4, 287–291. 25. U. Trebbe, M. Niggemann, K. Cammann, G. C. Fiaccabrino, M. Koudelka-Hep, S. Dzyadevich, and O. Shulga, A new calcium sensor based on ion-selective conductometric microsensors–membranes and features. Fresenius’ Journal of Analytical Chemistry, 2001, 371, 734–739. 26. S. K. Mikkelsen and G. A. Rechnitz, Conductometric transducers for enzyme-based biosensors. Analytical Chemistry, 1989, 61, 1737–1742. 27. U. Bilitewski, W. Drewes, and R. D. Schmid, Thick film biosensors for urea. Sensors and Actuators B, 1992, 7, 321–326. 28. D. C. Cullen, R. S. Sethi, and C. R. Lowe, Multi-analyte miniature conductance biosensor. Analytica Chimica Acta, 1990, 231, 33–40. 29. C. J. McNeil, D. Athey, M. Ball, W. On Ho, S. Krause, R. D. Armstrong, J. D. Wright, and K. Rawson, Electrochemical sensors based on impedance measurement
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
13
of enzyme-catalyzed polymer dissolution: theory and application. Analytical Chemistry, 1995, 67, 3928–3935. T. A. Sergeeva, N. V. Lavrik, A. E. Rachkov, Z. I. Kazantseva, S. A. Piletsky, and A. V. El’skaya, Hydrogen peroxide–sensitive enzyme sensor based on phtalocyanine thin film. Analytica Chimica Acta, 1999, 391, 289–297. H.-E. Endres and S. Drost, Optimization of the geometry of gas-sensitive interdigital capacitors. Sensors and Actuators, B, 1991, 4, 95–98. C. Besson, S. Vessillier, T. Gonzales, J. Saulnier, and J. Wallach, Conductimetric assay of pyroglutamyl peptidase activity. Analytica Chimica Acta, 1994, 294, 305–309. S. V. Dzyadevych, Amperometric biosensors. Modern technologies of creation and commercial analyzers. Biopolymers Cell, 2002, 18, 363–376. W. On Ho, S. Krause, C. J. McNeil, J. A. Pritchard, R. D. Armstrong, D. Athey, and K. Rawson, Electrochemical sensor for measurement of urea and creatinine in serum based on ac impedance measurement of enzyme-catalyzed polymer transformation. Analytical Chemistry, 1999, 71, 1940–1946. W. T. Chin and W. Kroontje, Conductivity method for determination of urea. Analytical Chemistry, 1961, 33, 1757–1760. P. Bourrelly and V. Bourrelly-Durand, Methode d’etude par conductometric differentielle de la cinetique de l’hydrolise enzymatique de l’uree. Journal of Chemical Physics, 1965, 65, 673–677. V. S. Andreev, V. I. Rozengart, and V. A. Torubarov, Registration of kinetics of enzymatic reaction by highfrequency method. Ukrainian Biochemical Journal, 1965, 37, 920–926. V. S. Andreev and A. V. Bashtanov, Differential conductometric device for registration of physicalchemical processes. Zavodskaya laboratoriya, 1968, 34, 1546–1548. M. Hanss and A. Rey, Application de la conductometrie a l’etude des reactions enzymatiques. Systeme uree-urease. Biochimica et Biophysica Acta, 1971, 227, 630–638. A. J. Lawrence and G. R. Moores, Conductimetry in enzyme studies. European Journal of Biochemistry, 1972, 24, 538–546. A. A. Shul’ga, S. V. Dzyadevich, A. P. Soldatkin, S. V. Patskovsky, V. I. Strikha, and A. V. El’skaya, Thinfilm conductometric biosensor for glucose and urea determination. Biosensors and Bioelectronics, 1994, 9, 217–223. A. A. Shul’ga, S. V. Dzyadevich, A. P. Soldatkin, S. V. Patskovsky, V. I. Strikha, Conductometric biosensors for glucose and urea based on microfabricated thinfilm interdigitated array-electrodes. Biologi Italiani, 1993, 23(6), 40–45. A. A. Shul’ga, S. V. Dzyadevich, A. P. Soldatkin, S. V. Patskovsky, N. F. Starodub, V. I. Strikha, and A. V. El’skaya, Thin-film conductometric enzymatic biosensor for glucose and urea determination in blood. Russian Journal of Electrochemistry, 1993, 29, 860–864. S. V. Dzyadevich, A. P. Soldatkin, A. A. Shul’ga, V. I. Strikha, and A. V. El’skaya, Conductometric biosensor for organophosphorus pesticides determination. Journal of Analytical Chemistry, 1994, 49, 789–792.
14
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
45. B. F. Y. Hin, R. S. Sethi, and C. R. Lowe, Multi-analyte microelectronic biosensor. Sensors and Actuators A, 1990, 1, 550–554. 46. A. M. Nyamsi Hendji, N. Jaffrezic-Renault, C. Martelet, S. V. Dzyadevich, A. A. Shul’ga, A. P. Soldatkin, and A. V. El’skaya, Enzyme biosensor based on micromachined interdigitated conductometric transducer: application to the detection of urea, glucose, acetyl and butyrylcholine chlorides. Sensors and Actuators, B, 1994, 21, 123–129. 47. S. V. Dzyadevich, G. A. Zhylyak, A. P. Soldatkin, and A. V. El’skaya, Conductometric urease microbiosensor based on thin-films interdigitated electrodes for urea determination. Biopolymers Cell, 1996, 12, 53–57. 48. M. M. Castillo-Ortega, D. E. Rodriguez, J. C. Encicas, M. Plascencia, F. A. Mendez-Velarde, and R. Olayo, Conductometric uric acid and urea biosensor prepared from electroconductive polyaniline-poly(n-butyl methacrylate) composites. Sensors and Actuators, B, 2002, 85, 19–25. 49. W. Limbut, P. Thavarungkul, P. Kanatharana, P. Asaawatreratanakul, C. Limsakul, and B. Wongkittisuksa, Comparative study of controlled pore glass, silica gel and Poraver for the immobilization of urease to determine urea in a flow injection conductimetric biosensor system. Biosensors and Bioelectronics, 2004, 8, 813–821. 50. A. Steinschaden, D. Adamovic, G. Jobst, R. Glatz, and G. Urban, Miniaturised thin film conductometric biosensors with high dynamic range and high sensitivity. Sensors and Actuators, B, 1997, 44, 365–369. 51. W.-Y. Lee, K. S. Lee, T.-H. Kim, M.-C. Shin, and J.K. Park, Microfabricated conductometric urea biosensor based on sol-gel immobilized urease. Electroanalysis, 2000, 12, 78–82. 52. N. F. Sheppard, D. J. Mears, and A. Guiseppi-Elie, Model of an immobilized enzyme conductimetric urea biosensor. Biosensors and Bioelectronics, 1996, 11, 967–979. 53. A. M. Gallardo Soto, S. A. Jaffari, and S. Bone, Characterisation and optimisation of AC conductimetric biosensors. Biosensors and Bioelectronics, 2001, 16, 23–29. 54. A. Senillou, N. Jaffrezic, C. Martelet, and S. Cosnier, A laponite clay-poly(pyrrole-pyridinium) matrix for the fabrication of conductometric microbiosensors. Analytica Chimica Acta, 1999, 401, 117–124. 55. A. P. Soldatkin, S. V. Dzyadevich, Y. I. Korpan, V. N. Arkhipova, G. A. Zhylyak, S. A. Piletsky, T. A. Sergeeva, T. L. Panasyuk, and A. V. El’skaya, Biosensors based on conductometric detection. Biopolymers Cell, 1998, 14, 268–277. 56. S. V. Dzyadevych, V. N. Arkhypova, A. V. El’skaya, N. Jaffrezic-Renault, C. Martelet, and A. P. Soldatkin, Conductometric enzyme biosensors for substrates or inhibitors analysis. Current Topics in Analytical Chemistry, 2001, 2, 179–186. 57. P. Jin, A. Yamaguchi, O. F. Asari, S. Matsuo, J. Tan, and H. Misawa, Glucose sensing based on interdigitated array microelectrode. The Analytical Science, 2001, 17, 841–846. 58. A. P. Soldatkin, A. V. El’skaya, A. A. Shul’ga, A. S. Jdanova, S. V. Dzyadevich, N. JaffrezicRenault, C. Martelet, and P. Clechet, Glucose sensitive
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
conductometric biosensor with additional NAFION membrane: reduction of influence of buffer capacity on the sensor response and extension of its dynamic range. Analytica Chimica Acta, 1994, 288, 197–203. S. V. Dzyadevych, O. P. Soldatkin, V. N. Arkhypova, A. A. Shul’ga, G. V. El’ska, Conductometric enzyme glucosensor. Searching the ways of increasing of analytical characteristics. Ukrainian Biochemical Journal, 1995, 67(6), 53–59. S. V. Dzyadevich, V. N. Arkhipova, A. P. Soldatkin, A. V. El’skaya, and A. A. Shul’ga, Glucose conductometric biosensor with potassium hexacyanoferrate (III) as an oxidizing agent. Analytica Chimica Acta, 1998, 374, 11–18. S. V. Dzyadevich, A. A. Shul’ga, A. P. Soldatkin, A. M. Nyamsi Hendji, N. Jaffrezic-Renault, and C. Martelet, Application of conductometric for sensitive detection of pesticides biosensor based on the cholinesterases. Electroanalysis, 1994, 6, 752–758. V. N. Arkhypova, S. V. Dzyadevych, O. N. Schuvaylo, A. P. Soldatkin, A. V. El’skaya, N. JaffrezicRenault, and C. Martelet, Concept of multibiosensors for determination of different toxic compounds based on enzyme inhibitor analysis. Biopolymers Cell, 2001, 17, 70–77. V. N. Arkhypova, S. V. Dzyadevych, A. P. Soldatkin, A. V. El’skaya, N. Jaffrezic-Renault, H. Jaffrezic, and C. Martelet, Multibiosensor based on enzyme inhibition analysis for determination of different toxic substances. Talanta, 2001, 55, 919–927. S. V. Dzyadevych, A. P. Soldatkin, and J.-M. Chovelon, Assessment of the toxicity of parathion and its photodegradation products in water samples using conductometric enzyme biosensors. Analytica Chimica Acta, 2002, 459, 33–41. S. V. Dzyadevych and J.-M. Chovelon, A comparative photodegradation studies of methyl parathion by using Lumistox test and conductometric biosensor technique. Materials Science and Engineering C, 2002, 21, 55–60. S. V. Dzyadevych, A. P. Soldatkin, V. N. Arkhypova, A. V. El’skaya, J.-M. Chovelon, C. Georgiou, C. Martelet, and N. Jaffrezic-Renault, Early-warning electrochemical biosensor system for the environmental monitoring based on enzyme inhibition effect. Sensors and Actuators, B, 2005, 105, 81–87. G. A. Zhylyak, S. V. Dzyadevich, Y. I. Korpan, A. P. Soldatkin, and A. V. El’skaya, Application of urease conductometric biosensor for heavy-metal ion determination. Sensors and Actuators, B, 1995, 24-25, 145–148. V. N. Arkhipova, S. V. Dzyadevich, A. P. Soldatkin, and A. V. El’skaya, Enzyme biosensors for determination of penicillin based on conductometric planar electrodes and pH-sensitive field effect transistors. Ukrainian Biochemical Journal, 1996, 68, 27–32. O. A. Biloivan, S. V. Dzyadevych, O. P. Soldatkin, N. F. Starodub, G. V. El’ska, Enzymosensor based on trypsin and conductometric planar electrodes for determination of bulks and peptid’s substrates in solution. Ukrainian Biochemical Journal, 1997, 69(2), 14–18. S. V. Dzyadevych, V. N. Arkhypova, Y. I. Korpan, A. V. El’skaya, A. P. Soldatkin, N. Jaffrezic-Renault, and C. Martelet, Conductometric formaldehyde sensitive
CONDUCTOMETRIC ENZYME BIOSENSORS biosensor with specifically adapted analytical characteristics. Analytica Chimica Acta, 2001, 445, 47–55. 71. T. Mai Anh, S. V. Dzyadevych, M. Chau Van, N. JaffrezicRenault, N. Duc Chien, and J.-M. Chovelon, Conductometric tyrosinase biosensor for the detection of diuron, atrazine and its main metabolites. Talanta, 2004, 63, 365–370.
15
FURTHER READING A. J. Lawrence, Conductimetric enzyme assays. European Journal of Biochemistry, 1971, 18, 221–225.
24 Chemical and Biological Field-Effect Sensors for Liquids – A Status Report Arshak Poghossian and Michael J. Sch¨oning Institute of Nano- and Biotechnologies and Research Center J¨ulich, Aachen University of Applied Sciences, J¨ulich, Germany
1 INTRODUCTION
Semiconductor-type field-effect devices (FED) based on the electrolyte–insulator–semiconductor (EIS) concept, that is, ISFET (ion-sensitive fieldeffect transistor), capacitive EIS sensor, and LAPS (light-addressable potentiometric sensor) are currently one of the basic structural elements of chemical and biological microsensors with new functional and application possibilities. They provide a lot of potential advantages such as small size and weight, robustness, fast response time, high reliability, batch processing capability, and so on. On the other hand, the possible field of application of these three kinds of FEDs reaches from medicine, biotechnology, and environmental monitoring over food and drug industries up to defense and security purposes including antibioterrorism and biological warfare agents field. The miniaturization of FEDs and their compatibility with advanced microfabrication technology also make them very attractive for the integration into microfluidic platforms in order to build up miniaturized analytical systems such as micro total analysis system (µTAS), “lab on chip”, and electronic tongue devices. This paper gives a status report on research and development of chemical sensors and biosensors
based on ISFET, capacitive EIS, and LAPS structures. In planning this review, we have chosen to mainly focus upon developments occurring during the last 5 years, from the beginning of 2001 to the end of 2005 (for the early works, the interested reader is referred to Refs. 1–5).
2 PRINCIPLE OF FIELD-EFFECT-BASED (BIO-)CHEMICAL SENSORS
Figure 1 shows a typical structure of a capacitive EIS sensor (a), a LAPS (b), and an ISFET (c). These sensor structures are obtained by replacing the metallic gate of the metal–insulator– semiconductor (MIS) capacitance and the insulated-gate field-effect transistor (IGFET), respectively, by an electrolyte solution and a reference electrode. Since (bio-)chemical FEDs are very sensitive for any kind of electrical interaction at or near the gate insulator/electrolyte interface, in general, nearly each (bio-)chemical reaction leading to chemical or electrical changes at this interface can be measured. Therefore, the ISFET, capacitive EIS sensor or LAPS must be coupled with the respective chemical or biological recognition element. Changes in the chemical composition will induce changes in the electrical surface charge of the gate insulator and in the
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Impedance analyzer
RE
V= bias
Analyte
V~
Gate insulator Depletion layer
Capacitance
2
C1
C2
C3
Si Bias voltage
(a) Feedback circuit
Analyte PC Gate insulator Depletion layer
Photocurrent
Potentiostat RE WE
C1
C2
C3
Si Modulated light source
Bias voltage
(b) Chemical or biological recognition element Drain current
VG
RE Analyte
Gate insulator n+ Source
(c)
Channel
n+ Drain p-Si
C1
C2
C3
Id
VDS
Gate voltage
Figure 1. Setup and typical sensor response of a capacitive EIS sensor (a), a LAPS (b), and an ISFET (c); RE: reference electrode; WE: working electrode; PC: computer; c1,2,3 : analyte concentration. For operation, the gate or bias voltage is applied via a reference electrode (e.g., Ag/AgCl liquid-junction electrode), to set the working point of the sensor as well as to provide a stable potential in the solution. In the case of the EIS sensor setup, an additional small alternating voltage V∼ (∼10–50 mV) is applied to the system in order to measure the capacitance of the EIS sensor. To detect the variation of the capacitance of the depletion layer, the LAPS is illuminated with a modulated light beam, which induces an ac photocurrent to be measured as the sensor signal.
potential drop at the electrolyte/insulator interface that consequently modulates the current in the ISFET’s channel, the capacitance of the EIS sensor or the photocurrent of the LAPS (see Figure 1, right). Generally, the following basic mechanisms of potential generation can be
considered: a pH or ion-concentration change, enzymatic reactions, adsorption of charged macromolecules (e.g., polyelectrolytes, deoxyribonucleic acid (DNA)), affinity binding of molecules (e.g., antigen–antibody affinity reaction, DNA hybridization), and potential changes that are
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS
coming from living biological systems as a result of more sophisticated (bio-)chemical processes (e.g., action potentials of nerve cells).
3 FET-BASED (BIO-)CHEMICAL DEVICES 3.1
pH ISFET
Different oxides and nitrides, like SiO2 , Si3 N4 , SiOx Ny (oxynitride), Al2 O3 , Ta2 O5 , ZrO2 , SnO2 , TiO2 , WO3 , PbTiO3 , AlN, TiN, GaN, Al2 O3 – ZrO2 , and Al2 O3 –Ta2 O5 double oxides, hydrogenated diamond, hydrogenated amorphous Si (a-Si:H), and so on, have been proven as pHsensitive material for FEDs.6–25 However, sometimes these results have been “rediscovered” from results that have already been obtained more than 10–30 years ago. At present, Si3 N4 , Al2 O3 , and Ta2 O5 serve as pH-sensitive gate insulator materials in commercial ISFETs. Other more exotic materials such as AlN, TiN, PbTiO3 , WO3 , and so on, sometimes show nearly Nernstian sensitivity, but have been only rarely studied. Ta2 O5 is considered as the best pH-sensitive material for field-effect sensors, combining a practically ideal Nernstian pH sensitivity, minimal drift, and hysteresis.5,6,13 In addition, as discussed in Section 4, it seems that Ta2 O5 is also the best corrosion-resistant pH-sensitive material. pH ISFETs are now a commercial reality: they are available from more than 20 companies such as Honeywell (USA), Orion (USA), Horiba (Japan), Endress + Hausser (Germany), Mettler Toledo (Switzerland), Sentron Europe (The Netherlands), and other producers of electrochemical sensors. Early problems with ISFETs, that is, drift, temperature instability, light sensitivity, encapsulation, and packaging, can be considered as satisfactorily solved, at least by leading companies producing pH-ISFET sensors.25 Today, commercially available ISFET sensors are exceptionally stable (practically drift free), fully temperature compensated, rugged, and reliable, and exhibit performances comparable to those of pH glass electrodes.25 Moreover, different ISFET fabrication technologies with integrated readout interfaces (e.g., constant current–voltage driver, calibration circuitry, electronics for drift and temperature compensation) have been realized using complementary metal-oxide-semiconductor
3
(CMOS) processes.10,26,27 However, to our knowledge, up to now no ISFET with integrated readout interface has been commercialized. Resistance to breakage is the most obvious feature of solid-state field-effect sensors compared to the pH glass electrode. Therefore, nowadays, in many in-line process-monitoring systems in biotechnology, food, pharmaceutical and cosmetic industries, the breakable pH glass electrode is gradually being replaced by nonglass, unbreakable pH sensors based on ISFETs.25 However, the main problem with those applications is the required cleaning-in-place (CIP) suitability of ISFET devices. According to Ref. 25, commercially available pH ISFETs have a limited lifetime in CIP solutions due to the destruction or dissolution of the pH-sensitive layer during the cleaning of the process vessels using highly caustic media and high temperatures. 3.2
Enzyme-modified FETs (EnFET)
Most reported enzyme-modified field-effect transistors (EnFETs) are built-up of pH-sensitive ISFETs, where hydrogen ions are produced or consumed by the enzymatic reaction. A multitude of EnFETs differing in their sensor design or gate material, enzyme–membrane composition, or immobilization method have been reported for the detection of glucose, urea, penicillin, organophosphorus pesticides, creatinine, phenolic compounds, glycoalkaloids, and so on.28–44 Some recently developed EnFETs are summarized in Table 1. For an exact measurement, a pH ISFET/EnFET differential arrangement is often employed (see e.g., Refs. 32 and 39), where the pH ISFET contains a blank enzyme-free membrane and acts as a reference device. Intensive work during the last years focused on the improvement of EnFET characteristics to circumvent problems, which actually prevent the successful commercialization of EnFETs (e.g., dependence of the sensor response on buffer capacity, ionic strength, and pH of the test sample; restricted dynamic range, nonlinearity; relatively slow response and recovery times; operating and storage stability; reproducibility; dependence on enzyme-immobilization method; the incompatibility of most used enzyme-containing layer deposition and patterning methods with silicon technology).
4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS Table 1. Recently developed EnFETs with enzyme system used and analyte to be detected
Analyte Glucose
Urea Penicillin Creatinine Organophosphate compounds Fluorine-containing organophosphates Phenolic compounds Glycoalkaloids
Enzyme system
References
Glucose oxidase Glucose oxidase/MnO2 powder Glucose oxidase/MnO2 nanoparticles Glucose oxidase/SiO2 nanoparticles Urease Penicillinase Creatinine deiminase Organophosphate hydrolase
28 29 30 31 32–37 12, 13, 38, 39 40 41
Organophophorus acid anhydrolase
42
Tyrosinase Butyryl cholinesterase
43 44
In general, the main disadvantage of field-effect transistor (FET)-based urea and glucose biosensors for biomedical applications is a rather narrow dynamic range (a few millimolars) implying a necessary dilution of the biological sample before the measurement. To extend the dynamic range of a glucose-sensitive EnFET, a glucose-oxidase membrane has been doped with MnO2 powder that catalyzes hydrogen peroxidase (as by-product of the glucose-oxidation reaction).29 As a result, the additional product oxygen is produced, which can be recycled for the glucose-oxidation reaction. In this way, the upper detection limit of the biosensor was extended up to 20 mM. Hence, the developed EnFET can be applied for the glucose detection in undiluted blood samples. A glucose-sensitive EnFET with an extended dynamic range, a good reproducibility, and stability has been realized by the co-immobilization of glucose oxidase and MnO2 nanoparticles, where the MnO2 nanoparticles act as an oxidant rather than a catalyst.30 With the same background, a urea-sensitive EnFET with an extended dynamic range of up to 80 mM was developed taking advantage of a recombinant urease with a genetically modified enzymatic active site.34 The concept of in-situ electrochemical generation of OH− ions in the enzyme membrane was used to improve the recovery time of a glucose-sensitive EnFET.28 A highly sensitive, low detection limit, and long lifetime penicillin-sensitive EnFET was developed in Ref. 39 by adsorptive immobilization of the enzyme penicillinase on a Ta2 O5 -gate ISFET. An extreme low detection limit of 5 µM and a penicillin sensitivity of about 120 mV mM−1 were
achieved by using an optimized buffer solution. The main advantages of the adsorptive immobilization technique are its simplicity, cheapness, and the possibility of a subsequent enzyme (sensor) regeneration. For the fabrication of truly inexpensive and thus, disposable biosensors it is of great interest to integrate the enzyme–membrane preparation in the whole ISFET-fabrication process. A CMOS-compatible, disposable EnFET/ISFET structure with an integrated pseudoreference electrode has been fabricated in Ref. 32. In addition, extended-gate FET structures that are simple in fabricating and packaging have been applied for the development of EnFETs.33 Although EnFETs have a long history and much effort and investment has been done, the transfer of EnFETs from scientific laboratories to real life remains rather slow. Up to now, no EnFET has been commercialized for a wider range of applications. 3.3
DNA-modified FEDs
The possibility of a label-free detection of DNA hybridization utilizing FEDs offers a next generation of DNA chips with direct electrical readout for a fast, simple, and inexpensive real-time analysis of nucleic acid samples. Therefore, in recent years, a considerable research effort has been devoted to the label-free electronic detection of biomolecules (DNA, proteins) by their intrinsic molecular charge using FEDs.45–60 In most cases, the experimentally observed sensor response is interpreted in a way that during the binding event (hybridization of immobilized single-strand DNA (ssDNA)
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS cDNA C
A
T
C
G
G
A
A
ssDNA
A A G C C T T
A
G
G
C
T
A
C
dsDNA
C A T C G G A A
A G C C T T
C A A-T G-C C-G C-G T-A T-A
Hybridization
Gate insulator n+
5
C A A-T G-C C-G C-G T-A T-A
Gate insulator n+
p-Si
n+
n+
p-Si
Figure 2. Schematic structure of a DNA-modified FET and principle of DNA-hybridization detection (ssDNA: single-stranded DNA; dsDNA: double-stranded DNA; cDNA: complementary DNA).
with its complementary target molecule (cDNA); see Figure 2) the charge associated with the target molecule effectively changes the charge applied to the gate of the FED. As a result, the operating characteristics of the FED, that is, the flat-band voltage and capacitance of the EIS sensor or the threshold voltage and drain current of the FET device, will also change. Although the discussed subject is highly interesting, there are still insufficient correct theoretical models for a clear understanding of both the functioning mechanism of these sensors and the source of the experimentally observed signal generation. Moreover, the reported results are very diverse and controversial (e.g., the observed values of the response signal varies from several millivolts45,51 up to 1.9 V46 ). Furthermore, it is not understandable, why a much higher signal has been observed for a sensor with a less density of immobilized ssDNA (1.45 V with 3.8 × 108 molecules/cm2 49 ) compared to a sensor with a densely packed ssDNA (3 mV with 5 × 1013 molecules/cm2 45 ). Or to give a second example, it is not understandable, what the reason is for the much higher biosensor signals, which are observed when floating-gate transistors47–50 or devices without a necessary reference electrode46,47,49 have been used; or, why a DNA-FET without a reference electrode can deliver a reliable sensor signal? 46,47,49 Field-effect sensors are basically surface-charge (potential) measuring devices and are principally able to measure the charge of adsorbed
macromolecules or the charge change due to a hybridization event. At the same time, however, owing to the so-called counterion screening effect and the nonideality of the molecular layer, the realization of these devices for a direct electrostatic detection of charged macromolecules by their intrinsic molecular charge in relatively high-ionic-strength solutions such as physiological solutions is problematic. This subject has been critically discussed for immunospecies and protein molecules and the DNA-hybridization detection in Refs. 2, 61, and 62, respectively, where the estimated hybridization signals were in the range of several millivolts.62 To enhance the sensor signal, the biosensor must be operated in a very low ionicstrength solution (1013 ssDNA/cm2 ). Even under these conditions, the theoretical basis of the sometimes experimentally observed large sensor signals still remains unclear. To overcome the described problems, recently, an alternative mechanism based on the detection of the DNA hybridization-induced redistribution of the ion concentration within the intermolecular spaces and/or the alteration of the ion sensitivity of the field-effect device has been proposed.62 Here the theoretical calculations predict a substantial change in the charge redistribution within the intermolecular spaces of the immobilized DNA induced upon the hybridization event that are enough to obtain a detectable signal of about ∼25–35 mV, when using FEDs.62
6
3.4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
‘‘Cell/transistor’’ Hybrid
“Cell/transistor” hybrids are of great interest for a large variety of applications including the monitoring of electrical communication within neuronal networks, transmission paths of ionic channels, or use as biosensor for pharmaceutical agents, toxic substances, or pollutants. The state of a single cell or a cell system can be monitored by means of various methods that can be distinguished into two basic families: 1. The first type utilizes the energic metabolism of cells; signal parameters such as changes in the extracellular acidification rate, concentration of ions, oxygen consumption, CO2 production, and other metabolic products caused by different external stimuli (e.g., by addition of drugs or toxic agents to the medium) can be detected by the various underlying FEDs. The parallel and noninvasive measurement of the different parameters allows a clear interpretation of different effects on living cells, because cells respond to external stimuli with a parallel activation of different signaling pathways. Examples are a cell-monitoring system with different microsensors (array of pH ISFETs and cell-potential FETs with different gate areas63 ), and a completely automated cell-monitoring microsystem (12 spatially distributed ISFETs) with space and time resolution.64 The experiments demonstrate the capability of the developed system for in-vitro toxicity-screening applications by monitoring changes in the extracellular acidification rate after applying the toxic agents. A further approach describes a multisensor silicon needle consisting of two ISFETs and a temperature sensor that has been realized for the simultaneous measurement of the H+− , K+− concentration and temperature of a myocardial tissue.9 2. The second type of methods utilizes specific features of electrogenic cells such as neuronal and muscle cells and refers on extracellular potential measurements. Electrogenic cells generate spontaneous or triggered action potentials that can be measured by coupling them to FEDs. The activity of a neuron leads to ionic and displacement currents flowing through the cell membrane. This results in
an extracellular voltage drop along the narrow cleft between the cell membrane and the gate insulator of the ISFET that finally, modulates the drain current of the ISFET.3,65–67 Different approaches and equivalent electricalcircuit models have been suggested to describe the signal transfer from electrogenic cells to a FET device in order to explain the recorded signal behavior in terms of shape and amplitude (see e.g., Refs. 65–68 and references therein). However, all these models cannot fully explain some kinetic components in the transistor signal.68 Experiments with HEK-293 cells with voltage-gated K+ channels show, for instance, that variations in the ion concentration in the small cleft between the cell and the FET change the surface potential of the underlying gate insulator and, thus, can play a major role in the recorded biosensor signal.68 In addition to neuronal cells, cardiac muscle cells represent an interesting electrically active system. Their beating frequency can be strongly influenced by cardiac stimulants and relaxants. Such a bioelectronic device, built-up of an array of 16 FETs, has been adapted for recording effects of different drugs on cardiac myocytes.69 Further examples are a 16–channel backside contacted FET array,70 a depletionmode floating-gate FET fabricated by 0.5 µm CMOS technology,71 a large array (16 × 16) of floating-gate FETs,72 and an AlGaN/GaN FET with a high signal-to-noise ratio.73 Moreover, an ISFET-based sensor array, which includes both FETs for detecting the action potential and ISFETs coupled with different ion-sensitive membranes for the measurement of concentration of extracellular ions, like Na+ , K+ , and Ca2+ , has been realized for studying the relationship between the membrane potential and the influx/efflux of ions under external stimulation and drug treatment.74 The main advantage of cell-based ISFET biosensors, compared to, for example, patchclamp technique, is the possibility of a real-time, noninvasive, long-term monitoring of the state of living cells. Nonetheless, all these investigations are still in an experimental stage, far from becoming a commercially available product.
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS
(Bio-)chemical Sensor as a Physical Sensor
In multiparameter analysis systems, besides (bio-)chemical parameters it is often necessary to also measure physical parameters of the liquid, like temperature, flow rate, flow direction, and so on. A simple and mostly used way to construct an ISFET-based multiparameter detection system is the integration of an array of ISFETbased (bio-)chemical sensors with different wellestablished single-function physical sensors (see e.g., Refs. 9 and 75). Another concept for multiparameter detection, so-called (bio-)chemical and physical sensors based on an identical transducer principle, has
Four ISFET structures
Eight sensor functions
Multiparameter detection system
K+ ISFET H+ + H+ H Ion generator
Flow-direction sensor
ISFET1
Flow-velocity sensor
pH sensor
K+ sensor
Enzyme sensor
Reference electrode
Liquid-level sensor
Physical sensors
Temperature sensor
(Bio-)chemical sensors
been introduced in Ref. 12. In this approach, the same ISFET, which is well known as a (bio-)chemical sensor, also serves as a physical sensor and thus, the amount of obtained (bio-)chemical and physical information can be significantly higher than the number of sensors present in the system (“high-order” system).76–78 On the basis of this concept, an ISFET-based multiparameter system for the detection of three (bio-)chemical (pH, K+ , and penicillin concentration) and five physical quantities (temperature, flow velocity, flow direction, diffusion coefficient of ions, and liquid level) has been realized using only four ISFET transducers (see Figure 3).13
Diffusion-coefficient sensor
3.5
7
ISFET2 EnFET
Meander-gate
Printed circuit board
Au ion generator
ISFET chips
Figure 3. Schematic of an ISFET-based multiparameter system for detecting three (bio-)chemical (pH, K+ , penicillin) and five physical quantities (temperature, flow velocity, flow direction, diffusion coefficient of ions, liquid level) by using only four ISFET transducers. In this approach, the same ISFET, which is well known as a (bio-)chemical sensor, also serves as a physical sensor; the multifunctionality is achieved by means of different sensor configurations and/or operation modes.
8
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
4 CAPACITIVE EIS SENSORS
In general, owing to the same functional mechanism, applied materials, and deposition techniques, the sensitive characteristics obtained with EIS and ISFET sensors are quite similar. Therefore, in the following part, we focus on some innovative developments for the fabrication of EIS sensors, like the pulsed laser deposition (PLD) technique, CIP-suitable pH-sensitive materials, porous Si, and the integration of miniaturized EIS sensors with micromachined flow-through microcells. Recently, a capacitive EIS sensor based on PLDprepared Al2 O3 and Ta2 O5 films has been reported as a highly long-term stable pH sensor with nearly Nernstian behavior.79,80 In a further approach, the EIS sensor and a thick-film screen-printed reference electrode have been integrated into a hybrid sensor module.81 These results favor the employment of the PLD process as an alternative deposition method for thin-film pH and chemically sensitive gate insulator materials. Besides its compatibility with silicon planar technology, the main advantage of this method is the controlled deposition of even multicomponent compositions in a defined stoichiometry. As discussed in Section 3.1, commercial pH ISFETs have a limited lifetime in CIP solutions. In order to extend the application area of pH-sensitive FEDs, it is of strong interest to develop insulator materials combining both high corrosion-resistance properties and a high pHsensitive behavior. Therefore, the CIP suitability of EIS sensors with pH-sensitive Ta2 O5 films, prepared by thermal oxidation of Ta layers at an optimized temperature, has been tested under real CIP conditions.82 These experiments have shown, that even after 30 CIP cycles (each CIP cycle includes a cleaning in 4% NaOH solution at 80 ◦ C during 15 min and subsequently in 0.65% HNO3 solution at 80 ◦ C during 5 min), the sensors have a linear calibration curve with a nearly Nernstian sensitivity of 57 ± 1.5 mV/pH. Ellipsometric, video-microscopic, and scanning electron microscopy investigations do not show any visible change in the thickness or degradation of the Ta2 O5 films after the CIP cycles. These experiments could demonstrate the CIP suitability of EIS sensors with Ta2 O5 films, where a nonglass, unbreakable sensor can be placed in direct contact
with food for pH measurements without the risk of broken glass fragments.
4.1
Enzyme-modified EIS Sensor
At present, enzyme-modified capacitive EIS sensors have been realized for the detection of urea,83 penicillin,38,84,85 organophosphorus pesticides (paraoxan, parathion, diazinon, and dichlorvos),86,87 alliin,88 cyanide,89 and other analytes using the respective enzyme. In order to simultaneously detect both the variation of the pH and penicillin content, an EIS sensor array has been developed (see e.g., Ref. 79). In addition, the cross-sensitivity of a EIS penicillin sensor in combination with a diffusion barrier has been investigated with respect to four different kinds of penicillin, namely penicillin G, ampicillin, amoxicillin, and cloxacillin.38 Functioning and, therefore, also the problems related to enzyme-modified capacitive EIS sensors are similar to those of EnFETs, which have been discussed in Section 3.2. In addition, an impedance effect of the enzyme-containing membrane can influence the sensor signal.90 In spite of their very simple structure, generally, enzyme-modified capacitive EIS biosensors are less investigated than EnFETs.
4.2
Capacitive EIS Sensor Based on Porous Si
Capacitive EIS sensors based on porous Si exhibit the advantages of a protected embedment of chemical or biological receptor molecules inside the pores against a fast leaching out, and the enlargement of the effective sensor area due to the porous structure (see Figure 4). Porous pH sensors with an n-Si/SiO2 /Si3 N4 structure have been reported in Refs. 91 and 92: the pH sensitivity (54 mV/pH) is in good agreement with results obtained for nonporous (planar) sensors with the same Si3 N4 layer. Here, the effective sensor surface and thus, the capacitance of the porous pH sensor is larger by a factor of 30 with respect to a comparable nonporous EIS sensor. The possibility to use oxidized porous Si as a transducer material for ion-sensor applications has been demonstrated in Ref. 93. Surprisingly, an unusual super-Nernstian
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS
9
Chemical/biological receptors Si3N4 SiO2 Si SiO2 Contact
Si3N4
50 nm
Figure 4. Schematic and photo of a macroporous EIS sensor; the mean pore diameter varies between 0.5 and 3 µm.
sensitivity toward Na+ and Cu2+ ions has been found, that was interpreted by using a quantumeffect model. Besides the possibility of miniaturization, porous EIS sensors should offer the advantage of fixing biomolecules inside the pores just by means of a physical adsorption process. For example, two porous Si EIS biosensors for the determination of penicillin and triglycerides utilizing the adsorptively immobilized enzymes penicillinase and lipase, respectively, have been investigated.80,94
4.3
Micromachined Flow-through Cell with Wafer-level-integrated Capacitive EIS Sensor
In literature, two platforms have been applied to integrate EIS sensors into a flow-through cell. In a first and most often used approach, the EIS sensor represents a separate component in a homemade flow-through cell. In this way, for example, pHsensitive and penicillinase-modified Ta2 O5 -gate EIS sensors have been integrated into a flowthrough cell with a variable internal volume from 12 to 48 µl that is combined with a commercial flow-injection analysis (FIA) system.84,85 The second platform favors a monolithic waferlevel integration of the flow channel together with the EIS sensor structure. In this case, the sensor represents an integral part of the whole flow-through microcell. Here, a micromachined
flow-through microcell with integrated EIS sensor was realized by combining Si and SU-8 technologies;85,95 the flow-through micro-channel has been formed in a thick SU-8 layer directly onto an already prepared p-Si-SiO2 –Ta2 O5 EIS structure (see Figure 5). In order to extend the functional possibilities of this microcell, two thinfilm Pt microelectrodes have been deposited onto the same chip for additional amperometric and flow-velocity measurements. The EIS structures have been integrated in a pH- and penicillinsensitive configuration in flow-through and FIA mode, respectively, yielding a comparable sensor behavior as for the current single pH- and penicillin-sensitive EIS structures.85
5 LAPS
In contrast to the capacitive EIS sensor, where the measured value of the analyte concentration is an average value over the whole sensing surface in contact with the analyte, the LAPS measurement has the advantage of being spatially resolved. The measured area on the sensing surface is defined by the area of illumination, where the ac photocurrent to be measured has been generated.3,96,97 LAPS devices became popular in many chemical and biological applications such as the detection of bacterial growth, the measurement of cell metabolism, the study of mechanisms of drug action on cell physiology, and so
10
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS Inlet
Outlet
Plexiglass converplate
Passivation layer
Metal electrodes Sensor chip
SU-8 microchannels
(a)
Thin-film ion generator (Pt) Flow channel (~4 µl) EIS sensors
SU-8 Si chip
2 mm
(b)
(c)
Figure 5. Schematic flow-through FIA cell setup of the capacitive EIS sensor (a); micromachined flow-through microcells fabricated by combining Si and SU-8 technologies, where the microchannels have been formed as thick SU-8 layers directly onto the already prepared Si–SiO2 –Ta2 O5 EIS structures: microcell with EIS sensor and Pt ion-generator (b), and microcell with two separate microchannels and EIS sensor for multisensor and/or differential setup applications (c).
on (see e.g., Refs. 97 and 98). The first successfully commercialized system using the LAPS is the Cytometer Microphysiometer system, realized in the 1990s by the company Molecular Devices Corporation (USA).97,98
5.1
Different Types of LAPS
Thanks to the structural similarity between LAPS, capacitive EIS sensor and ISFET, many of the sensitive materials, membrane-deposition techniques, and enzyme-immobilization strategies already developed for ISFETs and EIS sensors are also applicable to LAPS devices. Traditionally, the LAPS is employed for pH recording (see e.g., Refs. 3, 96, and 99–102). Si3 N4 is the most frequently used pH-sensitive material in LAPS devices, although several alternative materials
such as PLD-deposited Ta2 O5 99–101 and Al2 O3 102 have also been proved as alternative pH-sensitive materials. An application of porous silicon for a pH LAPS was demonstrated in Ref. 99. In addition, a LAPS with a submicrometer spatial resolution has been fabricated using amorphous silicon.103,104 A concept for a submicron LAPS is theoretically analyzed in Ref. 105. In contrast to pH-sensitive LAPS, ion-sensitive and enzyme-modified LAPS are studied in less detail. For example, the application of LAPS as an ion sensor for the detection of different cations (Li+ , K+ , Cs+ , Ca2+ , and Mg2+ )106–108 and anions − (NO− 3 and SO4 ) has been demonstrated in Ref. 109. A LAPS for heavy-metal detection using a chalcogenide-glass membrane was developed in Refs. 110 and 111. An enzyme-modified LAPS has been realized for the detection of penicillin,99–101 urea, and butyrylcholine.112
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS
5.2
11
Chemical Imaging and Multilight LAPS
In order to achieve a pH distribution with spatial resolution along the LAPS sensor surface, either the light pointer can be scanned along the surface or multiple light pointers can be used. In the chemical imaging sensor, a map of a twodimensional distribution of the pH value or the ion concentration can be visualized by measuring the amplitude of the photocurrent at each point while the focused laser beam is scanned across the surface of the semiconductor.108,113–118 The detection of the metabolic activity of bacteria (Escherichia coli colonies) immobilized on a p-Si/SiO2 /Si3 N4 LAPS surface113 and the potentiometric imaging of a fluid inside a microchannel108 are two examples of a possible application for the chemical imaging sensor. In contrast to the chemical imaging setup, in the multilight LAPS the sensor surface is illuminated at multiple regions by using many light pointers; each light pointer is modulated at a different frequency.106,114,119 On the basis of the multilight LAPS concept, a novel microphysiometer has been proposed for the simultaneous measurement of several extracellular ion concentrations (H+ , Na+ , K+ , and Ca2+ )106,114 in order to study the influence of drugs on the metabolism of a suckling rat’s nephridium and cardiac muscle cells. An alternative approach is presented in Ref. 120, which allows the simultaneous assessing of the metabolic states of two or more cell populations cocultured on different parts of a single sensor surface. Another setup for a multilight LAPS was realized in Ref. 115. Here, defined positions have been illuminated sequentially by multiple light-emitting diodes (LEDs) using the same modulation frequency. In this way, a pen-shaped miniaturized LAPS device with integrated multisensor functions and sequential readout has been developed (see Figure 6). Such a pen-shaped LAPS can be directly dipped into the test sample to be measured. In further work, the authors have developed a handheld 16-channel pen-shaped LAPS with integrated signal processing unit121 and a LAPS card-based sensor setup.122 The general advantage of the multilight LAPS setup is the short time to take an “image” of the surface-potential distribution (16 sensor spots corresponding to 16 “pixels”). However, the number
Figure 6. Photographs of the portable pen-shaped LAPS device with 16 LEDs for defining the sensor spots (inlet, left) and chamber for the test sample (inlet, right).
of illuminated sensor spots per unit surface area is limited due to the geometrical restrictions of actually commercially available LEDs.
5.3
LAPS for Single-cell Measurements
The number of active measuring sites in ISFETbased cell-monitoring systems is limited by the number of FETs as well as by the fact that typically many cells either do not adhere or only partially adhere on top of the gate of the individual FET. LAPS overcomes the problem of a limited number of active measurement sites. By scanning the light pointer and illuminating the LAPS surface exactly below the cell of interest, it should be possible to record their electric activity.116 Although there are many cells cultured on the chip surface, only that cell illuminated by the focused light beam is interrogated. Therefore, several attempts have been made to record action potentials of single cells by means of a LAPS device.116–118 However, the observed signals were small, typically about 10 µV. Although these experiments have demonstrated the capability of the LAPS for detecting extracellular signals from single cells, its present low signal-to-noise ratio hinders the LAPS to be a competitive technique for extracellular signal recording. Nevertheless, with its conceptual advantage of a free addressability of the measurement point with a
12
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
spatial resolution of a few tens of micrometers, the LAPS is very attractive for single-cell investigations, especially when improving the potential sensitivity and signal-to-noise ratio. Another area is devoted to the application of LAPS-based systems for the detection of the metabolic activity of single living cells. Here, the main technical problem for extracellular acidification measurements of single cells is the need for a means of confining the produced protons to a small and defined “analyte” volume.
6 CONCLUSIONS AND PERSPECTIVES
A comparative study of an ISFET, a capacitive EIS sensor, and a LAPS100 has shown that, since these three sensor types are based on the same transducer principle and materials, they also have a comparable sensor behavior. Nevertheless, the sensor configuration (i.e., measuring setup) and the sensor preparation (i.e., technology) are specific for each type. For example, capacitive EIS sensors and LAPS are cheaper and easier to prepare due to their simplicity in the layout and the absence of a required encapsulation procedure. An attractive feature of the LAPS, compared to the EIS sensor and ISFET, is its addressability. The main disadvantage of the LAPS is the necessity of a light pointer and the light sensitivity. One common disadvantage of the capacitive EIS sensor and the LAPS is the dependence of the sensor signal on any series impedance in general, and on the electrolyte resistance, in particular. Finally, all three sensor types are suitable for multisensor applications. Despite the intensive research and tremendous amount of published works, generally, it can be concluded that a practical realization of (bio-)chemical FEDs and their transfer from scientific laboratories to real life proceeds still rather slowly. The study of the current state of (bio-)chemical FEDs reveals that some of them, like pH-sensitive, enzyme-modified or cell-based FEDs are at a well-developed stage, whereas others such as DNA-modified FEDs are still in the experimental stage or starting phase. Only very few (bio-)chemical FEDs, namely a pH-sensitive ISFET and a LAPS system for cell-acidification detection have been successfully commercialized
so far. Many improvements have been made in the last few years and it can be expected that (bio-)chemical FEDs sensitive to ions others than hydrogen will become commercially available in the near future. Hence, perspectives for research activities of (bio-)chemical FEDs can be expected in the following directions: • The development of biosensors based on organic FEDs.123 • FED-based DNA chips can be considered as a new tool for a label-free nucleic acid analysis. In this context, the detection of layerby-layer adsorbed polyelectrolytes by means of FEDs51,124,125 could be very useful as a model system for fundamental study effects induced in FEDs by the adsorption and binding of charged macromolecules, in particular, during the DNAhybridization event. • The replacement of the inorganic gate insulator of the FED by a molecular insulator layer directly having the reactive sites for a (bio-)chemical interaction. REFERENCES 1. P. Bergveld and A. Sibbald, Analytical and Biomedical Applications of Ion-Selective Field-Effect Transistors, Elsevier, Amsterdam, 1988. 2. G. F. Blackburn, Chemically Sensitive Field-Effect Transistors, in Biosensors: Fundamentals and Applications, A. P. F. Turner, I. Karube, and G. S. Wilson (eds), Oxford University Press, Oxford, 1987, pp. 481–530. 3. M. Grattarola and G. Massobrio, Bioelectronics Handbook: MOSFETs, Biosensors and Neurons, McGraw-Hill, New York, 1998. 4. M. J. Sch¨oning and A. Poghossian, Recent advances in biologically sensitive field-effect transistors (BioFETs). Analyst, 2002, 127, 1137–1151. 5. P. Bergveld, Thirty years of ISFETOLOGY: what happened in the past 30 years and what may happen in the next 30 years. Sensors and Actuators, B, 2003, 88, 1–20. 6. Y. G. Vlasov, Y. A. Tarantov, and P. V. Bobrov, Analytical characteristics and sensitivity mechanisms of electrolyte-insulator-semiconductor system-based chemical sensors—a critical review. Analytical and Bioanalytical Chemistry, 2003, 376, 788–796. 7. P. K. Shin and T. Mikolajick, Alkali and hydrogen ion sensing properties of LPCVD silicon oxynitride thin films. Thin Solid Films, 2003, 426, 232–237. 8. F. Yan, P. Estrela, Y. Mo, P. Migliorato, H. Maeda, S. Inoue, and T. Shimoda, Polycrystalline silicon ion sensitive field effect transistors. Applied Physics Letters, 2005, 86, 53901–53903.
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS 9. A. Errachid, N. Zine, J. Samitier, and J. Bausells, FETbased chemical sensor systems fabricated with standard technologies. Electroanalysis, 2004, 16, 1843–1851. 10. C. G. Jakobson, U. Dinnar, M. Feinsod, and Y. Nemirovsky, Ion-sensitive field-effect transistors in standard CMOS fabricated by post processing. IEEE Sensor, 2002, 2, 279–287. 11. J. C. Chou, C. Y. Weng, and H. M. Tsai, Study on the temperature effects of Al2 O3 gate pH-ISFET. Sensors and Actuators, B, 2002, 81, 152–157. 12. A. Poghossian, H. L¨uth, J. W. Schultze, and M. J. Sch¨oning, (Bio-)chemical and physical microsensor arrays using an identical transducer principle. Electrochimica Acta, 2001, 47, 243–249. 13. A. Poghossian and M. J. Sch¨oning, Detecting both physical and (bio-)chemical parameters by means of ISFET devices. Electroanalysis, 2004, 16, 1863–1872. 14. S. Yoshida, N. Hara, and K. Sugimoto, Development of a wide range pH sensor based on electrolyte-insulatorsemiconductor structure with corrosion-resistant Al2 O3 Ta2 O5 and Al2 O3 -ZrO2 double-oxide thin films. Journal of the Electrochemical Society, 2004, 151, H53–H58. 15. J. L. Chiang, S. S. Jan, J. C. Chou, and Y. C. Chen, Study on the temperature effect, hysteresis and drift of pH-ISFET devices based on amorphous tungsten oxide. Sensors and Actuators, B, 2001, 76, 624–628. 16. Y. L. Chin, J. C. Chou, T. P. Sun, H. K. Liao, W. Y. Chung, and S. K. Hsiung, A novel SnO2 /Al discrete gate ISFET pH sensor with CMOS standard process. Sensors and Actuators, B, 2001, 75, 36–42. 17. J. C. Chou and L. P. Liao, Study of TiO2 thin films for ion-sensitive field-effect transistor application with RF sputtering deposition. Japanese Journal of Applied Physics, 2004, 43, 61–65. 18. L. T. Yin, J. C. Chou, W. Y. Chung, T. P. Sun, and S. K. Hsiung, Study of indium tin oxide thin film for separative extended gate ISFET. Materials Chemistry and Physics, 2001, 70, 12–16. 19. S. S. Jan, Y. C. Chen, and J. C. Chou, Effect of Mg2+ dopant on the characteristics of lead titanate sensing membrane for ion-sensitive field-effect transistors. Sensors and Actuators, B, 2005, 108, 883–887. 20. J. L. Chiang, Y. C. Chen, J. C. Chou, and C. C. Cheng, Temperature effect on AlN/SiO2 gate pH-ionsensitive field-effect transistor devices. Japanese Journal of Applied Physics, 2002, 41, 541–545. 21. Y. L. Chin, J. C. Chou, Z. C. Lei, T. P. Sun, W. Y. Chung, and S. K. Hsiung, Titanium nitride membrane application to extended gate field effect transistor pH sensor using VLSI technology. Japanese Journal of Applied Physics, 2001, 40, 6311–6315. 22. G. Steinhoff, M. Hermann, W. J. Schaff, L. F. Eastman, M. Stutzmann, and M. Eickhoff, pH response of GaN surfaces and its application for pH-sensitive field-effect transistors. Applied Physics Letters, 2003, 83, 177–179. 23. J. C. Chou and Y. F. Wang, Temperature characteristics of a-Si:H gate ISFET. Materials Chemistry and Physics, 2001, 70, 107–111. 24. J. A. Garrido, A. H¨artl, S. Kuch, M. Stutzmann, O. A. Williams, and R. B. Jackmann, pH sensors based on hydrogenated diamond surfaces. Applied Physics Letters, 2005, 86, 73504–73506.
13
25. W. Oelßner, J. Zosel, U. Guth, T. Pechstein, W. Babel, J. G. Connery, C. Demuth, M. Grote Gansey, and J. B. Verburg, Encapsulation of ISFET sensor chips. Sensors and Actuators, B, 2005, 105, 104–117. 26. W. Y. Chung, C. H. Yang, Y. F. Wang, Y. J. Chan, W. Torbicz, and D. G. Pijanowska, A signal processing ASIC for ISFET-based chemical sensors. Microelectronics Journal, 2004, 35, 667–675. 27. P. A. Hammond and D. R. S. Cumming, Performance and system-on-chip integration of an unmodified CMOS ISFET. Sensors and Actuators, B, 2005, 111 – 112, 254–258. 28. K. Y. Park, S. B. Choi, M. Lee, B. K. Sohn, and S. Y. Choi, ISFET glucose sensor system with fast recovery characteristics by employing electrolysis. Sensors and Actuators, B, 2002, 83, 90–97. 29. L. T. Yin, J. C. Chou, W. Y. Chung, T. P. Sun, K. P. Hsiung, and S. K. Hsiung, Glucose ENFET doped with MnO2 powder. Sensors and Actuators, B, 2001, 76, 187–192. 30. X. L. Luo, J. J. Xu, W. Zhao, and H. Y. Chen, A novel glucose ENFET based on the special reactivity of MnO2 nanoparticles. Biosensors and Bioelectronics, 2004, 19, 1295–1300. 31. X. L. Luo, J. J. Xu, W. Zhao, and H. Y. Chen, Glucose biosensor based on ENFET doped with SiO2 nanoparticles. Sensors and Actuators, B, 2004, 97, 249–255. 32. W. Sant, M. L. Pourciel, J. Launay, T. Do Conto, A. Martinez, and P. Temple-Boyer, Development of chemical field effect transistors for the detection of urea. Sensors and Actuators, B, 2003, 95, 309–314. 33. J. C. Chen, J. C. Chou, T. P. Sun, and S. K. Hsiong, Portable urea biosensor based on the extended-gate field effect transistor. Sensors and Actuators, B, 2003, 91, 180–186. 34. A. P. Soldatkin, J. Montoriol, W. Sant, C. Martelet, and N. Jaffrezic-Renault, A novel urea sensitive biosensor with extended dynamic range based on recombinant urease and ISFETs. Biosensors and Bioelectronics, 2003, 19, 131–135. 35. F. Aouni, R. Mlika, C. Martelet, H. Ben Ouada, N. Jaffrezic-Renault, and A. P. Soldatkin, Modelling of the potentiometric response of ENFETs based on enzymatic multilayer membranes. Electroanalysis, 2004, 16, 1907–1911. 36. A. V. Rebriiev and N. F. Starodub, Enzymatic biosensor based on the ISFET and photopolymeric membrane for the determination of urea. Electroanalysis, 2004, 16, 1891–1895. 37. D. Niwa, K. Omichi, N. Motohashi, T. Homma, and T. Osaka, Organosilane self-assembled monolayermodified field-effect transistors for on-chip ion and biomolecule sensing. Sensors and Actuators, B, 2005, 108, 721–726. 38. A. Poghossian, M. Thust, P. Schroth, A. Steffen, H. L¨uth, and M. J. Sch¨oning, Penicillin detection by means of silicon-based field-effect structures. Sensors and Materials, 2001, 13, 207–223. 39. A. Poghossian, M. J. Sch¨oning, P. Schroth, A. Simonis, and H. L¨uth, An ISFET-based penicillin sensor with high
14
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS sensitivity, low detection limit and long lifetime. Sensors and Actuators, B, 2001, 76, 519–526. W. Sant, M. L. Pourciel-Gouzy, J. Launay, T. Do Conto, R. Colin, A. Martinez, and P. Temple-Boyer, Development of a creatinine-sensitive sensor for medical analysis. Sensors and Actuators, B, 2004, 103, 260–264. A. L. Simonian, A. W. Flounders, and J. R. Wild, FET-based biosensors for the direct detection of organophosphate neurotoxins. Electroanalysis, 2004, 16, 1896–1906. A. L. Simonian, J. K. Grimsley, A. W. Flounders, J. S. Schoeniger, T. C. Cheng, J. J. De Frank, and J. R. Wild, Enzyme-based biosensor for the direct detection of fluorine-containing organophosphates. Analytica Chimica Acta, 2001, 442, 15–23. S. V. Dzyadevych, T. M. Anh, A. P. Soldatkin, N. D. Chien, N. Jaffrezic-Renault, and J. M. Chovelon, Development of enzyme biosensor based on pHsensitive field-effect transistors for detection of phenolic compounds. Bioelectrochemistry, 2002, 55, 79–81. A. P. Soldatkin, V. N. Arkhypova, S. V. Dzyadevych, A. V. El’skaya, J. M. Gravoueille, N. Jaffrezic-Renault, and C. Martelet, Analysis of the potato glycoalkaloids by using of enzyme biosensor based on pH-ISFETs. Talanta, 2005, 66, 28–33. J. Fritz, E. B. Cooper, S. Gaudet, P. K. Sorger, and S. R. Manalis, Electronic detection of DNA by its intrinsic molecular charge. Proceedings of the National Academy of Sciences, 2002, 99, 14142–14146. M. W. Dashiell, A. T. Kalambur, R. Leeson, K. J. Roe, J. F. Rabolt, and J. Kolodzey, The Electrical Effects of DNA as the Gate Electrode of MOS Transistors, In: Proceedings IEEE Lester Eastman conference on High Performance Devices, Delaware, 2002, pp. 259–264. D. S. Kim, Y. T. Jeong, H. K. Lyu, H. J. Park, H. S. Kim, J. K. Shin, P. Choi, J. H. Lee, G. Lim, and M. Ishida, Fabrication and characteristics of a field-effect transistortype charge sensor for detecting deoxyribonucleic acid sequence. Japanese Journal of Applied Physics, 2003, 42, 4111–4115. J. K. Shin, D. S. Kim, H. J. Park, and G. Lim, Detection of DNA and protein molecules using an FETtype biosensor with gold as a gate metal. Electroanalysis, 2004, 16, 1912–1918. D. S. Kim, Y. T. Jeong, H. K. Lyu, H. J. Park, J. K. Shin, P. Choi, J. H. Lee, and G. Lim, An FET-type charge sensor for highly sensitive detection of DNA sequence. Biosensors and Bioelectronics, 2004, 20, 69–74. D. S. Kim, H. J. Park, H. M. Jung, J. K. Shin, Y. T. Jeong, P. Choi, J. H. Lee, and G. Lim, Fieldeffect transistor-based bimolecular sensor employing a Pt reference electrode for the detection of deoxyribonucleic acid sequence. Japanese Journal of Applied Physics, 2004, 43, 3855–3859. F. Uslu, S. Ingebrandt, D. Mayer, S. B¨ocker-Meffert, M. Odenthal, and A. Offenh¨ausser, Label free fully electronic nucleic acid detection system based on a fieldeffect transistor device. Biosensors and Bioelectronics, 2004, 19, 1723–1731. F. Pouthas, C. Gentil, D. Cote, G. Zeck, B. Straub, and U. Bockelmann, Spatially resolved electronic detection of biopolymers. Physical Review E, 2004, 70, 31906–31913.
53. T. Ohtake, C. Hamai, T. Uno, H. Tabata, and T. Kawai, Immobilization of probe DNA on Ta2 O5 thin film and detection of hybridized helix DNA using ISFET. Japanese Journal of Applied Physics, 2004, 43, L1137–L1139. 54. T. Uno, T. Ohtake, H. Tabata, and T. Kawai, Direct deoxyribonucleic acid detection using ion-sensitive fieldeffect transistors based on peptide nucleic acid. Japanese Journal of Applied Physics, 2004, 43, L1584–L1587. 55. T. Sakata, M. Kamahori, and Y. Miyahara, Immobilization of oligonucleotide probes on Si3 N4 surface and its application to genetic field-effect transistor. Materials Science and Engineering, 2004, C 24, 827–832. 56. T. Sakata, M. Kamahori, and Y. Miyahara, DNA analysis chip based on field-effect transistors. Japanese Journal of Applied Physics, 2005, 44, 2854–2859. 57. T. Sakata, S. Matsumoto, Y. Nakajima, and Y. Miyahara, Potential behavior of biochemically modified gold electrode for extended-gate field-effect transistor. Japanese Journal of Applied Physics, 2005, 44, 2860–2863. 58. T. Sakata and Y. Miyahara, Detection of DNA recognition events using multi-well field effect devices. Biosensors and Bioelectronics, 2005, 21, 827–832. 59. K. Y. Park, M. S. Kim, and S. Y. Choi, Fabrication and characteristics of MOSFET protein chip for detection of ribosomal protein. Biosensors and Bioelectronics, 2005, 20, 2111–2115. 60. P. Estrela, P. Migliorato, H. Takiguchi, H. Fukushima, and S. Nebashi, Electrical detection of biomolecular interactions with metal–insulator–semiconductor diodes. Biosensors and Bioelectronics, 2005, 20, 1580–1586. 61. P. Bergveld, A critical evaluation of direct electrical protein detection methods. Biosensors and Bioelectronics, 1991, 6, 55–72. 62. A. Poghossian, A. Cherstvy, S. Ingebrandt, A. Offenh¨ausser, and M. J. Sch¨oning, Possibilities and limitations of label-free detection of DNA hybridization with field-effect devices. Sensors and Actuators, B, 2005, 111 – 112, 470–480. 63. M. Lehmann, W. Baumann, M. Brischwein, H. J. Gahle, I. Freund, R. Ehret, S. Drechsler, H. Palzer, M. Kleintges, U. Sieben, and B. Wolf, Simultaneous measurement of cellular respiration and acidification with a single CMOS ISFET. Biosensors and Bioelectronics, 2001, 16, 195–203. 64. L. Lorenzelli, B. Margesin, S. Martinoia, M. T. Tedesco, and M. Valle, Bioelectrochemical signal monitoring of in-vitro cultured cells by means of an automated microsystem based on solid state sensor-array. Biosensors and Bioelectronics, 2003, 18, 621–626. 65. M. Voelker and P. Fromherz, Signal transmission from individual mammalian nerve cell to field-effect transistor. Small, 2005, 1, 206–210. 66. S. Ingebrandt, C. K. Yeung, M. Krause, and A. Offenh¨ausser, Neuron-transistor coupling: interpretation of individual extracellular recorded signals. European Biophysics Journal, 2005, 34, 144–154. 67. S. Martinoia and P. Massobrio, ISFET-neuron junction: circuit models and extracellular signal simulations. Biosensors and Bioelectronics, 2004, 19, 1487–1496. 68. G. Wrobel, R. Seifert, S. Ingebrandt, J. Enderlein, H. Ecken, A. Baumann, U. B. Kaupp, and
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
A. Offenh¨ausser, Cell-transistor coupling: investigation of potassium currents recorded with p-and n-channel FETs. Biophysical Journal, 2005, 89, 3628–3638. C. K. Yeung, S. Ingebrandt, M. Krause, A. Offenh¨ausser, and W. Knoll, Validation of the use of field effect transistors for extracellular signal recording in pharmacological bioassays. Journal of Pharmacological and Toxicological Methods, 2001, 45, 207–214. S. Ingebrandt, C. K. Yeung, W. Staab, T. Zetterer, and A. Offenh¨ausser, Backside contacted field effect transistor array for extracellular signal recording. Biosensors and Bioelectronics, 2003, 18, 429–435. A. Cohen, M. E. Spira, S. Yitshaik, G. Borghs, O. Shwartzglass, and J. Shappir, Depletion type floating gate p-channel MOS transistor for recording action potentials generated by cultured neurons. Biosensors and Bioelectronics, 2004, 19, 1703–1709. M. J. Milgrew, M. O. Riehle, and D. R. S. Cumming, A large transistor-based sensor array chip for direct extracellular imaging. Sensors and Actuators, B, 2005, 111 – 112, 347–353. G. Steinhoff, B. Baur, G. Wrobel, S. Ingebrandt, A. Offenh¨ausser, A. Krost, M. Stutzmann, and M. Eickhoff, Recording of cell action potentials with AlGaN/GaN field-effect transistors. Applied Physics Letters, 2005, 86, 33901–33903. P. Wang, G. Xu, L. Qin, Y. Xu, Y. Li, and R. Li, Cell-based biosensors and its application in biomedicine. Sensors and Actuators, B, 2005, 108, 576–584. C. Jimenez, L. Moreno, C. de Haro, F. X. Munoz, A. Florido, P. Rivas, A. Bratov, and C. Dominguez, Development of a multiparametric system based on solid-state microsensors for monitoring a nuclear waste repository. Sensors and Actuators, B, 2003, 91, 103–108. A. Poghossian, L. Berndsen, and M. J. Sch¨oning, Chemical sensor as physical sensor: ISFET-based flowvelocity, flow-direction and diffusion-coefficient sensor. Sensors and Actuators, B, 2003, 95, 384–390. A. Poghossian, J. W. Schultze, and M. J. Sch¨oning, Application of a (bio-)chemical sensor (ISFET) for the detection of physical parameters in liquids. Electrochimica Acta, 2003, 48, 3289–3297. A. Poghossian, J. W. Schultze, and M. J. Sch¨oning, Multi-parameter detection of (bio-)chemical and physical quantities using an identical transducer principle. Sensors and Actuators, B, 2003, 91, 83–91. M. J. Sch¨oning, A. Poghossian, T. Yoshinobu, and H. L¨uth, Semiconductor-based field-effect structures for chemical sensing. Proceedings of the SPIE, 2001, 4205, 188–198. M. J. Sch¨oning, “Playing around” with field-effect sensors on the basis of EIS structures, LAPS and ISFETs. Sensors, 2005, 5, 126–138. M. J. Sch¨oning, A. Simonis, T. Krings, H. L¨uth, and J. Wang, Evaluation of a chip-based thin-film /thick-film sensor hybrid for (bio-)chemical analysis. Electroanalysis, 2002, 14, 955–958. M. J. Sch¨oning, D. Brinkman, D. Rolka, C. Demuth, and A. Poghossian, CIP (cleaning-in-place) suitable “nonglass” pH sensor based on a Ta2 O5 -gate EIS structure. Sensors and Actuators, B, 2005, 111 – 112, 423–429.
15
83. H. Barhoumi, A. Maaref, M. Rammah, C. Martelet, N. Jaffrezic-Renault, C. Mousty, S. Cosnier, E. Perez, and I. Rico-Lattes, Insulator semiconductor structures coated with biodegradable latexes as encapsulation matrix for urease. Biosensors and Bioelectronics, 2005, 20, 2318–2323. 84. D. Rolka, A. Poghossian, and M. J. Sch¨oning, Integration of a capacitive EIS sensor into a FIA system for pH and penicillin determination. Sensors, 2004, 4, 84–94. 85. N. N¨ather, D. Rolka, A. Poghossian, M. KoudelkaHep, and M. J. Sch¨oning, Two microcell flow-injection analysis (FIA) platforms for capacitive silicon-based field-effect sensors. Electrochimica Acta, 2005, 51, 924–929. 86. J. Wang, R. Krause, K. Block, M. Musameh, A. Mulchandani, P. Mulchandani, W. Chen, and M. J. Sch¨oning, Dual amperometric-potentiometric biosensor detection system for monitoring organophosphorus neurotoxins. Analytica Chimica Acta, 2002, 469, 197–203. 87. M. J. Sch¨oning, M. Arzdorf, P. Mulchandani, W. Chen, and A. Mulchandani, A capacitive field-effect sensor for the direct determination of organophosphorus pesticides. Sensors and Actuators, B, 2003, 91, 92–97. 88. M. Keusgen, M. J¨unger, I. Krest, and M. J. Sch¨oning, Development of a biosensor specific for cysteine sulfoxides. Biosensors and Bioelectronics, 2003, 18, 805–812. 89. M. Keusgen, J. P. Kloock, D. T. Knobbe, M. J¨unger, I. Krest, M. Goldbach, W. Klein, and M. J. Sch¨oning, Direct determination of cyanides by potentiometric biosensors. Sensors and Actuators, B, 2004, 103, 380–385. 90. A. Poghossian, D. T. Mai, Y. Mourzina, and M. J. Sch¨oning, Impedance effect of an ion-sensitive membrane: characterisation of an EMIS sensor by impedance spectroscopy, capacitance-voltage and constant-capacitance method. Sensors and Actuators, B, 2004, 103, 423–428. 91. M. J. Sch¨oning, A. Simonis, C. Ruge, H. Ecken, M. M¨uller-Veggian, and H. L¨uth, A (bio-) chemical field-effect sensor with macroporous Si as substrate material and a SiO2 / LPCVD-Si3 N4 double layer as pH transducer. Sensors, 2002, 2, 11–22. 92. A. Simonis, C. Ruge, M. M¨uller-Veggian, H. L¨uth, and M. J. Sch¨oning, A long-term stable macroporous-type EIS structure for electrochemical sensor applications. Sensors and Actuators, B, 2003, 91, 21–25. 93. S. Zairi, C. Martelet, N. Jaffrezic-Renault, F. Vocanson, R. Lamartine, R. M’gaieth, H. Maaref, and M. Gamoudi, P-type porous-silicon transducer for cation detection: effect of the porosity, pore morphology, temperature and ion valency on the sensor response and generalisation of the Nernst equation. Applied Physics A, 2001, 73, 585–593. 94. R. R. K. Reddy, I. Basu, E. Bhattacharya, and A. Chadha, Estimation of triglycerides by a porous silicon based potentiometric biosensor. Current Applied Physics, 2003, 3, 155–161. 95. M. J. Sch¨oning, N. N¨ather, V. Auger, A. Poghossian, and M. Koudelka-Hep, Miniaturised flow-through cell with integrated capacitive EIS sensor fabricated at wafer level
16
96.
97.
98.
99.
100.
101.
102.
103.
104.
105. 106.
107.
108.
109.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS using Si and SU-8 technologies. Sens Actuators B, 2005, 108, 986–992. D. G. Hafeman, J. W. Parce, and H. M. McConnell, Light-addressable potentiometric sensor for biochemical systems. Science, 1988, 240, 1182–1185. J. C. Owicki, L. J. Bousse, D. G. Hafeman, G. L. Kirk, J. D. Olson, H. G. Wada, and J. W. Parce, The light-addressable potentiometric sensor: principles and biological applications. Annual Review of Biophysics and Biomolecular Structure, 1994, 23, 87–113. F. Hafner, Cytosensor microphysiometer: technology and recent applications. Biosensors and Bioelectronics, 2000, 15, 149–158. T. Yoshinobu, H. Ecken, A. Poghossian, H. L¨uth, H. Iwasaki, and M. J. Sch¨oning, Alternative sensor materials for light-addressable potentiometric sensors. Sensors and Actuators, B, 2001, 76, 388–392. A. Poghossian, T. Yoshinobu, A. Simonis, H. Ecken, H. L¨uth, and M. J. Sch¨oning, Penicillin detection by means of field-effect based sensors: EnFET, capacitive EIS sensor or LAPS. Sensors and Actuators, B, 2001, 78, 237–242. T. Yoshinobu, H. Ecken, A. Poghossian, A. Simonis, H. Iwasaki, H. L¨uth, and M. J. Sch¨oning, Constantcurrent-mode LAPS (CLAPS) for the detection of penicillin. Electroanalysis, 2001, 13, 733–736. A. B. Md. Ismail, T. Harada, T. Yoshinobu, H. Iwasaki, M. J. Sch¨oning, and H. L¨uth, Investigation of pulsed laser-deposited Al2 O3 as a high pH-sensitive layer for LAPS-based biosensing applications. Sensors and Actuators, B, 2000, 71, 169–172. T. Yoshinobu, M. J. Sch¨oning, F. Finger, W. Moritz, and H. Iwasaki, Fabrication of thin-film LAPS with amorphous silicon. Sensors, 2004, 4, 163–169. W. Moritz, T. Yoshinobu, F. Finger, S. Krause, M. Martin-Fernandez, and M. J. Sch¨oning, High resolution LAPS using amorphous silicon as the semiconductor material. Sensors and Actuators, B, 2004, 103, 436–441. Q. Zhang, Theoretical analysis and design of submicronLAPS. Sensors and Actuators, B, 2005, 105, 304–311. W. Yicong, W. Ping, Y. Xuesong, Z. Qingtao, L. Rong, Y. Weimin, and Z. Xiaoxiang, A novel microphysiometer based on MLAPS for drugs screening. Biosensors and Bioelectronics, 2001, 16, 277–286. Yu. Ermolenko, T. Yoshinobu, Yu. Mourzina, S. Levichev, K. Furuichi, Yu. Vlasov, M. J. Sch¨oning, and H. Iwasaki, Photocurable membranes for ion-selective lightaddressable potentiometric sensor. Sensors and Actuators, B, 2002, 85, 79–85. T. Yoshinobu, H. Iwasaki, Y. Ui, K. Furuichi, Yu. Ermolenko, Yu. Mourzina, T. Wagner, N. N¨ather, and M. J. Sch¨oning, The light-addressable potentiometric sensor for multi-ion sensing and imaging. Methods, 2005, 37, 94–102. Yu. Mourzina, Yu. Ermolenko, T. Yoshinobu, Yu. Vlasov, H. Iwasaki, and M. J. Sch¨oning, Anion-selective light-addressable potentiometric sensors (LAPS) for the determination of nitrate and sulphate ions. Sensors and Actuators, B, 2003, 91, 32–38.
110. Yu. Mourzina, T. Yoshinobu, J. Schubert, H. L¨uth, H. Iwasaki, and M. J. Sch¨oning, Ion-selective lightaddressable potentiometric sensor (LAPS) with chalcogenide thin film prepared by pulsed laser deposition. Sensors and Actuators, B, 2001, 80, 136–140. 111. H. Men, S. Zou, Y. Li, Y. Wang, X. Ye, and P. Wang, A novel electronic tongue combined MLAPS with stripping voltammetry for environmental detection. Sensors and Actuators, B, 2005, 110, 350–357. 112. Yu. G. Mourzina, T. Yoshinobu, Y. E. Ermolenko, Y. G. Vlasov, M. J. Sch¨oning, and H. Iwasaki, Immobilization of urease and cholinesterase on the surface of semiconductor transducer for the development of lightaddressable potentiometric sensors. Microchimica Acta, 2004, 144, 41–50. 113. T. Yoshinobu, H. Ecken, A. B. Md. Ismail, H. Iwasaki, H. L¨uth, and M. J. Sch¨oning, Chemical imaging sensor and its application to biological systems. Electrochimica Acta, 2001, 47, 259–263. 114. W. Yicong, W. Ping, Y. Xuesong, Z. Gaoyan, L. Rong, Y. Weimin, Z. Xiaoxiang, H. Jinghong, and C. Dafu, Drug evaluations using a novel microphysiometer based on cell-based biosensors. Sensors and Actuators, B, 2001, 80, 215–221. 115. T. Yoshinobu, M. J. Sch¨oning, R. Otto, K. Furuichi, Yu. Mourzina, Yu. Ermolenko, and H. Iwasaki, Portable light-addressable potentiometric sensor (LAPS) for multisensor applications. Sensors and Actuators, B, 2003, 95, 352–356. 116. B. Stein, M. George, H. E. Gaub, and W. J. Parak, Extracellular measurements of averaged ionic currents with the light-addressable potentiometric sensor (LAPS). Sensors and Actuators, B, 2004, 98, 299–304. 117. A. B. Md. Ismail, T. Yoshinobu, H. Iwasaki, H. Sugihara, T. Yukimasa, I. Hirata, and H. Iwata, Investigation on light-addressable potentiometric sensor as a possible cell–semiconductor hybrid. Biosensors and Bioelectronics, 2003, 18, 1509–1514. 118. G. Xu, X. Ye, L. Qin, Y. Xu, Y. Li, R. Li, and P. Wang, Cell-based biosensors based on lightaddressable potentiometric sensors for single cell monitoring. Biosensors and Bioelectronics, 2005, 20, 1757–1763. 119. Z. Qintao, W. Ping, W. J. Parak, M. George, and G. Zhang, A novel design of multi-light LAPS based on digital compensation of frequency domain. Sensors and Actuators, B, 2001, 73, 152–156. 120. B. Stein, M. George, H. E. Gaub, J. C. Behrends, and W. J. Parak, Spatially resolved monitoring of cellular metabolic activity with a semiconductor-based biosensor. Biosensors and Bioelectronics, 2003, 18, 31–41. 121. M. J. Sch¨oning, T. Wagner, C. Wang, R. Otto, and T. Yoshinobu, Development of a handheld 16 channel pen-type LAPS for electrochemical sensing. Sensors and Actuators, B, 2005, 108, 808–814. 122. T. Wagner, C. Raoa, J. P. Kloock, T. Yoshinobu, R. Otto, M. Keusgen, and M. J. Sch¨oning, LAPS card—a novel chip card-based light-addressable potentiometric sensor (LAPS). Sensors and Actuators, B, 2006, 118, 33–40. 123. H. E. Katz, Chemically sensitive field-effect transistors and chemiresistors: new materials and device structures. Electroanalysis, 2004, 16, 1837–1842.
CHEMICAL AND BIOLOGICAL FIELD-EFFECT SENSORS FOR LIQUIDS 124. T. Kassab, Y. Han, A. Poghossian, S. Ingebrandt, A. Offenh¨ausser, and M. J. Sch¨oning, Detection of layerby-layer absorbed polyelectrolytes by means of fieldeffect based capacitive EIS structures. Biomedizinische Technik, 2004, 49, 1034–1035.
17
125. G. Maruccio, P. Visconti, A. Biasco, A. Bramanti, A. D. Torre, P. P. Pompa, V. Frascerra, V. Arima, E. D’Amone, R. Cingolani, and R. Rinaldi, Nanoscaled biomolecular field-effect transistors: prototypes and evaluations. Electroanalysis, 2004, 16, 1853–1862.
25 Overview of Optical Biosensing Techniques Ibrahim Abdulhalim,1 Mohammad Zourob2 and Akhlesh Lakhtakia3 1
Department of Electrooptics Engineering, Ben-Gurion University of the Negev, Beer-Sheva, Israel, 2 Biosensors Division, Biophage Pharma, Montreal, Quebec, Canada and 3 Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA, USA
1 INTRODUCTION
The development of optical biosensing technology is an extremely important scientific and technological issue for (i) the diagnosis and monitoring of diseases, (ii) drug discovery, (iii) proteomics, and (iv) the environmental detection of pollutants and/or biological agents. Traditionally, a biosensor is derived from the coupling of a ligand–receptor binding reaction to a transduction mechanism.1 In an optical biosensor, either the reaction product effects a significant change in the response of a transducer to incoming light or the reaction produces an optical signal that is sensed by the transducer, possibly after amplification and/or conversion to some other form (Figure 1). Optical biosensing has a wide scope nowadays: it encompasses optical methods for medical diagnosis2 and imaging, the most well-known of which are a variety of techniques to measure blood glucose concentration noninvasively using optics, such as by analyzing optical scattering from tissue, polarimetric measurement through fluids in the eye, optical coherence tomography, and skin optical imaging. Optical biosensors have seen tremendous advancement during the last two decades due to major developments in optics in general and optoelectronic components in particular, for example,
miniature light sources, optical fibers, materials, and optoelectronic devices.3–11 These advances have been backed by advances in computers and nanotechnology, thereby allowing for smart, miniature, and highly efficient sensors to be developed. The eventual manufacturing of large-scale arrays composed of highly miniaturized biosensing elements that enable the real-time, parallel monitoring of multiple species is an important driving force in biosensor research. Biosensors can be based on a variety of chemical or physical phenomena. Optical sensing has many advantages over nonoptical sensing modalities. Optical sensing schemes are sensitive. Singlemolecule sensing is possible due to the possibility for single-photon detection. Several light signals at different frequencies can be sent over the same optical beam because they do not interfere with one another. By measuring differences in wavelengths, arrival times, or polarization states, optical signals can be readily multiplexed and demultiplexed. Some optical techniques, such as fluorescence, have intrinsic amplification in which a single label can lead to a million photons. Surfaceenhanced techniques such as surface-enhanced Raman scattering (SERS) from molecules located near metallic nanoparticles allow sensing of small concentrations and the ability to recognize specific
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Light source
Analyte (biological, biochemical, or biophysical)
Incident light
Optical transducer
Detector
Output light
Figure 1. Schematic showing the main parts of an optical biosensor.
analytes in the sample. In addition, some optical techniques, such as surface plasmon resonance (SPR) and null ellipsometry, are zero- or blackbackground techniques: the only source of the signal is due to the presence of the analyte species, thereby enabling high-sensitivity measurements. Finally, optical signals travel in an open path; no wires or other transmitting conduits are needed, although the use of optical fibers may be sometimes necessary for nonoptical reasons and may have certain advantages. This attractive feature enables remote measurements to be made. Several optical-sensing mechanisms exist. These mechanisms are identified in Table 1, along with the variables measured. Several variations are possible on each of these mechanisms as well as combination of more than one optical measurement in the same setup. Another classification scheme for optical sensors involves the categories of materials, surfaces, and arrays. For instance, Table 1. Summary of the optical techniques and measured parameters used in biosensing
Technique Ellipsometry and polarimetry Absorption spectroscopy Elastic scattering
SPR Phosphorescence Raman scattering Guided-wave resonance Evanescent wave Interference
in a classification schemes based on materials, we might categorize sensors as based on porous materials, nanoparticles, quantum dots, photonic crystals, and so on. The need to simultaneously measure many analytes is satisfied by multiplexing several sensing elements together. Arrays containing tens of thousands and even hundreds of thousands of sensing elements are commonplace in DNA microarrays. Fluorescence is the main transduction mechanism used in most such microarrays. Indeed, the ability of simultaneously sensing many analytes has revolutionized the thinking of researchers working on biosensing in general and optical biosensing in particular. Single-analyte measurement is no longer considered sufficient for most sensing applications, as it gives only partial information about the sample. A variety of interesting and novel methods for fabricating arrays, coupled with intelligent signal processing, have been developed. Our goal in this chapter is to provide an overview of optical techniques for biosensing. In order to avoid duplication of the subjects in this handbook, the reader is referred to the specific chapters that deal with particular optical techniques in detail; however, details are provided on optical techniques that are not covered elsewhere in this handbook. Additionally, some new enabling technologies such as miniaturized spectrometers, nanomaterials, porous materials, sculptured thin films, and microresonators are reviewed.
2 SPECTROSCOPIC TECHNIQUES
Measured parameters Polarization state Intensity, spectrum, polarization dependence Intensity; angular, wavelength, and polarization dependences; correlation length Intensity, phase, peak position, polarization Intensity, wavelength, polarization state, lifetime Intensity, peak position, polarization state Intensity, resonance frequency Mode profile, intensity, spectrum Intensity, phase
2.1
The Revolution of Parallel Miniature Optical Spectrometers
Miniature optical spectrometers, introduced originally by Ocean Optics, have revolutionized optical spectroscopy with their parallel-processing performance. Diffraction grating-based spectrometers are used along with charge-coupled devices (CCD) or pin-diode arrays (PDAs) for light detection. The early models were just a fraction of the cost of spectrometers then available on the market. Today, tens of companies produce and sell many different models of such spectrometers for a wide variety of applications.
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
Combined with fiber-optic technology, these spectrometers continue to be powerful and relatively inexpensive research tools. They are very compact and low in weight. They utilize universal serial bus (USB) technology, making them ideal for field use; for example, marine researchers have their spectrometers in waterproof housings to work with corals in their natural habitats, and other researchers use them for water-pollution measurements and plant-disease diagnostics in the field. The spectrometers have found applications in seas and oceans to measure fluorescence from marine organisms,12 around trees to measure the concentration of oxygen,13 in volcanoes for monitoring sulphur dioxide emissions,14 for analyzing the solar corona, and even to search for signs of life in simulated Mars missions.15,16 The majority of these spectrometers use the popular crossed Czerny–Turner configuration (Figure 2). Diffraction gratings are used to split incoming light into its spectral components. In order to avoid chromatic aberrations, curved mirrors are used to image the slit onto an array of detectors. For further miniaturization, some companies have started using gratings ruled on the curved mirror itself. The diffracted light falls upon the CCD array, and specialized software is then used to analyze and report spectral characteristics. As a grating’s efficiency is dependent on the (free space) wavelength, these spectrometers generally under-report the data at some wavelengths. Some diffraction gratings have their maximum efficiency at 500 nm and less, particularly
Curved mirror
Curved mirror
Diffraction gratings
Input slit
Detectors array
Figure 2. Schematic of the Czerny–Turner configuration commonly used in miniature spectrometers.
3
blue and red wavelengths. Ideally, one would use a reflectance standard along with the software to correct the spectrometer’s measurements, but that is not absolutely necessary. The data are easily exported to a spreadsheet program for further analysis. The use of fiber-optic cords offers advantages in that they can be tightly attached to the aperture (via SMA 905 terminated fibers), thereby protecting the internal parts of the spectrometer from any debris that may enter the spectrometer housing. The fiber-optic cords also offer other advantages. While it is apparent that cords are necessary for connecting optional accessories, the diameter of a cord is also a critical consideration. Very simply, the larger the diameter of the cord, the more light is transmitted. If high-intensity light saturates the CCD array and causes the reported intensity to be above the maximum allowed, a cord of smaller diameter could attenuate the signal sufficiently to allow measurement satisfactorily. Thus, an important consideration is the expected light intensity. In addition to the diameter of the fiber-optic cord acting as an intensity regulator, the diversity of available slit widths (5, 10, 25, 50, 100, and 200 µm) is beneficial. Low-light applications (such as the ones involving fluorescence measurements) require larger slit widths (say, 400 µm), while higher light intensity requires smaller slit widths. Low light-levels can be compensated by changing available integration time of a detector from a few microseconds to few hundreds of microseconds, and some spectrometers offer even up to a few seconds. On the other hand, optical resolution is a function of slit width and the resolving power of the diffraction grating. An optical resolution of just a fraction of a nanometer is possible, and subangstrom resolution maybe achieved almost effortlessly (with a 5-µm-wide slit and a dense diffraction grating) if the available light intensity is sufficient. However, the standard optical resolution is generally between 2 and 10 nm. High-resolution and highly sensitive spectrometers have just emerged for portable Raman spectroscopy, and are ideal for optical sensing in the field. The spectrometers can even be cooled and temperature-controlled for better stability and signal-to-noise ratio. Gratings with maximum efficiency in the near-infrared range are also possible: spectrometers for the 900–2200 nm range are available using CCDs
4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
made of InGaAs with prices ranging from US $15K–25K. Biosensing as well as other biomedical applications have benefited significantly from these revolutionary developments in optical spectroscopy. Jobin-Yvon has introduced the only in vivo, fiberoptic spectrofluorometer designed specifically for skin-fluorescence measurements. This device uses a quartz-fiber bundle, which selectively delivers light in the ultraviolet to the near-infrared range to the sample and collects the resulting fluorescence. It is used to measure direct fluorescence from the skin, to evaluate cosmetic and pharmaceuticals, and to quantify drug delivery in photodynamic therapy (PDT).17 As a portable fluorescence spectrometer, it is also used to quantify trace contaminants for environmental monitoring. Multichannel computerized operation is possible, as shown in Figure 3, with each channel monitoring a single sample. An industrial computer system is available from StellarNet (www.stellarnet-inc.com) capable of providing 24-hour, nonstop service for process monitoring. For optical metrology, this spectrometer offers a high speed of processing, thus enabling high-throughput ellipsometry and reflectometry systems for inspection of semiconductor nanoelectronic fabrication processes. With this background, it is clear that miniature fast spectrometers have helped biosensing research and development significantly. But, even though their typical unit cost (few thousands of US dollars
and above) is only a fraction of the typical unit cost of classical bulky spectrometers, they are still too expensive for widespread use in homes, offices, and clinics. However, they are suitable as components in optical sensing systems that are expensive anyway. A further 10-fold reduction of their unit cost would enable the incorporation of these spectrometers in widely used, compact sensing devices. Technologies that can help achieve that goal include microelectromechanical systems (MEMS), nanotechnology, acousto-optics, or highspeed tunable liquid crystal filters, but all of these potential technologies are still too expensive for widespread use. Commercially available small spectrometers for the near-infrared and the visible ranges are often grating-based systems, each with an engine that is a few inches in each dimension. One type of spectrometer uses a linear, variable-band-pass optical filter instead of a grating. Several miniature spectrometers exploiting the technology of MEMS have also been either constructed or proposed, including some Fourier-transform spectrometers. A new design developed by Axsun Technologies18,19 based on MEMS technology uses an optical channel monitor that is commonplace in the telecommunications industry for monitoring dense wavelength-division multiplexing traffic. The spectrometer is a MEMS-based tunable Fabry–Perot filter operating in either a pre- or
Fiber Channel 1 spectrometer
Channel 2 spectrometer
Processor
Channel 3 spectrometer
Channel N spectrometer
Figure 3. Schematic of a multichannel spectrometer system.
Coupling lens
Sample
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
postdispersive mode (i.e., with the wavelengthselective device positioned either before or after the sample).
2.2
Absorption Spectroscopy in Biosensing
Optical absorption-based biosensors exploit many mechanisms ranging from pure resonant absorption of biomolecules to the use of SPR, which is engendered by absorption in thin metal films. Since the middle of the twentieth century, opticalabsorption spectroscopy has been used to detect pollutants in water.20–22 The oxymeter, used for measurement of oxygen saturation in blood hemoglobin,23 was among the first clinical optical absorption-based sensors. Techniques to enhance the effects of absorption have been developed such as the attenuated total internal reflection (ATIR) technique used for evanescent-wave sensing in the cladding of optical fibers and planar waveguides (PWs). Several schemes for absorption spectroscopy exist. The easiest and most widely used is the spectrophotometry, wherein quantitative determination of the absorbing species (chromophores) is done using optical spectroscopy from the ultraviolet to the infrared ranges. Spectrophotometry is used to measure absorption of various enzyme assays, and for the detection of proteins, nucleic acids, and metabolites.24 In spectrophotometry, the reference measurement is taken first using a well-known calibration sample, and then the transmittance through the sample is measured to yield the absorbance. This technique is thus based on the Beer–Lambert law A = ζ CL Incident laser pulse
(1)
where A = − log10 T is the absorbance (a desirable name for A is insertion loss, but that term is not in vogue) and T is the transmittance, ζ is the molar extinction coefficient in cm−1 mol−1 l−1 , C is the concentration in mol−1 l−1 , and L is the optical path length in cm. Since the reflectance from the sample is not very sensitive to the concentration, it can be either ignored or measured just once and then taken into account. The Beer–Lambert law is premised on the assumption of uniform concentration and discounts any dependence on the angle of incidence of light. Variations in concentration can occur, for example, due to dimerization of chromophore molecules at higher concentrations (provided the ζ value for the dimer is different from that for the monomer), thus causing a nonlinear dependence of the absorbance A on the concentration C. Another approach is the cavity ring-down technique25 in which light pulses of duration shorter that the round-trip time inside a cavity are launched into the cavity, and the transmitted pulses after each round trip are monitored with a detector and an oscilloscope (Figure 4). Since each of the injected photons bounces back and forth in the resonant cavity for a certain interval of time before it is absorbed, the transmitted photons are delayed on the average by the ring-down time of the cavity, thereby shifting the phase of the transmitted light. This phase shift is inversely related to the sum of the reflection losses in the cavity, which can be determined to a precision of the order of 100 ppm/round trip. If the cavity is filled with an absorbing medium and the mirror loss is assumed to be negligible, the system becomes an absorption spectrometer. For a cavity length of 1 m, the 100-ppm/round trip precision translates to a minimum detectable absorbance of ≈10−6 cm−1 . The major limitation Transmitted train of pulses
Analyte inside cavity
Mirror
5
Mirror
Figure 4. Schematic of a cavity ring-down absorption spectroscopy setup.
6
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
on the precision of this technique is the noise in the transmitted beam, which affects the determination of the phase of the transmitted beam; practical limitations on the signal integration time restrict this approach to precision on the order of 50 ppm/pass. For a uniform absorber, the intensity of the transmitted signal decays exponentially with time as −t I = I0 exp (2) τ where I0 is a reference intensity, and τ = l/{c(1 − R + Y + αl)} is the decay constant with R being the cavity mirror’s reflectance, Y the mirror loss, α the absorption coefficient of the analyte, c the speed of light in free space, and l the cavity length supposed to be filled by the analyte. The absorbance is determined as l 1 1 A= (3) − 2.303c τ τe where τe is the decay time when the cavity is without the analyte. Alternatively the absorption coefficient of the analyte can be determined from the ratio between the two decay times if the mirrors losses are neglected or measured separately. The presence of noise in the transmitted beam is intrinsic to the foregoing technique. Fabry–Perot analysis shows that transmission will occur only when one of the laser modes (which occur at integer multiple frequencies of c/2l, where l is the length of the probe laser cavity) matches the frequency of one of the cavity modes. Since the two cavity lengths vibrate independently, the modes of the two cavities match up periodically to permit some transmission, but at random times. Although the phase shifts of the individual transmitted photons are fixed, the phase of the transmitted beam is obscured by the random-intensity noise. Anderson et al.26 addressed this problem by configuring the detection electronics in a triggered mode of operation and waiting for the modes to overlap before starting the data-acquisition procedure. The detector monitoring the transmitted beam triggers on the electronics when longitudinal mode-matching has occurred in one of these coincidences, thereby permitting the intensity to build up in the test cavity so that a signal is observed on transmission. Triggered by the “light is present” observation at the detector, a fast optical switch then shuts off the
probe laser beam, and the ring-down time is measured. The sum of the losses in the test cavity is related to the ring-down time. A precision of about 5 ppm has been demonstrated with this method. Although this reconfiguration of the electronic circuitry reduces the noise, the problem of the random mode-coincidences remains, producing thereby intensity fluctuations, intermittent delays between successive signals, and occasional nonexponential decays. O’Keefe et al.27 proposed a technique which allows measurement of the decay time of the test cavity, but avoids the problems associated with the requirement of longitudinal mode coincidences by using short optical pulses so that every pulse of the probe laser enters the cavity with no additional intensity fluctuation or time delay. This results in improved sensitivity and higher data rate, and also significantly relaxes the requirements on the stability of the system. Another enhancing technique is the ATIR technique or the evanescent-wave absorption technique. Use of ATIR in waveguides and optical fibers increases the interaction length between the evanescent field and the analyte present in the region of evanescence. Fibers made of crystalline silver halide have been used recently, as they are excellent for Fourier spectroscopy in the infrared range.28 This is because the fiber’s surface does not need any protection jacket and the fiber’s transmission range of 4000–600 cm−1 overlaps with the majority of spectral absorption lines of organic molecules. The sensors are based on evanescentwave spectroscopy (the region of evanescence is about 5–10 µm in thickness) wherein resonant absorption of the evanescent light results in dips in the fiber’s transmittance spectrum. The absorption spectral lines are characteristic of the analyte material and have modulation depths corresponding to analyte concentration. One of the major advantages of fiber-based ATIR sensors is that they enable remote chemical analysis in real time. Another significant advantage is that the mechanical robustness of fibers allows their use many times for probing.
2.3
Fluorescence Techniques
Fluorescence is the most widely used technique in analytic instruments. The large number of applications ranges from analytic measurement of metals
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
in aqueous environments to pH measurements in living cells. Fluorescence spectroscopy is also a powerful laboratory tool, both for applied research and for fundamental studies of physical processes in molecules and interactions between biomolecules such as proteins and nucleic acids. In clinical laboratories, the use of fluorescence is dominant for the determination of analyte concentration; indeed, fluorescence immunoassays have largely replaced radioimmunoassays. The high levels of sensitivity and dynamic range are responsible for this dominance, and many laboratories have reported the detection of single molecules. In fluorescence, the incident light is completely absorbed and the molecule is transferred to an excited state from which it can go to various lower states only after a delay quantified by the resonance lifetime.29 Dependence on the polarization state of the incident light is exhibited when the molecules lack spherical symmetry and there is some orientational order. A technique called polarized fluorometry is therefore becoming popular for detecting various analytes. Information on the lifetimes and the dynamic decay of the fluorescence signal provides another important tool, called time-resolved fluorescence, for detection and identification of analytes. Applications of fluorescence in biology and medicine are numerous and significant. Fluorescence techniques play a critical role in the description of biological processes at the molecular and cellular levels. Endogenous and exogenous fluorescent molecules are used as specific markers of metabolic status or disease processes. Fluorescent particles are used as contrast agents for the study of transport phenomena (e.g., blood) in biological media. The successes reported are for transparent media or in tissue when the analyte is located close to the surface. Most biological tissues scatter light so strongly, however, that even special techniques to remove multiply scattered light (such as two-photon and confocal microscopies) fail at depths greater than 0.5 mm below the tissue surface. A complicating factor is the strong attenuation of light as it passes through tissue that degrades the signal-to noise ratio of detected photons. Fortunately, development of fluorescent dyes (such as porphyrin and cyanine) that excite and reemit in the “biological window” at near-infrared wavelengths, where
7
scattering and absorption coefficients are relatively low, has opened new possibilities for deep fluorescence imaging in tissue.30 The instrumentation required is usually not very expensive and can be combined with other instruments such as confocal microscopes and spectrometers. The fluorescence setup consists of a light source, an excitation-wavelength filter, a sample holder, an emission-wavelength filter, and a photodetector that converts any emitted fluorescence photons into an electronic signal. The excitation wavelengths lie usually from the ultraviolet (200 nm) to the visible range (650 nm), corresponding to the electronic energy levels in the outer shells of atoms. Mercury arc lamps are usually satisfactory but can produce only a limited number of wavelengths in the ultraviolet to the visible range (e.g., 253.6, 302.2, 313.2, 365, 404.7, 435.8, and 546.1 nm). Most fluorometers use a high-pressure xenon arc lamp since it has a continuous high intensity spectrum between 200 and 1000 nm wavelength; and pulsed lamps with power of 75 W deliver high-enough peak power to enable high fluorescence signals. Fluorescence parameters that are characteristic of the fluorescent material and its surrounding can be used as signatures that can be correlated with the analyte. Examples include the lifetime and the fluorescence anisotropy. Fluorescence-lifetime imaging (FLIM) is becoming nowadays a popular technique for biological imaging. Time-resolved measurement techniques use pulsed exciting light (usually in the nanosecond range) and then analyze the decay of the emitted pulses. Another possibility is to modulate the exciting light at a certain frequency and then measure the delay time and demodulation of the emitted signal. For fluorescence-anisotropy measurement, linearly polarized excitation light is used while the intensity of emitted signal is measured with the analyzer axis parallel (I|| ) and perpendicular (I⊥ ) to the exciting polarization. The degree of fluorescence anisotropy is defined as a=
I|| − I⊥ I|| + 2I⊥
(4)
wherein the denominator on the right side represents the total fluorescence intensity. The fluorescence anisotropy is a measure of the rotational motion or dynamics of the analyte molecule, which
8
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
is important in obtaining flexibility parameters of local segments of macromolecules. Another technique is a combination between the polarization anisotropy and time-resolved measurement. This is called the polarized excitation and emission decay technique: the sample is excited using vertically polarized light, and the horizontally as well as the vertically polarized components of the fluorescence decay are separately recorded. The observed decay in the intensity of either of the linearly polarized component is due both to (i) the decrease in size of the population of excited molecules and (ii) the tumbling motion of each molecule, which reorients the transition dipole moment between absorption and emission and so changes the intensity of the light observed using a particular orientation of the emission polarizer. Hence, the polarized fluorescencedecay components contain information about the dynamical motion of the molecules. An interesting way to improve the fluorescence uses the evanescent-field technique relying on total internal reflection (TIR). First described by Hirschfeld31 in 1965, evanescent-field excitation was first used by Kronick and Little32 for fluorescence immunoassays. Total internal reflection fluorescence (TIRF) is a means to selectively excite light emission from fluorophores present near the surface of a waveguide and is relatively immune to the bulk effects. Several studies have been conducted on the use of TIRF for the detection of a single biomolecule. Ligler and coworkers33 investigated multianalyte fluorescence detection using PWs. Plowman et al.34 reported femtomolar sensitivity of fluoro-immunoassay using a dual-channel waveguide. A commercially available product, the Zeptosens reader (Zeptosens, Bayer Tech. Services GmbH) uses the evanescent
field to excite fluorescence in the near-interface region of a microarray chip and has sensitivity down to picomolar concentration (which is equivalent to zeptomoles of captured antibodies).35 There are several advantages to the use of PWs such as independence from the molecular weight of the analyte unlike refractive-index-based techniques, relative ease of preparation, possibility of creating arrays of different recognition molecules using different techniques (e.g., photolithography and ink-jet printing), and integration with fluidic systems. The major disadvantage of using TIRF systems is the requirement of a fluorescent label that should ideally be so selective as to not interfere with the binding interactions of the analyte. Labeling of the analyte with a fluorescent dye is a method for enhancing the signal from the analyte to ensure specific measurements or to ensure both specific detection and amplification of the signal from the specific analyte. Three different major assay formats (direct binding, competitive, and sandwich assays), as shown in Figure 5, have been used. In the direct assay format (Figure 5a), first the recognition molecules are immobilized on the surface of the waveguide and then a fluorescence signal is emitted when analyte molecules with fluorescent labeling bind to the recognition molecules. The competitive assay format (Figure 5b) is normally used in immunoassays for small molecules containing a single epitope (which is the part of a macromolecule that is recognized by the immune system, specifically by antibodies, B cells, or cytotoxic T cells). This format is classified into two subformats. In the first subformat, unlabeled and labeled analytes compete for binding to the immobilized recognition molecules on the waveguide’s surface, and the
Fluorescence-labeled antibody Fluorescence-labeled analyte Analyte Antibody (a)
(b)
(c)
Figure 5. Schematics of the main three formats of solid-phase immunoassays: (a) direct assay between the analyte in solution to the immobilized antibody (recognition molecules), (b) competitive assay between the fluorescently labeled (known concentration) and unlabeled (unknown concentration) antigen for the binding sites of the immobilized antibody, and (c) sandwich assay in which the analyte is put between immobilized antibody and secondary labeled antibodies.
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
decrease in the fluorescence signal is proportional to the proportion of the unlabeled analyte. The second requires that the recognition molecules be present in the solution and an analog of the analyte be immobilized on the optical waveguide. The sandwich assay format (Figure 5c) is commonly used for large molecules and requires secondary fluorescent recognition species. This fluorescent secondary recognition molecule is used to detect an analyte that has been captured by immobilized recognition element on the waveguide’s surface. The sandwich and the competitive assay formats are the most commonly used in the immunoassays using PWs. A technique using the evanescent field from a planar waveguide to excite fluorescence near interface region is called waveguide excitation fluorescence microscopy (WEFM ). The use of WEFM to imaging biointerfaces shows a 10-fold improvement in sensitivity in comparison to conventional fluorescence microscopy.36,37 WEFM has been integrated with other techniques such as AFM, SNOM, magnetic tweezers, and optical tweezers for dynamic in situ studies of single molecules at interfaces.
2.4
Raman Spectroscopy for Biosensing
Raman spectroscopy is a form of electronic (more accurately, vibronic) spectroscopy. Classically, the Raman effect arises when a photon incident on a molecule interacts with the molecule’s induced dipole moment, which is proportional to the polarizability change during the molecule’s vibration.38 Quantum mechanically, the interaction is viewed as a scattering event that occurs in 10 fs or less, and the scattering is described as an excitation to a virtual state lower in energy than a real electronic transition with nearly coincident de-excitation and a change in vibrational energy. The energy difference between the initial and final vibrational levels is called the Raman shift −1 ν¯ = λ−1 incident − λscattered
(5)
where λincident and λscattered are the wavelengths (in cm) of the incident and the Raman-scattered photons, respectively. Raman shifts range from a few hundred per centimeter (615–630 cm−1 for
9
ring deformation) to a few thousand per centimeter (3300–3400 cm−1 for bonded antisymmetric NH2 stretch in primary amines). Because only a small fraction of the incident photons is scattered inelastically (1 in 107 photons), the intensity of Raman scattering is low; hence, the accompanying thermal dissipation does not cause a measurable temperature rise in the analyte material. Various Raman techniques have been attempted to analyze blood, water, serum, plasma solutions, skin cancer, and the eye, but many problems remain before tissue diagnosis and blood chemicals analysis in vivo and in real time can be performed.39 Implementation problems include (i) instabilities in the laser wavelength and intensity and (ii) errors due to other chemicals in the tissue sample and long spectral acquisition times. Most importantly, the inherent difficulty of Raman spectroscopy is that its signals are very weak, Raman scattering having an intensity about a 1000th of that of Rayleigh scattering. With the replacement of slow photomultiplier tubes by faster CCD arrays and the manufacture of higher-power near-infrared laser diodes, that inherent difficulty is being alleviated. As mentioned in Section 2.1, miniature fiber-based spectrometers are already commercially available with a reasonable price in the range US $15K–25K. These spectrometers are suitable for Raman measurements. Although their resolution is relatively low (2–4 cm−1 ), they are useful for many biosensing applications because the Raman peaks of molecules in solution or in biological environment are broadened.40 For in vivo tissue studies, an infrared laser is generally used as the excitation source to minimize the background fluorescence. However, the inherent weakness of the Raman signal creates the needs for very high excitation intensity and relatively long signal-collection time; therefore, photothermal damage of the tissue is of great concern for in vivo measurements. Specificity is the most important requirement for an acceptable multicomponent blood analysis: signals from different blood constituents should be distinguishable from each other for independent quantification. Raman spectroscopy is well suited to this requirement. A significant advantage of Raman spectroscopy over near-infrared absorption spectroscopy is that the former’s spectrum
10
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
has distinct and pronounced peaks, thereby easing the task of separating signals produced by the analytes of interest. In contrast, absorption, scattering and polarimetry techniques in the near-infrared range should be thought of as purely empirical methods that rely heavily on statistical multivariate analysis. Another significant factor to consider is stability, that is, the sensitivity of the technique to disturbances or noises. The most significant noise that has been a problem for near-infrared absorption and scattering methods is the variability of the skin temperature. Raman spectra have been found to be not as sensitive to temperature changes as the near-infrared absorption and scattering spectra. Thus, Raman spectroscopy is one of the most promising techniques for accurate and reliable quantification of the many constituents of blood. The Raman spectra have sharp and distinct peaks, thereby allowing greater discrimination among closely spaced signals and, consequently, more accurate extraction of concentrations from the spectral data. However, Raman spectroscopy is still faced with several critical problems some of which are being overcome by the incorporation of miniature spectrometers and light sources and by using enhancement mechanisms explained in the following subsection. 2.5
Surface-enhanced Optical Phenomena
2.5.1 Nanoenhancement of Surface Plasmon Sensitivity (LSPR Technique)
Several research groups are now exploring alternative strategies for optical biosensing and chemical sensing based on the extraordinary optical properties of nanoparticles made of noble metals. Nanoscale chemosensors and biosensors can be realized through shifts in the localized surface plasmon resonance (LSPR).41–45 A LSPR nanobiosensor, based on LSPR spectroscopy, operates in a manner totally analogous to an SPR sensor by transducing small changes in the refractive index near a noble-metal surface into a measurable wavelength shift as follows46–48 : λmax = m(nadsorbate − nblank ) 2dadsorbate × 1 − exp − ld
(6)
Here, m is the refractive-index sensitivity of the sensor; nadsorbate and nblank are the refractive indices of the adsorbate (i.e., analyte) and the bulk environment prior to the sensing event, respectively; dadsorbate is the effective thickness of the adsorbate layer; and ld is the characteristic electromagnetic field decay length associated with the sensor. While the responses of the LSPR and SPR sensors can be described via the same equation, the sensitivities of the two techniques arise from different experimental parameters.49–52 Flatsurface SPR sensors have a large refractive-index sensitivity (∼2 × 106 nm/RIU, where RIU stands for “refractive index unit”), which is the chief component of their overall sensitivity.53 LSPR nanosensors have modest refractive-index sensitivity (∼2 × 102 nm/RIU),54 in contrast. Nevertheless both types of sensors have approximately equivalent sensitivity for a given adsorbate. In addition to the difference in refractive-index sensitivity, the electromagnetic field decay length ld is also different for SPR and LSPR sensors. SPR sensors have a decay length in the order of ∼200 nm. For LSPR nanosensors using noble-metal nanoparticles, a much shorter electromagnetic field decay length (∼6 nm) has been measured.55 The shorter decay length gives rise to the larger overall sensitivity of LSPR nanosensors. In contrast to the conventional SPR technology, LSPR technology promises multiplexed, high-throughput screening platforms in a highly miniaturized format, requiring small volumes (e.g., attoliters) of analyte solutions. The sensitivity is a few orders of magnitudes better than that of the conventional SPR sensors without metallic nanostructures. In addition, LSPR technology does not require precise controls of the angle of incidence and the ambient temperature, both of which are necessary for the conventional SPR technology. As the measurements are noninvasive in nature, the LSPR platforms are ideal for in vivo quantification of chemical species and the monitoring of dynamic processes inside biological cells. LSPR sensors can be divided into three broad groups: (i) those based on monitoring changes in the relative permittivity of the immediate environment, (ii) those based on changes in SP coupling, and (iii) those exploiting a combination of these two effects.56 The first group of
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
LSPR sensors were implemented for the detection of hexadecanethiol down to zeptomolar sensitivity by monitoring changes in the resonant Rayleigh scattering.57 The second and the third groups were demonstrated for many chemical and biosensing applications by monitoring the changes in LSPR band of metal nanostructures upon analyte binding, using standard spectrophotometric instruments in the transmission mode.58 The advantage of LSPR sensing in the transmission configuration over conventional SPR sensing is a simple experimental procedure that involves measurement just at one wavelength. This simplicity enables the development of disposable LSPR sensors for personal medicine and field applications.
2.5.2 Resonant Raman Effect (RRE) and Surface-enhanced Raman Scattering (SERS)
Four mechanisms are used to enhance the Raman signal59–61 : (i) stimulated Raman scattering due to the excitation of analyte molecules by a high energy pulse (optical electric field of strength ∼109 V cm−1 ); (ii) coherent anti-Stokes Raman scattering (CARS) due to excitation with two strong collinear laser beams having frequency difference equal to the frequency of a Raman peak; (iii) resonant Raman effect (RRE) caused by excitation with photon energies corresponding to resonant energies within the electronic spectrum of the analyte molecules; and (iv) SERS, when the analyte molecules in close proximity (fraction of nanometers) of metallic nanoparticles are excited. The last two mechanisms are the ones that are mostly used for optical biosensing. The RRE increases the intensity of some Raman-active vibrations by a factor of 102 –105 . This effect occurs when the excitation-laser frequency is chosen in such a way that it crosses the frequencies of excited electronic states and resonates with them. The enhancement factor increases when the molecular expansion along its axis of vibration is higher as it absorbs photons. Formally, one can think of the Raman transition probability being proportional to the elements of the polarizability tensor of a bound electron; as the exciting frequency approaches the
11
resonance frequency, these elements are enhanced in a Lorentz model of the bound electron. A common example of this mechanism is furnished by the ring-breathing (in-plane expansion) modes of porphyrins. Another mechanism, called vibronic enhancement, involves vibrations which couple two electronic excited states. In both mechanisms, the enhancement factors are nearly proportional to the intensities in the absorption spectrum of the adsorbate. The enhancement does not begin at a sharply defined wavelength. In fact, enhancement by factor of 5–10 is usually observed if the wave number of the exciting laser is only within a few hundred per centimeter below the electronic-transition wave number of the analyte molecule. SERS is the second relevant enhancement mechanism. The Raman scattering from a compound (or ion) adsorbed on or even within a few angstroms of a structured metal surface can be enhanced by factor of 103 –1014 compared to the case when it is in a solution. SERS is strongest on a silver surface, but is observable on gold and copper surfaces as well, and it is now known that the shape of the nanoparticle plays a crucial role in determining the enhancement factor. So far, the triangular-pyramid shape has been found to give the strongest enhancement. Although a complete understanding of SERS has not been achieved yet, two main mechanisms are widely accepted. The first, called chemical enhancement, involves enhancement of polarizability of the analyte molecule that may occur because of a charge-transfer effect or chemical bond formation between the metal surface and the analyte molecules. The second is due to the enhanced electromagnetic field produced at the surface of the metal when the wavelength of the incident light matches the SPR wavelength of the metal. Molecules adsorbed or in close proximity to the metal surface experience an exceptionally large electric field. Because the Raman effect is proportional to the fourth power of the field amplitude, the efficiency is enhanced by factors as large as 1014 . Molecular vibrational modes normal to the metal surface are most strongly enhanced in comparison to other vibrational modes. Electromagnetic simulations confirm that the electric field can be enhanced62,63 by a factor of 103 and
12
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
so the Raman signal is enhanced by a factor of 1012 . For a spherical nanoparticle whose radius is much smaller than the wavelength of light, the electric field is uniform across its dimensions, and the electrostatic (Rayleigh) approximation suffices to explain the enhancement. More generally, the field induced at the surface of a spheroidal nanoparticle (with major and minor semi-axes of lengths L and S) is related to the applied external field as ε1 (ω) − ε2 (ω) Einduced = Elaser (7) ε1 (ω) + χε2 (ω) where ε1 (ω) is the complex-valued, frequencydependent, relative permittivity scalar of the metal, ε2 (ω) is that of the ambient material, ω is the angular frequency, and χ is a geometrical factor that depends on the shape of the nanoparticle. The factor χ equals 2 for a sphere, but χ is larger than 2 for prolate spheroids (L > S) and less than 2 for oblate spheroids (L < S). SPR greatly increases the local field experienced by a molecule adsorbed on the surface of the nanoparticle, when Re{ε1 (ω) + χε2 (ω)} = 0. One can visualize this phenomenon by considering the nanoparticle as localizing the electric field of a dipole field centered in the sphere, which then decays with the dipole decay law away from the surface in all directions. In this sense, the nanoparticle acts as an antenna which amplifies the intensity of the scattered light. When χ is greater than 2, the plasmon resonance condition Re{ε1 (ω) + χε2 (ω)} = 0 is satisfied for a wavelength that lies to the red of that for a sphere (due to the fact that the real part of ε1 of metals is, according to equation (7), more negative for longer wavelengths). Of course, this also means that for oblate spheroids, the resonance is blueshifted relative to a sphere. However, the resonance described here refers to an incident field with the electric field polarized parallel to the axis of symmetry of the spheroidal nanoparticle. There is another plasmon resonance associated with the incident electric field polarized perpendicular to the symmetry axis. This resonance is identical in frequency to the parallel resonance for a sphere, but it shifts in the opposite direction for a spheroid, that is, blueshifting for prolate spheroids and redshifting for oblate
spheroids. The parameter χ for the two cases of parallel and perpendicular polarization is given by59 χ|| =
χ⊥ =
2 − 1 (8) ξ +1 −2 (ξ 2 − 1) ξ ln ξ −1 −2 −1 ξ +1 − 2ξ 2 ξ(ξ 2 − 1) ln ξ −1
(9)
where ξ = (1 − S 2 /L2 )−1/2 . The signal enhancement is so dramatic that very weak Raman peaks that are unnoticeable in spontaneous Raman spectra can appear prominently enough in the SERS spectra. Some trace contaminants can also contribute additional peaks. Moreover, because of chemical interactions with metal surfaces, certain peaks that are strong in conventional Raman spectra might not be present in the SERS spectra at all. The nonlinear character of signal intensity as a function of the concentration complicates things even further. Very careful consideration of all physical and chemical factors must be made while interpreting SERS spectra, which makes it extremely impractical. Because of such complications, the surfaceenhanced resonance Raman spectroscopy (SERRS) was developed. As it exploits the best features of both the SERS and the RRE, the resulting enhancement of the Raman signal intensity can be as high as 1014 . Additionally, the SERRS spectra resemble the regular RRE spectra, which make the former much easier to interpret. SERS was discovered with pyridine. Other aromatic nitrogen- or oxygen-containing compounds, such as aromatic amines or phenols, also display strong enhancement due to SERS. The enhancement can also be seen with other electron-rich analytes such as carboxylic acids. Although SERS allows easy observation of Raman spectra from solutions with concentration in the micromolar (10−6 ) range, slow adsorption kinetics and competitive adsorption limit its application in analytical chemistry. The SPR intensity is dependent on many factors, including the wavelength of the incident light and the morphology of the metal surface. The Raman excitation wavelength should match the plasma wavelength of the metal, which is about 382 nm
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
for a 5 µm silver particle but can be as high as 600 nm for larger ellipsoidal silver particles. The plasma wavelength shifts to 650 nm for copper and gold, the other two metals that are used for SERS at wavelengths in the range from 350 to 1000 nm. The best modality for SPR excitation is the use of either a nanoparticle ( nr , as θi is increased there is a value of θi for which the refracted wave propagates parallel to the interface (and is called a surface wave). As shown in Figure 6(b), then θr = 90◦ and the incidence angle is called the critical angle: −1
θc = sin
nr ni
(11)
When the angle of incidence exceeds the critical angle, no refraction occurs; instead, the light is
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
nr qr
·
ni
qi qi
qc
(a)
(b)
(c)
Figure 6. Illustration of total internal reflection (TIR) at a specularly smooth and planar interface of two different nonabsorbing materials.
reflected back into the medium of incidence. The TIR phenomenon is illustrated in Figure 6(c). No net flow of energy occurs across the interface under the TIR conditions.95 In order to satisfy the boundary conditions, the electromagnetic field must penetrate the optically rarer medium (Figure 7) and an evanescent wave is then generated with exponentially decaying amplitude: −z E = E0 exp (12) dp where z > 0 is the distance from the interface, and dp is the penetration depth: dp =
λ 2πni sin2 θi − (nr /ni )2
(13)
The penetration depth is of the order of few hundred nanometers or less, and can be controlled
Z
Evanescent wave
nr dp ni
15
by an appropriate choice of the ratio of refractive indices of the two materials, the angle of incidence, and the wavelength of the incident light. Another phenomenon important for sensing is frustrated total internal reflection (FTIR). When two materials of higher refractive index are separated by a thin layer of a material having lower refractive index, the energy of the incident light may flow through the gap between the two optically denser materials if the layer is much thinner than a wavelength. Thus, TIR is effectively frustrated, resulting in a partially or fully transmitted beam, depending on the size of the gap. This phenomenon is therefore called FTIR. It is also called optical tunneling, in analogy with the quantummechanical tunneling of a particle through an energy barrier. For angles of incidence much larger than the critical angle, the evanescent field decays very rapidly and a smaller gap is required to achieve significant optical tunneling. Therefore, for a specific angle of incidence, variation in the size of the gap can be used to control the efficiency of power transfer from one high-index material to the other.
3.3
Planar Waveguides
Waveguides are structures that confine and guide electromagnetic radiation (Figure 8). A PW, at its simplest, consists of a thin, transparent, dielectric film deposited on a substrate, and a cover may be put on top of the thin film. In order to achieve a true-guided mode in the thin film, the following condition must be satisfied: nf > max{ns , nc }. If ns = nc then the PW is referred to as being asymmetric; otherwise, it is symmetric. A further requirement is that the incidence angle θi must exceed the critical angle. Thus, the incidence angle must satisfy the following criteria: θi > sin−1 (ns /nf ) and θi > sin−1 (nc /nf ).
E
Cover nc Guiding layer n f
Standing wave
Figure 7. Schematic of the field structure when TIR occurs at a smooth planar interface.
Substrate ns
Figure 8. Schematic showing a standard PW, along with the transverse intensity profile of a guided mode.
16
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
nc
qi
qi
qi
nf Z
ns (a)
Radiation mode
(b)
Substrate mode
(c)
Guided mode
Figure 9. Ray patterns of (a) a radiation mode, (b) a substrate mode, and (c) a guided mode in a PW, where nf > ns ≥ nc .
Figure 9 illustrates the distinctions between radiation modes (no confinement), substrate modes and guided modes (full confinement). Numerous materials can be used for thin-film PWs, many of which can be spun, dip-coated, or deposited by various techniques on to a substrate. The dielectric film must be of sufficient thickness to support propagation of guided modes and should support a flow of electromagnetic energy only along the guiding structure not perpendicular to it. Light can be led into the thin film using a prism coupler, a grating coupler, an optical fiber, or an end-fire focusing coupler. The basic sensing principle of the PW sensor is to measure changes in the modal effective index neff due to refractiveindex changes in the cover. After propagating through the sensing section of the PW, light is coupled out and its intensity is measured by a detector to generate the sensorgram. A PW can be designed to support many modes propagating along the guide axis, each with a different modal wave number or modal effective index. These modes follow a zig-zag path as the light is totally reflected in the region between the boundary surfaces at angles θi greater than critical angle. However, not all arbitrary beams launched at incidence angles beyond the critical angle can propagate like guided modes. In a guided mode, the incident and reflected light at any position in the guide must be in phase, which requires the magnitude of the total phase change after a complete cycle of TIR to be an integer multiple of 2π . As the ratio λ/df between the wavelength λ and the waveguiding layer’s thickness df increases, the mode becomes more confined to the waveguide layer. This trend affects neff , which at the cutoff is equal to the smaller of the refractive indices of the cover and the substrate
(i.e., min{nc , ns }), and which becomes equal to the refractive index nf of the guiding layer for large λ/df .
3.4
Planar-waveguide Sensing Configurations
The fact that the (evanescent) fields of the guided mode extend into the cover and substrate of the PW makes the device useful for sensing purposes. The cover can be coated with a reagent, such as an indicator or recognition element. A chemical reaction within the region of the evanescent field is sensed due to the changes induced by the analyte on the sensor surface, such as scattering, absorption, fluorescence, or changes in the refractive index. PW sensors are attractive for several reasons. Their small size and flexibility facilitates the miniaturization of analytical instrumentation for in situ measurements. Their availability encouraged the development of remote and real-time continuous sensing in gaseous and liquids without sampling. They are immune to electromagnetic interference and resist hostile and hazardous environments. PW sensors can be fabricated from a variety of materials and can be relatively easily adapted for specific biomedical applications that require biocompatibility, sterilization, disposability, and the capability for in vivo measurements if needed. Light signals in PW sensors can be modulated in several ways to eliminate interference and facilitate the use of internal references. Their adaptability to act as electrodes allows integration with other techniques such as electric-field and magnetic-field techniques to provide more information, reduce nonspecific binding, shorten the analysis time, and improve the sensitivity.
OVERVIEW OF OPTICAL BIOSENSING TECHNIQUES
3.4.1 Asymmetric Waveguides
17
the critical angle on the substrate–spacer interface is coupled into the waveguiding layer via the evanescent field in the spacer, when the propagation constants in the substrate and waveguide match. For monochromatic light, this occurs over a very narrow range of incidence angles, typically spanning considerably less than 1◦ . Alternatively, it can be operated at a fixed incidence angle, and coupling occurs over a narrow range of wavelengths.102
Asymmetric PWs are useful as direct biochemical sensors, as the TIR boundary can be between the chemically selective waveguiding layer and the sample. The basic sensor is a thin waveguiding layer (on which the recognition element is immobilized) that is either deposited or engraved on either a polymer or a glass substrate. Air serves as the cover medium in the simplest case (Figure 10a). This basic device was introduced for sensing purposes by Tiefenthaler and Lukosz,96 who applied it as a humidity and gas sensor by detecting a change in the refractive index of the cover medium. Typical PW sensors use the high-index waveguide formed by depositing silicon nitride (Si3 N4 ),97 indium tin oxide (ITO),98 or tantalum pentoxide (Ta2 PO5 )99 on a glass or polymer substrate. The efficacy of such sensors have been demonstrated for label-free and fluorescence detection. The planar waveguide can be a multilayer and the integrated waveguide SPR biosensor.100 The resonant mirror (RM) was developed for immunosensors (e.g., the commercially available product IAsys from Affinity Sensors, a company in Cambridge, UK), as it is very sensitive to changes in the refractive index of the interfacial layer caused by the binding of macromolecules such as proteins to immobilized biorecognition species such as antibodies.101 In the RM sensor, FTIR is used to couple the light in and out of a high-index waveguiding layer. The RM is effectively a prism coupler where the air gap has been replaced by a low-index dielectric layer. Figure 10(b) shows the RM device structure, consisting of a highindex substrate (n = 1.72), a thin low-index spacer (about 550 nm of silica) and a very thin monomode waveguiding layer (about 80 nm of Si3 N4 ). The high-index resonant layer acts as both a waveguiding and a sensing layer. Light incident above
3.4.2 Grating Couplers
Some PW biosensors are called grating couplers. Either one or two gratings are fabricated on the surface of a PW by conventional lithographic or holographic techniques. Coupling occurs when the propagation vector of a guided mode is matched by the propagation vector of one of the diffracted orders from the grating. This is expressed mathematically as: neff = nc sin α + mλ/, where m is the diffraction order, is the grating period, and α is the propagation angle. In a grating coupler, a reversal of the direction of wave propagation transforms reciprocally the in-coupling into an out-coupling process. Therefore, in some sensor platforms, separate input and output grating couplers are employed. Grating couplers have been fabricated from semiconductors, dielectrics, and polymers. A grating coupler can be an integral part of the sensor. The term integrated optical waveguide (IOW ) sensor refers to a planar waveguide integrated with the grating couplers where the PW can be part of a flow-through cell. Brecht et al.101 concluded that the use of the IOW significantly improves the sensitivity of a sensing system by a factor of 100 compared with waveguides illuminated in discrete spots. The grating waveguide sensors have been demonstrated for single-sensing platforms as in
High-index waveguide Si3N4 Substrate (a)
Substrate (b)
Figure 10. Schematics of (a) a single-layer PW (b) and a multilayer PW that is an RM.
SiO2
18
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Microvacuum chips (OWLS, Microvacuum, Hungary) and Zeptosens high-throughput screening array platforms (Zeptosens, Bayer, Germany)103 for the detection of different bioanalytes using refractive index and fluorescence. A label-free PW with induced Bragg grating has been used in the detection of protein adsorption. Upon adsorption, the corrugation depth of the grating rises, which causes changes in the reflectance. Resonant grating waveguide structures (GWS) have also been used for biosensing. They are very sensitive to the adsorption/desorption of molecules on the waveguide surface and to any change of refractive index of the medium covering the surface of the sensor chip. When a resonant GWS is illuminated with an incident light beam, the diffracted light matches the guided-mode condition and interference with the m = 0 (specular) diffracted order causing resonant reflection backwards. This happens at a specific wavelength and incidence angle of the incident beam at which the resonance condition is satisfied, whereby the rediffracted beam destructively interferes with the transmitted beam, so that the incident light beam is completely reflected.104,105 The combination of a colorimetric resonant grating and photonic crystal embedded in the plastic surfaces of microtiter plates (96-, 384-, and 1536-well) has been developed by SRU Biosystems as a label-free, high-throughput, screening platform. The sensor can detect a shift in wavelength as low as half a picometer. Binding interactions can be quantified with proteins, cells, and small molecules. Sensitivity is quoted in the 0.05 µg ml−1 to 1 mg ml−1 range with molecular weights 15 nm, with a near-linear region resembling the response to SAMs with variable thickness.49 Since biorecognition interfaces usually comprise several layers, generally including biomolecules,60 thicker metal films with better-defined islands—providing higher distance sensitivity—should be preferred as platforms for LSPR biosensors. The possibility of spatial distribution of different analyte binding sites on the LSPR transducer should be considered. This is related to inhomogeneity of the LSPR transducer surface on the nanometer scale and the existence of regions with different local refractive index sensitivity.80 Different optical sensitivity for binding to the metal nanostructures and to the exposed hole bottom was demonstrated for nanohole arrays.18,67 Hence, in the application of equations (2) and (3) to the analysis of the response of heterogeneous LSPR substrates, the bulk refractive index sensitivity m represents a local value or a weighted average, depending on the particular spatial distribution of bound analyte. A basic question in LSPR spectroscopy concerns the quantitative relationship between surface coverage by the analyte (in the submonolayer regime) and the LSPR response. We addressed this issue by studying the sensitivity of the SP band to change in the fractional coverage of a self-assembled chromophoric monolayer on an Au island film up to a full monolayer coverage.14,15 Au island films similar to the one presented in Figure 2(c,d) were used as LSPR transducers, and the disulfide chromophoric molecule 1 (Figure 11b)15 or Co-tetraphenylporphyrin (CoTPP)14 (not shown) served as a target analyte. The spectral separation between the molecular absorption band and the Au SP band allowed independent determination of the fractional surface coverage and the SP change during formation of a SAM. The results in Figure 11(a,b) show a clear correlation between the two absorption
13
bands, increasing simultaneously during formation of a SAM of 1. As in the previous examples, the maximum change in SP extinction (the PIC) is red-shifted with respect to the maximum SP absorbance. The kinetics of the shift of both the wavelength of the SP band and the PIC are similar (Figure 11c), showing saturation upon completion of monolayer adsorption. The PIC and the chromophore absorption, the latter directly proportional to the fractional coverage by the SAM, are linearly correlated over the entire range of surface coverage (Figure 11d). The same linearity applies to the shift of the wavelength of the SP band (not shown). In biological sensing, the target analyte usually does not bind directly to the surface of the LSPR transducer (as in the previous example), but rather binds to specific receptors preassembled on the surface. Binding of an analyte to a receptor layer on an LSPR transducer was modeled by binding of Co-TPP to an imidazole-functionalized SAM on an Au island film.14,42 As in the corresponding case of formation of a SAM of 1 (Figure 11), a linear correlation was observed between the fractional surface coverage by the analyte, bound at a finite distance form the island surface via the imidazole receptor, and the optical response (PIC or shift of the wavelength of the SP band), up to a monolayer coverage. Similar detection of analyte binding at a distance from a metal island surface was reported for protoporphyrin IX binding to a silica coating on an Au island film.60 These examples provide experimental basis for application of LSPR transducers in quantitative analysis of analyte binding. The linear dependence of the optical signal on surface coverage by the analyte allows quantification of kinetic data and calculation of binding constants, either to the metal or to a receptor layer, using experimental schemes similar to those applied in the well-developed propagating SPR technique.
3.2
LSPR Biosensing: Case Studies
Biosensing applications have been realized for almost all the configurations of LSPR transducers described above. Such applications require preparation of a biorecognition interface exposing specific receptors to target analytes. In general, immobilization of receptors follows established
14
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS 0.06 49 min
0.18
49 min
225 s
0.05
Absorbance
Absorbance difference
45 s
0.16
15 s 0s
0.14
0.12
1 225 s
0.04 O
O
O
O
0.03
45 s S−S
0.02 15 s
0.01 0.1 400 500 600 Wavelength (nm)
300 (a)
Absorbance at 365 nm
Normalized signal
0.8 0.6 0.4 0.2 0 0
10
20
30
40
300
(b)
1
(c)
0
700
Adsorption time (min)
(d)
700
0.016 0.012 0.008 0.004 0
50
400 500 600 Wavelength (nm)
0
0.02
0.04
0.06
Plasmon intensity change
Figure 11. Transmission UV–vis spectroscopy for the formation of a monolayer of 1 (inset in b) on an Au island substrate (2.5 nm nominal thickness, evaporated on quartz, unannealed), adsorbed from 2 mM solution of 1 in chloroform. (a) Absolute spectra. (b) Difference spectra (obtained by subtraction of the 0-s spectrum from the other spectra in a); dashed line corresponds to the spectrum of a thick layer of 1, obtained by evaporation of a drop on quartz (original spectrum divided by 6). (c) Normalized quantities of the plasmon absorbance versus adsorption time: squares—shift of the wavelength of maximum absorbance (from a); circles—intensity of the maximum difference peak (plasmon intensity change, PIC) (from b). (d) Correlation between the PIC and the maximum absorbance of 1 at 365 nm (both from b). All UV–vis spectra were taken ex situ, after rinsing the sample with chloroform and ethanol and drying under a nitrogen stream. [Reprinted with permission from Kalyuzhny et al.15 Copyright 2001 American Chemical Society.]
schemes developed for the preparation of biorecognition interfaces (see e.g., a recent review81 ). Biotin–avidin interactions have been exploited extensively both for immobilization of specific receptors and as a model system for protein recognition. A popular scheme includes preparation of a carboxylate- or amine-terminated SAM on the surface of metal NPs or islands, followed by covalent coupling of biotin through an amide bond.13,35,72 An example of this type of experimental scheme is shown in Figure 12, where both the peak intensity and wavelength are sensitive to binding of avidin (see also data in Table 2). Immobilization of biological receptors on oxide surfaces can be carried out by converting an
amine-terminated siloxane SAM to a carboxylateterminated surface followed by coupling of a biomolecular receptor through an amide bond. This scheme and corresponding transmission spectra are shown in Figure 13 for the preparation of an immunoglobulin (IgG) protein recognition interface57 on silica-stabilized Au islands on glass.60 Specific recognition was demonstrated in experiments where bound mouse-IgG and rabbitIgG antigens were exposed to the corresponding specific and nonspecific IgG antibodies, showing highly specific protein recognition (Figure 14). In LSPR systems analyte binding can be detected by measurements in solution (in situ) as shown in Figure 12(b), or in air or nitrogen
LSPR SPECTROSCOPY IN BIOSENSING
15
Fabrication protocol 3
2
1
Glass
Silane SAM
AuCM – MPA
AuCM
Biotin - functionalized AuCM – MPA O
O
(1) (CH3CH2O)3Si
NH2 (2) HS
OH
(3) H2N
O
O
O
HN NH H H S
H N
(a)
O
Detection protocol Absorbance
0.20
Light
After protein binding Befor protein binding
0.15 0.10 0.05 0.00
Biomolecular binding event on sensor chip detected by absorbance change
(b)
300 400 500 600 700 800 900 Wavelength (nm)
Figure 12. (a) Schematic presentation of the steps involved in the fabrication of an immobilized colloidal Au sensor chip on glass. The glass substrate was functionalized with 3-aminopropyl trimethoxysilane (1) to provide an amine-terminated surface for binding of a monolayer of citrate-stabilized Au NPs (AuCM ). A SAM of mercaptopropionic acid (MPA) (2) on the Au NPs provides reactive carboxylic groups that can be further modified by biotin (3) to study specific binding of streptavidin. (b) Protein binding to the biotinylated transducer results in increase and red shift of the SP band. [Reprinted with permission from Nath and Chilkoti.72 Copyright 2002, American Chemical Society.] Au Glass
0.45
Binding of Anti-rabbit IgG
0.40
Immobilization of Rabbit IgG
0.35 OMe Si OMe
SH OH HO OH HO OH HO OH Si OH Si O Si OH Si OH
S
S
S
0.30 Extinction
MeO
MPTS
Anti-rabbit IgG
Functionalization of interface
0.25 0.20 0.15
S
0.10 0.05
Na2SiO3
Rabbit IgG 400
S
S
S
700
800
900 1000 O C
HO OH HO OH HO OH HO OH Si OH Si OH Si OH Si OH S
600
Wavelength (nm)
~ ~1.5-nm silica HO OH Si OH
500
O OH OH O OHO OH O OH C C C C
Modification of silica surface
S
Figure 13. Schematic chart showing the preparation of a protein recognition interface using a T-LSPR transducer based on a silica-encapsulated, 5.0-nm Au island film evaporated on silanized glass and annealed, followed by specific protein recognition. Inset: actual transmission UV–vis measurements, carried out ex situ. Conditions of silica coating as in Ref. 60. [Adapted from Bendikov et al.57 ]
16
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Table 2. Selected biorecognition systems based on LSPR detection
Configuration of LSPR sensor
Preparation of recognition interface
Experimental details
Au NPs (40 nm) in solution68
Au NPs covered with monoclonal anti-hFABP. Detection of binding of hFABP
Incubation of Au NP bioconjugate with analyte
Au NPs immobilized on silanized glass, NPs enlarged by electroless deposition33 Au NPs (13–50 nm) immobilized on silanized glass12,72
Au NPs functionalized with HSA or anti-HSA. Detection of binding of anti-HSA or HSA, respectively Au NPs functionalized with MPA followed by chemical coupling of biotin. Detection of streptavidin and anti-biotin monoclonal antibodies Au NPs functionalized with disulfide polymer carrying receptor (glucose, pepstatin fragment, phenylboronic acid). Binding of ConA, HIV-1 protease, glycoprotein Au NPs functionalized with BSA or HSA
Difference in extinction after binding of analyte in stagnant solution
Au NPs (40 nm) immobilized on silanized glass82–84
Au NPs (40 nm) immobilized on silanized glass85
Detection of binding of anti-BSA (pig) or anti-HSA (goat, vector) Au NP (9 nm) film Au NPs functionalized with immobilized on fiber cystamine followed by coupling after removal of normal of biotin or immobilization of cladding86 anti-SEB Au NPs immobilized on Binding of streptavidin to unclad silanized silica immobilized biotin fiber75 Au NPs immobilized on Binding of streptavidin to the end face of a fiber87 immobilized biotin Biotin–NA Continuous 20-nm-thick Ganglioside (GM1 ) Au film with 110-nm holes18 glycolipid—CT hybridization of oligo-DNA (15 mer)
Evaporated Au island film, 2.5–5.0 nm13,57 Au NPs (40 nm) immobilized on glass88
Binding of streptavidin to immobilized biotin; specific recognition of IgG proteins Streptavidin binding to biotinylated BSA immobilized on Au NPs.
Au NPs (40 nm) immobilized on glass89
SZ-BSA immobilized on Au NPs followed by exposure to anti-SZ
Ag film evaporated through an NSL mask on mica35,50
Ag islands functionalized with mixed MUA-C8 SAM followed by coupling of biotin. Anti-biotin recognized by surface-bound biotin
Sensitivity Ca. 20 ng ml−1 hFABP. Determination of affinity constants in the range 109 –1011 mol−1 Minimum detected concentration 100 µg ml−1 of HSA
Kinetics of extinction change during analyte binding in a stirred solution
Detection limit of streptavidin ca. 20 nM for 13-nm NPs and 1 nM for 39-nm NPs
Kinetics of extinction change during analyte binding and release
Detection limit of ConA, 1.9 nM; HIV-1, protease, 50 nM; glycoprotein (OVA), 100 nM
Measurements of transmission spectra in solution
Detection limit ca. 30 nM for anti-HSA
Laser intensity attenuation, measured using lock-in technique
Detection limit (S/N ratio of 3) 1 nM for streptavidin and 1.4 pM for anti-SEB
Binding of analytes in a flow cell
Detection limit (S/N ratio of 3) of streptavidin 0.1 nM; raw data not shown Optical response measured at 20 µg ml−1 Kinetics of analyte binding followed at a constant wavelength with collection of spectra. Concentration of analytes: 0.3 µM NA; 0.5 µM CT; 0.2 µM DNA Minimum detected concentration 10 nM for antirabbit IgG binding 5.0 meV shift of SP band upon exposure to 2 µM streptavidin in solution
Binding of analyte in stagnant solution Binding of analyte in stagnant solution. Measurements performed in situ
Binding of analyte in stagnant solution. Measurements performed ex situ in air Light scattering from single epiluminescent Au NPs immersed in stagnant solution. Light scattering from epiluminescent Au NPs immersed in stagnant solution Binding of analyte in liquid-displacement cell. Measurements in N2 and in PBS solution. Spectra smoothed due to mica interference
Detection limit 20 nM for Sz
Detection limit 1 pM for streptavidin
LSPR SPECTROSCOPY IN BIOSENSING
17
Table 2. (continued)
Configuration of LSPR sensor
Preparation of recognition interface
Experimental details
Ag islands functionalize with Ag film evaporated mannose-terminated SAM. through a NSL mask on ConA served as analyte mica90 Ag triangle NPs (25 nm Ag islands functionalized with thick) evaporated mixed MUA-C8 SAM followed through an NSL mask by coupling with anti-ADDL. onto mica91 ADDL binding measured directly and in sandwich configuration after binding of second anti-ADDL antibody
Sensitivity
Kinetics of binding and release Specific binding of ConA in solution (9 µM) in the presence of interfering BSA (15 µM) Binding of analyte in stagnant Detection limit for ADDL solution for 30 min. Spectra binding 10 pM (directly) and measured in N2 after drying 100 fM (sandwich configuration)
MPA: mercaptopropionic acid; HSA: human serum albumin; OVA: ovalbumin; BSA: bovine serum albumin; SEB: staphylococcal enterotoxin B; NA: neutroavidin; CT: cholera toxin; IgG: immunoglobulin G; ADDL: amyloid-β-derived diffusible ligand.
(ex situ) after drying13,35,55 as shown in Figures 13 and 14. The change in refractive index upon analyte binding is substantially higher in ex situ measurements, as the reference signal before binding is taken in air (nair = 1) compared to aqueous solution (nsol ≈ 1.33). Ex situ measurements are also well suited to disposable kits. On the other hand, affinity measurements and quantitative analysis of analyte adsorption and desorption kinetics can only be carried out in situ.
LSPR sensing covers a wide range of systems, from simple chips capable of detecting a target analyte, to sophisticated instrumental schemes allowing observation of analyte binding to single nanostructures. A representative selection of LSPR biorecognition systems is summarized in Table 2. Analysis of the data in Table 2 shows that detection limits reported by different groups vary by 3 orders-of-magnitude for similar model systems. The highest reported sensitivity was obtained using
Rabbit–anti-rabbit 0.4
Anti-rab Rabbit Extinction
Extinction
0.4
Rabbit–anti-mouse
0.3 0.2 0.1
Anti-mouse Rabbit
0.3 0.2 0.1
400
600
800
1000
400
Wavelength (nm)
0.4
Anti-mouse Mouse
0.3
Extinction
Extinction
800
1000
Mouse–anti-rabbit
Mouse–anti-mouse 0.4
600
Wavelength (nm)
0.2
Anti-rab Mouse
0.3 0.2 0.1
0.1 400
600 800 Wavelength (nm)
1000
400
600 800 Wavelength (nm)
1000
Figure 14. Specific recognition of IgG antibodies using an array configuration of T-LSPR sensors with bound IgG antigens. Preparation of the biorecognition interfaces and experimental conditions as in Figure 13. [Adapted from Bendikov et al.57 ]
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
an LSPR transducer based on multiple attenuated total reflection from an unclad optical fiber.86 This result corroborates with the high refractive index sensitivity of this special type of LSPR transducer (see Table 1, Ref. 75). Results showing extremely high sensitivity with LSPR transducers should be treated with caution, as the valid test for the maximum sensitivity is the bulk refractive index sensitivity, which does not vary dramatically (see Table 1). Let us consider as an example the detection limit of 100 pg ml−1 , reported for specific recognition of immunoglobulin proteins in a reflection LSPR sensing scheme (Figure 5).92 The experimental protocol consisted of placing a 100-nl drop with analyte on a ca. 1-mm2 spot of LSPR transducer with the immobilized receptor for 30 min, followed by washing to remove nonspecifically adsorbed molecules and measurement of reflectance changes. If we assume that all the analyte molecules in the drop were adsorbed and there was no desorption during washing before the optical measurements, one calculates that 10−14 g protein was adsorbed on the transducer spot. Taking the density of the protein as 1.35 g cm−3 ,93 gives a mean thickness of the analyte film of ca. 10−5 nm. Calculation of the expected response using equation (3) and data from Table 1 for refractive index sensitivity shows that detection of such a minute amount is below the instrument resolution. Other reports of extraordinary LSPR biosensitivity may be questioned for similar reasons. Table 2 shows a variety of schemes for the preparation of biorecognition interfaces and for analyte binding that have been successfully adapted to various LSPR sensing platforms. To date only one system, i.e., bioconjugated Au NPs in solution,68,94,95 has been tested with sufficient statistics (hundreds of samples) using standard protocols and equipment applied in quantitative bioanalysis. High-throughput screening (144 samples/h) was applied to the study of affinity constants for a variety of monoclonal antigen–antibody pairs95 and the results were validated by comparison with common SPR measurements. Results obtained with other LSPR biosensor systems at the current stage of development should be considered as proof-of-concept demonstrations. Intriguing results were reported with sandwich structures comprising avidin-functionalized Au NPs conjugated to a biotin-functionalized lipid
bilayer formed on a planar quartz surface.96 Avidin receptors on the immobilized NPs allow further binding of biotinylated biomolecules. The kinetics of hybridization of single-stranded 15-mer DNA with the immobilized complementary strand were measured with close to 1:1 hybridization efficiency. Of particular interest is the use of biorecognition systems comprising a small number of or even single NPs as the optical transducer. Monitoring the SP of an individual NP can push to the lower Streptavidin (analyte)
Biotin (acceptor) Biotinylated BSA
AU
AU
(a) Scattering intensity (arb units)
18
(b)
1.00 0.95 0.90 2.22 2.24 2.26 2.28 2.30 2.32 2.34 Photon energy (eV) Objective lens
(c)
Figure 15. Schematic representation of a biosensor based on light scattering from single Au NPs. (a) Au NPs are functionalized with biotinylated BSA which subsequently binds streptavidin. (b) Mie-type calculations for the three situations shown in (a). (c) Left: true-color photograph of a sample comprising functionalized Au NPs, obtained using dark-field illumination. Right: experimental setup enabling dark-field microscopy of single Au NPs immersed in a liquid. [Reprinted with permission from Raschke et al.88 Copyright 2003 American Chemical Society.]
Plasmon shift (meV)
LSPR SPECTROSCOPY IN BIOSENSING
0
−2
C = 1 × 10−6 mol l−1 −4
−6
C = 2 × 10−6 mol l−1 −10
0
10 20 30 40 Incubation time (min)
50
60
Figure 16. LSPR single-NP experiments: shift of the SP band versus incubation time for different streptavidin concentrations and control experiments. Upon addition of streptavidin at time t = 0 the NP resonance starts to red shift (green triangles and orange circles), while addition of potassium phosphate buffer leaves the resonance position unchanged (red squares). The streptavidin concentration was 1 × 10−6 and 2 × 10−6 mol l−1 for triangles and circles, respectively. Addition of 1 × 10−6 mol l−1 streptavidin to an NP coated with nonbiotinylated proteins shows no evidence of unspecific binding (blue asterisks). [Reprinted with permission from Raschke et al.88 Copyright 2003 American Chemical Society.]
limit the number of detected analyte molecules. Several groups have explored LSPR sensing using individual metal NPs.88,97–99 An example of a biorecognition experiment using single NPs as LSPR transducers is shown in Figure 15.88 Au NPs (40 nm diameter) were functionalized with biotinylated bovine serum albumin (BSA) and the sample was exposed to a solution of streptavidin. The kinetics of binding were measurable using the spectral shift of the SP maximum, while no change was observed in control experiments (Figure 16). This demonstration indicates that, in principle, LSPR spectroscopy can be performed on the single-NP level.
4 CONCLUSIONS
The current status of LSPR biosensing based on noble metal nanostructures was examined. Several classes of optical transducers supporting LSPR in the visible spectral region were discussed as potential biosensors. The general sensing scheme includes immobilization of a
19
biorecognition interface (specific receptors) on the LSPR transducer and measurement of changes in the SP band (intensity or wavelength) accompanying specific analyte binding. The reported results indicate that quantitative comparison of the sensitivity of different LSPR systems requires further experimental verification. Reduction of LSPR biological transducers to practice needs additional engineering and optimization; however, the relatively low cost of most LSPR transducers and corresponding experimental setups, and the relative simplicity of the measurements, point to possible applications such as disposable chips. The possibility of measuring single nanostructures may enable significant miniaturization.
ACKNOWLEDGMENTS
The authors gratefully acknowledge financial support from the Israel Science Foundation, Grant No. 307/03, and EU Project No. 028331 (PROSURF). REFERENCES 1. J. Homola, Present and future of surface plasmon resonance biosensors. Analytical and Bioanalytical Chemistry, 2003, 377, 528–539. 2. M. Moskovits, Surface-enhanced Raman spectroscopy: a brief retrospective. Journal of Raman Spectroscopy, 2005, 36, 485–496. 3. E. Hutter and J. H. Fendler, Exploitation of localized surface plasmon resonance. Advanced Materials, 2004, 16, 1685–1706. 4. P. B. Johnson and R. W. Christy, Optical constants of the noble metals. Physical Review B, 1972, 6, 4370–4379. 5. K. C. Grabar, R. G. Freeman, M. B. Hommer, and M. J. Natan, Preparation and characterization of Au colloid monolayers. Analytical Chemistry, 1995, 67, 735–743. 6. I. Doron-Mor, Z. Barkay, N. Filip-Granit, A. Vaskevich, and I. Rubinstein, Ultrathin gold island films on silanized glass. Morphology and optical properties. Chemistry of Materials, 2004, 16, 3476–3483. 7. H. X. Xu and M. Kall, Modeling the optical response of nanoparticle-based surface plasmon resonance sensors. Sensors and Actuators B: Chemical, 2002, 87, 244–249. 8. N. G. Khlebtsov, V. A. Bogatyrev, B. N. Khlebtsov, L. A. Dykman, and P. Englebienne, A multilayer model for gold nanoparticle bioconjugates: application to study of gelatin and human IgG adsorption using extinction and light scattering spectra and the dynamic light scattering method. Colloid Journal, 2003, 65, 622–635. 9. N. G. Khlebtsov, V. A. Bogatyrev, L. A. Dykman, and A. G. Melnikov, Spectral extinction of colloidal gold
20
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS and its biospecific conjugates. Journal of Colloid and Interface Science, 1996, 180, 436–445. P. Englebienne, A. Van Hoonacker, and M. Verhas, Surface plasmon resonance: principles, methods and applications in biomedical sciences. Spectroscopy: An International Journal, 2003, 17, 255–273. D. Eck, C. A. Helm, N. J. Wagner, and K. A. Vaynberg, Plasmon resonance measurements of the adsorption and adsorption kinetics of a biopolymer onto gold nanocolloids. Langmuir, 2001, 17, 957–960. N. Nath and A. Chilkoti, Label-free biosensing by surface plasmon resonance of nanoparticles on glass: optimization of nanoparticle size. Analytical Chemistry, 2004, 76, 5370–5378. M. Lahav, A. Vaskevich, and I. Rubinstein, Biological sensing using transmission surface plasmon resonance spectroscopy. Langmuir, 2004, 20, 7365–7367. G. Kalyuzhny, A. Vaskevich, G. Ashkenasy, A. Shanzer, and I. Rubinstein, UV/vis spectroscopy of metalloporphyrin and metallophthalocyanine monolayers selfassembled on ultrathin gold films. The Journal of Physical Chemistry B, 2000, 104, 8238–8244. G. Kalyuzhny, M. A. Schneeweiss, A. Shanzer, A. Vaskevich, and I. Rubinstein, Differential plasmon spectroscopy as a tool for monitoring molecular binding to ultrathin gold films. Journal of the American Chemical Society, 2001, 123, 3177–3178. I. Doron-Mor, H. Cohen, Z. Barkay, A. Shanzer, A. Vaskevich, and I. Rubinstein, Sensitivity of transmission surface plasmon resonance (T-SPR) spectroscopy: self-assembled multilayers on evaporated gold island films. Chemistry: A European Journal, 2005, 11, 5555–5562. A. J. Haes, C. L. Haynes, A. D. McFarland, G. C. Schatz, R. P. Van Duyne, and S. L. Zou, Plasmonic materials for surface-enhanced sensing and spectroscopy. MRS Bulletin, 2005, 30, 368–375. A. Dahlin, M. Zach, T. Rindzevicius, M. Kall, D. S. Sutherland, and F. Hook, Localized surface plasmon resonance sensing of lipid-membrane-mediated biorecognition events. Journal of the American Chemical Society, 2005, 127, 5043–5048. G. Bauer, F. Pittner, and T. Schalkhammer, Metal nano-cluster biosensors. Mikrochimica Acta, 1999, 131, 107–114. M. Himmelhaus and H. Takei, Cap-shaped gold nanoparticles for an optical biosensor. Sensors and Actuators B: Chemical, 2000, 63, 24–30. T. Endo, K. Kerman, N. Nagatani, Y. Takamura, and E. Tamiya, Label-free detection of peptide nucleic acid-DNA hybridization using localized surface plasmon resonance based optical biosensor. Analytical Chemistry, 2005, 77, 6976–6984. E. Hutter and M. P. Pileni, Detection of DNA hybridization by gold nanoparticle enhanced transmission surface plasmon resonance spectroscopy. The Journal of Physical Chemistry B, 2003, 107, 6497–6499. T. Schalkhammer, C. Lobmaier, F. Pittner, A. Leitner, H. Brunner, and F. R. Aussenegg, The use of metalisland-coated pH-sensitive swelling polymers for biosensor applications. Sensors and Actuators B: Chemical, 1995, 24, 166–172.
24. N. L. Rosi and C. A. Mirkin, Nanostructures in biodiagnostics. Chemical Reviews, 2005, 105, 1547–1562. 25. M. C. Daniel and D. Astruc, Gold nanoparticles: assembly, supramolecular chemistry, quantum-size-related properties, and applications toward biology, catalysis, and nanotechnology. Chemical Reviews, 2004, 104, 293–346. 26. W. D. Geoghegan, Immunoassays at the microscopic level–solid-phase colloidal gold methods. Journal of Clinical Immunoassay, 1988, 11, 11–23. 27. V. A. Bogatyrev, L. A. Dykman, B. N. Khlebtsov, V. K. Plotnikov, and N. G. Khlebtsov, Optical properties of colloidal gold-oligothymidine conjugates and their variations on hybridization with polyadenylic acid. Colloid Journal, 2005, 67, 413–421. 28. G. Chumanov, K. Sokolov, B. W. Gregory, and T. M. Cotton, Colloidal metal-films as a substrate for surfaceenhanced spectroscopy. The Journal of Physical Chemistry, 1995, 99, 9466–9471. 29. A. Doron, E. Katz, and I. Willner, Organization of au colloids as monolayer films onto ito glass surfaces–application of the metal colloid films as base interfaces to construct redox-active monolayers. Langmuir, 1995, 11, 1313–1317. 30. J. Schmitt, P. Machtle, D. Eck, H. Mohwald, and C. A. Helm, Preparation and optical properties of colloidal gold monolayers. Langmuir, 1999, 15, 3256–3266. 31. P. Hanarp, D. S. Sutherland, J. Gold, and B. Kasemo, Control of nanoparticle film structure for colloidal lithography. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2003, 214, 23–36. 32. Y. D. Jin, X. F. Kang, Y. H. Song, B. L. Zhang, G. J. Cheng, and S. J. Dong, Controlled nucleation and growth of surface-confined gold nanoparticles on a (3-aminopropyl)trimethoxysilane-modified class slide: a strategy for SPR substrates. Analytical Chemistry, 2001, 73, 2843–2849. 33. F. Frederix, J. M. Friedt, K. H. Choi, W. Laureyn, A. Campitelli, D. Mondelaers, G. Maes, and G. Borghs, Biosensing based on light absorption of nanoscaled gold and silver particles. Analytical Chemistry, 2003, 75, 6894–6900. 34. M. Gluodenis, C. Manley, and C. A. Foss, In situ monitoring of the change in extinction of stabilized nanoscopic gold particles in contact with aqueous phenol solutions. Analytical Chemistry, 1999, 71, 4554–4558. 35. J. C. Riboh, A. J. Haes, A. D. McFarland, C. R. Yonzon, and R. P. Van Duyne, A nanoscale optical biosensor: real-time immunoassay in physiological buffer enabled by improved nanoparticle adhesion. The Journal of Physical Chemistry B, 2003, 107, 1772–1780. 36. N. Hashimoto, T. Hashimoto, T. Teranishi, H. Nasu, and K. Kamiya, Cycle performance of sol-gel optical sensor based on localized surface plasmon resonance of silver particles. Sensors and Actuators B: Chemical, 2006, 113, 382–388. 37. A. G. Brolo, R. Gordon, B. Leathem, and K. L. Kavanagh, Surface plasmon sensor based on the enhanced light transmission through arrays of nanoholes in gold films. Langmuir, 2004, 20, 4813–4815. 38. P. A. Mosier-Boss and S. H. Lieberman, Comparison of three methods to improve adherence of thin gold films to
LSPR SPECTROSCOPY IN BIOSENSING
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
glass substrates and their effect on the SERS response. Applied Spectroscopy, 1999, 53, 862–873. C. A. Goss, D. H. Charych, and M. Majda, Application of (3-Mercaptopropyl)trimethoxysilane as a molecular adhesive in the fabrication of vapor-deposited gold electrodes on glass substrates. Analytical Chemistry, 1991, 63, 85–88. M. Wanunu, A. Vaskevich, and I. Rubinstein, Widelyapplicable gold substrate for the study of ultrathin overlayers. Journal of the American Chemical Society, 2004, 126, 5569–5576. M. Wanunu, A. Vaskevich, A. Shanzer, and I. Rubinstein, Divergent growth of coordination dendrimers on surfaces. Journal of the American Chemical Society, 2006, 128, 8341–8349. G. Kalyuzhny, A. Vaskevich, M. A. Schneeweiss, and I. Rubinstein, Transmission surface-plasmon resonance (T-SPR) measurements for monitoring adsorption on ultrathin gold island films. Chemistry: A European Journal, 2002, 8, 3850–3857. T. Karakouz, A. Tesler, M. Idelson, R. PopovitzBiro, A. Vaskevich, and I. Rubinstein, Refractive index sensitivity of the localized plasmon absorbance in gold island films. In preparation. V. L. Schlegel and T. M. Cotton, Silver-island films as substrates for enhanced Raman-scattering–effect of deposition rate on intensity. Analytical Chemistry, 1991, 63, 241–247. S. Norrman, T. Andersson, C. G. Granqvist, and O. Hunderi, Optical-properties of discontinuous gold-films. Physical Review B, 1978, 18, 674–695. J. C. Hulteen and R. P. Van Duyne, Nanosphere lithography–a materials general fabrication process for periodic particle array surfaces. Journal of Vacuum Science and Technology A: Vacuum, Surfaces, and Films, 1995, 13, 1553–1558. H. Takei, Surface-adsorbed polystyrene spheres as a template for nanosized metal particle formation: optical properties of nanosized Au particle. Journal of Vacuum Science and Technology B, 1999, 17, 1906–1911. F. A. Denis, P. Hanarp, D. S. Sutherland, and Y. F. Dufrene, Nanoscale chemical patterns fabricated by using colloidal lithography and self-assembled monolayers. Langmuir, 2004, 20, 9335–9339. M. D. Malinsky, K. L. Kelly, G. C. Schatz, and R. P. Van Duyne, Chain length dependence and sensing capabilities of the localized surface plasmon resonance of silver nanoparticles chemically modified with alkanethiol selfassembled monolayers. Journal of the American Chemical Society, 2001, 123, 1471–1482. A. J. Haes and R. P. Van Duyne, A nanoscale optical biosensor: sensitivity and selectivity of an approach based on the localized surface plasmon resonance spectroscopy of triangular silver nanoparticles. Journal of the American Chemical Society, 2002, 124, 10596–10604. T. R. Jensen, M. D. Malinsky, C. L. Haynes, and R. P. Van Duyne, Nanosphere lithography: tunable localized surface plasmon resonance spectra of silver nanoparticles. The Journal of Physical Chemistry B, 2000, 104, 10549–10556. A. J. Haes and R. P. Van Duyne, A unified view of propagating and localized surface plasmon resonance
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
21
biosensors. Analytical and Bioanalytical Chemistry, 2004, 379, 920–930. F. R. Aussenegg, H. Brunner, A. Leitner, C. Lobmaier, T. Schalkhammer, and F. Pittner, The metal island coated swelling polymer over mirror system (MICSPOMS)–a new principle for measuring ionic-strength. Sensors and Actuators B: Chemical, 1995, 29, 204–209. J. Prikulis, P. Hanarp, L. Olofsson, D. Sutherland, and M. Kall, Optical spectroscopy of nanometric holes in thin gold films. Nano Letters, 2004, 4, 1003–1007. A. B. Dahlin, J. O. Tegenfeldt, and F. Hook, Improving the instrumental resolution of sensors based on localized surface plasmon resonance. Analytical Chemistry, 2006, 78, 4416–4423. V. P. Drachev, V. C. Nashine, M. D. Thoreson, D. BenAmotz, V. J. Davisson, and V. M. Shalaev, Adaptive silver films for detection of antibody-antigen binding. Langmuir, 2005, 21, 8368–8373. T. A. Bendikov, A. Rabinkov, T. Karakouz, A. Vaskevich, and I. Rubinstein, Specific Recognition of IgG Proteins Using Transmission Localized Surface Plasmon Resonance (T-LSPR) Spectroscopy. In preparation. S. E. Roark and K. L. Rowlen, Thin Ag films—influence of substrate and postdeposition treatment on morphology and optical-properties. Analytical Chemistry, 1994, 66, 261–270. L. M. Liz-Marzan, M. Giersig, and P. Mulvaney, Synthesis of nanosized gold-silica core-shell particles. Langmuir, 1996, 12, 4329–4335. I. Ruach-Nir, T. A. Bendikov, I. Doron-Mor, Z. Barkay, A. Vaskevich, and I. Rubinstein, Silica-stabilized gold island films for transmission localized surface plasmon sensing. Journal of the American Chemical Society, 2007, 129, 84–92. A. V. Whitney, J. W. Elam, S. Zou, A. V. Zinovev, P. C. Stair, G. C. Schatz, and R. P. Van Duyne, Localized surface plasmon resonance nanosensor: a high-resolution distance-dependence study using atomic layer deposition. The Journal of Physical Chemistry B, 2005, 109, 20522–20528. G. D. Sockalingum, A. Beljebbar, H. Morjani, J. F. Angiboust, and M. Manfait, Characterization of island films as surface-enhanced Raman spectroscopy substrates for detecting low antitumor drug concentrations at single cell level. Biospectroscopy, 1998, 4, S71–S78. M. McMahon, R. Lopez, H. M. Meyer, L. C. Feldman, and R. F. Haglund, Rapid tarnishing of silver nanoparticles in ambient laboratory air. Applied Physics B: Lasers and Optics, 2005, 80, 915–921. T. Okamoto and I. Yamaguchi, Field enhancement by a metallic sphere on dielectric substrates. Optical Review, 1999, 6, 211–214. L. S. Jung, C. T. Campbell, T. M. Chinowsky, M. N. Mar, and S. S. Yee, Quantitative interpretation of the response of surface plasmon resonance sensors to adsorbed films. Langmuir, 1998, 14, 5636–5648. M. M. Miller and A. A. Lazarides, Sensitivity of metal nanoparticle surface plasmon resonance to the dielectric environment. The Journal of Physical Chemistry B, 2005, 109, 21556–21565. T. Rindzevicius, Y. Alaverdyan, A. Dahlin, F. Hook, D. S. Sutherland, and M. Kall, Plasmonic sensing characteristics
22
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS of single nanometric holes. Nano Letters, 2005, 5, 2335–2339. P. Englebienne, Use of colloidal gold surface plasmon resonance peak shift to infer affinity constants from the interactions between protein antigens and antibodies specific for single or multiple epitopes. The Analyst, 1998, 123, 1599–1603. P. Englebienne, A. Van Hoonacker, M. Verhas, and N. G. Khlebtsov, Advances in high-throughput screening: biomolecular interaction monitoring in real-time with colloidal metal nanoparticles. Combinatorial Chemistry and High Throughput Screening, 2003, 6, 777–787. S. Underwood and P. Mulvaney, Effect of the solution refractive-index on the color of gold colloids. Langmuir, 1994, 10, 3427–3430. A. C. Templeton, J. J. Pietron, R. W. Murray, and P. Mulvaney, Solvent refractive index and core charge influences on the surface plasmon absorbance of alkanethiolate monolayer-protected gold clusters. The Journal of Physical Chemistry B, 2000, 104, 564–570. N. Nath and A. Chilkoti, A colorimetric gold nanoparticle sensor to interrogate biomolecular interactions in real time on a surface. Analytical Chemistry, 2002, 74, 504–509. T. Okamoto, I. Yamaguchi, and T. Kobayashi, Local plasmon sensor with gold colloid monolayers deposited upon glass substrates. Optics Letters, 2000, 25, 372–374. Y. G. Sun and Y. N. Xia, Increased sensitivity of surface plasmon resonance of gold nanoshells compared to that of gold solid colloids in response to environmental changes. Analytical Chemistry, 2002, 74, 5297–5305. S. F. Cheng and L. K. Chau, Colloidal gold-modified optical fiber for chemical and biochemical sensing. Analytical Chemistry, 2003, 75, 16–21. F. Meriaudeau, T. Downey, A. Wig, A. Passian, M. Buncick, and T. L. Ferrell, Fiber optic sensor based on gold island plasmon resonance. Sensors and Actuators B: Chemical, 1999, 54, 106–117. F. Meriaudeau, A. Wig, A. Passian, J. P. Lauret, and T. L. Ferrell, Multiple gold island layers on a fiber core: a promising sensing device. Optical Engineering, 2001, 40, 658–660. A. J. Haes, S. L. Zou, G. C. Schatz, and R. P. Van Duyne, A nanoscale optical biosensor: the long range distance dependence of the localized surface plasmon resonance of noble metal nanoparticles. The Journal of Physical Chemistry B, 2004, 108, 109–116. N. Nath and A. Chilkoti, Label free colorimetric biosensing using nanoparticles. Journal of Fluorescence, 2004, 14, 377–389. A. J. Haes, S. L. Zou, G. C. Schatz, and R. P. Van Duyne, Nanoscale optical biosensor: short range distance dependence of the localized surface plasmon resonance of noble metal nanoparticles. The Journal of Physical Chemistry B, 2004, 108, 6961–6968. D. Chen, G. Wang, and J. Li, Interfacial bioelectrochemistry: fabrication, properties and applications of functional nanostructured biointerfaces. The Journal of Physical Chemistry C, 2007, 111, 2351–2367. S. Morokoshi, K. Ohhori, K. Mizukami, and H. Kitano, Sensing capabilities of colloidal gold modified with a selfassembled monolayer of a glucose-carrying polymer chain on a glass substrate. Langmuir, 2004, 20, 8897–8902.
83. H. Kitano, Y. Makino, H. Kawasaki, and Y. Sumi, Selfassembled monolayer of a pepstatin fragment as a sensing element for aspartyl proteases. Analytical Chemistry, 2005, 77, 1588–1595. 84. H. Kitano, Y. Anraku, and H. Shinohara, Sensing capabilities of colloidal gold monolayer modified with a phenylboronic acid-carrying polymer brush. Biomacromolecules, 2006, 7, 1065–1071. 85. K. Fujiwara, H. Watarai, H. Itoh, E. Nakahama, and N. Ogawa, Measurement of antibody binding to protein immobilized on gold nanoparticles by localized surface plasmon spectroscopy. Analytical and Bioanalytical Chemistry, 2006, 386, 639–644. 86. L. K. Chau, Y. F. Lin, S. F. Cheng, and T. J. Lin, Fiberoptic chemical and biochemical probes based on localized surface plasmon resonance. Sensors and Actuators B: Chemical, 2006, 113, 100–105. 87. K. Mitsui, Y. Handa, and K. Kajikawa, Optical fiber affinity biosensor based on localized surface plasmon resonance. Applied Physics Letters, 2004, 85, 4231–4233. 88. G. Raschke, S. Kowarik, T. Franzl, C. Sonnichsen, T. A. Klar, J. Feldmann, A. Nichtl, and K. Kurzinger, Biomolecular recognition based on single gold nanoparticle light scattering. Nano Letters, 2003, 3, 935–938. 89. M. P. Kreuzer, R. Quidant, G. Badenes, and M. P. Marco, Quantitative detection of doping substances by a localised surface plasmon sensor. Biosensors & Bioelectronics, 2006, 21, 1345–1349. 90. C. R. Yonzon, E. Jeoung, S. Zou, G. C. Schatz, M. Mrksich, and R. P. Duyne, A comparative analysis of localized and propagating surface plasmon resonance sensors: the binding of concanavalin A to a monosaccharide functionalized self-assembled monolayer. Journal of the American Chemical Society, 2004, 1326, 12669–12676. 91. A. J. Haes, L. Chang, W. L. Klein, and R. P. Van Duyne, Detection of a biomarker for Alzheimer’s disease from synthetic and clinical samples using a nanoscale optical biosensor. Journal of the American Chemical Society, 2005, 127, 2264–2271. 92. T. Endo, K. Kerman, N. Nagatani, H. M. Hiepa, D. K. Kim, Y. Yonezawa, K. Nakano, and E. Tamiya, Multiple label-free detection of antigen-antibody reaction using localized surface plasmon resonance-based coreshell structured nanoparticle layer nanochip. Analytical Chemistry, 2006, 78, 6465–6475. 93. H. Fischer, I. Polikarpov, and A. F. Craievich, Average protein density is a molecular-weight-dependent function. Protein Science, 2004, 13, 2825–2828. 94. P. Englebienne, A. Van Hoonacker, and J. Valsamis, Rapid homogeneous immunoassay for human ferritin in the Cobas Mira using colloidal gold as the reporter reagent. Clinical Chemistry, 2000, 46, 2000–2003. 95. P. Englebienne, A. Van Hoonacker, and M. Verhas, Highthroughput screening using the surface plasmon resonance effect of colloidal gold nanoparticles. The Analyst, 2001, 126, 1645–1651. 96. L. Olofsson, T. Rindzevicius, I. Pfeiffer, M. Kall, and F. Hook, Surface-based gold-nanoparticle sensor for specific and quantitative DNA hybridization detection. Langmuir, 2003, 19, 10414–10419.
LSPR SPECTROSCOPY IN BIOSENSING 97. J. J. Mock, D. R. Smith, and S. Schultz, Local refractive index dependence of plasmon resonance spectra from individual nanoparticles. Nano Letters, 2003, 3, 485–491. 98. A. D. McFarland and R. P. Van Duyne, Single silver nanoparticles as real-time optical sensors with zeptomole sensitivity. Nano Letters, 2003, 3, 1057–1062.
23
99. A. Curry, G. Nusz, A. Chilkoti, and A. Wax, Substrate effect on refractive index dependence of plasmon resonance for individual silver nanoparticles observed using darkfield micro-spectroscopy. Optics Express, 2005, 13, 2668–2677.
27 Picoscopes, New Label-Free Biosensors Petr I. Nikitin Natural Science Center of General Physics Institute, Russian Academy of Sciences, Moscow, Russia
1 INTRODUCTION
Optical biosensors and biochips are widely used in many areas of science, technology, and medicine.1 Label-free or “reagentless” optical methods for detecting bio- and chemical interactions have a number of advantages such as ability to monitor reactions in real time and good reliability of results obtained using fewer operations. Among such methods, one can mention those based on the surface plasmon resonance (SPR),2 whose sensitivity and/or signal-to-noise ratio can be substantially improved by using phase peculiarities of light reflected under SPR,3–7 the reflection interference spectroscopy,8 and a variety of waveguide methods (“resonance mirror”,9 grating coupler,10 planar Mach-Zehnder interferometer,11 etc.). The phase SPR methods can be used not only for enhancement of assay sensitivity,3,4 but also for label-free readout of microarrays or biochips. For example, methods such as phasepolarization contrast with spectral SPR,5 dark-field SPR microscopy,6 and SPR interferometry7 can be used in multichannel sensor systems designated for high-throughput screening or simultaneous detection of several analytes. Although the phase detection SPR methods for gas measurements3 may yield very high threshold sensitivity, for example, relative change of refractive index n/n ≈ 4 × 10−8 , the resolution of SRP and other refractometric biosensors in studies of liquids is restricted. This is due to the
strong dependence of the refractive index of liquids upon temperature (10−4 per 1 ◦ C). The thermal drift can be partially compensated by the differentiation of the biosensing schemes or by a reference channel in two- or multichannel detection schemes.7 The sensitivity to bulk refractive index also limits the dynamic range of the sensors and the type of reagents to be used. Typically, changing of buffers and reagents brings n ≈ 0.02, whereas biochemical reactions yield only n ≈ 10−5 –10−7 . Besides, it is difficult to provide simultaneously high sensitivity and wide dynamic range of the measurements with the use of compact and affordable refractometric sensors. All the above-mentioned label-free methods employ thin-film structures with precisely deposited metal films (gold or silver) or dielectric films. Such structures and related biochips are rather expensive to be single used. Regeneration of the biochips increases maintenance cost of the equipment, which, in practice, is not lower than that of many label-based biosensing instruments. For medical diagnostics, biochip regeneration is practically unacceptable because of contamination issues. Preferably, in order to compete in medical applications with nitrocellulose immunochromatographic strips, the biochips should be comparable with them in cost. The goal of our research was the development of a new type of label-free biosensors that would be free from the above-mentioned drawbacks of this class of instruments.
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
2 GLASS SLIP AS A BIOCHIP
R 0.06
For the first time, we proposed the use of simple microscopic glass slips without deposition of any metal or dielectric films as the labelfree biochips.12–15 A sufficiently thick (tens or hundreds of micrometers) slip is an acceptable Fabry–Perot interferometer provided that a small area on its surface is used for each biosensor channel. The developed technology employs interference between two beams reflected from the biochip as shown in Figure 1: the beam 5 reflected from the bottom surface of the glass slip 2 and the beam 6 reflected from the upper surface with the biological recognition layer 3. The result of the interference naturally depends on the phase thickness of the slip with the recognition layer. The calculated reflection spectrum of the glass plate with thickness d = 50 µm and refractive index n = 1.5 for radiation of a superluminescent light emitting diode (SLD) with the central wavelength λ = 850 nm and spectral width 30 nm is shown in Figure 2. The spectrum is a periodic function with maxima (minima) separated by spectral intervals that are equal to ν = c/2nd, where c is the speed of light. During a biological reaction, some components of the solution adhere to the surface of the biomolecular layer (or separate recognition spots), whereas others detach from it. This causes a change of the phase difference between the interfering waves 5 and 6 (Figure 1) and, accordingly, a change of the reflection spectrum (Figure 2). These changes are measured to judge about the binding reaction. As seen in Figure 2, the reflection from a transparent glass slip is small (only 6% at maxima and 2% at minima). So, to make an affordable disposable biochip
4
3 2 1
5
6
Figure 1. Scheme of interference pattern formation: 1: air; 2: glass slip; 3: recognition layer; 4: biological solution; 5 and 6: reflected beams.
0.04
0.02
0
830
850
870
l (nm)
Figure 2. Reflection spectrum of a glass slip of 50 µm thickness irradiated by SLD (λ = 850 nm, spectral width = 30 nm).
of a simple glass slip, we developed new low-noise methods and readout devices.
3 SPECTRAL CORRELATION METHODS
Label-free biosensors named Picoscope were developed for simultaneous real-time detection of several biological agents by measuring picometerrange changes of the thickness of different recognition spots or channels on the biochip surface. The sensors are based on spectral correlation (SC) methods, which use correlation signals between two coupled interferometers.13–15 A Picoscope optical scheme is shown in Figure 3. The first interferometer 4 is the above-mentioned glass slip (biochip) with recognition spots 8 or flow channels, or wells. The second one is a scanned interferometer 2 that employs periodical modulation of path difference of the interfering beams. For example, this could be a scanned Fabry–Perot interferometer, whose base (i.e., distance L between its two mirrors) is periodically changed by a piezoelectric driver. Other interferometers, for example, Michelson, Mach-Zahnder, including planar or fiber-optical interferometers can also be used. These interferometers are scanned by variation of the path difference of interfering beams (or arms) of the interferometers. As shown in Figure 3, the radiation from SLD 1 passes through the scanned interferometer 2, then through a semitransparent mirror 3 and is incident onto a glass plate 4. The radiation reflected from the plate is directed by the
PICOSCOPES, NEW LABEL-FREE BIOSENSORS I (ru) 6
8
7
3
4 5 6
4
2
3
9
2
1 Figure 3. Scheme of the Picoscope : 1: SLD; 2: scanned interferometer; 3: semitransparent mirror; 4: glass slip; 5: optics; 6: CCD camera; 7: fluidic system; 8: recognition spots or wells; 9: computer.
0
830
850
870
l (nm)
Figure 4. Spectra of radiation recorded by each pixel of the CCD camera for two bases of the Fabry–Perot interferometer L1 = 75 µm (low curve) and L2 = 75.212 µm (upper curve).
U (ru) 5 4
semitransparent mirror to photodetectors or a CCD camera 6. For the scanned Fabry-Perot interferometer, the initial base L is adjusted to optical thickness of the biochip L ≈ n · d. Then a scanned piezodriver periodically modulates L by a linear law with amplitude of a few micrometers. Figure 4 shows the calculated spectral distribution of the radiation incident on the photodetector or each pixel of the CCD camera for two bases L1 = 75 µm and L2 = 75.212 µm while using a 50-µm glass slip as the biochip. In the second case, the total light intensity incident to each photodetector integrated over the spectrum of the radiation source is much higher than that in the first case. The situation repeats with further increase of the base of the scanned interferometer by a half of wavelength, and so on. Precise calculations for scanning the base L between 72 and 78 µm give the resulting output signal U from each photodetector shown in Figure 5. The relative phase shift of this sinusoidal correlation dependence U (L) is used to calculate the change of thickness d of the biochip due to the biochemical reaction and sensogram generation. To realize the method, we developed hardware and software. Under the control of the software, the hardware generated different voltages
3 2 1 0
73
74
75
76
77 L (µm)
Figure 5. Dependence of correlation signal from each photodetector on the interferometer base.
applied to the piezodriver of the scanned interferometer. The resulting images from the CCD camera or signals from several photodiodes were processed by the software. In some software versions, the signals from the CCD pixels corresponding to each spot with the recognition layer were averaged over the spot area to minimize the electronic noise. In other versions, the software provided spatially resolved measurements within each spot. The area between the spots (or biological channels) was used for the reference measurements of the glass thickness changes due to, for example, temperature variations while changing the solutions to eliminate temperature drifts.
4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
4 TWO-CHANNEL BIOSENSORS
1.4 PBS
1.2
Second Ab
1 ∆d (nm)
Several different types of biosensing instruments were designed based on the SC methods. They were tested for various biochemical applications. One type of the devices is designated for twochannel detection of biochemical reactions in flow mode.14 It has one additional reference channel for elimination of temperature drifts. Three separate photodiodes are used to record intensity of the light reflected from each channel. This type of the device, named Affinoscope , is intended for testing affinity between various biological and chemical agents (Figure 6). It uses a microscopic glass slip of 100 µm thickness as the biochip. A special cuvette forms two flow channels with sensing areas 1 mm in diameter. By immobilization of antibodies (Ab) on the biochip surface using different interface chemistry, for example, via biotin–streptavidin bridge,12 various proteins,14 and food pathogen15 were detected by Affinoscope . An example of detection of Listeria monocytogenes by biotinylated first Ab is given in Figure 7. The antigen (Ag) was in the form of cell suspension at concentration of C = 107 cells/ml. The Ag layer thickness averaged over the sensor’s surface of d ∼ 80 pm was surely detectable. The sensor’s response increased significantly by d = 400 pm by binding of the second native Ab. Control experiments with different combinations of nonand specific Abs and Ags showed very low (∼1%) nonspecific binding (NSB) for the used interface chemistry and reagents. The biochips were regenerated by glycine buffer (GB) with pH 2.2.
PBS
Ag
PBS
First Ab
0.8 0.6
GB
PBS Streptavidin
0.4 0.2 0
0
1000 2000 3000 4000 5000 6000 7000
−0.2
t (s)
Figure 7. Sensogram of detection of Listeria monocytogenes at concentration 107 cell/ml.
The detailed investigations showed that in the flow mode the used Ab mainly captured smallscale Ag in the form of soluble protein and fragments of cell membrane rather than whole bacteria. To increase the sensor’s response, the cell culture was heat-treated and ultrasonically processed for cell lysis. Besides, the surface density of the Ab was varied by changing the density of the biotin and streptavidin molecules to optimize specific capturing of the whole cells and soluble proteins. As a result, the cell concentration as low as 104 cells/ml was surely detected by the second Ab by recording of d = 90 pm (Figure 8). This detection limit is 2 orders of magnitude better than that of the standard enzyme-linked immunosorbent assay (ELISA) with the same Abs. The observed detection limit for bacteria is much lower than that measured by the high-end “BIAcore 2000” SPR instrument.16 Employment of the second Ab permits one to confirm specificity 2.5 2
PBS
Ag PBS
∆d (nm)
1.5
Second PBS Ab
PBS First Ab
1 0.5 0
−0.5
Figure 6. Photo of two-channel Affinoscope .
0
1000
2000
3000
4000
5000
6000
7000
Streptavidin
t (s)
Figure 8. Sensogram of detection of Listeria monocytogenes at concentration 104 cell/ml by second antibody.
PICOSCOPES, NEW LABEL-FREE BIOSENSORS
5
of reactions and to achieve the detection limit for food pathogens, which is 3–2 orders of magnitude better than that reported for the SPR devices.16,17
5 MACRO- AND MICROARRAYS
Another attractive application of the technology is label-free readout of macro- and microarrays for high-throughput screening and for biomedical diagnostics. First of all, we tested less favorable case for readout of macroarrays with very large surface area, namely, for detection of parallel reactions in 96 separate wells in a standard ELISA plate format of 127 × 85 mm2 .13,14 A photo of the Picoscope prototype for this format is shown in Figure 9. A glass plate of 50 µm thickness was used as the bottom of the ELISA frame. A special design of the frame was tested to avoid deflection of the plate while pouring and removing the solutions and to prevent vaporization of the solutions to minimize temperature drifts. The large surface area of the ELISA plate was illuminated by a low-power SLD ( 0 S1′
S 2′ ∆CL2 < 0 S0
S1
S2
∆CL1 > 0 CL1 (P1)
Figure 2. Momentary and dynamic innate immune system status. Here CL1 and CL2 are imaginary parameters of respiratory burst based on CL kinetics information (set of CL parameters characteristic for the case). The points denoted as S1, S1′ , S0, S2, S2′ depict different momentary innate immune status; CL1 and CL2 are measures of its dynamic change in case of two different scenarios.
classified into three groups: physical, biological, and temporal. The physical parameters consist of cell numbers (phagocytes, erythrocytes), stimulant concentration (particle/cell ratio), volume-tosurface ratio, mixing (sample oxygenation and phagocytosis synchronization), pH of the buffer used, and temperature. These parameters allow for calibration and the investigator must keep them constant at some predetermined value to avoid a multiparametric interpretation. The other two groups form a “phagocytosis-inherent space”
(cell reserve of reagents participating in respiratory burst-related chain of reactions as well as the existing primed status of phagocytes) and cannot be controlled. Instead, we propose that the change in the phagocyte respiratory burst caused by some well-controlled shift of these parameters can be measured.50 Briefly, having in mind the limited capacity of the phagocyte to restore its ability to generate ROS (inherent irreversibility of phagocytes), it is not the same if the phagocyte respiratory burst follows the S1–S0–S2′ or S1′ –S0–S2 trajectory (both from an estimative and prognostic point of view). It would also be important to have blood samples under properly controlled assay conditions in order to correlate their experimentally or clinically defined functional states with those obtained by the component analysis of CL kinetics.
4.2
CL Probes
The most common lumigenic substances are luminol (5-amino-2, 3-dihydro-1,4-phthalazinedione) and lucigenin (bis-N -methylacridinium nitrate), with emission at 460 and 510 nm, respectively. There are three main characteristics of the luminolamplified CL reaction in neutrophils: (i) a peroxidase (usually MPO originating from azurophil granules) is required, (ii) oxygen metabolites produced by the NADPH oxidase are needed, and
6
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
(iii) both intracellular and extracellular reactions are measured.44 Lucigenin-dependent chemiluminescence (LDCL), in contrast to LCL, is independent of MPO and is postulated to be the consequence of superoxide production by an NADPH oxidase located in the plasma membrane of PMN.
4.3
Stimulants, Mechanisms
The stimulants can be divided into different groups depending upon their solubility and interaction with the membrane receptors. Most of the used soluble receptor stimulants (n-formylmethionyl-leucyl-phenylalanine [nFMLP], phorbol 12-myristate 13-acetate [PMA], conA, plateletactivating factor PAF, Ca-ionophore A23187) interact with the PMN membrane by a nonreceptor mechanism. Arachidonic acid (AA) can also stimulate without interaction with receptors. Zymosan and latex beads belong to the group of particle stimulants. When luminol is used as a CL probe, the PMN CL response is highly dependent on the type of stimulus employed. Contact with opsonified particles results in recognition via complement and/or immunoglobulin G (IgG) Fc receptors of the phagocyte, activation of redox metabolism, phagocytosis, specific and azurophilic degranulation, and formation of phagolysosomes. On the other hand, the liquid stimulant PMA causes specific degranulation with full activation of granulocyte redox metabolism, but little or no azurophilic degranulation, and no formation of phagolysosome.3 Depending on the mutual correlation between these processes, different phases of LCL kinetics are formed, which are clearly visible in some cases or may remain invisible, which makes the LCL test difficult for clinical use (Table 2).
4.4
Requirements for a Dedicated Luminescent Sensor of the Phagocyte Respiratory Burst
A typical chemiluminescent instrument71,72,64,9 consists of a sample compartment and a measuring section, which are functionally connected and operate under computer control. The correct interpretation of data, as well as data reproducibility, depends on many requirements that need to be satisfied by the instrument design. First, in order to achieve precise, repeatable, and comparable results, the measurement of biological samples must be performed at equivalent conditions regarding both the device and the samples. Such a measurement is better realized if the samples are tested simultaneously, in a multichannel mode.9 Of course, data is also acquired “simultaneously” just in case the maximal time shift inside the cycle of measurement is sufficiently lower than the time interval for a significant chemiluminescent intensity change. Second, high-sensitivity measurements allow for routine work at lower reagent concentrations. Since common CL measurements occur in the visible portion of the spectrum, the detector of choice is usually a photomultiplier tube (PMT). The most important characteristic of the PMT is the quantum efficiency of the photocathode (the ratio between the number of electrons emitted by the photocathode to the number of incident photons). However, the photomultiplier noise (dark current) is the ultimate limiting factor in determining sensitivity. Because photon-counting offers 100 times better signal-to-noise ratio in comparison to DC registration and thus higher sensitivity, it is probably the most widely used form of detection in CL. Third, as in all cell activities, the emitted CL is strongly temperature dependent,3 requiring the samples to have precise temperature control.
Table 2. Multiphase CL response of CL probes and its correlation to extracellular and intracellular ROS production
fMLP Lum-NADPH and MPO-dep (H2 O2 ; O2 − ) Luc-NADPH-dep (O2 − )
2 peaks first extracellular second intracellular 1 peak extracellular
PMA
Zymosan
1 peak extracellular + intracellular(predominant part)
1 peak extracellular + intracellular
1 peak 50% extracellular + ?
1 peak ?% extracellular
PHAGOCYTE LUMINESCENT SENSOR
4.5
Sensor Design, Block Diagram
The chemiluminescent system allows us to measure intracellular as well as extracellular reactions,11 and when using a previously described component analysis,48 it is possible to simultaneously assess the contribution of both the extracellular and the intracellular CL emitted from the same cellular system. On the other hand, optical-fiberbased biosensors have demonstrated their ability to detect biological entities with high sensitivity due to the intimacy between the specific biological interactions and the fiber core coupling with minimal signal losses.5,12,51,52,41,39 Moreover, it has also been shown that a silica surface stimulates circulating blood phagocytes, which then produce a CL pattern similar to the first extracellular phase of the well-known nFMLP pattern.69 By applying the drop of blood on the end face of an optical fiber, an increased surface-to-volume ratio is obtained, improving phagocytosis and light-capturing conditions. The instrument design described here allows for the assessment of both the extracellular and intracellular parts of the CL response simultaneously. This fiber-based chemiluminescent sensor will provide timely and clinically relevant diagnostic and management information for patients undergoing an infection. A new proprietary fiber-based luminometer dedicated to phagocyte activity assessment has been evaluated as a putative tool for rapid, sensitive, reproducible, and inexpensive
7
measurement of the in vivo inflammation state of circulating phagocytes, and the evaluation of the patient status during infection and is described here.
4.6
Detectors, Signal Capturing and Transduction: a Dual Role of Optical Fibers
A luminometer was designed with the following basic characteristics: (i) computerized control of photodetection, (ii) photon-counting mode measurement of a six-fiber sample module, (iii) simultaneous transfer of the measured data to a serial port (allowing for data acquisition by an external computer), (iv) direct data recording into the computer memory while placing the graphs in parallel on the computer screen, and (v) printing of collected data. To realize all of the aforementioned characteristics, a specific driver was developed using LabWindows. A block diagram of the six-channel luminometer is shown in Figure 3. It consists of a thermostated fiber holder module (1) containing six fibers (2); a photon-counting PMT detector (3); a DC power supply (4); a stepper driver (5); a programmable logic controller (PLC) (6); a personal computer (7); a step motor (8); a position sensor (9); a thermocontroller (10); and a rotating diskshutter (11). The rotating disk-shutter contains a 2 4
1
10
4
8
5
6
9
3
12 11
7 4
(a)
(b)
Figure 3. External view (a) and block diagram (b) of the luminometer. 1: thermocontrolled fiber holder; 2: fibers; 3: photon-counting PMT detector; 4: power supply; 5: interface; 6: PLC; 7: computer; 8: step motor; 9: position sensor; 10: thermocontroller; 11: rotating disk-shutter with a hole 12. [Courtesy of Elsevier Science.]
8
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
hole that is positioned under the sample fiber at the moment it is under measurement, thus exposing only one fiber at a time. The fiber holder (sample compartment) was designed to offer optimal conditions for the adequate measurement of CL emitted by the phagocytes lying at the fiber end-face surface. The latter takes place in a sample cuvette or well (Figure 1d). The disk-shutter located in a light-tight space, can be rotated 60◦ around its axis by the step motor and by a worm gear (not shown) with a preciseness of 0.025◦ . When the orifice of the rotating shutter is positioned under one of the six fibers, it is then in optical contact with the naked PMT head of the photon-counting detector and the emitted photons are transmitted to the PMT surface and counted for a predetermined time interval. The subsequent turn of the shutter by a 60◦ angle positions the next fiber for measurement and the cycle is thus repeated. The position sensor’s feedback is used to ensure the correct function of the shutter positioning. Neither the samples nor the detector change their position during the measurement. Such an arrangement is space thrifty, ensures constant thermoregulation with the fiber holder, and provides a minimal optical path between the light-emitting samples and the light detector thus allowing optimum light collection. The measuring section consists of a photon-counting PMT detector that responds to light emission with electric impulses, the number of which correlates with the number of photons emitted, that is, light intensity.
of the buffers, and so on. This is why investigators must keep them constant at some predetermined value to avoid a multiparametric interpretation. However, the expected change in the CL pattern with the change of a known and described parameter validates the dynamic properties of the instrument and also provides an additional option to compare it with the competing instruments and protocols.
5 FEATURES OF THE PHAGOLUM DEVICE
5.2
A number of experiments show the instrument’s ability to provide relevant data of the phagocyte functional status, as well as check the extent of the recorded kinetic data that helps analyze the extra- to intracellular proportion of the phagocytereleased ROS when using the same emitting system during phagocytosis. Phagocyte CL kinetics is known to depend on many parameters3,40,50,30 : the PMNs’ functional status, the stimulating agent, the dilution of the blood, the stimulating particle/cell ratio, temperature, mixing, surface-to-volume ratio, storage duration, hydrophilic/hydrophobic properties of the sample surface, pH, metal content
It has been reported that the storage temperature47 as well as the temperature change that occurs in vitro, greatly affects the shape of the CL kinetics.4 Moreover, thermal shock or feverassociated temperatures enhance the neutrophils’ ROS production.61 That is why most of the luminometers designed to measure phagocyte ROS production are temperature controlled. Figure 5 shows two representative CL kinetics recorded at 20 and 37 ◦ C after 2-h blood preincubation at room temperature. A two-phase pattern is visible in both curves. The first peak appears at the
5.1
Zymosan Concentration as a CL Kinetics Parameter
Zymosan concentration (cell/particle ratio) may be misleading if not considered as a factor of CL patterns. Figure 4 shows the effect of zymosan concentration on the recorded CL pattern, demonstrating one of the known parameters that affects phagocyte activity assessment (and therefore needs a proper use depending on the goal). The samples are identical regarding both the volume and the concentration of the cells and the rest of the reagents used. The only difference is in the zymosan concentration, and therefore, in the amount of collisions and attempts of phagolysosomes forming. That in turn changes the amount and ratio of the ROS produced extra- and intracellularly during phagocytosis, thus causing clear difference in the observed CL kinetics pattern (Figure 4). The component analysis shows that the most significant change is increased emission of extracellular ROS with the increase of zymosan concentration. CL Kinetics at Room and Physiological Temperatures
PHAGOCYTE LUMINESCENT SENSOR
PMN 250000 ml−1 PMN 25000 ml−1 PMN 2500 ml−1
1e + 6 Capacity (counts)
9
1e + 5
1e + 4 0
3 4 2 Zymosan (mg ml−1)
1
(a) 2000
1.8
1800
1.6
1600
Relative units
CL (cps)
1.4
0.5 mg ml−1 4.0 mg ml−1
1400 1200 1000 800
1.2
400
0.4
200
0.2 5
10
(b)
15 Time (min)
20
25
0
30
Relative phagocapacity Velocity
0.8 0.6
0
6
1
600
0
5
0.5 mg ml−1
4 mg ml−1
(c)
Figure 4. CL kinetics pattern dependence on standard and extracellularly forced emitting system. Sample content: luminol E-4M: 10 µl; blood 1/30: 10 µl; opsonized zymosan (OZ): 12/1.5 mg ml−1 , correspondingly, 10 µl. (a) CL capacity dependence on zymosan concentration—some saturation is observed at 2 mg ml−1 zymosan. (b) CL patterns recorded after stimulation with low and high zymosan concentrations. (c) Derived CL information.
1100 1000 20 °C 37 °C
CL intensity (cps)
900 800 700 600 500 400 300 200 100 0
0
10
20
30
40 50 Time (min)
60
70
80
90
Figure 5. Representative CL kinetics recorded at room temperature (20 ◦ C, filled circle) and mammalian physiological temperature (37 ◦ C, empty circle). Both kinetics show a two-phase pattern CL. Sample content: luminol E-4M: 10 µl; blood 1/30: 10 µl; OZ: 12 mg ml−1 , correspondingly, 10 µl.
same time—approximately 3 min after the stimulation of the phagocytes in both kinetics (see the arrows). As for the second visible peak, it appears 10 and 45 min after stimulation for the kinetics recorded at 37 and 20 ◦ C respectively. Some stimulants are known to show two-phase CL patterns, but it is not characteristic in the case of zymosan-induced CL kinetics. In fact, there is one exceptional CL pattern at physiological conditions—the “frustrated state” of the phagocytes (see Ref. 50) featuring two peaks in its CL kinetics. It has been demonstrated by others that the presence of silica in the sample compartment causes an additional stimulation to PMNs.69 As mentioned in the preceding text, the front-end surface of the silica optical fibers in our system also serves as our cuvette bottom. Indeed,
10
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
both the size of the target to be phagocytized and the silica material lead to one very important feature of the use of this biosensor, the clear indication of the extracellularly produced light and the time of its appearance. This is intended to simplify the analysis and understanding of the mechanisms of PMNs activation, as well as the assessment of the extra- and intracellularly generated ROS using the same light-emitting system as was described in our earlier works (see CL assay). Note also the initial CL peak presence as well as the time of its appearance in all experiments in this article. Another contributor to the increased relative CL participation of the earlier stage processes (mainly extracellularly generated ROS) is the high quantity of zymosan used in the experiment.50 5.3
Assessment of Primed Phagocytes – Blood Storage (Aging of Blood) as a Priming Factor
The exposure of human neutrophils to low concentrations of chemotactic agents and other biologically active substances shows an increased response when stimulated again with the same or another agent (priming—see e.g., Ref. 15). Moreover, attempts have been made to correlate the primed activity of circulating PMNs with the severity of disease and outcome.66,73 That is why it was essential to check the sensitivity of our system and approach to assess
the priming by the functional status of phagocytes as reflecting the organism’s readiness for defense, therefore providing a high predictive value.45 Representative changes in CL pattern are shown in Figure 6 after different durations of blood post-venipuncture storage. The data, measured in a time interval of several hours, describes the aging of the same blood sample. It illustrates the potential of the computerized multisample temperature controlled luminometer for a fiber array–based biosensor as a rapid and sensitive measuring tool for the in vivo state of inflammation of circulating phagocytes. It is commonly accepted that aging of blood leads to the priming of phagocytes; therefore, it is a useful model for the investigation of the functional status of the phagocyte.76 Indeed, the analysis of the kinetics shows that the noticeable decrease in the peak time of the observed CL pattern can be represented as a continuous shift of the derived CL parameter’s effectiveness and velocity. As defined earlier,50 the effectiveness is calculated as the ratio of intracellular CL to extracellular CL due to phagocytosis (Figure 1b). As for the velocity, it is derived as the ratio of the total CL emitted due to phagocytosis to the light emitted during processes not directly related to phagocytosis. Figure 6(b) shows the quantitative estimation of both the effectiveness and the velocity of the emitted during respiratory burst ROS after storage at room temperature for 0, 1, 2, 3
600
Relative units
CL (cps)
400
Effectiveness Velocity
2.5
After 0 h After 1 h After 2 h After 4 h
500
300
2 1.5 1
200 0.5 100 0 0 (a)
10
20 Time (min)
30
After 0 h
40
After 1 h
After 2 h
After 4 h
(b)
Figure 6. (a) CL patterns recorded with the aging of same blood as a model for primed phagocyte activation. (b) Effect of aging on derived CL kinetic parameters. Sample content: luminol: E-4M: 10 µl; blood 1/30: 10 µl; OZ: 12 mg ml−1 , correspondingly, 10 µl.
PHAGOCYTE LUMINESCENT SENSOR
and 4 h, respectively. Indeed, the proper oxidative action of phagocytes is expected to be fast and efficient because of ROS created mostly during the phagocytosis timeframe and inside the cell. Because of the priming that takes place during the aging of blood, the phagocyte functional status undergoes continuous transition from “resting”50,76 state (high efficiency, low velocity) to “standby” state (decreased efficiency, higher velocity). Thus, the aging model illustrates the device potential to monitor the phagocyte functional status change with the changes in humoral-immune modulators.
5.4
fMLP priming
Figure 7(a) shows a representative change in CL pattern caused by a prior storage of tested blood with low concentration (3 nmol) of chemoattractant fMLP. Next, Figure 7(b) compares the quantitative estimation of both the effectiveness and the velocity of respiratory burst after fMLP priming recorded in parallel with a nonprimed sample from the same blood. It is interesting to see that at these conditions, fMLP priming causes an increase in velocity and some decrease in the effectiveness of respiratory burst owing to the changes in phagocytosis-related components of the CL (the
amount of extracellularly produced ROS is more significantly increased).
5.5
Glucose Level as a Priming Agent
It is known that the lack of glucose leads to exhausting of phagocytosis. Part of zymosanstimulated CL kinetics is glucose dependent, and the other part is not dependent on the glucose content. The glucose-independent part is considered to be directly connected to AA metabolism.13 The same work shows a 30–40% decrease of CL in the case of glucose lack for 2 h, without influence on the cell’s vitality. In other words, a change of glucose level during phagocyte stimulation is a priming factor and could provide additional information for the connection between CL and phagocytosis. Data shown in Figure 8 demonstrate CL kinetics of samples incubated with presence/absence of glucose (5.56 mmol l−1 ) for 3 h and after stimulation with relatively low zymosan concentration in the presence/absence of glucose in the standard CL system. In general, the phagocytosis-related part of CL kinetics shows a clear dependence on glucose concentration. The decrease of glucose level in the phagocyte media both prior and during the stimulation primes phagocyte respiratory burst readiness (Figures 8a and b).
Dependence of CL pattern on prior storage of phagocytes with fMLP
Dependence of derived CL parameters on prior storage of phagocytes with fMLP 7
4000
6
Without fMLP With 3 nmol l−1 fMLP
2000
Without fMLP 3 nmol l−1 fMLP
5 Relative units
3000 CL (cps)
11
4 3 2
1000
1 0
0 0 (a)
2
4
6 8 Time (min)
10
12
Effectiveness
Velocity
(b)
Figure 7. (a) Chemiluminescence response after incubation of diluted whole blood samples in vitro for 5 min at 37 ◦ C in the presence (•) and absence (o) of 3 nmol l−1 of fMLP. (b) Effect of fMLP on derived CL kinetic parameters. Sample content: luminol E-4M: 10 µl; blood 1/30: 10 µl; OZ: 12 mg ml−1 , correspondingly, 10 µl.
12
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS Dependence of CL component capacity on the presense of glucose in CL system
Effect of the glucose on phagocyte activity 3000
18 000 16 000
2500
Gi+Gp+ Gi+Gp− Gi−Gp+ Gi−Gp−
Capacity (cps)
CL (cps)
2000
14 000
1500 1000
12 000
Gi+Gp+ Gi+Gp− Gi−Gp+ Gi−Gp−
10 000 8000 6000 4000
500
2000
0
0 0
10
(a)
20 Time (min)
30
1+2
40
3
(b)
Figure 8. Comparison of the in vivo relative contribution of phagocytosis-related and nonrelated reactive oxygen species produced after incubation of diluted whole blood samples in vitro with D-glucose (priming of phagocytes). (a) Chemiluminescence pattern after incubation of diluted whole blood samples in vitro in the presence (+) or absence (−) of 5.56 mmol l−1 of D-glucose during the incubation (Gi) or during the measurement (Gp). (b) Comparison of the in vivo relative contribution of phagocyte extracellularly and intracellularly produced reactive oxygen species produced after incubation of diluted whole blood samples in vitro in the presence (+) or absence (−) of 5.56 mmol l−1 of D-glucose during the incubation (Gi) or during the measurement (Gp).
5.6
Circulating Phagocyte Activation
During the respiratory burst, upon stimulation with either soluble or particulate matter, PMNs generate reactive oxygen species and emit CL as a result of metabolic activation. The measurement of
CL has been demonstrated to be a useful tool for the in vitro assessment of the opsonophagocytic function of PMNs. Furthermore, the analysis of CL kinetics can be used to characterize the functional state of the PMNs.76,50 Figure 9(a) shows a typical pattern of a healthy “resting” state and
Derived respiratory burst parameters Infected (versus) healthy response 2000
4
Relative units
1500 CL (cps)
Effectiveness Velocity
Infected Control
1000
500
2
1
0 0 (a)
3
10
20 Time (min)
30
0
40
Candida
Control
(b)
Figure 9. (a) Simultaneously recorded representative CL responses ± CV of healthy (control) and infected patient blood (Candida albicans). According to their phagocyte functional status the healthy and infected responses can be classified50 as “resting” and “activated” correspondingly. Sample content: luminol E-4M: 10 µl; blood 1/30: 10 µl; OZ: 12 mg ml−1 , correspondingly, 10 µl. (b) Effect of circulating phagocyte status on derived CL kinetic parameters.
PHAGOCYTE LUMINESCENT SENSOR
an infected “activated” one, recorded while using operator-blinded blood samples obtained from two patients, one of whom was a patient suffering from a urinary tract infection. Both patterns were recorded after a 2-h preincubation of the blood. The healthy blood possesses, in general, a higher capacity for in vitro priming as compared to the blood of an infected patient, which had already undergone priming. This is why a relative prolongation of storage time could result in bringing both patterns of CL kinetics closer. However, Figure 9 shows that even at such a condition, the observed CL patterns differ sufficiently. It is clear from the very different CL patterns that this device is useful in distinguishing healthy from infected blood. Further analysis emphasizes this by showing the differences in the patient respiratory burst depicted in Figure 9(b). The data implies that during development of any phagocyte-related disease it will be possible to monitor the overall change in the patient’s humoral-immune-modulated phagocyte functional status. Our preliminary data (not shown), also depict unique CL kinetic patterns of PMNs obtained from patients infected by different types of organisms. Thus, this instrument has the potential of elucidating the nature of an infecting agent at a stage where definite characterization by bacteriological and/or serological methods is still unavailable. This valuable information could assist decision making regarding early resumption of antibiotic treatment and the type of chemotherapeutic agent used.
5.7
5.8
Derived respiratory burst parameters 8
Follow-up CAPD Follow-up CAPD+Diabetes
Relative units
CL (cps)
6
1500 1000
Priming in Biological Space II: Priming of the Surrounding Media
Of course, the approach is applicable not only to blood samples. For many disorders interfering with the innate immune system, other phagocyte-containing liquids could be relevant. The chemical composition of human cerebrospinal fluid is considered to reflect brain metabolism and
3000
2000
Priming in Biological Space I: Diabetes as a Glucose-level Priming Agent during Recurrent Therapeutic Procedures
One of the major causes of diabetic complication is hyperglycemia, but the exact mechanism of its detrimental effect is not clear. An impaired production of oxygen-derived free radicals (e.g., superoxide anion (O2 − ) and hydrogen peroxide-myeloperoxidase-halide (H2 O2 -MPO-Cl− ) antimicrobial system) by neutrophils from poorly controlled diabetic patients was demonstrated recently.63 This was supported by our observations of continuous ambulatory peritoneal dialysis (CAPD) patients. Some of these patients show an elevated glucose level, which in turn acts as an additional priming parameter (Figure 10). Figure 10(a) demonstrates different CL patterns between two groups of follow-up CAPD patients due to different levels of glucose in their blood. More detailed analysis shows (Figure 10b) an impaired phagocytosis-related part of respiratory burst.
CL kinetics patterns of CAPD patients
2500
13
Effectiveness Velocity
4
2 500 0
0 0 (a)
10
20 30 Time (min)
40
50
FUP
FUP+Diabetes
(b)
Figure 10. (a) Comparison between representative CL kinetics of two groups of CAPD follow-up patient with different blood glucose levels. (b) Derived CL information demonstrates impaired phagocytosis in the case of elevated glucose levels.
14
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
there is experimental evidence78 of a decrease in sulfhydryl groups and increased content of products of lipid peroxidation—both affecting the oxidative stress—this implies that cerebrospinal fluid also can be a parallel sample source for CL assessment in some autoimmune disorders like multiple sclerosis, stroke, Parkinson’s disease, Alzehimer’s disease, and physiological aging. Figure 11(a) compare simultaneously the activation level of whole blood and peritoneal liquid phagocytes. The striking difference in the phagocytosis-related part of demonstrated CL patterns suggests possible existence of some factor in the peritoneal liquid that is blocking this part of CL emission. Such a thesis is supported by the data reported in.79 Next, Figure 11(b) shows consecutive daily CL records of phagocyte activity in blood and peritoneal liquid taken from patient with peritonitis which illustrates the ability of the approach to monitor the innate immune system change with the patient treatment during the course of particular infection, in that case—peritonitis.
5.9
Priming in Temporal Space: CAPD Follow-up and Suppressed Patients Demonstrate Different Aging of Circulating Phagocytes
As reported earlier,76 the storage of blood interferes with the phagocyte level of dynamic
activation in many pathologies. Indeed, the initial phagocyte functional status, their transitional direction characteristics, in particular, clinical picture, taken together with the limited potential of phagocyte to restore its oxidative functionality, change the phagocyte functionality to a different extent and balance over time. Figure 12 shows different patterns of phagocyte response change after 2-h blood storage of CAPD follow-up patients and the response obtained from patients with suppressed phagocyte activity after surgical intervention.
5.10
Case Separation and Monitoring
All the analytical and technological steps mentioned earlier can be used for better diagnostic and prognostic evaluation during patient hospitalization and medical treatment. Figure 13 summarizes our results for phagocyte activity of CAPD patients. The method uses internal reference and comparison between primed and nonprimed phagocytes. Standard parameters (area under the CL kinetics, initial slope, time to peak), as well as CL kinetics component information were used to assess the normalized (by cell count) phagocyte activity. Component analysis was performed to derive more phagocytosis-related information. CL1 and CL2 are linear combinations made by the best-distinguishing CL parameters
Peritonitis—parallel CL data from peripheral blood and peritoneal liquid phagocytes
Peritoneal liquid: consecutive measurements
1600
1 Peritonitis blood Peritonitis liquid
1400
CL (relative units)
CL (cps)
1200 1000 800 600 400
PL Run 1 PL Run 2 PL Run 3 PL Run 4
0.8 0.6 0.4 0.2
200 0 0 (a)
10
20 30 Time (min)
40
0 50
0 (b)
10
20 30 Time (min)
40
50
Figure 11. Peritonitis blood versus peritoneal liquid—direct comparison between the local and overall state of phagocyte activation. (a) Different pattern of CL kinetics due to significant decrease of phagocytosis-related part of the respiratory burst. (b) Consecutive daily CL records of phagocyte activity in peritoneal liquid taken from patient with peritonitis.
PHAGOCYTE LUMINESCENT SENSOR Follow-up and suppressed CL patterns recorded without and after 2 h aging
Derived CL parameters 5
1600 1400
Follow-up Follow-up + 2 h Suppressed Suppressed + 2 h
1000
Effectiveness Velocity
4 Relative units
CL (cps)
1200
15
800 600 400
3 2 1
200 0
0 0
(a)
10
20 30 Time (min)
40
50
FUP
FUP+2h
Supp
Supp+2h
(b)
Figure 12. (a) CL patterns recorded with the aging of follow-up (FUP) and suppressed blood as a model for primed phagocyte activation. (b) Effect of aging on derived CL kinetic parameters. Sample content: luminol E-4M: 10 µl; blood 1/30: 10 µl; OZ: 12 mg ml−1 , correspondingly, 10 µl.
(Table 3 the list of parameters used for data separation shown on Figure 13). Standard statistical procedure was used to select the parameters and determine CL1 and CL2. As a result there is no overlapping between the groups shown in Figure 13. Moreover, Figure 13 illustrates the ability of the approach to monitor changes in the innate immune system with the patient treatment during the course of a particular infection, in this case—peritonitis. Daily measurements were preformed several times and CL information was used to characterize the innate immune status. 5.11
Effect of Pharmaceutical Substances on Respiratory Burst of Circulating Phagocytes
As shown earlier, the phagocyte luminescent concept can be applied to follow disease activity or early infection—before antibodies are detectable—and provide diagnostic and prognostic information during the course of patient treatment. It can be easily adapted to provide toxicology information about interaction between phagocytes and allergens, microbial and industrial pollutants, as well as for assessment of immunomodulating activity of pharmacological products. Figure 14 demonstrates the in vitro effect of IgE and aspirin on phagocyte respiratory burst activity.
Table 3. List of some parameters for group separation and case monitoring
Parameter nonPhagoSA RelCapSP RelPtimeSP
VelSP ExtraS RelNoPhagoSA
ExtraSA BkgSP NoPhagoS VelSA BkgSA RelPtimeSA
ExtraSP EffS SlopeS SlopeSP
Definition Nonphago-related CL of aged sample Capacity of primed sample divided by capacity of standard sample Peak time of primed sample divided by peak time of standard sample Velocity of primed sample Extracellular phagocytosis-related emission of standard sample Nonphago-related CL of aged sample divided by nonphago-related CL of standard sample Extracellular phagocytosis-related emission of primed sample Background CL of primed sample Nonphago-related CL of standard sample Velocity of aged sample Background CL of aged sample Peak time of primed sample divided by peak time of aged sample Extracellular phagocytosis-related emission of primed sample Effectiveness of standard sample Peak of standard sample divided by time to reach it Peak of primed sample divided by time to reach it
16
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS Infection—case monitoring 4
2 Run 1 Run 2 CL2
0
−2
FUP TINF PER SUPR DIAB TRANS
Run 3 −4
−6 −6
−4
−2
0
2 CL1
4
6
8
10
Figure 13. Group separation and case monitoring by discriminant analysis. Every case is shown by a small circle. Large circles depict mean canonical group coordinates. For case monitoring see text. FUP: follow-up case; TINF: case of tunnel infection; PER: case peritonitis; SUP: case of patient suppressed after transplantation; DIAB: case of diabetes mellitus; TRANS: case during treatment.
Effect of IgE on phagocyte activity
1600
Effect of aspirin on phagocyte activity
4000
1400 Control IgE-1U IgE-5U
1000
Control Aspirin 1.5 mmol
3000 CL (cps)
CL (cps)
1200
800 600 400
2000
1000
200 0 0 (a)
10
20 Time (min)
30
0
40 (b)
0
5
10
15 20 Time (min)
25
30
35
Figure 14. Examples of pharmaceutical agents testing—aspirin and IgE.
6 PROSPECTS
The PhagoLum biosensor system is based on an approach and a device with the potential to be an automated system for the assessment of circulating phagocyte functional status by the luminescence emitted during stimulated respiratory burst. Of course, to become efficient such an automated system will require relevant associated
databases. Several directions for applications seem to be very promising: (i) to follow disease activity or early infection in hospitals (before antibodies are detectable) in humans for both medical diagnosis and prognosis; (ii) to study the effects of pharmacological agents on the metabolism of granulocytes; (iii) to provide information about toxic interaction between phagocytes and new biomaterials.
PHAGOCYTE LUMINESCENT SENSOR
REFERENCES 1. R. C. Allen and L. D. Loose, Phagocytic activation of a luminol-dependent chemiluminescence in rabbit alveolar and peritoneal macrophages. Biochemical and Biophysical Research Communications, 1976, 69, 245–252. 2. R. C. Allen, R. L. Stjemholm, and R. H. Sieel, Evidence for the generation of electronic excitation state(s) in human polymorphonuclear leukocytes and its participation in bactericidal activity. Biochemical and Biophysical Research Communications, 1972, 47, 679–684. 3. R. C. Allen, Phagocytic leukocyte oxygenation activities and chemiluminescence: a kinetic approach to analysis. Methods in Enzymology, 1986, 133, 449–493. 4. B. Andersen and A. Brendzel, Use of a unique chemiluminescence spectrometer in a study of factors influencing granulocyte light emission. Journal of Immunological Methods, 1978, 19, 279–287. 5. P. Arenkov, V. Berezin, and N. Starodub, Chemiluminescence fiber optic immunosensor for detecting antibodies to the influenza virus. Ukra¨ıns’ky˘ı biokhimichny˘ı zhurnal, 1991, 63(4), 99–103. 6. D. A. Bass, J. W. Parce, L. R. DeChatelet, I. Szejda, M. C. Seeds, and M. Thomas, Flow cytometric studies of oxidative product formation by neutrophils: a graded response to membrane stimulation. Journal of Immunology, 1983, 130, 1910–1917. 7. P. Bellavite, The superoxide-forming enzymatic system of phagocytes. Free Radical Biology and Medicine, 1988, 4, 225–221. 8. J. G. Bender and D. E. Van Epps, Analysis of the bimodal chemiluminescence pattern stimulated in human neutrophils by chemotactic factors. Infection and Immunity, 1983, 41, 1062–1070. 9. P. Bochev, B. Bechev, and M. Magrisso, Six-sample multiplexing computerized analyzer for integral and spectral luminescence measurements. Analytica Chimica Acta, 1992, 256, 29–32. 10. E. P. Brestel, Co-oxidation by hypochlorite and hydrogen peroxide—implications for neutrophil chemiluminescence. Biochemical and Biophysical Research Communications, 1985, 126, 482–488. 11. G. Briheim, O. Stendahl, and C. Dahlgren, Intra- and extracellular events in luminol-dependent chemiluminescence of polymorphonuclear leukocytes. Infection and Immunity, 1984, 45, 1–5. 12. M. Cattaneo, K. Male, and J. Luong, A chemiluminescence fiber-optic biosensor system for the determination of glutamine in mammalian cell cultures. Biosensors and Bioelectronics, 1992, 7(8), 569–574. 13. K. Cheung, A. Archibald, and M. Robinson, The origin of chemiluminescence produced by neutrophils stimulated by opsonized zymosan. Journal of Immunology, 1983, 130, 2324–2329. 14. H. J. Cohen and M. E. Chovaniec, Superoxide generation bydigitonin-stimulated guinea pig granulocytes. A basis for acontinuous assay for monitoring superoxide production and for the study of the activation of the generating system. Journal of Clinical Investigation, 1978, 61, 1081–1087. 15. A. M. Condliffe, E. Kitchen, and E. R. Chilvers, Neutrophil priming: pathophysiological consequences and
16.
17.
18.
19.
20.
21.
22.
23.
24. 25.
26.
27.
28.
29.
30.
17
underlying mechanisms. Clinical Science, 1998, 94, 461–471. C. Dahlgren and O. Stendahl, Role of myeloperoxidase in luminol-dependent chemiluminescence of polymorphonuclear leukocytes. Infection and Immunity, 1983, 39, 736–741. C. Dahlgren, Difference in extracellular radical release after chemotactic factor and calcium ionophore activation of the oxygen radical generating system in human neutrophils. Biochimica et Biophysica Acta, 1987, 930, 33–38. C. Dahlgren, Analysis of luminol-dependent chemiluminescence from granule depleted neutrophil cytoplast reveals two different light emitting mechanisms. Journal of Bioluminescence and Chemiluminescence, 1988a, 2, 25–33. C. Dahlgren, Effects on extra- and intracellular localized, chemotactic induced, oxygen radical production in neutrophils following modulation for ligand-receptor interaction. Inflammation, 1988b, 12, 335–349. C. Dahlgren, Is lysosomal fusion required for the granulocyte chemiluminescence reaction? Journal of Free Radicals in Biology and Medicine, 1989, 6, 399–403. C. Dahlgren, P. Follin, A. Johansson, R. Lock, H. Lundqvist, and A. Walan, Chemiluminescence as a means of following the function of phagocytic cells. Trends Photochemistry And Photobiology, 1991, 2, 427–443. S. M. Dauphinee and A. Karsan, Lipopolysaccharide signaling in endothelial cells. Laboratory Investigation, 2006, 86(1), 9–22. L. R. DeChatelet, G. D. Long, P. S. Shirley, D. A. Bass, M. J. Thomas, F. W. Henderson, and M. S. Cohen, Mechanism of the luminol-dependent chemiluminescence of human neutrophils. Journal of Immunology, 1982, 129, 1589–1593. M. A. DeLuca (ed), Methods in Enzymology, Academic Press, New York, 1978, Vol. LVII. M. A. DeLuca and W. D. McElroy (eds), Methods in Enzymology, Academic Press, Orlando FL, 1986,Vol. 133, Part B. S. W. Edwards, Luminol and lucigenin-dependent chemiluminescence of neutrophils: role of degranulation. Journal of Clinical and Laboratory Immunology, 1987, 22, 35–39. A. Emmend¨orffer, M. Hecht, M. L. Lohmann-Matthes, and J. Roesler, A fast and easy method to determine the production of reactive oxygen intermediates by human and murine phagocytes using dihydrorhodamine 123. Journal of Immunological Methods, 1990, 131, 269–275. H. Faden and N. Maciejewski, Whole blood luminoldependent chemiluminescence. Journal of the Reticuloendothelial Society, 1981, 30, 219–226. J. Faldt, M. Ridell, A. Anna Karlsson, and C. Dahlgren, The phagocyte chemiluminescence paradox: luminal can act as an inhibitor of neutrophil NADPH-oxidase activity. Luminescence, 1999, 14, 153–160. I. Ginsburg, R. Misgav, D. F. Gibbs, J. Varani, and R. Kohen, Chemiluminescence in activated human neutrophils: role of buffers and scavengers. Inflammation, 1993, 1(3), 227–243.
18
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
31. L. Glasser and R.L. Fiederlein, The effect of various cell separation procedures on assays of neutrophil function. A critical appraisal. American Journal of Clinical Pathology, 1990, 93, 662–669. 32. B. L. Granger, M. L. Flenniken, D. A. Davis, A. P. Mitchell, and J. E. Cutler, Yeast wall protein 1 of Candida albicans. Microbiology, 2005, 151(Pt 5), 1631–1644. 33. O. Haller, G. Kochs, and F. Weber, The interferon response circuit: induction and suppression by pathogenic viruses. Virology, 2006, 344(1), 119–130. 34. M. B. Hallet and A. K. Campbell, Two distinct mechanisms for stimulation of oxygen-radical production by polymorphonuclear leukocytes. The Biochemical Journal, 1983, 216, 459–465. 35. D. C. Hargreaves and R. Medzhitov, Innate sensors of microbial infection. Journal of Clinical Immunology, 2005, 25(6), 503–510. 36. P. A. Hyslop and L. A. Sklar, A quantitative fluorimetric assay for the detection of oxidant production by polymorphonuclear leukocytes: its use in the simultaneous fluorimetric assay of cellular activation processes. Analytical Biochemistry, 1984, 141, 280–286. 37. T. W. Jungi and E. Peterhans, Change in die chemiluminescence reactivity pattern during in vitro differentiation ol human monocytes to macrophages. Blut, 1988, 56, 213–220. 38. S. J. Klebanoff and R. A. Clark, The Neutrophil: Function and Clinical Disorders, Amsterdam, Elsevier, Biomedical Press, 1978. 39. T. Konry, A. Novoa, S. Cosnier, and R.S. Marks, Development of an ‘electroptrode’ immunosensor: indium–tin– oxide–coated optical fiber tips with an electropolymerized thin film with conjugated cholera toxin B subunit. Analytical Chemistry, 2003, 75, 2633–2639. 40. L. Liu, H. Elwing, A. Karlsson, G. Nimeri, and C. Dahlgren, Surface-related triggering of the neutrophil respiratory burst. Characterization of the response induced by IgG adsorbed to hydrophilic and hydrophobic glass surfaces. Clinical and Experimental Immunology, 1997, 109(1), 204–210. 41. Y. Liu, J. Ye, and Y. Li, Rapid detection of Esherichia coli O157:H7 inoculated in ground beef, chicken carcass, and lettuce samples with an immunomagnetic chemiluminescence fiber-optic biosensor. Journal of Food Protection, 2003, 66, 512–517. 42. R. Lock and C. Dahlgren, Characteristics of the granule chemiluminescence reaction following an interaction between human neutrophils and Salmonella typhimurium bacteria. APMIS Acta Pathologica, Microbiologica, et Immunologica Scandinavica, 1988, 96, 299–305. 43. R. Lock, A. Johansson, K. Orselius and C. Dahlgren, Analysis of horseradish peroxidase-amplified chemiluminescence produced by human neutrophils reveals a role for the superoxide anion in the light emitting reaction. Analytical Biochemistry, 1988, 173, 450–455. 44. H. Lundqvist and C. Dahlgren, Isoluminol-enhanced chemiluminescence: sensitive method to study the release of superoxide anion from human neutrophils. Free Radical Biology and Medicine, 1996, 20(6), 785–792. 45. E. G. Maderazo, C. L. Woronick, S. D. Albano, S. P. Breaux, and R. M. Pock, Inappropriate activation deactivation and probable autooxidative damage as a
46.
47.
48.
49.
50.
51.
52.
53.
54.
55. 56. 57. 58.
59.
60.
61.
mechanism of neutrophil locomotory defect in trauma. Journal of Infectious Diseases, 1986, 154, 471–477. P. N. Madianos, Y. A. Bobetsis, D. F. Kinane, Generation of inflammatory stimuli: how bacteria set up inflammatory responses in the gingiva. Journal of Clinical Periodontology, 2005, 32(Suppl 6), 57–71. U. Magnusson and H. Holst, Assaying granulocyte phagocytosis by chemiluminescence: effect of storage time and temperature of blood samples. Zentralblatt fur Veterinarmedizin. Reihe B, 1998, 45(4), 217–222. M. Magrisso, M. Alexandrova, P. Bochev, B. Bechev, V. Markova, and I. Benchev, Model components of luminol chemiluminescence generated by PMNL. Journal of Biochemical and Biophysical Methods, 1995a, 30, 257–269. M. Magrisso, B. Bechev, P. Bochev, V. Markova, and M. Alexandrova, A new approach for analysis of chemiluminescent kinetics of activated phagocytes in blood. Journal of Bioluminescence and Chemiluminescence, 1995b, 10, 77–84. M. Magrisso, M. Alexandrova M. Markova, B. Bechev, and P. Bochev, Functional States of Polymorphonuclear Leukocytes Determined by Chemiluminescent Kinetic Analysis. Luminescence, 2000, 15, 143–151. R.S. Marks, E. Bassis, A. Bychenko, and M.M. Levine, Chemiluminescent optical fiber immunosensor to cholera antitoxin. Optical Engineering, 1997, 36(12), 3258–3264. R. Marks, A. Margalit, A. Bychenko, E. Bassis, N. Porat, and R. Dagan, Development of a chemiluminescent optical fiber immunosensor to detect Streptococcus pneumoniae antipolysaccharide antibodies. Applied Biochemistry and Biotechnology, 2000, 89(2–3), 117–126. J. A. Metcalf, J. I. Gallin, W. M. Nausseef, R. K. Root, Laboratory Manual of Neutrophil Function. Raven Press, New York, 1986. M. A. Model, L. S. Ganelina, and R. F. Todd III, A microscopic study of Fc gamma RIII-mediated respiratory burst in neutrophils. Immunobiology, 1998, 199(1), 39–50. H. Z. Movat, The Inflammatory Reaction, Amsterdam, Elsevier, 1985. L. Packer (ed), Methods in Enzymology, Academic Press, Orlando, FL, 1984, Vol. 105. L. Packer (ed), Methods in Enzymology, Academic Press, New York, 1994, Vol. 233. E. Piva, S. De Toni, A. Caenazzo, M. Pradella, F. Pietrogrande, and M. Plebani, Neutrophil NADPH oxidase activity in chronic myeloproliferative and myelodysplastic diseases by microscopic and photometric assays. Acta Haematologica, 1996, 96(4), 264–265. R. K. Root and M. S. Cohen, The microbicidal mechanisms of human neutrophils and eosinophils. Reviews of Infectious Diseases, 1981, 3, 565–598. G. M. Rosen, S. Pou, C. L. Ramos, M. S. Cohen, and B. E. Britigan, Free radicals and phagocytic cells. FASEB Journal, 1995, 9, 200–209. A. J. Rosenspire, A. L. Kindzelskii, and H. R. Petty, Cutting edge: fever associated temperatures enhance neutrophil responses to lipopolysaccharide: a potential mechanism involving cell metabolism. Journal of Immunology, 2002, 169(10), 5396–5400.
PHAGOCYTE LUMINESCENT SENSOR 62. F. Rossi, The O2¨y-forming NADPH oxidase of the phagocytes: nature, mechanisms of activation and function. Biochimica et Biophysica Acta, 1986, 853, 65–89. 63. N. Sato, H. Shimizu, K. Suwa, Y. Shimomura, M. Mori, and I. Kobayashi, Myeloperoxidase activity and generation of active oxygen species in leukocytes from poorly controlled diabetic patients. Diabetes Care, 1992, 15, 1050–1052. 64. P. E. Stanley, Commercially available luminometers and imaging devices for low-light level measurements and kits and reagents utilizing bioluminescence or chemiluminescence: survey update 5. Journal of Bioluminescence and Chemiluminescence, 1997, 12(2), 61–78. 65. P. Stevens, D. J. Winston, and K. Van Dyke, In vitro evaluation of opsonic and cellular granulocyte function by luminol-dependent chemiluminescence: utility in patients with severe neutropenia and cellular deficiency states. Infection and Immunity, 1978, 22, 41–51. 66. D. L. Stevens, A. E. Bryant, J. Huffman, K. Thompson, and R. C. Allen, Analysis of circulating phagocyte activity measured by whole blood luminescence: correlations with clinical status. Journal of Infectious Diseases, 1994, 170(6), 1463–1472. 67. M. Suematsu, C. Oshio, S. Miura, and M. Tsuchiya, Real-time visualization of oxyradical burst from single neutrophil by using ultrasensitive video intensifier microscopy. Biochemical and Biophysical Research Communications, 1987, 149, 1106–1110. 68. M. Tarpey and I. Fridovich, Methods of detection of vascular reactive species: nitric oxide, superoxide, hydrogen peroxide, and peroxynitrite, Circulation Research, 2001, 89, 224–236. 69. M. Tuomala, M. R. Hirvonen, M. Holopainen, and K. Savolainen, Stimulation of human polymorphonuclear leukocytes by consecutive doses of quartz and interactions of quartz with fMLP. Toxicology and Applied Pharmacology, 1993, 118(2), 224–232. 70. P. van der Valk and C. J. Herman, Leukocyte functions. Laboratory Investigation, 1987, 56(2), 127–137.
19
71. K. Van Dyke (ed), Bioluminescence and Chemiluminescence: Instruments and Applications, CRC, Boca Raton, Florida, 1986, Vol. 1. 72. K. Van Dyke and V. Kastranova (eds), Cellular Chemiluminescence, CRC, Boca Raton, Florida, 1987, Vol. 1. 73. C. H. Wakefield, P. D. Carey, S. Foulds, J. R. Monson, P. J. Guillou, Polymorphonuclear leukocyte activation. An early marker of the postsurgical sepsis response. Archives of Surgery, 1993, 128(4), 390–395. 74. K. H. Western and V. Videm, Donor neutrophil function after plateletpheresis, Transfusion, 2000, 40(11), 1414–1418. 75. E. Wiener, Impaired phagocyte antibacterial effector functions in beta-thalassemia: a likely factor in the increased susceptibility to bacterial infections. Hematology, 2003, 8(1), 35–40. 76. J. M. Zgliczynski, E. Kwasnowska, T. Stelmaszynska, E. Olszowska, S. Olszowski, and J. M. Knapik, Functional states of neutrophils as suggested by whole blood chemiluminescence. Acta Biochimica Polonica, 1988, 35, 330–342. 77. G. A. Zimmerman, A. D. Renzetti, and H. R. Hill, Functional and metabolic activity of granulocytes from patients with adult respiratory distress syndrome. Evidence for activated neutrophils in the pulmonary circulation. The American Review of Respiratory Disease, 1983, 127, 290–300. 78. V. Calabrese, R. Bella, D. Testa, F. Spadaro, A. Scrofani, V. Rizza, and G. Pennisi, Increased cerebrospinal fluid and plasma levels of ultraweak chemiluminescence are associated with changes in the thiol pool and lipidsoluble fluorescence in multiple sclerosis: the pathogenic role of oxidative stress. Drugs under Experimental and Clinical Research, 1998, 24(3), 125–131. 79. I. Daniels, K. S. S. Bhatia, C. J. Porter, M. A. Lindsay, A. G. Morgan, R. P. Burden, and J. Fletcher, Hydrogen peroxide generation by polymorphonuclear leukocytes exposed to peritoneal dialysis effluent. Clinical And Diagnostic Laboratory Immunology, 1996, 3(6), 682–688.
31 Applications of the Electrogenerated Luminescent Reactions in Biosensor and Biochip Developments Christophe A. Marquette and Lo¨ıc J. Blum Laboratoire de G´enie Enzymatique et Biomol´eculaire, Universit´e Claude Bernard Lyon 1, Villeurbanne Cedex, France
Luminescent transitions of excited molecules or atoms to a state of lower energy are characterized by electromagnetic radiations dissipated as photons in the ultraviolet, visible, or near-infrared regions. These luminescent reactions are classified according to the energy source involved during the excitation step. Thus, the most classical lightemission reactions are referred to as bioluminescence (from in vivo systems), chemiluminescence (from a chemical reaction), electrogenerated luminescence (from an electrochemical reaction), and photoluminescence (from UV, visible, or nearIR radiations). More trivial reactions were also described as pyroluminescence (from flameexcited metal atoms), radioluminescence (from irradiation by x rays or γ rays), sonoluminescence (from ultrasonication of dissolved substance), and thermoluminescence (from solids subjected to mild heating). Luminescence measurements consist in monitoring the rate of production of photons and, thus, the light intensity depends on the rate of the luminescent reaction. Consequently, light intensity is directly proportional to the concentration of a limiting reactant involved in a luminescence reaction. With modern instrumentation, light can be measured at a very low level, and this allows the
development of very sensitive analytical methods based on these light-emitting reactions. Electrogenerated luminescence sensors have been developed with the aim of combining the sensitivity of light-emitting reactions with the convenience of sensors. Fiber optics associated with a sensitive light detector appeared to be convenient transducers for designing biosensors. In addition to these fiber optics–based sensors several luminescence analytical systems, including imaging systems and on-chip biosensing, have been described and are reported hereafter.
1 ELECTROGENERATED LUMINESCENT REACTIONS 1.1
Electrochemiluminescent (ECL) Reaction
Chemiluminescence reactions are generally oxidoreduction processes and the excited compound that is the reaction product, has a different chemical structure from the initial reactant. Several hundreds of organic and inorganic compounds are at the origin of chemiluminescence reactions, which can occur in liquid or solid phases, or at solid–liquid or solid–gas interfaces.1–3
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Here, electrochemiluminescence (ECL) refers to liquid phase light-emitting reactions based on the oxidation of 5-amino-2,3-dihydrophthalazine-1,4dione (luminol).4 Figure 1(a) shows the overall reactions in aqueous medium. The luminol oxidation leads to the formation of an aminophthalate ion in an excited state, which emits light when returning to the ground state. The quantum yield of the reaction is low (≈0.01) and the emission spectrum shows a maximum at 425 nm.5 The electrochemical oxidation of luminol is usually considered as the second most efficient way of triggering the reaction, behind the horseradish peroxidase (HRP) biocatalyzed one. In a mechanistic study of this ECL reaction, Sakura6 had proposed that luminol was first oxidized at the electrode surface and then reacted, mole to mole, with hydrogen peroxide (Figure 1a). The theoretical ratio:
H +, e –
Diazaquinone Luminol (LH–) + 425 mV vs Pt pH 8–9
Oxidant
Luminol endoperoxide
h n 425 nm N2 [3-Aminophthalate]
[3-Aminophthalate]*
I (µA)
(a) 3.5 3 2.5 2 1.5 1 0.5 0 −0.5 −1 0 (b)
0.2
0.4 0.6 E (V)
0.8
1
Figure 1. (a) Schematic representation of the electrocatalyzed chemiluminescent reaction and (b) a typical cyclic voltammogram of luminol in aqueous solution.
(photon produced)/(H2 O2 consumed) is then 1, while it is only 0.5 for the peroxidase-catalyzed reaction.
1.2
Electroluminescent (EL) Reactions
Electroluminescent (EL), here, refers to the electrogenerated oxidoreduction reactions of compounds, different from the luminol or luminol derivatives, leading to a photon emission. The main competitors to luminol ECL in the field of analytical chemistry are the ruthenium complexes7,8 and more generally the metal chelate systems. Nevertheless, other molecules such as 9,10-diphenylanthracene,9 phenothiazine,10 and pyrene11 were demonstrated as EL. One of the most attractive characteristics of these metal chelate complexes is their ability to be regenerated in their native form after having completed the light-emission reaction sequence. A single molecule could then theoretically generate more photons than the luminol in the destructive ECL. The most widely used and studied of these metal chelate complexes is the tris(2,2′ bipyridyl)-ruthenium (II) also named Ru(bpy)3 2+ (Figure 2a). Indeed, since it was first reported as an EL compound in 1972,12 Ru(bpy)3 2+ has become the most thoroughly studied ELactive molecule.7,13 This domination of the field is mainly due to its strong luminescence, its solubility in both aqueous and nonaqueous media at room temperature, and of course its ability to undergo, as mentioned in the preceding text, reversible one-electron transfer reaction (Figure 2b). Figure 2(c) presents the most widely used system for triggering EL of Ru(bpy)3 2+ in aqueous solution. First, the ruthenium complex is electrooxidized in a one-electron reaction at the electrode surface (polarized at a potential between +1 and +1.5 V). Concomitantly, the classical coreagent tripropylamine (TPA)14 is also oxidized and deprotonated to generate a radical species that will reduce the oxidized metal complex (Ru(III)(bpy)3 3+ ), leading to an excited state of the reduced ruthenium complex that emits a photon (617 nm) while returning to the ground state.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
3
0.8 0.6 2+ N N
N
2Cl−
I (µA)
0.4
N
0 −0.2
Rh N
0.2
−0.4
6H2O
−0.6
N
−0.8 0.5
0.7
(b)
(a)
0.9
1.1 E (V)
1.3
1.5
h ν 617 nm
Ru(II)(bpy)32+*
Ru(II)(bpy)32+
e− +1000∼1500 mV pH 7
e−
Ru(III)(bpy)33+ TPA
TPA.
TPA+
(c)
H+
Figure 2. (a) Structure of the tris(2,2′ -bipyridyl)-ruthenium (II) also named Ru(bpy)3 2+ , (b) a typical cyclic voltammogram of the Ru(bpy)3 2+ in aqueous solution, and (c) schematic representation of the electroluminescent reaction of Ru(bpy)3 2+ with tripropylamine (TPA).
2 BIOSENSOR APPLICATIONS
The main interest of luminol ECL in biochemical and clinical analysis is the possibility of coupling this light-emitting reaction with enzyme-catalyzed reactions generating hydrogen peroxide. Simple auxiliary H2 O2 -generating reactions as well as multienzymatic systems leading to the production of hydrogen peroxide can then be used for the specific detection of different metabolites. Moreover, avoiding the use of fragile enzymes for the catalysis of the chemiluminescent reaction could lead to more stable and reproducible sensors. Consequently, regarding the sensitivity of hydrogen
peroxide detection, the electrogenerated chemiluminescence of luminol will be more efficient than the peroxidase-catalyzed reaction.15 Because of this high sensitivity for hydrogen peroxide, most of the applications of ECL will be dedicated to the detection of hydrogen peroxide-generating label enzymes and hydrogen peroxide enzymatic precursors. Less often, this electrogenerated reaction will be used for the detection of luminol-labeled molecules. On the contrary, the majority of the bioanalytical applications of the metal chelate complexes are based on the labeling of biomolecules with the complex itself. Indeed, by attachment of a suitable group to the bipyridine moieties, Ru(bpy)3 2+ can
4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
be linked to biologically interesting molecules such as proteins or nucleic acids. The structures of three of these modified complexes are presented in Figure 3.
O
2+
OH OH O
N N
N Ru N
N N
(a) O
P
N(iPr)2 2+ CN
Enzyme-based Biosensors
2.1.1 Luminol-based Systems
As mentioned in the preceding text, an original and unusual way to obtain highly sensitive hydrogen peroxide detection is the ECL of luminol. On the basis of this electro-optical process, flow-injection analysis (FIA) H2 O2 sensors was developed.15–17 ECL was generated using a polarized glassy carbon electrode (GCE) versus a platinum pseudoreference electrode and integrated in a FIA system that could take advantage of the use of optical fibers to separate the detector and the flow system15,18–20 (Figure 4). The optimization of the reaction conditions showed that an applied potential of +425 mV versus a platinum pseudoreference electrode enabled the realization of a sensitive H2 O2 sensor while avoiding fouling of the working electrode. An optimum pH measurement of 9 was found and, moreover, the pH dependence of the ECL sensor
CH3
N N
2.1
Detection system
N Ru
N
N N
FO (b) 2+
O O O
FC
N
SL
O R
N N
N Ru
N
N N
(c)
Figure 3. Structures of (a) bifunctional Ru(bpy)3 2+ derivative, (b) TrisBipyidyl Ruthenium (TBR)-label used by Perkin– Elmer, and (c) label used by Igen (R=CH3 , Origen TAG-NHS Ester ) and Boehringer Mannheim (R=H).
Applied potential
GCE
Figure 4. Flow cell for electrochemiluminescence measurements. GCE: glassy carbon electrode; SL: sensing layer; FO: liquid-core single optical fiber.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
appeared less pronounced than when using immobilized HRP as the sensing layer (SL). Under optimum conditions, hydrogen peroxide measurements could be performed in the range 1.5 pmol–30 nmol. This ECL H2 O2 sensor then exhibited slightly higher performances than membranebased HRP chemiluminescent FIA biosensors.21 For the development of glucose and lactate ECL FIA biosensors,15 the hydrogen peroxide ECL sensor could be associated with the catalytic action of glucose oxidase and lactate oxidase. The oxidases were immobilized on synthetic preactivated membranes brought into contact with the GCE. The glucose or lactate electro-optical biosensor was then able to detect the target analyte with detection limits of 150 and 60 pmol, respectively. In each case, glucose and lactate measurements could be performed over 4 decades of concentration. These biosensors were tested for glucose and lactate measurements in sera, and for lactate measurements in whey solutions. The agreement between the results of the present method and those of reference methods was good. For glucose analysis in serum, the coefficient of variation for 53 repeated measurements performed over a 10-h period was 4.8% while for lactate analysis, 80 assays performed over a 15-h period gave a coefficient of variation of 6.7%. Thus, the ECL-based biosensors gave the possibility to sensitively detect glucose and lactate in complex matrices without pretreatment of the samples. A flow-injection fiber-optic ECL biosensor for choline was also developed.19,20 Choline oxidase was immobilized by physical entrapment in a photo-cross-linkable poly(vinyl alcohol) polymer (PVA-SbQ) after adsorption on weak anionexchanger beads (diethylaminoethyl (DEAE)-Sepharose). In this way, the SL was directly created at the surface of the working GCE. The optimization of the reaction conditions and of the physicochemical parameters influencing the FIA biosensor response allows the measurement of choline concentration with a detection limit of 10 pmol. The DEAE-based system also exhibited a good operational stability since 160 repeated measurements of 3 nmol of choline could be performed with a variation coefficient of 4.5%. A cholesterol FIA biosensor has also been described as an application of the H2 O2 ECL sensor.18 In that work, the luminol ECL, previously studied in aqueous media, was implemented in
5
Veronal buffer added with 0.3% triton X-100 (v/v), 0.3% PEG, and 0.4% cholate to enable the solubilization of the cholesterol and then its efficient oxidation catalyzed by the immobilized cholesterol oxidase. The ECL reaction occurred, thus, in a micellar medium and the performances of the H2 O2 ECL sensor were investigated. The calibration curve obtained for hydrogen peroxide exhibited a detection limit of 30 pmol and ranged over 3 decades at least. These performances compared well with those previously obtained in nonmicellar media.16 The presence of surfactant compounds in the ECL measurement buffer appeared, thus, to have little effect on the H2 O2 ECL sensor performances. In optimized conditions, the determination of free cholesterol could be performed with a detection limit of 0.6 nmol and a calibration curve ranging over 2 decades of concentration at least.
2.1.2 Ru(bpy)3 2+ -based Systems
Most of the Ru(bpy)3 2+ applications in enzymebased biosensors involve dehydrogenase-type enzymes.22,23 A FIA detection method for glucose was presented which was based on oxidation of glucose by glucose dehydrogenase with concomitant conversion of NAD+ to nicotinamide adenine dinucleotide (NADH) followed by EL detection of NADH as a coreactant (Figure 5).22 Here, because the reduced form of NADH contains a tertiary amine, NADH can be used as a coreactant for Ru(bpy)3 2+ . The mechanism of the reaction of NADH with the ruthenium complex is then similar to that of the TPA/Ru(bpy)3 2+ mentioned in the preceding text. In a work from Martin et al., glucose dehydrogenase was immobilized via glutaraldehyde cross-linking to controlled pore glass beads bearing aminogroups to form an immobilized enzyme reactor used upstream to a platinum working electrode modified with Nafion film entrapping the EL reagent. Conditions for optimum enzyme reactor efficiency and EL detection were determined and reported for pH (about 6.51) and NAD+ concentration (l–2.5 mM). At the optimum conditions, a working curve was constructed where the upper limit for glucose detection was dependent on
6
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
Substrate + NAD+
hn 617 nm
[Ru(bpy)32+]*
dehydrogenase
Product + NADH + H+
H+
NAD+ NAD•
Ru(bpy)32+
Ru(bpy)33+
NADH•+
e−
NADH
e− +1.5 V vs Ag/AgCl
Figure 5. Electroluminescence-based reaction mechanism for the detection of NAD+ dehydrogenase substrates. [Reprinted from Martin and Nieman22 , with permission from Elsevier.]
NAD+ concentration and the lower detection limit was 10 µM glucose. On the basis of a principle equivalent to the glucose EL biosensor presented in the preceding text, an FIA system was developed for ethanol.24 The electrogenerated luminescence of ruthenium tris(2,2′ -bipyridine) was then used to detect a reduced form of NADH produced following the action of an immobilized alcohol dehydrogenase (ADH). The enzyme was used immobilized onto aminosilane modified glass beads packed in a column separated from the measurement flow cell. This latter was composed of a glassy carbon working electrode and a specially design stainless steel counterelectrode. The NADH determination was optimum under a flow rate of 1.5–2.0 ml min−1 and an applied voltage of +1.6 V. The NADH calibration plot showed a linear behavior in the concentration range from 10 to 250 µM, and the lower detection limit was 10 µM. Thus, using these optimized conditions, ethanol was detected in the FIA system as 0.01% (v/v) in buffer and with a detection ranging over 2 decades at least. Another interesting EL reaction based on Ru(bpy)3 2+ is described in Figure 6, where the coreactant of the metal complex is the oxalate (C2 O4 2− ).25 In this EL reaction, the oxalate is oxidized to form a strongly reducing carbon dioxide anion radical, which promotes Ru(bpy)3 2+ EL. In this case, the reaction occurring between oxalate
and hydrogen peroxide, leading to the production of water and CO2 , could be used to detect the presence of H2 O2 . Thus, this ability of the H2 O2 to diminish the oxalate-promoted EL was shown to be a possible basis for glucose detection using glucose oxidase.23 An inverse relationship was observed between glucose concentration and EL intensity but on a very narrow range, from 1 to 12 mM. Moreover, this system was not set up as a real biosensor, that is, including immobilized enzyme, and then consumed large quantities of reagent. Similar to the work presented above for the detection of glucose through oxalate/Ru(bpy)3 2+ EL, cholesterol has been detected using cholesterol oxidase.23 Again, a correlation between the EL intensity decrease and the cholesterol concentration was found on a limited range, from 2 to 10 mM.
2.2
Immunosensors
2.2.1 Luminol-based Systems
Luminol- and luminol-derivative-labeled antibodies16 are possible reagents for ECL immunosensors. Nevertheless, only few works were published with such labels,16,26 mainly because of the difficulties encountered when attempting to
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
Substrate + O2 H2O2 + C2O42−
oxidase
7
Product + H2O2
H2O + CO2
hn 617 nm CO2 C2O4• −
C2O42−
Ru(bpy)32+
CO2• − [Ru(bpy)32+]*
Ru(bpy)33+
Ru(bpy)33+
e−
Ru(bpy)32+
e− +1.5 V vs Ag/AgCl
Figure 6. Electroluminescence-based reaction mechanism for the detection of oxidase substrates through EL inhibition.
achieve their attachment through standard chemical reactions that retain the integrity of the protein. Indeed, the luminol molecule possesses only an aromatic amine as an available functional group, which is not easily covalently linked to biomolecules. Luminol derivatives having more reactive functions, such as N -(4-aminobutyl)N -ethylisoluminol (ABEI), were then tested as labels27 but were found to exhibit lower lightemission properties when grafted directly to proteins. Luminol-labeled antibodies were prepared using glutaraldehyde as a cross-linking agent and used in a 2,4-dichlorophenoxyacetic acid (2,4-D) competitive ECL immunosensor (Figure 7). 2,4-D was covalently immobilized at a GCE surface, via a 6-carbon spacer arm, by a procedure allowing to obtain stable immobilized antigens that could be then stored dry, used, and regenerated 50 times without loss of binding capacity. The luminol ECL detection was performed in an FIA system. The optimum conditions were found to be an oxidation potential of +500 mV versus a platinum
Glassy carbon electrode e− H2O2 Lu
hn 425 nm
O=C NH CH2
H2C H2C H2C NH H2C
CH2 CH2 C-O O Cl
Cl
Figure 7. Electrochemiluminescence-based immunosensor for 2,4-D using luminol-labeled antibodies and direct immobilization of the antigen at a glassy carbon electrode surface.
pseudoreference electrode, in the presence of 600 µM H2 O2 . Under these conditions, luminol could be detected in the range 5.5 fmol–55 nmol. Luminol-labeled anti-2,4-D antibodies were tested
8
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
for 2,4-D immunodetection. The corresponding ECL immunoassays exhibit a detection limit of 0.2 µg l−1 of free 2,4-D. The overall time taken for the experiment was 50 min and a linear range from 0.2 to 200 µg l−1 was obtained. Numerous works by Wilson et al.28–30 were published about the use of the ECL reaction of luminol for the achievement of immunosensors. These works were based on glucose oxidase–labeled antibodies used to locally generate the hydrogen peroxide required for the ECL reaction (Figure 8). Thus, antibodies to atrazine were labeled with glucose oxidase and used in enzyme-linked immunoassays.28 Transparent aminosilanized indium tin oxide (ITO)-coated glass electrodes were derivatized with aminodextran covalently modified with atrazine caproic acid. The labeled antibodies were used to investigate the derivatized electrodes in an ECL flow-injection analyzer. ECL immunoassay for atrazine in the range 0–1 µg l−1 showed that it was possible to detect the target molecule at concentrations as low as 0.1 µg l−1 . In a similar way, an ECL enzyme immunoassay for 2,4,6-trinitrotoluene (TNT) was reported.29 The deposition of a reusable immunosorbent dextran surface anchored to a gold surface in the flow cell by chemiadsorbed thiol groups was described as an SL. Antibodies were again labeled with the enzyme glucose oxidase and used in competitive immunoassays in which the separation step was carried out by concentrating unbound antibodies
H2O2
on the immunosorbent surface. Hydrogen peroxide generated by the enzyme label when glucose was pumped through the flow cell was then subsequently detected using luminol ECL. The light intensities obtained were inversely proportional to the concentration of TNT in the sample in the range 2.3–100 µg l−1 . More trivial electroassisted chemiluminescence enzyme immunoassays for TNT and pentaerythritol tetranitrate (PETN) were described.30 Haptens corresponding to these explosives were covalently attached to high-affinity dextran-coated paramagnetic beads. The beads were mixed with the corresponding Fab antibodies fragments and the sample. After adding HRP-labeled antispecies-specific antibody, the mixture was pumped into an electrochemiluminometer where beads were magnetically concentrated on the working electrode. The amount of analyte in the sample was determined by measuring light emission when H2 O2 was generated electrochemically in the presence of luminol and p-iodophenol (Figure 9). The detection limits obtained here for TNT and PETN were 0.11 and 19.8 µg l−1 , respectively.
hn 425 nm
Luminol
Horseradish peroxidase
Glucose
Glucose oxidase
O2
Magnetic bead
H2O2
hn 425 nm
Antigen coating
Luminol Antigen coating e−
ITO electrode
Figure 8. Electrochemiluminescence-based immunosensor for TNT using glucose oxidase as label and antigen-coated dextran-modified ITO.
Electrode Magnet
Figure 9. Electroassisted chemiluminescence-based immunosensor for TNT using horseradish peroxidase as a label and antigen-coated dextran-modified magnetic beads.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
2.2.2 Ru(bpy)3 2+ -based Systems
Electroluminescence-based immunoassays use, in most cases, Ru(bpy)3 2+ as a label, which allows measurements in aqueous solutions at a pH optimal for immunoreactions. The nondestructive mechanism of the Ru(bpy)3 2+ ECL also enhances theoretically the signal intensity by producing more than one photon per label. A variety of different Ru(bpy)3 2+ -based labels were published and used for basic research or commercial applications. N-hydroxysuccinimide (NHS) esters of Ru(bpy)3 2+ are used as monofunctional31,32 and bifunctional derivatives (Figure 3).33–35 Monofunctional NHS esters are used in commercial analyzers such as Elecsys and Origen (Origen TAG-NHS Ester ). These particular derivatives offer the advantage of a spacer between the biomolecule and the label. They also have no tendency to cross-link the biomolecules compared with the bifunctional NHS ester. The large majority of the studies performed using Ru(bpy)3 2+ as a label were conducted with magnetic beads as the solid support, brought into contact with the necessary electrode only during the EL-triggering step (Figure 10). Indeed, EL instrumentations were developed on the basis of the use of photomultiplier tube, particularly by ORIGEN , to measure EL labels present at the surface of magnetically responsive beads. The beads were designed to be easily coated with either antibodies or antigen and act as a binding solid phase.36 Thus, sensitive detections of various biotoxins and bacterial spores using the commercial ORIGEN analyzer were achieved by capture Introduce the magnetic beads Capture of the beads bearing and the assay components the reacted components
9
on antibody-conjugated micron sized magnetic beads followed by binding of Ru(bpy)3 2+ -labeled secondary antibodies.31 Femtogram sensitivity levels were obtained for all biotoxins tested including botulinus A, cholera β subunit, ricin, and staphylococcal enterotoxoid B when using this immunomagnetic ECL approach. Assays for Bacillus anthracis spores were also successfully performed with a detection limit of at least 100 spores. Sensitive immunoassays for digoxin, thyrotropin, carcinoembryonic antigen, and α-fetoprotein were also developed using a similar system.37,38 ECL assays with Ru(bpy)3 2+ -derivatized antibodies or antigens were also achieved by performing the assay directly on an electrode surface.39 Biomolecules were immobilized onto disposable screen-printed (SP) gold electrodes via selfassembled monolayers of thiols or Fc-specific binding protein G. Finally, an alternative to the classical Ru(bpy)3 2+ EL reaction-triggering via TPA was proposed by Michel et al.40 The use of carbon interdigidated microelectrode arrays allowed a coreactant-free detection of labeled proteins via the annihilation reaction of Ru(bpy)3 + and Ru(bpy)3 3+ by redox-cycling. Nevertheless, up to now no specific applications were developed on the basis of such an electrochemical triggering system. 2.3
DNA Sensors
2.3.1 Luminol-based Systems
The luminol derivative ABEI was used to label a known oligonucleotide sequence, subsequently Applying the potential and measuring the EL signal
Photodetector
Electrode Magnet
Figure 10. Illustration of the process involved in EL analytical systems based on magnetic beads.
Release of the beads and wash out
10
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
used as a DNA probe for identifying a target single-stranded DNA.41 The developed system consisted of a platinum working electrode modified with electropolymerized polypyrrole. This electrogenerated polymer was used to immobilize the probe nucleic acid sequence, subsequently involved in the specific hybridization reaction. The hybridization events were evaluated by the ECL measurements of the ABEI-labeled, hybridized target sequence. The results showed that only a complementary sequence could form a double-stranded DNA with the DNA probe and give a strong ECL response, while a threebase mismatch sequence and a noncomplementary sequence gave no ECL signal. The intensity of the ECL was linearly related to the concentration of the complementary sequence in the range 9.6 × 10−11 –9.6 × 10−8 mol l−1 . Figure 11 depicts the configuration of a DNA detection system42 close to the system described above for the immunodetection of explosives by Wilson et al.—that is, based on electroassisted chemiluminescence. Nucleic acid probes were assembled on an Au-electrode using thiol-derivate sequences. The resulting monolayer-functionalized electrode was
hn 425 nm
Luminol
O2
H2O2
e−
Gold electrode
Figure 11. Schematic representation of the detection of nucleic acid sequence through the electroassisted chemiluminescence. The red sphere represents the doxorubicin intercalator.
then treated with the complementary target sequence, leading to the double-stranded DNA assembly on the electrode surface. This hybridized system was further treated with doxorubicin, a well-known specific intercalator of doublestranded CG base-pair-containing DNA sequences.43 The electrochemical reduction of the intercalator led to the electrocatalyzed reduction of O2 to H2 O2 , which in the presence of luminol enabled the catalysis of the chemiluminescent reaction by the free HRP. This electroassisted chemiluminescent reaction then enabled the detection, down to picomolar levels, of the target nucleic acid sequence.
2.3.2 Ru(bpy)3 2+ -based Systems
Likewise for immunoassays based on the EL of ruthenium complexes, commercial instrumentation is available for the detection of DNA or messenger RNA by reverse transcription polymerase chain reaction (PCR) (QPCR System 5000 from Perkin–Elmer44,45 and the Origen Analyzer from Igen46 ). The PCR products are usually immobilized on a solid support (magnetic beads) using biotin–streptavidin chemistry and the DNA (RNA) targets are detected with a Ru(bpy)3 2+ label down to attomolar levels. In a recent publication, m-RNA isolated and amplified from as low as 10 melanoma cells, were detected in a background media composed of 107 cells.46 Other target genes were evaluated such as the human immunodeficiency virus 1 (HIV-1) gag gene31,47 and the cystic fibrosis F-508 deletion mutation.47,48 The results obtained from these assays demonstrated the possibility to detect 10 copies of the HIV-1 gag genes or cystic fibrosis F-508 mutations in 1 ng of human DNA. A more integrated approach for DNA detection was proposed by Xu et al.49,50 Here, singlestranded DNA were immobilized directly on aluminum(III) alkanebiphosphate–modified electrodes and used as a probe sequence for the hybridization of an Ru(bpy)2 2+ -labeled target. In this case, only simple poly(dT) sequences were used and detected specifically, as proof of concept.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
(IDA) chelating beads (glucose oxidase only) or on DEAE anion-exchanger beads, and spotted on the surface of a glassy carbon foil (25 mm2 ) entrapped in PVA-SbQ photopolymer. The chip measurement was achieved by applying a +850 mV potential between the GCE and a platinum pseudoreference for 3 min, while capturing a numeric image of the multifunctional biosensing chip with a chargedcoupled device (CCD) camera. The use of luminol supporting beads (DEAESepharose) included in the SL was shown to enable the achievement of spatially well-defined signals and to solve the hydrogen peroxide parasite signal that appeared between contiguous spots using the free luminol found in the solution.51,52 The detection limits of the different biosensor were found to be 1 µM for glutamate, lysine, and uric acid, 20 µM for glucose, and 2 µM for choline and lactate. The detection ranges were 1–25 µM (uric acid), 1–0.5 µM (glutamate and lysine), 20–2 µM (glucose), and 2–0.2 µM (choline and lactate).
3 BIOCHIP APPLICATIONS
ECL and EL biochips presented here are either arrays of ECL or EL biosensors, or miniaturized biosensors integrating detector and/or microfluidics systems. The basis of the recognition and the biomolecules labeling are then equivalent to the methods described in the preceding text, and certain redundancies might appear.
3.1
11
Enzyme-based Biochips
3.1.1 Luminol-based Systems
A multifunctional biosensing chip was designed on the basis of the ECL detection of enzymatically produced hydrogen peroxide (Figure 12). Six different oxidases specific for choline, glucose, glutamate, lactate, lysine, and urate were noncovalently immobilized on iminodiacetic acid Sample Injection
Pt
W
S
Platinum pseudo-reference Analyte 1
hh n
Analyte 6
hhn Applied potential
GCE
H2O2 e−
H2O2
e−
Glassy carbon (a)
(b)
Figure 12. (a) Schematic representation of the sensing layer organization and reaction. The green sphere represents the luminol supporting beads. (b) Organization of the electrochemiluminescent multifunctional biosensing chip. GCE: glassy carbon electrode; Pt: platinum pseudoreference electrode; S: silicone spacer; W: Plexiglas window.
The ECL chip was used for the detection of glucose, lactate, and uric acid in human serum matrix. Good correlations between measured and expected values were found without the need of internal calibration of the sample, demonstrating the potential of the ECL multifunctional biosensing chip. This electrochemiluminescent biochip was extended to trienzymatic SL based on kinase-oxidase activities for the detection of acetate. A reaction sequence using acetate kinase, pyruvate kinase, and pyruvate oxidase was shown to enable the production of H2 O2 in response to acetate injection in the range of 10 µM–100 mM.53 On the basis of a similar entrapment concept of enzyme and luminol in PVA-SbQ photopolymer, a microarray of nine SP graphite electrodes (Figure 13a) was used to develop multiparametric ECL biochips.54 ECL cyclic-voltametric experiments were performed to characterize the ECL reaction triggered at the surface of the SP carbon electrode array and an optimum luminol oxidation potential of +650 mV versus platinum pseudoreference was found with a reproducibility of 4.4%. The different SLs were obtained through the entrapment of glucose oxidase, lactate oxidase, and choline oxidase in PVA-SbQ (poly(vinyl alcohol)-bearing styrylpyridinium groups) photopolymer deposited at the surface of three from the nine electrode arrays (Figure 13c). Therefore, ECL measurements were performed to establish calibration curves by using a CCD camera. In the present case, the achieved biosensor arrays
8, 4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
1
12
(a)
22, 3 Pseudoreference hn 425 nm
Analyte
Enzyme
Applied potential H2O2
Luminol
e−
(b)
Lox Gox Chox (c)
Glucose
Lactate
Choline
1 cm
Figure 13. (a) Schematic representation of an electrochemiluminescent active biochip based on screen-printed electrodes array, (b) principle of the electrochemiluminescent enzyme-based biochip, and (c) distribution of the different sensing layers at the biochip surface and electrochemiluminescent images of the biochip in the presence of either glucose, lactate, or choline. Chox: choline oxidase; Gox: glucose oxidase; Lox: lactate oxidase; Lu: luminol.
allowed the simultaneous detection and quantification of L-lactate, D-glucose, and choline with detection limits between 3 and 10 µM.
Working electrode Plexiglass Outlet
Inlet
SU-8 spacer
Silicon chip
Encapsulation epoxy Photodiodes
Printed-circuit board
Figure 14. Schematic representation of an integrated microfluidic electroluminescent silicon chip.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
3.1.2 Ru(bpy)3 2+ -based Systems
The fabrication of a miniaturized silicon device integrating both the electrode and the photodetector was presented by Fiaccabrino et al.55 Thus, a 5 × 6 mm chip composed of two identical cells (one active and one reference) and two photodiodes for the differential measurements of the EL reaction was developed. An interdigidated gold electrode rested on each photodiode, only separated by a reflective layer. This device could detect the ruthenium complex with a detection limit of 0.5 µM. The same cell using a platinum interdigidated electrode instead of a gold electrode was used to detect codeine with immobilized Ru(bpy)3 2+ .56 Using SU-8 technology, the silicon device could be modified to create a flow cell57 (Figure 14). A polymer thickness of 300 µm resulted in a cell volume of about 2.25 µl and allowed the detection of ruthenium complex concentrations as low as 50 µM in FIA. In a modified system, glucose could be detected by coupling the EL flow cell with a microenzymatic reactor with a detection limit of 50 µM.58 A submicrolitre EL detector was also described by Arora et al.59 A flow cell with an effective volume of 100 nl containing two platinum thinfilm electrodes was produced with poly(methyl methacrylate) (PMMA) and placed directly on the window of a photomultiplier tube. A very low detection limit of 5 × 10−13 M was achieved for Ru(bpy)3 2+ but no biosensing applications were proposed.
3.2
Immunochips and DNA Chips
3.2.1 Luminol-based Systems
A new active support for ECL biochip preparation has been developed on the basis of graphitemodified polydimethyl siloxane elastomer (PDMS).60 The addressed inclusion of Sepharose beads at the surface of the obtained elastomeric electrode generated local highly specific surfaces. This electrode structure was characterized by electrochemical and imaging methods and an increase factor of the surface area equal to 50 was found. This was due to the texturing of the surface
Glucose
13
H2O2
Glucose oxidase
hn 425 nm
Luminol
Conducting elastomer
(a) Glucose
H 2 O2
Glucose oxidase
hn 425 nm
Luminol
Conducting elastomer
(b)
Figure 15. The different electrochemiluminescent biochip formats based on conducting elastomer. (a) Nucleic acid–based biochip and (b) immunobiochip.
generated by the presence of the Sepharose beads (Figure 15). This new material was used to design biochips based on the ECL reaction of luminol in the presence of enzymatically produced H2 O2 . Using beads bearing biomolecules such as oligonucleotides or antigen in conjunction with glucose oxidase–labeled DNA or antibody, sensitive
14
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
biochips could be obtained with detection limits of 1011 and 1010 molecules, respectively.
3.2.2 Ru(bpy)3 2+ -based Systems
The only real EL biochip platform based on the Ru(bpy)3 2+ electroluminescence was developed by Meso Scale Diagnostics. It is based on disposable electrodes, on which the assays are carried out, made of SP carbon ink and placed within
the wells of multiwell plates.61 A variety of plate formats are available (96, 384, and 1536) and particularly, the 96-well format could be obtained as multispot with a patterned microarray within each well (Figure 16). Thus, 4, 7, or 10 differentiated spots could be present at the same time at the bottom of a single well. The electroluminescence of the labeled proteins or nucleic acids is then triggered by the well-bottom graphite electrode and collected with either a CCD camera or a series of photodiodes. An example of multiple simultaneous
CCD imaging camera
hn 617 nm
hn 617 nm
e−
Rubpy
Antigen 1 coating
e−
Rubpy
Antigen 2 coating
hn 617 nm
e−
Rubpy
Antigen 3 coating
Screen-printed electrode (a)
(b)
96 wells: 4, 7, and 10 spots/well
Figure 16. Surface organization (a) of one of the Meso Scale Diagnostics plate well bottom for the EL detection of protein or DNA. Electroluminescent images (b) obtained with 4, 7, or 10 multiwalled 96-well plates.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS
assays within one well is show in Figure 16(a). Different antigens are immobilized in the different subwells and are subsequently recognized by different Ru(bpy)3 2+ -labeled antibodies from the sample. A large number of assays were performed on this platform, from ligand–receptor binding assays to kinase assays and DNA–protein binding assays. For example, multiplexed cytokine immunoassays were used to simultaneously detect four human cytokines (IL-1β, IL-6, TNF-α, and INF-γ ) with detection limits in the 1–10 pg ml−1 range. Peptide arrays were also developed with SH2 domains of EGF receptor to precisely determine the binding site of the protein on its receptor.
REFERENCES 1. U. Isacsson and G. Wettermark, Chemiluminescence in analytical chemistry. Analytica Chimica Acta, 1974, 68(2), 339–362. 2. D. H. Stedman and M. E. Fraser, Analytical Applications of Gas Phase Chemiluminescence, in Chemi-and Bioluminescence, J. G. Burr (ed), Marcel Dekker, New York, 1985, pp. 439–468. 3. L. J. Blum, Bio-,Chemi-Luminescent Sensors, World Scientific, Singapore, 1997. 4. G. P. Jirka, A. F. Martin, and T. A. Nieman, pH and concentration response surfaces for the luminol–H2 O2 electrogenerated chemiluminescence reaction. Analytica Chimica Acta, 1993, 284(2), 345–349. 5. D. F. Roswell and E. H. White, The Chemiluminescence of Luminol and Related Hydrazides, in Methods in Enzymology, S. Fleischer and B. Fleischer (eds), Academic Press, London, 1978, pp. 409–423. 6. S. Sakura, Electrochemiluminescence of hydrogen peroxide-luminol at a carbon electrode. Analytica Chimica Acta, 1992, 262(1), 49–57. 7. W.-Y. Lee, Tris(2,2-bipyridyl) ruthenium (II) electrogenerated chemiluminescence in analytical science. Mikrochimica Acta, 1997, 127(1–2), 19–39. 8. A. W. Knight and G. M. Greenway, Occurrence, mechanisms and analytical applications of electrogenerated chemiluminescence. A review. The Analyst, 1994, 119(5), 879–890. 9. F. E. Beideman and D. M. Hercules, Electrogenerated chemiluminescence from 9,10-diphenylanthracene cations reacting with radical anions. Journal of Physical Chemistry, 1979, 83(17), 2203–2209. 10. A. Kakhr, Y. Mugnier, and E. Laviron, Electrochemical reduction of phenothiazine and fluorobenzene at low temperature. Electrochimica Acta, 1983, 28(12), 1897–1898. 11. A. J. Bard, Electrogenerated Chemiluminescence, Marcel Dekker, New York, 2004.
15
12. N. E. Tokel and A. J. Bard, Electrogenerated chemiluminescence. IX. Electrochemistry and emission from systems containing tris(2,2′ -bipyridine)ruthenium(II) dichloride. Journal of the American Chemical Society, 1972, 94(8), 2862–2863. 13. R. Wilson, H. Akhavan-Tafti, R. de Silva, and A. P. Schaap, Comparison between acridan ester, luminol, and ruthenium chelate electrochemiluminescence. Electroanalysis, 2001, 13(13), 1083–1092. 14. J. K. Leland and M. J. Powell, Electrogenerated chemiluminescence: an oxidative-reduction type ECL reaction sequence using tripropyl amine. Journal of the Electrochemical Society, 1990, 137(10), 3127–3131. 15. C. A. Marquette and L. J. Blum, Luminol electrochemiluminescence-based fibre optic biosensors for flow injection analysis of glucose and lactate in natural samples. Analytica Chimica Acta, 1999, 381, 1–10. 16. C. A. Marquette and L. J. Blum, Electrochemiluminescence of luminol for 2,4-D optical immunosensing in a flow injection analysis system. Sensors and Actuators, B: Chemical, 1998, 51, 100–106. 17. M. F. Laespada, J. P. Pavon, and B. M. Cordero, Electroluminescent detection of enzymatically generated hydrogen peroxide. Analytica Chimica Acta, 1996, 327(3), 253–260. 18. C. A. Marquette, S. Ravaud, and L. J. Blum, Luminol electrochemiluminescence-based biosensor for total cholesterol determination in natural samples. Analytical Letters, 2000, 33(9), 1779–1796. 19. V. C. Tsafack, C. A. Marquette, B. Leca, L. J. Blum, and V. C. Tsafack, An electrochemiluminescence-based fibre optic biosensor for choline flow injection analysis. The Analyst, 2000, 125(1), 151–155. 20. C. A. Marquette, B. D. Leca, and L. J. Blum, Electrogenerated chemiluminescence of luminol for oxidase-based fibre-optic biosensors. Luminescence, 2001, 16(2), 159–165. 21. L. J. Blum, Chemiluminescent flow injection analysis of glucose in drinks with a bienzyme fiberoptic biosensor. Enzyme and Microbial Technology, 1993, 15(5), 407–411. 22. A. F. Martin and T. A. Nieman, Glucose quantitation using an immobilized glucose dehydrogenase enzyme reactor and a tris(2,2′ -bipyridyl) ruthenium(II) chemiluminescent sensor. Analytica Chimica Acta, 1993, 281(3), 475–481. 23. F. Jameison, R. I. Sanchez, L. Dong, J. K. Leland, D. Yost, and M. T. Martin, Electrochemiluminescencebased quantitation of classical clinical chemistry analytes. Analytical Chemistry, 1996, 68(8), 1298–1302. 24. I. Rubinstein, C. R. Martin, and A. J. Bard, Electrogenerated chemiluminescent determination of oxalate. Analytical Chemistry, 1983, 55(9), 1580–1582. 25. K. Yokoyama, S. Sasaki, K. Ikebukuro, T. Takeuchi, I. Karube, Y. Tokitsu, and Y. Masuda, Biosensing based on NADH detection coupled to electrogenerated chemiluminescence from ruthenium tris(2,2′ -bipyridine). Talanta, 1994, 41(6), 1035–1040. 26. L. Dong and M. T. Martin, Enzyme-triggered formation of electrochemiluminescent ruthenium complexes. Analytical Biochemistry, 1996, 236(2), 344–347. 27. L. S. Hersh, W. P. Vann, and S. A. Wilheim, A luminol assisted competitive binding immunoassay of human
16
28.
29.
30.
31.
32.
33.
34.
35.
36.
37. 38.
39.
40.
41.
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS immunoglobulinG. Analytical Biochemistry, 1979, 93, 267–271. R. Wilson, M. H. Barker, D. J. Schiffrin, and R. Abuknesha, Electrochemiluminescence flow injection immunoassay for atrazine. Biosensors and Bioelectronics, 1997, 12(4), 277–286. R. Wilson, C. Clavering, and A. Hutchinson, Electrochemiluminescence enzyme immunoassay for TNT. The Analyst, 2003, 128(5), 480–485. R. Wilson, C. Clavering, and A. Hutchinson, Electrochemiluminescence enzyme immunoassays for TNT and pentaerythritol tetranitrate. Analytical Chemistry, 2003, 75, 4244–4249. G. F. Blackburn, H. P. Shah, J. H. Kenten, J. Leland, R. A. Kamin, J. Link, J. Peterman, M. J. Powell, A. Shah, and D. B. Talley, Electrochemiluminescence detection for development of immunoassays and DNA probe assays for clinical diagnostics. Clinical Chemistry, 1991, 37(9), 1534–1539. J. H. Kenten, J. Casadei, J. Link, S. Lupold, J. Willey, M. Powell, A. Rees, and R. Massey, Rapid electrochemiluminescence assays of polymerase chain reaction products. Clinical Chemistry, 1991, 37(9), 1626–1632. A. W. Knight and G. M. Greenway, Relationship between structural attributes and observed electrogenerated chemiluminescence (ECL) activity of tertiary amines as potential analytes for the tris(2,2-bipyridine)ruthenium(II) ECL reaction: a review. The Analyst, 1996, 121(11), 101R–106R. E. Terpetschnig, H. Szmacinski, and J. R. Lakowicz, Fluorescence polarization immunoassay of a highmolecular-weight antigen based on a long-lifetime RuLigand complex. Analytical Biochemistry, 1995, 227(1), 140–147. E. Terpetschnig, H. Szmacinski, H. Malak, and J. R. Lakowicz, Metal-ligand complexes as a new class of longlived fluorophores for protein hydrodynamics. Biophysical Journal, 1995, 68(1), 342–350. M. C. Kibbey, D. MacAllan, and J. W. Karaszkiewicz, Novel electrochemiluminescent assays for drug discovery. Journal of the Association for Laboratory Automation, 2000, 5(1), 45–48. D. R. Deaver, A new non-isotopic detection system for immunoassays. Nature, 1995, 377(6551), 758–760. D. L. Gatto-Menking, H. Yu, J. G. Bruno, M. T. Goode, M. Miller, and A. W. Zulich, Sensitive detection of biotoxoids and bacterial spores using an immunomagnetic electrocheminescence sensor. Biosensors and Bioelectronics, 1995, 10(6–7), 501–507. K. F¨ahnrich, C. K. O’Sullivan, and G. G. Guilbault, Disposable Biosensor Based on Electrogenerated Chemiluminescence (ECL), In: Europt(r)ode V , Lyon, 2000, p. 149, 2000 April 16–19. P. E. Michel, N. F. de Rooij, M. Koudelka-Hep, K. A. Fahnrich, C. K. O’Sullivan, and G. G. Guilbault, Redox-cycling type electrochemiluminescence in aqueous medium. A new principle for the detection of proteins labeled with a ruthenium chelate. Journal of Electroanalytical Chemistry, 1999, 474(2), 192–194. M. Yang, C. Liu, K. Qian, P. He, and Y. Fang, Study on the electrochemiluminescence behavior of ABEI and its
42.
43. 44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
application in DNA hybridization analysis. The Analyst, 2002, 127(9), 1267–1271. F. Patolsky, E. Katz, and I. Willner, Amplified DNA Detection by electrogenerated biochemiluminescence and by the catalyzed precipitation of an insoluble product on electrodes in the presence of the doxorubicin intercalator. Angewandte Chemie International Edition, 2002, 41(18), 3398–3402. F. Arcamone, Doxorubicin: Anticancer Antibiotics, Academic Press, New York, 1983. A. M. Siddiqi, V. M. Jennings, M. R. Kidd, J. K. Actor, and R. L. Hunter, Evaluation of electrochemiluminescence-and bioluminescence-based assays for quantitating specific DNA. Journal of Clinical Laboratory Analysis, 1996, 10(6), 423–431. K. Motmans, J. Raus, and C. Vandevyver, Quantification of cytokine messenger RNA in transfected human T cells by RT-PCR and an automated electrochemiluminescencebased post-PCR detection system. Journal of Immunological Methods, 1996, 190(1), 107–116. C. D. O’Connell, A. Juhasz, C. Kuo, D. J. Reeder, and D. S. B. Hoon, Detection of tyrosinase mRNA in melanoma by reverse transcription-PCR and electrochemiluminescence. Clinical Chemistry, 1998, 44(6), 1161–1669. S. R. Gudibande, J. H. Kenten, J. Link, K. Friedman, and R. J. Massey, Rapid, non-separation electrochemiluminescent DNA hybridization assays for PCR products, using 3′ -labelled oligonucleotide probes. Molecular and Cellular Probes, 1992, 6(6), 495–503. J. DiCesare, B. Grossman, E. Katz, E. Picozza, R. Ragusa, and T. Woudenberg, A high-sensitivity electrochemiluminescence-based detection system for automated PCR product quantitation. Biotechniques, 1993, 15(1), 152–157. X. Xu and A. J. Bard, Immobilization and hybridization of DNA on an aluminum(III) alkanebisphosphonate thin film with electrogenerated chemiluminescent detection. Journal of the American Chemical Society, 1995, 117(9), 2627–2631. X. Xu, H. C. Yang, T. E. Mallouk, and A. J. Bard, Immobilization of DNA on an aluminum(III) alkanebisphosphonate thin film with electrogenerated chemiluminescent detection. Journal of the American Chemical Society, 1994, 116(18), 8386–8387. C. A. Marquette and L. J. Blum, Self-containing reactant biochips for the electrochemiluminescent determination of glucose, lactate and choline. Sensors and Actuators, B: Chemical, 2003, 90(1–3), 112–117. C. A. Marquette, A. Degiuli, and L. J. Blum, Electrochemiluminescent biosensors array for the concomitant detection of choline, glucose, glutamate, lactate, lysine and urate. Biosensors and Bioelectronics, 2003, 19(5), 433–439. C. A. Marquette, D. Thomas, A. Degiuli, and L. J. Blum, Design of luminescent biochips based on enzyme, antibody, or DNA composite layers. Analytical and Bioanalytical Chemistry, 2003, 377(5), 922–928. B. P. Corgier, C. A. Marquette, and L. J. Blum, Screenprinted electrode microarray for electrochemiluminescent measurements. Analytica Chimica Acta, 2005, 538(1–2), 1–7.
APPLICATIONS OF THE ELECTROGENERATED LUMINESCENT REACTIONS 55. G. C. Fiaccabrino, N. F. de Rooij, and M. Koudelka-Hep, On-chip generation and detection of electrochemiluminescence. Analytica Chimica Acta, 1998, 359(3), 263–267. 56. P. E. Michel, P. D. van der Wal, G. C. Fiaccabrino, N. F. de Rooij, and M. Koudelka-Hep, Reagentless sensor integrating electrodes, photodetector, and immobilized Co-substrate for electrochemiluminescence-based assays. Electroanalysis, 1999, 11(18), 1361–1367. 57. P. E. Michel, G. C. Fiaccabrino, N. F. de Rooij, and M. Koudelka-Hep, Integrated sensor for continuous flow electrochemiluminescent measurements of codeine with different ruthenium complexes. Analytica Chimica Acta, 1999, 392(2–3), 95–103. 58. E. L’Hostis, P. E. Michel, G. C. Fiaccabrino, D. J. Strike, N. F. de Rooij, and M. Koudelka-Hep, Microreactor and electrochemical detectors fabricated using Si and EPON SU-8. Sensors and Actuators, B: Chemical, 2000, 64(1–3), 156–162. 59. A. Arora, A. J. de Mello, and A. Manz, Sub-microliter electrochemiluminescence detector—a model for small volume analysis systems. Analytical Communications, 1997, 34, 393–395. 60. C. A. Marquette and L. J. Blum, Conducting elastomer surface texturing: a path to electrode spotting: application
17
to the biochip production. Biosensors and Bioelectronics, 2004, 20(2), 197–203. 61. E. N. Glezer, K. Jonhson, J. D. Debad, M. Tsionsky, B. Jeffrey-Coker, C. Clinton, A. Kinshbaugh, J. K. Leland, M. Billadeau, S. Leytner, S. Altunata, G. B. Sigal, J. L. Wilbur, H. A. Biebuyck, and J. N. Wholstader, Electrochemiluminescent Microarrays: A New Tool for Drug Discovery and Life Science Research, In: 224th ACS National Meeting, Boston, 2002 August 18–22.
FURTHER READING M. L. Calvo-Munoz, A. Dupont-Filliard, M. Billon, S. Guillerez, G. Bidan, C. Marquette, and L. Blum, Detection of DNA hybridization by ABEI electrochemiluminescence in DNA-chip compatible assembly. Bioelectrochemistry, 2005, 66(1–2), 139–143. F. Patolsky, Y. Weizmann, and I. Willner, Redox-active nucleic-acid replica for the amplified bioelectrocatalytic detection of viral DNA. Journal of the American Chemical Society, 2002, 124(5), 770–772.
32 Dual Polarization Interferometry: A Real-Time Optical Technique for Measuring (Bio)molecular Orientation, Structure and Function at the Solid/Liquid Interface Graham H. Cross,1 Neville J. Freeman2 and Marcus J. Swann2 1
Department of Physics, Durham University, Durham, UK and 2 Farfield Scientific Ltd., Crewe, UK
1 INTRODUCTION
The challenging task of understanding and measuring the function of proteins and other biological molecules in all their structural and environmental complexity is one that has spawned a myriad of different techniques in the field of biological sciences. These span those providing exquisite detail in terms of molecular structure of a static system, such as X-ray crystallography or neutron reflection, to those providing dynamic measurements of protein function such as protein–protein interaction kinetics measured by some of the biosensing techniques covered in this handbook. Such measurements provide pieces of a jigsaw puzzle which need to be combined with others to provide a full picture. Of course, the function of any (bio)molecule is critically dependent on its environment, structure, and molecular arrangement. As such the ability to provide information linking these different areas is of fundamental interest.
Here we look at one such technique, dual polarization interferometry (DPI). This optical, surface analytical technique provides a multiparametric measurement of molecules at a surface to give information on molecular dimension (layer thickness) and packing (layer refractive index (RI), density) and surface loading and stoichiometry (mass). This combines the analytical nature of neutron reflection with the real-time, bench-top accessibility associated with biosensors and can be used to provide a link between a molecule’s structure and its function. This chapter puts DPI in the context of other multiparametric optical techniques, outlines the principles behind the technology and its implementation together with some simple examples to provide validation of the measurement. The basic experimental and data analysis methods are covered together with examples of areas of application of the technology specifically where structural information complements more conventional “mass” dependent measurement.
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
The growth, metabolism, and replication of cells, the basic building blocks of life, depend very heavily on proteins. Proteins, while expressed within the cells themselves, perform a vast range of functions of both an intracellular and intercellular nature. The profile of expressed proteins within a cell depends upon the metabolic status of the cell, its age, and its local environment. Understanding the role that proteins play in the status of the cell is crucial to the understanding of the diseased state and is therefore of great importance within medical and pharmaceutical studies of disease. In order to understand the complex nature of protein function and the many different roles a protein may have, real-time measurements of proteins and their behavior is required. The structure of a protein is determined by its primary, secondary, and tertiary structures, the latter two being noncovalent in nature and a range of interactions contribute to the final structure such as hydrogen bonding, electrostatic interactions, and dispersion forces. These interactions occur between adjacent amino acid groups and may be mediated or altered by solvation, the ionic strength of the solution, and a range of other environmental factors such as pH or temperature. Measurement of the dimensions of a protein layer as a function of these variables can probe these factors, and a particular protein’s sensitivity. The dynamic nature of the secondary and tertiary structures of proteins enables them to undergo structural changes in response to stimuli which are often related to their functional roles. Structurally distinct regions within a protein are often associated with specific functions and these structures may be conserved to undertake similar functions across a range of different proteins. Finally, these individual peptides often interact with other peptides to perform specific functions or to provide structural integrity and this final structure of the protein is known as the quaternary structure. The shape of the protein, especially its external surface often provides pockets or clefts, which offer specific binding sites for small molecules or other proteins. Interaction with these sites is often described as “specific binding” and is associated with the activation or regulation of the activity of the protein. Probing the dimensional aspects of these interactions with other proteins, peptides, and other
ligands and small molecules can provide information about the specific nature of the interactions. For example, determining how the layer structure of an oriented immobilized protein changes on binding its partner can indicate the location on the protein at which the binding site resides. Indeed for protein–small molecule complexes, interactions can be dominated by structural or conformational changes. This is particularly relevant to the pharmaceutical industry. Small molecules may interact in a range of ways with a protein, dependent on the nature and number of the binding sites. Determining a structural signature for a small molecule binding can differentiate between different modes of interaction. These may include molecules binding to the same site with different degrees of specificity, whether they promote a specific conformational change or not and those binding to unrelated sites. The behavior of proteins at interfaces (e.g., surfaces) is of particular relevance. The first aspect of this, is that many biologically important processes are interfacial in nature, with for example, the interaction of molecules at membrane surfaces whether above, within, or across the membrane bilayer. Lipid layers and vesicles can be immobilized and quantified and their interactions then probed through both mass and structural changes. The second aspect relates to the many of the ways we use proteins, many of which require us either to immobilize them or to prevent them from being immobilized at a surface. So following the structural evolution during the construction of a protein surface to be used for example, as a diagnostic test can help understand the factors influencing the functioning or otherwise of the proteins within the immobilized assembly. Alternatively, the detailed characterization of a surface and its interaction with proteins can be a valuable tool to understand the mechanisms of biofouling or protein resistant surfaces. 2 TECHNOLOGY 2.1
Overview of Optical Techniques
Spectroscopic ellipsometry is the most familiar method by which the optogeometrical properties of thin films may be deduced.1,2 This technique analyzes the state of polarization of light reflecting from multilayer reflective samples and
DUAL POLARIZATION INTERFEROMETRY
uses the laws of electromagnetism (formulated as Maxwell’s equations applied to reflection and refraction at the layer interfaces) to resolve the layer thicknesses and RIs of the layers. The analysis requires the experimenter to choose a specific structural model from which the corresponding expected data may be calculated and to which the observed data may be compared via an error minimization process. There is, alongside this, an increasing interest in optical-guided wave techniques that are able to determine the average thickness and density of ultrathin layers that bind to an optical waveguide surface.3–6 In any optical waveguide structure, the light field is not wholly confined within the physical boundaries of the guiding medium but, rather, decays exponentially away from the boundaries. This part of the optical field is known as the evanescent (vanishing) field. If a layer is added to or removed from the original waveguide surface or an existing layer changes its thickness or density the position of the boundary at which the light begins this exponential decay is altered. Such changes alter the speed of propagation (“phase velocity”) of the whole field. Here, the measured changes to the phase velocity of optical fields probing the layers are interpreted through application of Maxwell’s equations to guided optical fields in multilayer samples. These developments follow on from the widely used and commercialized evanescent wave methods, notably surface plasmon resonance (SPR) spectroscopy,7,8 where the information provided is limited to the change in a mass-related parameter (resonance angle shift) at the optical surface as a function of time. Thus the measurement of the kinetics of molecular binding events (association and dissociation constants) and concentration measurements have been the two main applications for evanescent wave sensors until recently. While there is still a great need for such simple information obtained with the speed and sensitivity that evanescent wave methods offer, the variety of examples given in the preceding text require a greater degree of information to distinguish different possible behavior of the proteins being studied. Structural changes have been inferred from single parameter measurements, as for example, the observed “mass” response will be enhanced or suppressed by the effect of the conformational change. This can only be interpreted however
3
with a view to what the expected mass change should have been and there are too many potential contributions from other factors to make this a viable approach in general. Most instances can be attributed to effects such as binding-promoted surfactant adsorption/desorption or ion motion due to pI changes of the protein or related immobilization matrix effects. In short, a single measurement by evanescent wave techniques cannot be interpreted for layer structure with any confidence without some additional information. With low noise instrumentation, evanescent wave techniques have the capability to resolve layer dimensional changes at the sub-angstrom level and RI increments of 10−6 . Here we describe methods that use an additional independent evanescent wave measurement to remove these ambiguities (dual mode evanescent wave spectroscopy (DMEWS)). With dual mode evanescent wave methods, the capability inherent in spectroscopic ellipsometry is shared by the guided wave techniques, the layer model assumptions are similar but the methods of data analysis are distinctly different. While in ellipsometry the data is analyzed using a multiparameter model fit using the χ 2 statistic and its minimization, in guided wave methods the phase velocity change data is carried through directly to the layer parameters required to produce such a change. It will be important to remember however that all methods require the correct choice of model in order to be successful. We will concentrate on the experimental determination of the two optogeometrical parameters (average RI and thickness) of a uniform, isotropic thin film: In ellipsometry, the corresponding model system is a “three-phase” model.1,2 In an optical waveguide measurement we need two independent sources of experimental data. These might comprise a data pair taken from two evanescent fields at different optical wavelengths or, to reduce the uncertainty that optical dispersion might introduce two orthogonally polarized fields at a fixed wavelength. Although there are developments possible with the former, we will concentrate on the application of the latter method. Here one can distinguish between methods that rely on the measurement of modal phase matching conditions (resonance techniques) and reflectometry, and that implemented in DPI which relies on the interferometric detection of optical field phase changes.
4
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
In the coupled plasmon waveguide resonance (CPWR) technique9 the waveguide structure comprises a Kretschman-type SPR arrangement10 but with the addition of a thin low RI dielectric film (silicon dioxide) above the silver film. As in conventional SPR the interface between the metal and the emergent dielectric confines an optical field with transverse magnetic (TM) polarization. Light incident through the coupling prism is therefore polarized in the plane of incidence (“p” polarized.) However, this simple layer addition allows the combined system to also confine a transverse electric (TE) field which can be coupled to by incident light polarized orthogonal to the plane of incidence (“s” polarized). Each of the two modes has an electric field component (the evanescent field) whose amplitude decays exponentially, but at different rates, beyond the dielectric layer into the medium of interest. In experiments the incidence angle is scanned with high angular resolution and the reflected light intensity is measured around a range of angles spanning the angles of minimum reflection. At these coupling angles, the tangential phase velocity of light in the prism matches that of the waveguide field and the real part of the effective index of the structure may be related to the angle. Thus as layers are added to the system this “phase matching” condition changes and the resonance angle changes. From a full analysis of data taken from the reflection spectrum it is, in principle, possible to use transfer matrix methods11 and multiparameter fitting procedures to give the layer thickness and the RI and extinction coefficient for each of the two polarizations. One important example where this has been applied is to lipid bilayer systems where the optical properties normal to the bilayer are distinctly different to those in the layer plane.5 In optical waveguide light mode spectroscopy12,13 (OWLS), the waveguide structure is a simple slab waveguide into which light may be coupled using a grating to achieve phase matching. As in all such coupling methods, light entering the structure must have a tangential phase velocity (determined by coupling angle) equal to that of the optical waveguide field. The dielectric slab waveguide is designed to support fields (or “modes” from hereon) of each polarization, TE, and TM. At an appropriate angle of incident light when the phase matching condition is achieved,
the waveguide mode is excited and propagates light along to the waveguide end facet where it is collected by a photodiode and recorded. These angles will change according to changes above the original surface as described earlier. Since the TE and TM resonance angles are separated from each other, it is difficult to implement a system that can track the resonance minima of two such modes in real time. Thus for real-time studies the technique works by following the shift in resonance angle for a single polarization. Kinetic studies are therefore possible. The layer optogeometrical properties can however be extracted after the layer has bound by measuring the resonance shifts for both polarizations individually. These will provide the effective index changes for TE and TM polarized modes and using numerical solution of the eigenvalue equations for the waveguide system and making the assumption that the only unknown layer is uniform and isotropic its thickness and index can be found. In cases where this assumption is not strictly valid the thickness and RI data will have inaccuracies and the extent to which these might be introduced in OWLS and other studies has been examined by Mann.14
2.2
Dual Polarization Interferometry (DPI)
Interferometers comprise devices with two optical paths that detect the change in optical path length experienced by an optical field passing through the sensing path of the interferometer. Sensitivity is governed by, among other things, the interaction length and the signal-to-noise ratio of the detection scheme. Typically, integrated optical interferometers are configured in the Mach–Zehnder format15 by creating channel waveguiding regions in the top surface of an optical dielectric stack. DPI uses a much-simplified interferometer based on slab waveguides, with the reference slab waveguide buried beneath the sensing waveguide simply as part of the multilayer fabrication process (see Figure 1a). Coherent light broadly illuminates the stack end facet, exciting all possible modes (guided and radiation) in the structure but only the guided modes propagate more than 50 µm or so along the path.16 Upon exiting the structure (after 20 mm or so depending on sample length) the light from the two modes diffracts into free space. Because the
DUAL POLARIZATION INTERFEROMETRY
waveguides are so close together (only ∼4 µm apart) the diffracted wavefront generates the wellknown pattern of Young’s interference fringes in the far-field only a few millimeters from the end facet. Changes in the optical properties of the sensing waveguide (e.g., layer changes) translate into variations in the phase of the sensing mode field and then into interference pattern intensity shifts, as captured by a high-resolution camera. The optical tolerances are so forgiving that macroscopic movements of the input coupling beam on the order of hundreds of micrometers cause no change in the interference pattern. These loose tolerances
5
allow the stack to be inserted and removed from the optical train without precision alignment, an essential characteristic for a disposable measurement platform. The waveguide stack is designed so that both measurement and reference arm support single modes in both TE and TM polarizations, enabling two optical phase change measurements to be made and the polarization of the light is alternately switched on a 2-ms cycle using a ferroelectric liquid crystal rotator. Direct measurement of the phase change is obtained by continuously monitoring the relative phase position of the fringe pattern by performing Camera
Biological layer Fringe positions move as the biological layer is modified or altered
Sensing waveguide
Laser (light source)
Reference waveguide Output from both waveguides combine to generate interference fringes in the far field
(a) Response
Sensor response
Time
x
Polarization 1
x
Polarization 2
t Maxwell’s equations
Equivalent homogeneous layer Thickness
Polarization 2 Polarization 1 (b)
Absolute RI (density)
Figure 1. (a) Schematic diagram of a laser illuminated measurement chip, with fringes produced on the camera as the propagated light diffracts out of the end of the waveguides and interferes. [Reprinted with permission M J Swann et al., copyright 2004, Elsevier.] (b) Schematic diagram of the process of conversion of measured phase data to calculated layer thickness and refractive-index values. [Reprinted with permission from Farfield Scientific Ltd.]
6
TRANSDUCER TECHNOLOGIES FOR BIOSENSORS
a Fourier transformation relating intensity to position. The application of a standard transfer matrix approach provides evaluation of the guided modes for the structure. This allows inclusion of an arbitrary number of layers in a model, each of which is represented by its own layer matrix. Calibration of the bare chip using two liquids of known RIs as upper layers to generate TE and TM phase changes provides an initial structure (waveguide layer thicknesses and indices.) A successive approximation method provides the optimum refractive-index value and the thickness of the waveguide layer. Subsequent phase changes are analyzed similarly by including a model uniform isotropic thin layer on the surface of the waveguide (see Figure 1b) and resolving its properties (index and thickness). This calculated layer corresponds to the single homogeneous layer equivalent to the layer deposited experimentally. The layer density can be determined from the RI (see subsequent text) and knowing size and density, one can also trivially calculate the total mass, surface concentration, number of protein molecules, molecular footprint, and other useful parameters. Because experimentally, layers of different proteins or other materials can be deposited sequentially, it can sometimes be the case that a layer has been deposited which is not expected to change during the ongoing course of the experiment. This layer can then be incorporated as a fixed layer in the optical multilayer structure, and subsequent changes calculated as a new layer on top. Where this analysis does not produce realistic results, this can indicate the subsequent layer is either penetrating or expanding the sublayer. Where the assumption of a uniform isotropic layer is not valid alternative approaches can be used and have been applied successfully, for example, to lipid bilayer systems.
2.3
Experimental Setup
The measurement chip is located on a thermal block which is held at the preset temperature of between 10 and 40 ◦ C with a stability of ±2 mK. The illumination is provided by a 632.8 nm He–Ne laser. Running buffer or samples are exposed to the two fluidic channels of the measurement chip via a
fluidic manifold supplied by an high pressure liquid chromatography (HPLC)-type sample injection system driven by a syringe pump. The fluidics can be configured to allow a variety of sample volumes and viscosities to be introduced, and the sample flow may be maintained or stopped to allow samples to incubate on the chip surface. While experiments are generally undertaken in aqueous phase, most water miscible solvents can be used. The RI range for the running buffer is 1–1.49. For thin layers however there is no upper limit on the layer RI. Measurement chips are calibrated at the beginning of a measurement with solutions of known RI, typically 80% ethanol/water and pure water. This allows the sensitivity of the phase response and hence waveguide parameters to be calculated. Measurement of the phase change between the water and the running buffer also allows the RI of the buffer to be checked. These values are then used for the chip structure in the subsequent data analysis.
2.3.1 Typical Experimental Approach
Experiments usually start from either an unmodified silicon oxynitride chip, or using one that has been chemically modified with an alkoxysilane (mono)layer. Typical surface functionalities are amine, thiol, or hydrophobic—either trimethylsilane or octadecylsilane, from which a variety of immobilization methodologies may be employed. Proteins may be physisorbed, either via hydrophobic, hydrophilic, or electrostatic interaction, or can be coupled to the surface with a cross-linker. This may be direct covalent coupling with a bifunctional cross-linker, or via an intermediate coupling layer, such as via biotinylation of an amine chip, coupling of streptavidin followed by binding of a biotinylated protein. As much of the protein layer immobilization as possible is generally undertaken on the instrument this ensures that all layers can be properly quantified and the final surface structure can be accurately calculated from the measured phase shift starting from the initial bare or modified surface. Once the layer thickness and RI have been measured, values for the protein layer density, mass, molecular footprint and so on, can be calculated.
DUAL POLARIZATION INTERFEROMETRY
2.3.2 Typical System Performance
Experimentally observed errors are somewhat dependent on the nature of the layers being measured. For a typical protein monolayer the errors are shown in Table 1. In thickness the resolution equates to less than one-tenth of an atomic bond length. Where layers are very diffuse, or the layer thickness approaches the 1/e2 extent of the original evanescent field (∼100 nm for an aqueous system with layer RI < 1.40) these may be larger.
2.4
Data Analysis
The RI increments of proteins are quite consistent1,17,18 with typical values in the region of 0.186 g cm−3 and it is therefore possible to determine the mass of material deposited on the sensor surface in a similar way to that of de Feijter19 using equations (1) and (2) ρL =
ρp (nL − ns ) (np − ns )
mL = ρ L τ L
(1) (2)
where ρL is the adsorbed layer density, ρp is the protein density, nL is the adsorbed layer RI, np is the protein RI, ns is the solution (bulk) RI, mL is the mass loading per unit area, and τL is the adsorbed layer thickness. From the mass loading it is straightforward to calculate the area per molecule according to equation (3) Mw A= (3) N a mL where A is the area per molecule, Mw is the protein molecular weight, and Na is Avogadro’s number.
7
By using the measured values for the RI of the bulk solution the volume fraction of the layer occupied by protein (φp ) can also be calculated using equation (4): φp =
(n2L − n2s ) (n2p − n2s )
After calculation of the parameters described in the preceding text, it is possible to draw inferences not only regarding the gross structures of the deposited protein layers but also the likely orientation of the protein molecules within the layers.
2.5
Measurement Validation and Model Examples
Validation of the measurements has been made using a variety of methods, generally however the use of measurements of protein dimensions as a validation methodology should be used with caution, as proteins may conform significantly onto solid surfaces and this degree of deformation can itself be a function of protein surface loading. Comparative measurements of protein systems against other techniques have been made. cf. C receptor protein as a comparison with atomic force microscopy (AFM),20 bovine serum albumin (BSA) structures as compared with neutron reflection21 or streptavidin immobilized on a biotinylated surface22 and an antibody oriented on a protein G surface (shown subsequent text) compared with its X-ray crystallographic dimensions. In the two latter cases, where the comparative technique (X-ray crystallography) is not a surface method, the proteins are captured via a specific interaction which means that both are specifically
Table 1. Typical percentage CV in phase, temperature, and “resolved” layer values from DPI measurements
Parameter TM phase measurement Mass (ng mm−2 ) Thickness Density Temperature
Typical layer value
Accuracy (±)
8 rad 2 ng mm−2 5 nm 0.4 g cm−3 20 ◦ C
0.25% 8.8) the trigonal form can react with OH− to form the more stable negatively charged tetrahedral state, which binds cis-diols more readily. Consequently, the requirement of conventional boronates for pH values >8.8 to effect binding of glucose presents a challenge for monitoring this metabolite in biological samples. However, more recently, new boronic acid derivatives that bind glucose at physiological pH values have been developed and these have been used in sensor hologram fabrication. Acrylamide-based hydrogels containing the monomer 3-acrylamidophenylboronic acid (3APB) have been fabricated and the chemical composition of the films optimized for glucose detection using embedded reflection holograms.35,36 Maximum sensitivity was observed at a functional monomer concentration of 20 mol%. The sensor holograms display a monotonic redshift in diffraction wavelength as a function of glucose concentration across the normal glucose concentration range (2–10 mM) at physiological pH and ionic strength values. It is believed that glucose diffuses into the holographic matrix from
the bulk medium and binds to the pendant boronic acid groups. This effectively decreases the pKa of the boronic acid–glucose complex by stabilizing the charged tetrahedral phenylboronate anion and the presence of these charged groups within the polymer generates a Donnan potential resulting in an osmotic pressure that causes the hologram to imbibe more water, swelling and thereby shifting the diffraction wavelength toward the red end of the spectrum. The reaction of glucose with boronic acids is unusual since the covalent bond formed between the two molecules is reversible in aqueous media and when the glucose is removed from the bathing medium, the hologram contracts and returns to its original diffraction wavelength.37 These observations suggest that the sensor is suitable for continuous real-time sensing of dynamic changes in glucose concentration. Current work in glucose detection is focused on assessing the performance of these sensor holograms for the detection of glucose in complex biological media and their incorporation into a suitable sensor format for continuous glucose monitoring in human subjects.38 The sensor responds to glucose in the presence of lactate (Figure 4).
4.9
Lactate
L-lactate is a metabolite generated during anaerobic metabolism and is a useful indicator in the food industry, fermentation, clinical diagnostics, and exercise performance in sports medicine. The synthetic receptors based on phenylboronates are
20 10 ∆lmax (nm)
8
Lactate
0 −10 −20 −30 −40
2
4
6 mM
8
10
12
Glucose/4mM Lactate Glucose
−50
Figure 4. Response of a 11.9 mol% 3-APB (3-acrylamidophenylboronate)/ 9.2 mol% DAPA (N -[3-(dimethylamino)propyl]acrylamide)/ 2.9 mol% MBA (N ,N ′ -Methylenebisacrylamide)/acrylamide copolymer hologram to glucose and lactate on PBS buffer pH 7.4 at 30 ◦ C.
HOLOGRAPHIC SENSORS
known to bind with bidentate chelating ligands to form 5- and 6-membered cyclic esters.38 The boronates bind saccharides and other carbohydrates as well as o-diphenols, o-hydroxy acids, dicarboxylic acids, and α-hydroxy acids such as L-lactate. Consequently, a significant challenge has been to devise a boronate analogue, which selectively binds lactate compared to glucose. Preliminary studies have demonstrated that a hologram containing a reduced concentration (5 mol%) of 3-acrylamidophenylboronate (3-APB) shows an improved sensitivity for L-lactate over glucose.39
4.10
Bacteria
There is substantial interest in use of small-scale bioprocesses (∼nanoliter to microliter volumes) for fast, reliable, and inexpensive high-throughput biology to enable both early process development studies and the screening of drug candidates for drug discovery. However, in such small volumes and in a parallel format, it is impossible to use standard industrial sensors because of their relatively large dimensions. The sensor holograms have been successfully used to monitor the metabolic products of small-scale ( PG .
into the channel. This positive capillary pressure would not be likely to push the liquid out of the PDMS channel however, because the receding contact angle would likely be less than 90◦ . Another interaction between surface tension and channel geometry is a surface tension stop as shown in Figure 8. A desirable (or unintentional) pressure difference can be sustained by an area discontinuity.
5.10
Absorptive and Wicking-driven Flow
Uptake of fluid by capillary pumping of a hydrophilic polymer is the mechanism for fluid movement in the lateral flow strip, a qualitative immunological assay in a passive handheld format, such as a home pregnancy test. Most current lateral flow strip assays are based on membranes composed of cellulose nitrate or cellulose acetate depending on the degree of protein binding desired. A similar functionality is obtainable from hydrogels, cross-linked polymer networks surrounded by an aqueous solution. When wetted by a solvent, the chains in the network are solvated, but do not mix due to the cross-linking, which provides an elastic restoring force to counter swelling. Hydrophilic polymers can be divided into categories based on the relaxation time of the polymer and the diffusion time of the solvent. One parameter is the Deborah number, the ratio of the rates of solvent penetration and polymer relaxation, De = λD/δ 2 , in which λ is the characteristic polymer relaxation time from swelling stresses, D is the diffusion coefficient of the solvent, and δ is the diffusional distance at time = λ.63 Case I transport (De ≪ 1) occurs when the diffusion time is much slower than the polymer relaxation time, leaving diffusion as the controlling mechanism. This is typical in nonswelling systems. In case II transport (De ≫ 1), the rate limiting process is polymer relaxation. In other hydrogels,
De is on the order of 1 and the two processes occur on the same time scale, leading to anomalous transport behavior. When De is very large, this is sometimes referred to as super case II behavior. A simple description of the time-dependent swelling of a polymer is Mt /M∞ = kt n , in which Mt /M∞ is the fractional uptake (or release) of solvent normalized by the equilibrium conditions and k and n are constants dependent on solvent diffusion coefficient and type of transport process.64–66 Figure 9 plots the fractional uptake of solvent for various transport types. For diffusion-controlled (type 1) transport, n = 0.5, while for polymer relaxation controlled (type II) transport, n = 1. Anomalous transport has 0.5 < n < 1, and super type II has n > 1. The uptake of solvent and solute in swelling polymeric systems can be numerically modeled by the species conservation equation67,68 if the solvent velocity u is known from the solution of the momentum conservation equations, from Darcy’s equation for fluid flow through a porous medium and the permeability and porosity of the medium, or from experimental data to determine the constants. For example, data from experiment with 5 µm pore size nitrocellulose polyether sulfone exhibits Type I transport behavior69 in which wetting speed is proportional to the square root of wetting time. Wetting time data for a typical lateral flow strip based on Whatman nitrocellulose “Purabind” is shown in Figure 10.
n > 1, Super II n = 1, Type II 0.5 < n < 1, Anomalous n = 0.5, Type I
5 4.5 4 3.5
Wt /W∞
16
3 2.5 2 1.5 1 0.5 0 0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time
Figure 9. Fractional uptake of solvent versus time for various transport types.
INTRODUCTION TO MICROFLUIDIC TECHNIQUES
Plasma Water
400
Wetting time (s)
350 300 250 200 150 100 50 0 0
2.5
5
7.5
10
12.5
Nominal pore size (µm)
Figure 10. Wetting times for 45-mm-long lateral flow strip of Whatcom nitrocellulose “Purabind” for water at 25 ◦ C. The data is also scaled by viscosity ratio to predict behavior with blood plasma as solvent.
6 USING APPROPRIATE STRATEGIES FOR DEVELOPMENT OF MICROFLUIDIC DEVICES
Microfluidics brings unique capabilities to assays and sensors in bioscience applications, such as: lower cost assays, more rapid results, smaller reagent volumes, and less hazardous waste. But fluids in the microscale do not display the familiar behavior of fluids in the human-scale world. Understanding the physics of microfluidics is one key to successful development of microdevices. It is a nearly impossible task to replicate macroscale fluid-management strategies in the microscale and far more fruitful instead to leverage its unique behaviors and inherent capabilities. In the microscale, inertial and gravitational forces are often much less important than surface tension and viscous forces. Alternatives to pressure-driven flows exist and are robust and reliable, such as electrokinetics and wicking. Dimensionless parameters can be utilized to show which physics will be dominant in your device. Then mathematical modeling becomes an effective way to estimate performance in the deterministic world of microfluidics. In this way, a new microfluidic device has the potential to become a novel solution in its application area.
17
REFERENCES 1. B. H. Weigl, R. L. Bardell, and C. R. Cabrera, Lab-ona-chip for drug development. Advanced Drug Delivery Reviews (invited review article), 2003, 55(3), 349–377. 2. E. Cabuz, J. Schwichtenberg, B. DeMers, E. Satren, A. Padmanabhan, and C. Cabuz, MEMS-Based Flow Controller for Flow Cytometry Honeywell International , 2002 http://content.honeywell.com/sensing/solutions/markets/ medical/88PAD HH 2002 twopage fullpaper.pdf. 3. U. Bonne et al., Microsensor Housing, US Patent 6,322,24, Issued November 27, 2001. 4. http://www.darpa.mil/MTO/bioflips/presentations/2001-1/ index.html, (DARPA BioFlip Program), 2001. 5. P. Belgrader, M. Okuzumi, F. Pourahmadi, D. Borkholder, and M. A. Northrup, A microfluidic cartridge to prepare spores for PCR analysis. Biosensors and Bioelectronics, 2000, 14, 849–852. 6. B. H. Weigl, R. L. Bardell, T. Schulte, D. C. Cullen, and J. Demas, Modeling of microscale processes speeds development and enhances performance of medical diagnostic product. In Vitro Diagnostics Technology, 10, June 2004, 41–49, invited. 7. B. H. Weigl, A. Holobar, W. Trettnak, and O. S. Wolfbeis, Optical triple sensor for measuring pH, oxygen and carbon dioxide. Journal of Biotechnology, 1994, 32, 127–138. 8. B. H. Weigl, New Assays and Separations Based on Laminar Fluid Diffusion Interfaces—Results From Field Trials for Cell Analysis, HTP Screening, and Medical Diagnostics, in Micro Total Analysis Systems 2002, D. J. Harrison and A. Van den Berg (eds), Kluwer Academic Publishers, Dordrecht, 2002. 9. P. Mitchell, Microfluidics—downsizing large-scale biology. Nature Biotechnology, 2001, 19, 717–721. 10. J. Benditt, Ten Emerging Technologies that will Change the World, MIT Technology Review: January/February, 2001. 11. L. J. Kricka, Microchips, microarrays, biochips and nanochips: personal laboratories for the 21st century. Clinica Chimica Acta, 2001, 307(1–2), 219–223. 12. Frost and Sullivan, U.S. Point-Of-Care Retail Diagnostics Markets, 3/15/2003, available from hyperlink http://www.marketresearch.com (this is a market research report), 2003. 13. R. L. Chien and J. W. Parce, Multiport flow-control system for lab-on-a-chip microfluidic devices. Fresenius Journal of Analytical Chemistry, 2001, 371(2), 106–111, (eng). 14. L. Bousse, C. Cohen, T. Nikiforov, A. Chow, A. R. Kopf-Sill, R. Dubrow, and J. W. Parce, Electrokinetically controlled microfluidic analysis systems. Annual Review of Biophysics and Biomolecular Structure, 2000, 29, 155–181. 15. M. T. Cronin, T. Boone, A. P. Sassi, H. Tan, Q. Xue, S. J. Williams, A. J. Ricco, and H. H. Hooper, Plastic microfluidic systems for high throughput genomic analysis and drug screening. Journal of the Association for Laboratory Automation, 2001, 6(1), 74–78. 16. M. T. Cronin, M. Pho, D. Dutta, F. Frueh, L. Schwarcz, and T. Brennan, Utilization of new technologies in drug trials and discovery. Drug Metabolism and Disposition, 2001, 29(4), 586–590.
18
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
17. A. J. Ricco, T. D. Boone, Z. H. Fan, I. Gibbons, T. Matray, S. Singh, H. Tan, T. Tian, and S. J. Williams, Application of disposable plastic microfluidic device arrays with customized chemistries to multiplexed biochemical assays. Biochemical Society Transactions, 2002 30(2), 73–78. 18. W. W. Weber and M. T. Cronin, Pharmacogenetic Testing, in Encyclopedia of Analytical Chemistry, R. A. Meyers (ed), John Wiley & Sons, Sussex, 2000, pp. 1506–1531. 19. M. T. Boyce-Jacino, J. E. Reynolds, T. T. Nikiforov, Y. H. Rogers, C. Saville, T. C. McIntosh, P. Goelet, and M. R. Knapp, High volume molecular genetic identification of single nucleotide polymorphisms using genetic bit analysis: application to human genetic disease. American Journal of Human Genetics, 1994, 55(3), 18–22. 20. B. H. Weigl, R. L. Bardell, N. Kesler, and C. J. Morris, Lab-on-a-chip sample preparation using laminar fluid diffusion interfaces—computational fluid dynamics model results and fluidic verification experiments. Fresenius Journal of Analytical Chemistry, 2001, 371(2), 97–105. 21. A. Y. Fu, C. Spence, A. Scherer, F. H. Arnold, and S. R. Quake, A microfabricated fluorescence-activated cell sorter. Nature Biotechnology, 1999, 17, 1109–1111. 22. J. W. Hong and S. R. Quake, Integrated nanoliter systems. Nature Biotechnology, 2003, 21, 1179–1183. 23. M. T. Taylor, P. Belgrader, R. Joshi, G. A. Kintz, and M. A. Northrup, Fully automated sample preparation for pathogen detection performed in a microfluidic cassette. Micro Total Analysis Systems, 2001, 670–672. 24. M. A. Northrup, L. Christel, W. A. McMillan, K. Petersen, F. Pourahmadi, L. Western, and S. Young. A New Generation of PCR Instruments and Nucleic Acid Concentration Systems, PCR Applications—Protocols for Functional Genomics, 1999, 105–125. 25. T. Thorsen, S. J. Maerkl, and S. R. Quake, Microfluidic large scale integration. Science, 2002, 298, 580–584. 26. P. Jandik, B. H. Weigl, N. Kesler, J. Cheng, C. J. Morris, T. Schulte, and N. Avdalovic, Initial study of using laminar fluid diffusion interface for sample preparation in HPLC. Journal of Chromatography A, 2002, 954, 33–40. 27. M. Gad-el-Hak, The fluid mechanics of microdevices—the freeman scholar lecture. Journal of Fluids EngineeringTransactions of the ASME, 1999, 121, 5–33. 28. O. Reynolds, An experimental investigation of the circumstances which determine whether the motion of water in parallel channels shall be direct or sinuous and of the law of resistance in parallel channels. Philosophical Transactions of the Royal Society of London, 1883, 174, 935–982. 29. P. Wu and W. A. Little, Measurement of friction factors for the flow of gases in very fine channels used for microminiature Joule-Thomson refrigerators. Cryogenics, 1983, 23, 273–277. 30. J. O. Hinze, Turbulence, 2nd Edn, McGraw-Hill, 1987, pp. 1–4. 31. A. Hatch, A. E. Kamholz, K. R. Hawkins, M. S. Munson, E. A. Schilling, B. H. Weigl, and P. Yager, A rapid diffusion immunoassay in a t-sensor. Nature Biotechnology, 2001, 19(5), 461–465. 32. D. J. Beebe and G. M. Walker, An evaporation-based microfluidic sample concentration method. Lab on a Chip, 2002, 2, 57–61.
33. National Academy of Science, Summary, the Impact of Materials—from Research to Manufacturing, http://www. national-academies.org. 2002. 34. A. Folch and M. Toner, Microengineering of cellular interactions. Annual Review of Biomedical Engineering, 2000, 2, 227. 35. J. P. Brody, P. Yager, R. Goldstein, and R. H. Austin, Biotechnology at low Reynolds numbers. Biophysical Journal, 1996, 71, 3430–3441. 36. R. H. Carlson, J. P. Brody, S. Chan, C. Gabel, J. Winkleman, and R. H. Austin, Self-sorting of white blood cells in a lattice. Physical Review Letters, 1997, 79, 2149–2152. 37. E. Altendorf, E. Iverson, D. Schutte, B. H. Weigl, T. Osborn, R. Sabeti, and P. Yager, Optical Flow Cytometry Utilizing Microfabricated Silicon Flow Channels, in Advanced Techniques in Analytic Cytology (BIOS 96) SPIE Proceedings, SPIE (formerly the International Society for Optical Engineering), Vol. 2678. 38. T. A. J. Duke and R. H. Austin, Microfabricated sieve for the continuous sorting of macromolecules. Physical Review Letters, 1998, 80, 1552–1555. 39. J. B. Knight, A. Vishwanath, J. P. Brody, and R. H. Austin, Hydrodynamic focusing on a silicon chip: mixing nanoliters in microseconds. Physical Review Letters, 1998, 80, 3863–3866. 40. R. H. Austin, J. P. Brody, E. C. Cox, T. Duke, and W. Volkmuth, Stretch genes: aligning single molecules of DNA. Physics Today, 1997, 32–36. 41. J. S. Kuo, Interfacing Chip-Based Nanofluidic-Systems to Surface Desorption Mass Spectrometry, JIN Reports, Pacific Northwest National Laboratory, 2003, http://www. pnl.gov/nano/institute/2003reports/. 42. T.-C. Kuo, D. M. Cannon Jr, M. A. Shannon, J. V. Sweedler, and P. W. Bohn, Hybrid three-dimensional nanofluidic/microfluidic devices using molecular gates. Sensors and Actuators A, 2003, 102/3, 223–233. 43. Y. Gogotsi, C. M. Megaridis, H. Bau, J.-C. Bradley and P. Koumoutsakos, Carbon Nanopipes for Nanofluidic Devices and In-situ Fluid Studies, In: NSF Nanoscale Science and Engineering Grantees Conference, National Science Foundation, Arlington, Virginia, 2003, Dec 16–18. 44. L. Wanli, J. O. Tegenfeldt, L. Chen, R. H. Austin, S. Y. Chou, P. A. Kohl, J. Krotine, and J. C. Sturm, Sacrificial polymers for nanofluidic channels in biological applications. Nanotechnology, 2003, 14, 578–583. 45. R. L., Panton, Incompressible Flow, John Wiley & Sons, 1984, p. 154. 46. F. M. White, Viscous Fluid Flow, 2nd Edn, John Wiley & Sons, 1991, p. 120. 47. O. C. Jones Jr, An improvement in the calculation of turbulent friction in rectangular ducts. Journal of Fluids Engineering-Transactions of the ASME, 1976, 98, 173–181. 48. N. Malmstadt, P. Yager, A. S. Hoffman, and P. S. Stayton, A smart microfluidic affinity chromatography matrix composed of poly(N-isopropylacrylamide)—coated beads. Analytical Chemistry, 2003, 75, 2943–2949. 49. G. H. Seong, J. Heo, and R. M. Crooks, Measurement of enzyme kinetics using a continuous-flow microfluidic system. Analytical Chemistry, 2003, 75, 3161–3167.
INTRODUCTION TO MICROFLUIDIC TECHNIQUES 50. M. J. MacDonald, C. F. Chu, P. P. Guilloit, and K. M. Ng, A generalized Blake-Kozeny equation for multisized spherical particles. AIChE Journal, 1991, 37, 1583–1588. 51. A. Manz, C. S. Effenhauser, N. Burggraf, D. J. Harrison, K. Seiler, and K. Fluri, Electroosmotic pumping and electrophoretic separations for miniaturized chemical analysis systems. Journal of Micromechanics and Microengineering, 1994, 4, 257–265. 52. T. S. Stevens and H. J. Cortes, Electroosmotic propulsion of eluent through silica-based chromatographic media. Analytical Chemistry, 1983, 55, 1365. 53. G. V. Sherbert, The Biophysical Characterisation of the Cell Surface, Academic Press, 1978. 54. R. C. Boltz and T. Y. Miller, A Citrate Buffer System for Isoelectric Focusing and Electrophoresis of Living Mammalian Cells, in Electrophoresis ’78, N. Catsimpoolas (ed), Elsevier North Holland Biomedical Press (New York), 1978, 345–355. 55. L. E. Locascio, C. E. Perso, and C. S. Lee, Measurement of electroosmotic flow in plastic imprinted microfluid devices and the effect of protein adsorption on flow rate. Journal of Chromatography A, 1999, 857, 275–284. 56. S. L. R. Barker, D. Ross, M. J. Tarlov, M. Gaitan, and L. E. Locascio, Control of flow direction in microfluidic devices with polyelectrolyte multilayers. Analytical Chemistry, 2000, 99A, 5925–5929. 57. Y. Liu, J. C. Fanguy, J. M. Bledsoe, and C. S. Henry, Dynamic coating using polyelectrolyte multilayers for chemical control of electroosmotic flow in capillary electrophoresis microchips. Analytical Chemistry, 2000, 72, 5939–5944. 58. S. Wang, C. E. Perso, and M. D. Morris, Effects of alkaline hydrolysis and dynamic coating on the electroosmotic flow in polymeric microfabricated channels. Analytical Chemistry, 2000, 72, 1704–1706.
19
59. F. Bianchi, F. Wagner, P. Hoffmann, and H. H. Girault, Electroosmotic flow in composite microchannels and implications in microcapillary electrophoresis systems. Analytical Chemistry, 2001, 73, 829–836. 60. R. F. Probstein, Physicochemical Hydrodynamics, 2nd Edn, Wiley-Interscience, 2003, 305–306. 61. P. M. Ball, Spreading it about. Nature (London), 1989, 338, 624. 62. K. Kaski, Europhysics News, 1995, 26, 23. 63. C. S. Brazel and N. A. Peppas, Dimensionless analysis of swelling of hydrophilic glassy polymers with subsequent drug release from relaxing structures. Biomaterials, 1999, 20, 721–732. 64. N. A. Peppas and R. W. Korsmeyer, Dynamically Swelling Hydrogels in Controlled Release Applications, in Hydrogels in Medicine and Pharmacy, N. A. Peppas (ed), CRC Press Boca Raton, 1987, Vol 3. 65. A. R. Berens and H. B. Hopfenberg, Diffusion and relaxation in glassy polymer powders: 2. Separation of diffusion and relaxation parameters. Polymer, 1978, 19(5), 489–496. 66. S. J. Kim, K. J. Lee, I. Y. Kim, and S. I. Kim, Swelling kinetics of interpenetrating polymer hydrogels composed of poly(vinyl alcohol)/chitosan. Macromolecular Science, 2003, A40(5), 501–510. 67. D. Berger and D. C. T. Pei, Drying of hygroscopic capillary porous solids: a theoretical approach. International Journal of Heat and Mass Transfer, 1973, 16, 293–302. 68. H. B. Hopfenberg and H. L. Frisch, Transport of organic micromolecules in amorphous polymers. Polymer Letters, 1969, 7, 405–409. 69. S. Krishnamoorthy, V. B. Makhijani, M. Lei, M. G. Giridharan, and T. Tisone, Computational Studies of MembraneBased Test Formats. In: Technical Proceedings 2000 International Conference on Modeling and Simulation of Microsystems, MSM , San Diego, CA, 2000, 590–593.
42 Practical Aspects of Microfluidic Devices: Moving Fluids and Building Devices Bernhard H. Weigl,1 Ron L. Bardell2 and Catherine Cabrera3 1
Department of Bioengineering, University of Washington, Seattle, WA, USA, 2 MicroPlumbers Microsciences LLC, Seattle, WA and Minneapolis, MN, USA and 3 Biosensor and Molecular Technologies, MIT Lincoln Laboratory, Lexington, MA, USA
1 MOVING FLUIDS IN A MICROFLUIDIC DEVICE
“Fluid” and “fluid movement” lie at the heart of microfluidics. Various types of fluid motion are required, including moving fluids in a continuous stream, dispensing a controlled bolus of liquid, and getting the fluid into the device from the macroscopic world in the first place. Each of these topics is discussed below. Techniques for controlling the motion of fluids in microfluidic devices can be as simple as harnessing the hydrostatic pressure of a column of liquid in a tube placed above a port in the device or as complicated as a microfabricated multistage pump. A discussion of laminar flow in general and the theoretical basis for several means of generating laminar flow in a microfluidic device can be found in Chapter 41, Introduction to Microfluidic Techniques.
1.1
be used to perform “valveless” switching to inject fixed amounts of a fluid into a flow stream.1 By controlling the voltage and duration of the applied field, the fluid volume dispensed can be precisely and reproducibly controlled (see Figure 1). Alternate approaches rely on pressure-driven flow to actuate fluid dispensation. Microscale fluid movers are small enough to be included on-chip in a microfluidic device, though the additional cost may not be justified in a single-use chip. They can be fabricated from a number of different materials, including silicon, brass, glass, and polymers, and usually function by changing volume, usually by flexing a wall. They typically have some type of valving to direct the fluid flow and are driven by electrical energy. They can be categorized in a number of different ways, by: purpose, output type, working fluid, and physical mechanism. We may call them micropumps, though not all of them conform to the usual concept of a machine, especially those in which the physical mechanism has no moving parts whatsoever.
Dispensing Discrete Volumes
Many biomedical microfluidic applications require dispensing precise volumes of fluids, perhaps to another fluid, a dried reagent pellet, or to or from the outside world. Electroosmotic flow (EOF) can
1.2
Micropumps
Depending on their purpose, micropumps can be divided into continuous-flow pumps and batchflow pumps. We can categorize micropumps by
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
MINIATURIZED, MICRO AND PARTICLE SYSTEMS + Source +
Injection bolus
Injection bolus
Source
Waste −
Waste
Analysis (a)
Analysis (b)
−
Figure 1. Electrokinetic injection of fluid sample. (a) Formation of injection bolus. (b) Switch voltage and direct bolus to analysis region.
output type, either as a displacement source or as a pressure source. A displacement source changes the volume of the pump to “displace” a chosen amount of fluid; pressure is not controlled. For accurate delivery in an open-loop control system, the pump should be paused after fluid displacement until fluidic compliances (e.g., bubbles, flexible membranes) discharge and fluid motion ceases. A pressure source can provide continuous flow in an open-loop control system, but needs an integrated flow sensor in a closed-loop feedback system to deliver a precise volume of fluid. Flow-rate sensors (as shown in Figure 2) can be fluid-resistance based (viscosity dependent) or calorimetric (specific-heat dependent) and are calibrated for the particular fluid, but can self-adjust to temperature variation if temperature sensing is integrated. If the sample is of unknown or variable composition (e.g., blood), one approach to achieve dispensed volume accuracy is to avoid pumping the actual fluid and instead pump a “pusher” fluid, such as water or FluorinertTM . Disadvantages include contamination and/or dilution of the sample as the sample/pusher interface is stretched by the parabolic velocity profile of pressure-based flow and interchange by diffusion occurs. This is ameliorated if the pusher fluid and sample are immiscible. Air can be used as a pusher fluid or as a small bubble separating pusher and sample fluids, but its compliance lengthens time constants for starting/stopping flow and, if dispensing pressure is low enough, surface tension effects (e.g., variation in capillary pressure at the liquid meniscus) may make control of fluid position difficult. Micropumps can be separated into several categories by their physical mechanism: electromechanical, pneumatic, capillary, and osmotic. Most
(a)
(b)
Figure 2. Liquid mass flow meter (a) for flow rates from 50 nl s−1 to 40 µ l s−1 based on calorimetric sensing. Sensor body with control electronics is 35 × 70 mm (Sensirion AG, Z¨urich, CH). (b) A differential pressure sensor for resistance-based flow metering. Sensor body is 6 × 7 × 8 mm (Honeywell Sensing and Control, Freeport, IL).
of these are pressure sources, but electromechanical micropumps can be designed as displacement sources and pneumatic or capillary micropumps can function as displacement sources if the backpressure opposing fluid motion is very small. Electromechanical micropumps form the most varied category, but in shear number of units the capillary type is most common due to the ubiquity of lateral flow strips.
1.3
Continuous-flow Micropumps
Electromechanical micropumps typically use electrical energy to expand and/or contract the pump body and use check valves to direct the resulting flow downstream. There are three types of check valves: fixed, passive, and active. Fixed valves have no moving parts. Their inner channel is shaped to dissipate more viscous energy in
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
upstream than in downstream flow. Turning off a fixed-valve micropump does not stop the flow; a separate valve must be actuated. A major advantage is that they can pass particles nearly the size of their throat (typically> 50 µm) without wear. Passive and active valves typically employ moving flaps that seal against the valve seat. The clearance between the seat and the flap is generally quite small and the seat and flap can be damaged by particles. The hydrodynamic forces moving the flap of a passive valve are small and can be overwhelmed by high fluid viscosity or surface tension, especially once the valve seat is wetted but the pump body contains air. An active valve can overcome surface tension at the expense of mechanical and control complexity. A robust micropump is self-priming and bubble tolerant. Its compression ratio should be large enough to pump air against the upstream pressure drop. The most common actuator for electromechanical micropumps is a piezoelectric disk bonded to the flexible wall(s) of the pump body.2–7 These are often referred to as membrane pumps. Figures 3 and 4 show examples of electromechanical micropumps using fixed valves and actuated by piezoelectric disks. Some designs indirectly couple the piezoelectric actuator to the flexible wall by means of a lever to produce proportionally larger motion8,9 (Figure 5).
3
(a)
(b) Figure 4. (a) Prototype piezoelectric-disk fixed-valve micropump in plastic with 12-mm-diameter chamber by Bartels Mikrotechnik GmbH, Dortmund. (b) Piezoelectrically activated diaphragm micropump for liquids and gases by thinXXS Microtechnology AG, Zweibruecken.
Figure 3. Piezoelectric-disk fixed-valve micropump with a 3-mm-diameter chamber and 0.114-mm-wide curly channels that operate as leaky check valves, providing a higher pressure drop in the reverse (left-to-right) than in the forward direction. Highest efficiency is achieved when the actuation frequency is set at the system resonance of the pump. [Reprinted with permission Morris and Forster14 copyright 2003, IEEE.]
Another actuation method for electromechanical micropumps is electrostatic force obtained by applying opposite charges on parallel conducting plates, one of which is rigid and the other is a flexible wall of the pump body.10,11 This electric field can cause electrokinesis of particles in the fluid being pumped, which may be an undesirable effect. Electromechanical pumps that utilize electromagnetic actuators, similar to audio speaker technology, have also been developed.12,13 Though bulkier than an electrostatic actuator or piezoelectric disk, this design offers low voltage–high current operation, which may be more appropriate for in vivo use.
4
MINIATURIZED, MICRO AND PARTICLE SYSTEMS 5 6
2 3
4 8
1
7
Figure 5. Piezoelectric-lever micropump with passive valves. Illustrated are pump membrane (1), chamber (2), inlet (3) and outlet (4) check valves, membrane stiffener block (5), top (6) and bottom (7) glass cover plates, and membrane-motion limiters (8). During operation, the membrane is pushed up against the top glass cover plate by a piezoelectric lever actuator (not shown) that pokes through the center hole and pushes against the membrane stiffener block, which is bonded to the membrane. [Reprinted with permission Maillefer et al.9 copyright 1999, IEEE.]
Acrylic pump chambers
Glass backplate
Figure 6. Illustration of actuation of electrothermal peristaltic pump. [After Grosjean.15 ] The working fluid (e.g., air) is contained in the crosshatched volumes below the membrane. Heaters in each volume operate in sequence to heat and expand the working fluid, pressing and sealing the membrane against the upper chamber surface and moving the pumped fluid from chamber to chamber. Arrows show pumped fluid direction.
Other novel electromechanical pumps have been developed, including an electrothermal peristaltic pump.15 In this design, three sequential chambers are sequentially compressed by a flexible membrane as illustrated in Figure 6. Each chamber has its own heater. The movement of the membrane is accomplished by heating air to lower its density and increase its volume. The actuation frequency is low, but it can be operated as a self-priming pump.
1.4
Batch-flow Micropumps
All the electromechanical micropumps described thus far have been continuous-flow pumps. However, it is possible to develop batch-flow pumps, based on a chamber with one fluid connection and flexible walls that are actuated by any of the
electromechanical methods described above. They can be operated reversibly with fluid intake on volume expansion and fluid output during contraction. A ubiquitous form of batch pump is the printhead of ink-jet printers (e.g., Bio-DotTM ), which are excellent at placing precise drops a sample solution on a solid or liquid surface. Not all micropumps are electromechanical. Some examples include thermally induced bubbles.16 For example, Liu et al. have developed batch-flow pneumatic micropumps,17 which use electrothermal actuation. A second design is an electrochemical pump that produces gas by electrolysis. It is more efficient, but cannot be operated reversibly. A third design uses phase-change actuation, essentially boiling the working fluid. A capillary micropump is a batch-flow, nonreversible device usually implemented as an absorbent pad of nitrocellulose as in a lateral flow strip. The fluid flow rate is controlled by the cross section of the pad, the duration of flow by its length. When a single, very-low, flow rate (e.g., 1 ml h−1 ) is desired, an osmotic pump may be appropriate (e.g., ALZET 1003D). These are essentially fluid-displacement pumps driven by chemical potential. They consist of an inner chamber containing the fluid to be pumped and an outer chamber with a rigid semipermeable outer wall and a flexible impermeable inner wall. On exposure to water, the concentrated salt solution in the outer chamber expands and the inner chamber is compressed, forcing out the pumped fluid.
1.5
Macro-to-micro Interface for Transferring Fluids
The interface between the macroscopic and microscopic worlds is a major design challenge of microfluidic devices.18,19 The simplest solution is the open reservoir with pipette access to introduce or remove sample. Examples include microarrays and the Agilent LabChip and 2100 Bioanalyzer system. Another approach is to epoxy plastic or glass tubing directly to the device, but machining holes in glass or silicon, positioning the tube over the hole, or selecting an epoxy that will bond to a silicone polymer without clogging the hole make this approach nonoptimal. A better approach is to insert the microfluidic chip into a void within a
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
5
Figure 8. Manual microsyringe pump (Stoelting Co., Wood Dale, IL). (a)
A more automated fluid delivery system would load only the sample fluid by pipette and load all reagents from off-card syringe pumps through small-bore rigid tubing. Figure 8 shows a syringe pump that, while manual, is inexpensive and accurate. Automatic computercontrolled dispensing from multiple syringe pumps is available with pump modules like those shown in Figures 9 and 10. (b)
Figure 7. (a) An illustration of a combined electromicrofluidic packaging architecture. The fluidic printed wiring board (FPWB), the largest object, has fluidic connections (the black holes) as well as electrical connections. The electromicrofluidic dual in-line package plugs into the FPWB. The microfluidic integrated circuit (MIC) is the smallest object shown.20 (b) A pneumatic–microfluidic manifold that connects air and liquid lines to inlet/outlet ports on the edge of a clear plastic laminate microfluidic card that is held along that edge by a spring-loaded clamp. The air lines operate pneumatically actuated valves on the card to direct the fluid flow (Micronics, Inc., Redmond, WA).
soft lithography or plastic laminate structure that provides a macro–micro interface, such as wells that align with access holes in the chip. Once the fluid connection positions become fixed in a final microdevice design, a standardized layout fixture that supplies multiple interfaces: fluidic, electrical, vacuum, and hydraulic connections between microchip and reader instrument is advantageous. An example is the Caliper sipper chip that has a glass capillary attached to the single fluid inlet port. Figure 7 shows two more examples.
2 MICROCONSTRUCTION TECHNIQUES AND DEVICE EXAMPLES
Fabrication of prototype microdevices is more like watchmaking than like automobile engine repair. Features that are 10 µm wide are barely visible to the unaided eye; they look like scratches. Electrical and fluid inlet and outlet connections are often susceptible to damage. Dust control is essential, usually by deploying ionizers at the assembly bench; it is generally impossible to remove dust from a completed structure, even with compressed air. Achieving uniform surface energy throughout the internal passages of the microdevice is often critical to proper filling of the channels with a polar liquid or aqueous solution. Care must be taken to avoid resting microdevice parts on any surface that has molecules eager to migrate to the higher-energy surfaces of your microdevice, thus lowering the surface energy and making wetting more difficult. There is a wide range of technologies available for constructing microdevices: traditional lithography, soft lithography, or machining of plastics,
6
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
(a)
(a)
(b)
Figure 9. Syringe pump modules that can deliver microliter fluid volumes. (Kloehn, Ltd, Las Vegas, NV (a), Harvard Apparatus, Inc., Holliston, MA (b).)
metals, or glass by laser ablation or computernumerical-control (CNC) mill to produce finished parts or primary molds. Polymer films can be laser-cut, knife-cut, or stamp-cut with patterns that when stacked, aligned, and bonded together create a three-dimensional laminate structure that can contain fluidic channels, pneumatically activated valves, flex circuits to support sensors, porous membranes, and so on. Almost any object can be positioned within a molded part or a laminate. (b)
2.1
Traditional Lithographic Techniques
Microelectromechanical systems (MEMS), of which microfluidics is a subcategory, began as a offshoot of the computer chip processing industry, in which hard substrates, most commonly silicon, are the primary construction material. The basic paradigm of standard lithographic techniques involves the use of electromagnetic radiation, typically ultraviolet (UV) light, to transfer a pattern to a light-sensitive polymer (also known as photoresist), which undergoes a chemical response upon exposure to the radiation (see Figures 11–13). The exposed areas are defined by means of a photomask, usually consisting
Figure 10. Complete microsyringe pump systems. If multiple fluids can be injected simultaneously at the same flow rate into your microfluidic device, the system in (a) offers complete control of flow rates and volumes (Harvard Apparatus, Inc., Holliston, MA). If up to four fluids need to be controlled independently, the device in (b) controls flow rates, volumes, and timing, as well as automatic reloading of the syringes from supply bottles (Micronics, Inc., Redmond, WA). It also offers multiple pneumatic lines that can be switched independently between positive, negative, and ambient pressure. These are used to control the pneumatic liquid-control valves on the disposable plastic laminate cards that fit in the manifold (foreground).
of a patterned thin chromium film on a quartz plate; areas to be exposed are defined by open areas in the mask plate. After light exposure, the
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
polymer-coated disk must be developed, similar to developing a photograph, such that the areas exposed to the electromagnetic radiation behave in an opposite manner to unexposed areas; one set of areas polymerizes and remains on the surface while the other set is washed away. The item being machined then undergoes either material removal (e.g., etching of exposed substrate) or material addition (e.g., metal deposition) Finally, the remaining photoresist is removed, typically with a solvent, and the substrate is ready for the next round of machining. Traditional lithography methods are amenable to parallel processing and can produce mechanically strong and chemically resistant devices, with feature sizes as small as 0.1 µm. However, these methods are also expensive, require significant chemical processing equipment outlays, and often require toxic chemicals. In addition, the processes are often time consuming and not amenable to rapid design iterations. New methods of MEMS fabrication, categorized as “soft lithography”, address many of these concerns. For many BioMEMS applications, traditional lithographic techniques may be most applicable for creating a stamp or mold that is then used to produce large numbers of devices via soft lithography.
7
Apply or grow resist Expose to radiation
Add layer (optional)
Develop resist
Substrate (silicon or silicon oxide)
Traditional lithography process
Etch or deposit material
Remove remaining resist
Figure 11. General process of traditional lithography.
In that context, there are several common modifications to traditional lithography when used for bioMEMS applications. Given the relatively large feature size required for bioMEMS devices, the mask that contains the pattern can be as simple as an overhead transparency on which the design has been drawn by hand. The desired feature size is the single most important consideration in selecting a mask and illumination source, with budget and time Light Mask
Add photoresist
Bare substrate (silicon wafer)
Expose
Develop
Etch Remove photoresist
Figure 12. The photolithographic process, using positive photoresist followed by an etch step, is shown.
8
MINIATURIZED, MICRO AND PARTICLE SYSTEMS Anisotropic etch
(100) Surface orientation
Isotropic etch
Any orientation
(110) Surface orientation Figure 13. Etch patterns based on silicon type.
considerations coming a close second. Typically ultraviolet (UV) light is used, although electromagnetic waves with narrower wavelengths, such as X rays, have been used to achieve a finer resolution. In traditional photolithography, the photoresist layer thickness is typically on the order of 1 µm and is completely removed by the end of a micromachining cycle. In contrast, by using special photoresists at a thickness on the order of hundreds of microns, the photoresist itself can be patterned and used as a mold. For example, SU-8 photoresist can be layered up to 450 µm thickness and can be used to achieve aspect ratios of 15:1.21 SU-8 components can be used as molds for soft lithography or assembled together to form a microfluidic device. A complete integrated microfluidic system for mass spectrometry has been constructed using SU-8 to form multiple layers of channels sandwiched between silicon and Pyrex wafers.22 Two excellent sources of additional information on traditional lithography are VLSI Technology 23 and Introduction to Microelectronic Fabrication.24
2.2
lithography as “an elastomeric stamp with patterned relief structures on its surface that is used to generate patterns” and claim reproduction of feature sizes as small as 30 nm and up to 500 µm. A commonly used elastomer is a silicone rubber, polydimethylsiloxane (PDMS), but other elastomers are available, such as polyurethanes, polyimides, and phenol formaldehyde polymers. They describe a variety of soft lithography techniques: cast molding, replica molding (REM), microcontact printing (µCP), microtransfer molding (µTM), micromolding in capillaries (MIMIC), and solventassisted micromolding (SAMIM), but the most commonly used techniques are injection molding and embossing. In hot embossing (see Figure 14) a hard negative mold is pressed into a heat-softened thermoplastic polymer, which on cooling retains the pattern. In soft embossing a flexible rubber mold is filled with hot polymer under pressure (see Figure 15). Unlike hard polymers, the deformability (elasticity) of soft polymers like PDMS can be problematic for accurate registration between parts and limits the aspect ratio of features to between 0.2 and 2.0 to ensure structural integrity. A useful reference for practical tips is My Little Guide to Soft Lithography (or Soft Lithography for 27 ˚ Dummies) from Krogh and Asberg. Commonly used materials for primary molds are SU-8, a negative photoresist, which can be patterned using the standard lithographic processes21
Soft Lithography
Though well-developed, traditional lithographic techniques are expensive and require a relatively elaborate fabrication facility. Soft lithography attempts to address these concerns and offer additional flexibility, such as the ability to work with nonplanar surfaces and to use primary molds that are either positive (i.e., the mold is the same shape as the final part) or negative (i.e., the void in the mold is the same shape as the final part). A foundation paper with many useful references is from Xia and Whitesides.25 They define soft
Figure 14. SEM image of pyramids with a 30-µm base width fabricated in PMMA by hot embossing with a silicon mask formed by standard lithography with wet-chemical etching.26 [Reprinted with permission Lin et al.26 copyright 1998, Springer Verlag.]
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES 200 µm
and used as the primary mold in any of the soft lithography processes; and high-temperature epoxy, which can be used as a stamp for hot embossing.28 Mold wear can be reduced by making parts from negative secondary molds that were formed from a primary positive mold.
2.3
(a) 2 mm
(b)
(c)
5 mm
Figure 15. (a) Soft embossing of heated plastic resins in a hard rubber mold can produce high aspect ratio (height/width) features, such as these micropillars; (b) vertical features such as these posts do not need the draft angle that is required in injection molding since the mold can flex when pulling the part; (c) features can be placed on both sides of the part. [Photos courtesy of Edge Embossing LLC, Medford, MA.]
9
Machining, Cutting, and Laser Ablation
Tools for direct shaping of materials for microdevice constructions include lasers, knife plotters, and miniature CNC milling machines. All use a software “mask” that can be easily altered. Miniature CNC mills (e.g., Sherline 5400 CNC Mill with tolerance ∼25µm, Taig CNC Mill with tolerance ∼15µm) are low-cost (US $2000) means to shape metals, glass, silicon, polymers, and plastic sheets with small-diameter diamond-tipped bits (e.g., 50, 75, 100, 125 µm diameter). Figure 16 shows an example of machined acrylic sheet. Small knife plotters (e.g., SummaCut D60 vinyl cutter with tolerance ∼25 µm) are available for a similar price to cut plastic films up to 0.030 in. (30 mils, 800 µm) thick. There are several types of lasers capable of shaping parts for microdevices. Lasers operating at IR or visible light wavelengths provide lower resolution and remove material by melting, vaporization, and pyrolysis (thermal decomposition of chemical bonds). Lasers that operate in the UV range provide high-resolution photolytic (direct decomposition of chemical bonds due to absorption of single or multiple photons) micromachining. At high power density, both pyrolytic and photolytic decomposition create a rapid rise in pressure and temperature that ejects material in a process called ablation. The energy is absorbed into a depth of the material, depending on the first-order absorption cross section of the material. The absorbed energy is converted to heat with a temperature profile that decays exponentially into the surrounding material. The material directly exposed to the radiation heats and vaporizes rapidly, ejecting gas and particles while cooling the surrounding material. As the laser is pulsed, it moves in steps to adjacent uncut material, and the laser is pulsed again. The laser repetition rate, the pulse width, and the pulses per inch all affect the quality of the cut. (L. Levine, private communication.30 ) Figure 17 compares cut quality
10
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
Figure 16. Close-up of CNC-machined fluidic prototype device.29 Machining is typically a more time-intensive process than laser ablation, but the edges of the machined features are sharper and the substrate material properties have not been altered by the high temperature characteristic of the ablation process.
Laser type
CO2
Tripled YAG
KrF excimer
18
>4000
Cut quality
Cutting time (s)
0.7
Figure 17. Trade-off between cut quality (a function of laser type, power, and spot size) and cutting time with Mylar as the substrate. [From Photomachining Inc.] Table 1. Comparison of wavelengths and spot sizes of various lasers. The smaller spot sizes give increased precision, but also require increased processing time
Laser type
CO2
YAG
Tripled YAG
KrF excimer
Color Wavelength (µm) Approximate spot size (µm)
IR 10.6 120
Near IR 1.06 12
Near UV 0.36 4
UV 0.25 3
From Laserod Inc.
and cutting time of different laser types when set up for high speed cutting. Table 1 compares the wavelengths and spot sizes of different types of lasers. CO2 lasers (e.g., Universal Laser M-360) can cut or engrave any of a wide variety of materials, such as: aluminum, brass, titanium, stainless steel, glass, silicon, and polymer or plastic sheet. The Nd:YAG (neodymium:yttrium–aluminum–garnet) laser is typically used to cut or weld metals. The tripled YAG can also cut some polymers, such as: Kapton, polycarbonate, and polyimide. An excimer laser is compatible with many polymers, including: fluorinated ethylene propylene (FEP), Kapton, parylene, polyethylene terephthalate (PET, Mylar),
polymethyl methacrylate (PMMA, acrylic), polycarbonate, polyester, polyethylene, polyimide, polyurethane, polyvinyl alcohol (PVA), and Teflon. Table 2 lists the preferred sizes and unique properties of the most commonly used materials. Cut edge quality depends on laser type. The ablation process of the UV lasers can produce a very clean edge, while the IR lasers tend to melt, instead of vaporize, the material.
2.4
Laminate Technologies
A very useful and adaptable construction technique for microfluidic devices is the concept of building
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
11
Table 2. Commonly available plastic sheet and film stock from suppliers/distributors: McMaster-Carr, CS Hyde Industries, and Sheffield Plastics
Material Polyester (PET or Mylar)
Available thickness (µm)
Polycarbonate Acrylic, clear (cast)
12.5, 25, 50, 75, 100, 125, 250 75, 125, 250 500, 1000, 1250, 1500
Acrylic, clear (extruded)
1000, 1500
Acrylic, black (extruded) Polyimide
250, 500, 1000, 1500 25, 50, 75
COP Silicone
50, 125, 250 125, 250, 500, 1000
Polypropylene (clear) Polypropylene (hazy) Polyethylene (LMW, HMW, or UHMW) PVDF (clear) FEP (clear) Urethane Acetal (white, black)
45 250, 500, 1500
Unique properties Okay optical properties, autofluorescence, most widely available in a variety of film stocks Good optical quality, low fluorescence Good optical quality, thickness highly variable, can come in UV grade transparent to 350 nm Good optical quality, grazes readily with exposure to even dilute alcohol Orange color, often thermally bonded at high temperature, not very suitable for CO2 laser Thickness variability around 20% Makes dust when cutting, requires air—assist to avoid flaming Usually low-quality material Readily available in food grade
50, 75 50, 75 50, 75+ 250, 500, 1000, 1500
COP: cyclic olefin polymer. LMW: low molecular weight. HMW: high molecular weight. UHMW: ultrahigh molecular weight. PVDF: polyvinylidene fluoride.
up a device by stacking layers, each of which has its own planar pattern of channels and chambers to hold fluids or simply holes (vias) to allow fluid communication between neighboring layers. (This stacking technique was intensively developed by engineers in the 1950s and 1960s for fluidic amplifiers in the flight control circuits of jet aircraft.) A basic microfluidic device can comprise a primary layer into which a pattern defining the fluid channels has been etched, machined, cast, or embossed and a second layer that serves as a cap to close off the channels. The layer materials can be any combination of silicon, glass, metals, polymers, or plastic films. If the pattern passes completely through the primary layer, then a third layer is needed as a capping layer. The choice of two-layer or three-layer construction depends mainly on the choice of pattern-making method. Adding additional layers allows overlapping channels and complex features like valves, mixing structures, and pumps, or the opportunity for easy parallelization.
2.5
Plastic Film and Sheet Stock (L. Levine, private communication)
Thinner materials are classified as films; those thicker than 0.010 in. are generally considered sheets. Other than custom stock, few materials are available in thicknesses between 0.010 and 0.015 in. Sheet materials up to 0.080 in. thick can be readily handled for laser cutting and lamination. Film thickness can be defined either by gauge (i.e., the thickness in inches multiplied by 100), or by “mil” thickness (i.e., 1 mil is equivalent to 0.001 in. = 25.4 µm and 40 mils (0.040 in.) = 1 mm). The thickness tolerance of film stock is generally 5–10%. Extruded sheet stock is similar, but the thickness tolerance of cast sheet stock can be as much as 25% of material thickness (e.g., cast acrylic). Other material characteristics of potential importance are the addition of flame retardants, typically brominated compounds, that may become chemical interferents in an assay and the autofluorescence of the material that may interfere with
12
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
high-sensitivity fluorescence detection. Materials specified as medical or food grade are often the better choice. The most widely available material is polyester film (i.e., PET, Mylar ). Widely used in the graphics industry, it can be purchased with surface treatments that enhance the adhesion of aqueousbased inks (usually corona treated), and is available in heat-sealable grades with a thin layer of low molecular weight (MW) polymer, usually a polyolefin or a polyvinylidene dichloride/PVA, or oligomers of PET itself.
2.6
Layer Bonding
The appropriate bonding method depends on the type of bond (permanent or reversible), the tolerance of the material to high temperatures, the deformability of the material, and solubility of the bonding agent to the fluids of interest (e.g., water, solvents). Some fabrication technologies, such as injection molding, plastic-film laminates, and soft lithography, have their own native bonding methods. Plastic-film laminates are typically fabricated by alternating layers of plastic film or rigid sheet with pressure-sensitive adhesive (PSA) film. For increased durability during assembly, the adhesive film is often a composite structure of a polyester carrier film coated with PSA on both sides, essentially double-sided sticky tape. One drawback is that, if fluid channels are cut in one of the layers to be bonded, or in the PSA film itself, the adhesive will be exposed to the fluid, which introduces a potential incompatibility to the use of organic solvents or the long-term storage of aqueous solution. Figure 18 shows a stack-up of layers to form a microfluidic card. Injection-molded parts are usually assembled by solvent bonding, diffusion bonding, ultrasonic bonding, or laser bonding, though PSA can be used as well. For devices implemented in PDMS by soft lithography, a low-temperature permanent bond is obtained by treating the bonding surfaces with an oxygen plasma to raise their surface energy. If the surfaces are placed in contact with each other within 3 min, a permanent bond is formed between the layers of PDMS (a rubbery transparent polymer) and glass, silicon, or itself. If a reversible
bond to glass is desired, the glass surface should be cleaned with isopropyl alcohol (IPA) to improve adhesion. A PSA film or double-stick tape (e.g., Adhesives Research, ARcare 7841) can also be used to bond PDMS to many materials, such as metals or printed-circuit boards. A reversible bond can be created between rigid materials that can tolerate a temperature of 100 ◦ C, including metal, glass, and ceramic, by using a wafer-mounting wax, such as Crystalbond 509, which is transparent in thin sections, dissolves in acetone, and, though it has a flow point at 120 ◦ C, is workable at 100 ◦ C. No bonding agent at all is necessary if the layers have optically smooth surfaces; they can be pressed together mechanically in a jig to form a watertight seal if they are completely dry when assembled. Permanent bonds can be formed with a lowviscosity epoxy (e.g., µ < 100 cP), which will wick over the smooth surface between parts, but be prevented from filling wider gaps by surface tension. Silicon and glass can be permanently bonded by anodic bonding, in which a high temperature (∼400 ◦ C) and a DC electrical potential (250 < V < 1000) cause sodium ions to migrate across the boundary from the glass to create a permanent electrostatic bond. Silicon can be bonded to silicon at lower temperatures by using an intermediate layer of sputtered lithium borosilicate.31
2.7
Surface Modification
In materials with naturally hydrophilic surfaces, channels that are designed to carry liquids are easily wetted if their width-to-height aspect ratio is near unity. Variations in surface energy will affect the wetting speed as the liquid proceeds through the channel, but the channel will wet as long as the meniscus contact angle remains less than 90◦ . However, even slight variations in surface energy between different locations along the wall can make wetting of channels with aspect ratios greater than 5 quite difficult (see Figure 19). There are several ways to achieve uniform surface energy. Sheet PET can be purchased with a hydrophilic surface coating, but manufacturing procedures are needed to ensure that the material does not contact any surface that will change its surface energy, such as a surface shedding hydrophobic particles or contaminated with skin
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
Glass (Pyrex) Layer 1: cover slip insert 0.125-mm PET
Layer 2: Layer 3: 0.025-mm adhesive 3.175-mm PMMA
Layer4: 0.100-mm ACA
13
Layer5: 0.125-mm PET
Figure 18. Layout of a microdevice built up from five plastic laminate layers and a glass insert. Layers 1 and 5 are polyethylene terephthalate (PET), layer 2 is a pressure-sensitive adhesive (PSA), layer 3 is polymethyl methacrylate (PMMA), and layer 4 is a sandwich of PSA with a PET center layer.
oils. These cautions should also be followed during assembly of parts that have undergone plasma treatment. Both oxygen and fluorine plasmas can create a temporary uniform hydrophilic surface. A direct-current corona torch forms hydroxyl groups on the surface of PET that, even after several days, can halve the contact angle of aqueous solutions. Another method to achieve uniform hydrophilic surfaces is to use stable coatings of hydrophilic polymers (e.g., AST HydroLast) of the channel walls after device assembly.
2.8
Integration of Heterogeneous Materials
It is possible to integrate a wide variety of materials as inserts into engineered voids in microdevices. Calculated tolerance stackups of inserts and part thicknesses should ensure that the insert is thinner than the void into which it fits. Any fluid inlet and outlet ports to the insert should be located on one side of the insert and pressed against or bonded to a mating surface of the surrounding part.
(a)
(b)
Figure 19. Wetting of two high width-to-height aspect ratio channels with hydrophilic walls. Channel walls in (a) have uniform surface energy; channel walls in (b) have variation in surface energy and, even though they are hydrophilic, air pockets are likely to form.
Thin-film membranes, proton-exchange membranes, and membranes for ultrafiltration, nanofiltration, and microfiltration can be integrated into a plastic-laminate or soft lithography system (see Figure 20). Applications such as ligand assays, enzymatic processes sensitive to trace metals, electrophoresis, degassing of aqueous solutions, and fluorometry are enabled in this way. Microdevices, such as lab-on-a-chip type devices, typically employ optically clear materials, but the feature size is small. Spherical or cylindrical lenses can be integrated to enable visual inspection with the unaided eye. Electronics can be integrated as polymer thickfilm (PTF) flexible circuits (i.e., flex circuits), which are screen-printed thick-film conductive inks on a low-cost polyester dielectric substrate. Multilayer circuits are produced with dielectric materials as insulating layers, and double-sided circuits with printed through-hole technologies. Lead-free, silver-loaded isotropic conductive adhesive provides both electrical and mechanical connection of active and passive surface-mounted components for applications such as optoelectronics, electrokinetics, and liquid crystal displays. Nanogen has been a leader in the field of integrating polymeric materials and IC chips for biological applications. In a collaboration with Genoptix and UC Irvine, researchers developed a microfluidic device that successfully isolated bacteria from a spiked blood sample and provides an excellent example of a heterogeneous microfluidic device, consisting of a variety of materials each machined in a different way (see Figure 21).32 The bottom layer, made of polycarbonate, contains
14
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
Glass cover plate
PSA with microfluidic channel
Kapton with flip-chip bonded DEP chip
(a)
PSA Fluidic inlet/outlet
Polycarbonate substrate Figure 21. Schematic of a heterogeneous microfluidic device used to extract bacteria from a blood sample. [Reprinted with permission Huang et al.32 copyright 2003, Springer Science and Business media.]
(b) Antibody–dye conjugate
Test zone
Control zone Cylindrical lens
Vent Injection port Sample receiver (c)
1 cm
Evaporation port Vent or absorbent pad
Figure 20. A wide variety of materials can be integrated into microdevices: (a) porous membranes (the white circles) can sequester live bacteria in neighboring wells (NASA GeneSat card); (b) elastomers (the gray disks) can form the flexible membranes of pneumatically actuated valves (ALine, Inc., Redondo Beach, CA); (c) a cylindrical lens inserted in a micro lateral flow strip device to magnify test and control zones for viewing with the unaided eye (MicroPlumbers Microsciences LLC, Seattle, WA).
machined fluidic channels. The next layer, a PSA, adheres the bottom layer to the heart of the device, a dielectrophoresis (DEP) chip. The DEP layer is itself heterogeneous, consisting of a flexible polyimide layer patterned with electrical circuits and a silicon chip with electrodes fabricated using repeated cycles of traditional UV photolithography
and sputtered deposition of a titanium–tungsten adhesion layer followed by deposition of a platinum layer. The electrodes are 50 µm squares, spaced 50 µm apart. The DEP chip is attached to the polyimide layer via “bump” bonding with silver epoxy. To complete the device, a glass cover plate is sealed to the top by a second PSA layer with cutouts for fluidic channels. 2.9
Immobilization of Biological Material
There are a number of ways to immobilize bioactive molecules on solid substrates in microdevices. Proteins can be directly adsorbed on a surface, but surface hydrophobicity, charge, and chemical makeup can affect both their stability and orientation. Protein–material interactions may result in decreased protein activity and nonspecific protein adsorption can change the intended biological activity. Another approach, in addition to adsorption and covalent coupling, is tethering with an intermediate linker molecule such as polyethylene
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
oxide (PEO) chains that reduce nonspecific protein adsorption and denaturing of the target protein that is caused by interactions with the substrate. Compared to direct adsorption, tethered proteins have greater mobility, which avoids steric hindrance of binding processes and allows clustering of ligand-bound receptors within the cell membrane, which is known to be a requirement for activation of some intracellular signaling pathways. Substrate materials include polystyrene, PMMA, polyurethane, polyamide, and hydrophobized glass. Mesoporous silica can be integrated into a plastic-laminate or soft lithography structure. With negatively charged functional groups on its surface, a favorable chemical environment is created for proteins. The Pacific Northwest National Laboratory (PNNL) has demonstrated that high concentrations of an active enzyme can be immobilized in a mesoporous, functionalized silica structure while exhibiting higher activity than they would in aqueous solution. DNA can also be immobilized onto a surface, most commonly in the form of a DNA chip used for hybridization microarray assays. DNA chips can be functionalized in two different ways. Various DNA probes can be synthesized off-chip and applied to the chip surface using a linker chemistry to immobilize the DNA. Alternatively, the DNA can be synthesized in situ, base pair by base pair, an approach commercialized by Affymetrix. Photolithographic techniques are used to selectively protect and expose different regions of the chip, typically 18- to 20-µm squares. A solution containing a single deoxynucleotide (A, C, T, or G) linked to a removable protection group is then washed over the entire chip. Exposed squares participate in linking reactions, thus extending one base pair, while the rest of the chip remains unmodified.
2.10
Separation Matrices
Hydrogels can be cast into wells in plasticlaminate and soft lithography structures. This includes agarose gel separation matrix designed for the separation of nucleic acid (NA) fragments. Applications include separation of base pairs ranging in length from 100 to 1200 and separation of polymerase chain reaction (PCR) products.
15
3 PUTTING IT ALL TOGETHER – AN EXAMPLE OF THE DEVELOPMENT OF A FULLY INTEGRATED MICROFLUIDIC DEVICE
One of us (Weigl), with collaborators (collaborators are Micronics, Inc, for microfluidic card development, Yager Group, University of Washington, for dry-down reagent development, and Dr. Tarr, Washington University, for clinical validation and support), is developing a lab-on-a-card platform to identify enteric bacterial pathogens in patients presenting with acute diarrhea, with special reference to infections that might be encountered in developing countries.33,34 Component functions that are integrated on this platform include on-chip capture and lysing of pathogens, multiplexed NA amplification and on-chip detection, sample processing to support direct use of clinical specimens, and dry reagent storage and handling. All microfluidic functions are contained on the lab card. This new diagnostic test will be able to rapidly identify and differentiate Shigella dysenteriae serotype 1, Shigella toxin–producing Escherichia coli, E. coli 0157, Campylobacter jejuni, and Salmonella and Shigella species. The multiplex disposable enteric card (DEC) test (see schematic in Figure 22) will be an automated, rapid, easy-to-use, point-of-care platform to simultaneously detect multiple enteric pathogens causing disease with similar symptoms. It will provide a mechanism for accurate, point-of-care diagnosis with a rapid turnaround time for results. The method is based on laminate microfluidic lab card technology developed at Micronics and the University of Washington. This technology has been used in many different applications ranging from diffusion-based separation and detection to projects involving flow cytometry on a chip and NA–based amplification and detection techniques. The individual microfluidic subcircuits of the DEC card were initially designed and validated with both pathogen isolates as well as stool samples before integration of the subcircuits into a single disposable unit. The subcircuits are (i) capture and lysis of pathogens, (ii) NA extraction, (iii) NA amplification, and (iv) visual detection of amplified NA. A subcircuit card that can purify NAs from lysed leukocytes or bacteria (Figure 23) is loaded with specimen in a lysis solution, which allows RNA
16
MINIATURIZED, MICRO AND PARTICLE SYSTEMS Feces extract
Salmonella Shigella STEC
Campylobacter E. coll O157:H7 Positive control Immunocapture NA amplification NA LFS detection
Figure 22. Schematic of DEC approach showing a combination of pathogen capture and lysis, nucleic acid extraction, PCR, and visual detection of amplicons.
•
DNA/RNA capture from lysate sample
Lot no. NCI – 25
Micronic
Wash Air
Wash
Lyse
Purified sample
Capture filter •
Washing of DNA/RNA
Raw sample
To waste
•
Removal of DNA/RNA from card
Elute
Figure 23. Credit-card-sized microfluidic lab card that automates NA extraction.
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
17
Rapid PCR amplification breadboard and lab card 8 min for 35 cycles (nonoptimized)
(a)
(b)
(c)
Figure 24. Microfluidics-enabled rapid PCR amplification lab card and breadboard designed by Micronics, Inc. Prototype Thermal Electric Cooler (a), lab card with PCR reaction chambers (b), and thermal couple trace showing 60-s reverse 60 ◦ C reverse transcription followed by 16-s PCR cycles (c).
5′
3′
5′ B
B 5′
1
2
3
1
2
3
4
5
6
7
8
F 3′
3′ I F
E
ssDNA is captured by labeled probes
SA–MPs bind biotinlabeled probe
DNA-bound SA–MP complexes migrate through membrane. Anti-FITC IgG binds complexes
Absorbent pad facilitates wicking
5′
3′
ssDNA, amplified product
Capture probes
B
F
SA-coated microparticle
Anti-FITC antibody
Figure 25. Visual lateral flow-based amplicon detection process developed by Micronics Inc.
4
5
6
7
8
18
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
from the sample to bind to silica. An on-card silica filter and microfluidic valves provide fluid control to automate the RNA binding, washing, drying, and elution steps. The card was initially validated in experiments in which 106 white blood cells were processed using commercial kits or suspended in lysis buffer prior to loading on the lab card. Realtime PCR assays determined that the microfluidic card solutions were detected a few cycles earlier than the control RNeasy solutions. This card has also been validated for lysis and detection of gramnegative bacteria isolated from feces. Similarly, the NA amplification subcircuit has been designed (Figure 24) and tested for each of the pathogens. The PCR product generated by the subcircuit shown in Figure 24 is detected using a microfluidic visual amplicon detection method. Analogous to immunochromatographic strip tests, this method allows multiplexed detection of PCR-amplified products without instrumentation or software. As amplicons bound to colored microparticles accumulate on an antibody stripe immobilized in the microchannel, a colored line becomes visually apparent, indicating successful NA amplification and the presence of target (Figure 25). Procedural control lines have been included and results are available in a few minutes. This example demonstrates typical subcomponents that have to be integrated to form a fully functional microfluidic device. Further, this example is quite representative of the status of integrated microfluidic analysis devices today—many are in development in both corporate and academic settings, but few, if any, are currently in production or use. Given the enormous progress that has been made in the microfluidics and lab-on-a-chip fields over the 15 years since its inception, and the enormous and broad efforts that currently go into microfluidics research, the authors believe that fully integrated microfluidic devices will indeed become mainstream analytical tools within the decade.
REFERENCES 1. A. Manz, C. S. Effenhauser, N. Burggraf, D. J. Harrison, K. Seiler, and K. Fluri, Electroosmotic pumping and electrophoretic separations for miniaturized chemical analysis systems. Journal of Micromechanics and Microengineering, 1994, 4, 257–265.
2. F. Forster, R. Bardell, M. Afromowitz, and N. Sharma, Design, Fabrication and Testing of Fixed-valve Micropumps, In: Proceedings of the ASME Fluids Engineering Division, 1995 IMECE , San Francisco in November 1995, Vol. 234 pp. 39–44. 3. A. Olsson, P. Enoksson, G. Stemme, and E. Stemme, A Valve-less Planar Pump in Silicon, In: The 8th International Conference on Solid-state Sensors and Actuators, and Eurosensors IX , Stockholm, Sweden, 1995 June 25–29, pp. 291–294. 4. T. Gerlach and H. Wurmus, Working principle and performance of the dynamic micropump. Sensors and Actuators A (Physical), 1995, 50(1–2), 135–140. 5. Y. H. Mu, N. P. Hung, and K. A. Ngoi, Simulation and Optimization of a Piezoelectric Micro-pump, In: International Conference of ASME , Anaheim, California, 1998, November 15–20. 6. R. Linnemann, M. Richter, A. Leistner, and P. Woias, A Full Wafer Mounted Self-priming and Bubble-tolerant Piezoelectric Silicon Micropump, In: Proceedings of the Actuator ’98 Conference, Bremen, Germany, 1998, June 17–19, pp. 78–81. 7. P. Woias, R. Linnemann, M. Richter, A. Leistner, and B. Hillerich, A Silicon Micropump with a High Bubble Tolerance and Self-priming Capability, in Micro Total Analysis Systems, J. Harrison and A. Van den Berg (eds), Kluwer Academic Publishers, Dordrecht, 1998, pp. 383–386. 8. C. R. Tamanaha, L. J. Whitman, and R. J. Colton, Hybrid macro-micro fluidics system for a chip-based biosensor. Journal of Micromechanics and Microengineering, 2002, 12, N7–N17. 9. D. Maillefer, S. Gamper, B. Frehner, P. Balmer, H. van Lintel, and P. Renaud, A High Performance Silicon Micropump for an Implantable Drug Delivery System Technical Digest MEMS’99 , 1999, pp. 541–546. 10. A. Richter and R. Zengerle, Properties and Applications of a Micro Membrane Pump with Electrostatic Drive, In: 3rd International Conference of New Actuators (ACTUATOR ’92), Bremen, Germany, 1992, pp. 28–33. 11. M. T. A. Saif, E. Alaca, and H. Sehitoglu, Analytical modeling of electrostatic membrane actuator for micro pumps. Journal of Microelectromechanical Systems, 1999, 8(3), 335–345. 12. C. Yamahata and G. Gijs, Integrated Plastic Micropumps with Magnetic Actuation, In: NanoTech 2003, 7th Annual European Conference on Micro & Nanoscale Technologies for the Biosciences, Montreux, Switzerland, 2003 November 25–27. 13. H. J. Yoon, J. M. Jung, J. S. Jeong, and S. S. Yang, Micro devices for a cerebrospinal fluid (CSF) shunt system. Sensors and Actuators A, 2004, 110, 68–76. 14. C. J. Morris and F. K. Forster, Low-order modeling of resonance for fixed-valve micropumps based on first principles. Journal of Microelectromechanical Systems, 2003, 12, 325–334. 15. C. Grosjean and Y. C. Tai, A Thermopneumatic Peristaltic Micropump, In: 1999 International Conference on Solidstate Sensors and Actuators (Transducers ’99), Sendai, Japan, 1999 June, pp. 1776–1779. 16. J. H. Tsai and L. W. Lin, Micro-to-macro fluidic interconnectors with an integrated polymer sealant.
PRACTICAL ASPECTS OF MICROFLUIDIC DEVICES
17.
18. 19.
20.
21.
22.
23. 24. 25.
26.
Journal of Micromechanics and Microengineering, 2001, 11, 577–581. R. H. Liu, J. Yang, R. Lenigk, J. Bonanno, and P. Grodzinski, Self-contained, fully integrated biochip for sample preparation, polymerase chain reaction amplification, and DNA microarray detection. Analytical Chemistry, 2004, 76, 1824–1831. D. Ross and L. E. Locasio, Rapid microfluidic mixing. Analytical Chemistry, 2002, 74, 45–51. J. Liu, C. Hansen, and S. R. Quake, Solving the “Worldto-Chip” interface problem with a microfluidic matrix. Analytical Chemistry, 2003, 75, 4718–4723. P. Galambos and G. Benavides, Electrical and Fluidic Packaging of Surface Micromachined Electro-microfluidic Devices, In: SPIE Micromachining and Microfabrication Conference, San Jose, California, 2000 September. H. Lorenz, M. Despont, N. Fahmi, N. LaBianca, P. Renaud, and P. Vettiger, SU-8: a low-cost negative resist for MEMS. Journal of Micromechanics and Microengineering, 1997, 7, 121–124. J. Carlier, S. Arscott, V. Thomy, J. C. Fourrier, F. Caron, J. C. Camart, C. Druon, and P. Tabourier, Integrated microfluidics based on multi-layered SU-8 for mass spectrometry analysis. Journal of Micromechanics and Microengineering, 2004, 14, 619–624. S. M. Sze, VLSI Technology, McGraw-Hill Science/ Engineering/Math, 1988. R. C. Jaeger, Introduction to Microelectronic Fabrication, 2nd Edn, Prentice Hall, 2001, p. 232. Y. Xia and G. M. Whitesides, Soft lithography. Angewandte Chemie International Edition, 1998, 37, 550–575. L. Lin, Y. T. Cheng, and C. J. Chiu, Comparative study of hot embossed micro structures fabricated by laboratory
27.
28.
29. 30. 31.
32.
33.
34.
19
and commercial environments. Microsystem Technologies, 1998, 4, 113–116. ˚ L. Krogh and P. Asberg, My Little Guide to Soft Lithography (or Soft Lithography for Dummies), Link¨oping University, website, http://www.ifm.liu.se/∼petas/mikrosystem/Links/Material− files/Soft− Lithography− for− Dummies.pdf, 2003. T. Koerner, L. Brown, and R. D. Oleschuk, Prototyping of Polymeric Microfluidic Devices with Hot Embossing, In: Third Canadian Workshop on MEMS , Ottawa, Canada, 2003 August 22. Photo Courtesy of MicroPlumbers Microsciences LLC , http://www.microplumbers.com, 2007. Private communication from L. Levine, ALine Inc, Redondo Beach, CA, Hyperlink http://www.alineinc.com/. A. Gerlach, D. Maas, D. Seidel, H. Bartuch, S. Schundau, and K. Kaschlik, Low-temperature anodic bonding of silicon to silicon wafers by means of intermediate glass layers. Microsystem Technologies, 1999, 5, 144–149. Y. Huang, J. M. Yang, P.J. Hopkins, S. Kassegne, M. Tirado, A. H. Forster, and H. Reese, Separation of simulants of biological warfare agents from blood by a miniaturized dielectrophoresis device. Biomedical Devices, 2003, 5(3), 217–225. P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. R. Tam, and B. H. Weigl, Microfluidic diagnostic technologies for global public health. Nature, 2006, 442(7101), 412–418. B. H. Weigl, J. Gerdes, P. Tarr, P. Yager, L. Dillman, R. Peck, S. Ramachandran, M. Lemba, M. Kokoris, M. Nabavi, F. Battrell, D. Hoekstra, E. J. Klein, and D. M. Denno, Fully integrated multiplexed Lab-on-achip assay for enteric pathogens. Proceedings of the SPIE, 2006, 6112, 1–11.
43 Polymer-Based Microsystem Techniques Matthias Schuenemann1 and Erol C. Harvey1,2 1
MiniFab (Aust) Pty. Ltd., Scoresby, Victoria, Australia and 2 Faculty of Engineering and Industrial Science, Swinburne University of Technology, Hawthorn, Victoria, Australia
1 INTRODUCTION
The need for point-of-care or point-of-use biosensors and bioanalytical devices in health care, in the food industry as well as for environmental testing has been a major factor in the development of low-cost microfluidic devices. The advantages of miniaturization (e.g., rapid analysis times, higher achievable analytical performance facilitated by changed fluid dynamics, low sample/reagent volumes, cost-effective reagent usage, reduced sample wastage, and reduced contamination and crosscontamination) have been attracting increasing attention from research groups as well as from commercial device manufacturers.1–3 Point-of-care devices enable diagnostic procedures at or close to the actual point of application (point-of-care, point-of-test). Extensive cleaning and test preparation procedures for setting up point-of-care tests are neither acceptable nor practical. For this reason, disposable devices are favored by the end user. On the other hand, it is very rarely economically feasible to integrate all technical components and subsystems required for complex sophisticated tests into disposable products. Therefore, biosensor systems tend to be divided into disposable chips or cartridges (or similar technical solutions) and a nondisposable host device (or instrument). The very essence of a successful design of a commercially viable pointof-care analysis system lies in the smart division
between the disposable cartridge and the reusable instrument, and the careful design of the interface between the two. Figure 1 shows a microscope slide-sized disposable biosensor chip with integrated micromixers, delay lines, passive valves, and electrochemical sensors. Economical restrictions dictate cost-efficient materials and technologies for the manufacturing of disposable cartridges. Many of the already published approaches to highly miniaturized bioanalytical systems are realized either in silicon4,5 or in glass,6,7 mostly relying on adapted silicon micromachining technologies and resulting in highly integrated miniaturized devices. Unfortunately, the respective manufacturing technologies as well as the utilized materials are rather expensive. For many potential applications of miniaturized bioanalytical devices, such high production costs cannot be justified. Polymer materials have been demonstrated to be a very versatile and cost-effective material choice, and the use of polymer-based microsystem techniques leads to bioanalytical devices with a very competitive cost of ownership per test.
2 MANUFACTURING OF LOW-COST POLYMER MICROFLUIDIC BIOCHIPS
One of the greatest challenges in developing lowcost disposable polymer microfluidic biosensors
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
Figure 1. Disposable biosensor chip with integrated micromixers, delay lines, passive valves, and electrochemical sensors.
is to select a set of manufacturing processes that can be readily scaled in volume for each of the development stages, and that from the start, uses materials that will form the final device. This is essential in order to make cost effective the process of developing a biosensor that will pass regulatory compliance to become a commercial product. Assuming that the bioassay has already been demonstrated at the laboratory scale, and that the appropriate reporting mechanism has been chosen (e.g., electrochemical, fluorescence, gravimetric), the first stage is to demonstrate the incorporation of these processes within a microfluidic system. The fabrication processes used for this stage must allow rapid iteration of design. Only a few working units, typically fewer than 10, are required to demonstrate the proof-of-concept. If the assay involves several discrete processing steps requiring specialist fluidic solutions, for example, passive valving, filtering, or metering, these steps may be demonstrated separately to show proof-of-principle. We generally refer to this as a Stage 0 development and use it to give confidence that the assay can be transferred to a disposable biosensor. Although desirable that the materials used to fabricate the proof-of-concept devices are the same as those to be used in the final manufacture, this becomes an essential requirement for the next stage. In Stage 1 each of the individual components of the disposable biosensor are brought together to demonstrate a working assay and explore its performance in terms of sensitivity,
selectivity, specificity, and performance variation. This information is compared to the performance of the assay at the laboratory bench if the design requirement was set on the basis of the benchscale performance. Often the microfluidic design must be further improved as sources of manufacturing variation are identified. These might include variation in channel dimensions, electrode surface areas, optical interference, or variation in component adhesion. The manufacturing process typically produces batches of 10–100 units for this stage and could be considered “prototyping” since the devices are manufactured either individually or, perhaps using thermal embossing techniques in small batches of 10–50 per sheet.8 For Stage 2 the manufacturing process must incorporate some element of batch or volume manufacturing as the biosensor testing process should now obtain statistically significant performance variation data on several thousand devices. Clearly there is less opportunity, and hopefully less need, for design change; and high-speed replication processes such as injection molding can be used. The tooling for this replication step must be robust enough to produce repeatable results and can be created in a number of ways including nickel electroforming from a master, electrodischarge machining, or diamond-tipped precision milling.9 Stage 2 development is used to eliminate sources of variation in the manufacturing process such as polymer shrinkage, warpage, component misalignment during assembly, or irreproducible bonding.
POLYMER-BASED MICROSYSTEM TECHNIQUES
After this stage the biosensors may be ready for early stage field trials so that Stage 3 must be able to produce runs of up to 10 000 devices in a few months. If the developer aims for a lowcost disposable polymer biosensor there should be a clear manufacturing path to achieve production volumes of 105 –107 devices per year; clearly only achievable with high-speed replication, minimal assembly, and considerable process automation. This is best achieved using highly integrated polymer devices.
3 POLYMER MATERIALS
The selection of a suitable polymer material for a disposable biosensor cartridge is highly dependent on the intended application, the sample preparation, amplification, and detection process as well as the design complexity of the device. Almost every bioanalytical application will introduce specific technical demands for the chosen polymer material. Nonetheless, many applications share a common set of general requirements.10 Most polymer-based microfluidic chips and cartridges are manufactured from thermoplastic polymers. Thermoplastic polymers are characterized by first- and second-order thermal transitions. Whereas the first-order transition temperature usually corresponds to melting and allotropic transformation and is usually well outside the thermal working range for most materials, the secondorder transition temperature characterizes the point above which the fixed molecular structure of the material is partially broken down by a combination of thermal expansion and thermal agitation.11 Most thermoplastic polymers soften above this glass transition temperature, Tg , and exhibit a rubberlike behavior. Only when a material is highly crystalline are mechanical properties maintained above the glass transition temperature. Certain biological processes such as polymerase chain reaction (PCR) are performed at temperatures as high as 96 ◦ C, while other devices potentially need to survive hostile storage temperatures. The polymer material has to be selected so that the fabricated device does not deform or disintegrate at the maximum specified temperatures. Microfluidic structures and highly miniaturized fluidic devices are characterized by a surface-tovolume ratio that is at least 1 or 2 orders of
3
magnitude higher than in conventional laboratory equipment. The large surface-to-volume ratio leads to a much stronger interaction of sample material and reagents with the substrate material, compared with standard laboratory equipment. The diagnostic integrity of a device is therefore heavily influenced by biocompatibility of the polymer material as well as by material properties such as water vapor permeability, gas permeability, and water absorption. Biocompatibility of the substrate materials with the assay is an especially important requirement for a good analytical performance. Protein adsorption and cell adhesion are common phenomena interfering with bioanalytical processes. Protein adsorption is affected by factors such as surface energy/tension, surface charge, roughness, crystallinity, and entropy. Additionally, additives embedded in the polymer material such as plasticizers or UV stabilizers might lyse from the substrate material when subjected to elevated temperatures or contact with fluidic samples and reagents and inhibit biochemical reactions. The increased surface-to-volume ratios as well as the small amounts of target molecules characteristic of microfluidic systems make them particularly susceptible to these processes.12 A major drawback of polymers is their relatively high water vapor permeability rates, making polymers more difficult to use in bioanalytical devices compared to glass and metals. Any loss of water or water vapor from a sample or reagent may lead to a drift in pH or osmolarity and interfere with the analytical process on the device. Again, the large surface-to-volume ratio in miniaturized bioanalytical devices increases their susceptibility to this. The rate of potential fluid loss depends upon material properties such as water vapor permeability and water absorption rates as well as on design factors such as the diffusion path length from the microfluidic structure to the external environment. Detection of reaction products forms a fundamental part of the bioanalytical device. Many detection methods rely on optical techniques, such as detecting fluorescent dyes attached to proteins or optical measurement of spots on microarrays.1,13 An optically transparent material is essential for devices that depend on such detection techniques. However, many polymer materials are characterized by background fluorescence
4
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
(or autofluorescence), resulting either from fluorescence intrinsic to the bulk polymer or from additives, impurities, or degradation products. The small sample and reagent volumes in miniaturized biosensors as well as the comparatively small amounts of fluorophores attached to the respective targets will result in small fluorescence signals. Background fluorescence decreases the signal-tonoise ratio significantly. It is therefore important to analyze, understand, and consider autofluorescence properties for successful system design as well as for suitable material selection.14 A well-informed selection of a suitable polymer material is critical not only to the performance but also to the manufacturability of the biosensor or biochip. During development and prototyping, significant research efforts have to be invested to overcome technical challenges resulting from imperfect material behavior. Having to change materials en route from Stage 1 to Stage 3 is likely to impose considerable cost as well as significant time delays in any development project. Manufacturing-related material requirements may include laser machinability, low shrinkage during injection molding, and sufficient resistance to cleaning agents. The material should be commercially available and reasonably well introduced into the market. A material with only a single supplier can potentially become temporarily or permanently unavailable, putting the commercial prospects of a device in a highly dangerous position. The selection of a rarely used material may also delay regulatory approval processes. Polymeric materials like polycarbonate (PC),15 polydimethylsiloxane (PDMS),16 polyethylene terephthalate (PET),17 polymethyl methacrylate (PMMA),18 cyclic olefin copolymer (COC),19 and polyimide (PI)20 have commonly been used to fabricate prototypes of microfluidic bioanalytical devices. Less frequently, other standard organic polymeric materials such as polyethylene (PE), polyetheretherketone (PEEK), polystyrene (PS), polyamide (PA), polyetherimide (PEI), liquid crystal polymer (LCP), polypropylene (PP), polybutylene terephthalate (PBT), polyoxymethylene (POM), polyphenylene ether (PPE), and polysulfone (PSU) have successfully been used in microtechnology, mainly in a research environment outside the miniaturized biosensor domain.8 PMMA and PC meet the basic material requirements for miniaturized biosensors, are widely
available and comparatively easy to machine and are therefore favored by the research community.10,12 Another material receiving increasing attention is PET,17,21,22 its main advantages being its widespread use in the packaging and printing industry and the extensive fabrication and processing knowledge base. PDMS is another transparent, elastomeric material that is used in soft lithography and molding processes for prototyping.23 Although quite suitable for the fabrication of prototypes, its disadvantages include limited mechanical strength, limited ability to bond to other polymer materials, and very high water vapor permeability. A newer group of materials, COC, has great potential in microanalytical processes for its high chemical stability, low permeability rates, and very good optical properties,24 but suffers from high costs and limited availability. Table 1 shows a qualitative overview of the suitability of selected polymer materials for biochip manufacturing. The analysis shows that there is no one ideal polymer able to meet all requirements. Rather, the device or package designer has to find a technically and economically valid compromise for his specific application, taking into account such factors as the temperature regime of the analytical process, the storage conditions for the disposables prior to their use, the emission wavelength of the utilized fluorophores, the design complexity, and the fabrication technology most suitable for the expected production volumes.12
4 MICROSTRUCTURING OF POLYMERIC MATERIALS
One of the greatest challenges in developing lowcost disposable polymer microfluidic biosensors is to select a set of manufacturing processes that can be readily scaled in volume for each of the development stages. For this reason direct machining, laser cutting, and thermal embossing are attractive methods for development since each is able to structure a bulk polymer sheet that is available in high volume with excellent uniformity. Also, in the case of thermoplastics, the sheet form of the polymer can be considered to have properties similar to that achieved by injection molding (although in some specific details this may not be a valid assumption as UV laser cutting can leave
Water absorption
Thermal stability
Autofluorescence n/a n/a n/a n/a
n/a n/a n/a
n/a
Machinability
Cost
Cutting/blanking Micromilling
Injection molding
Laser excimer (λ = 248 nm)
Laser 3 ω Nd:YAG (λ = 355 nm)
Laser CO2 (λ = 10.6 µm)
Optical transparency
Material characteristics and machinability of polymer materials can vary significantly depending on manufacturer and material grade, due to the influence of additives such as plasticizers or UV stabilizers. Especially laser micromachining processes are highly sensitive to the presence of such additives. Figures in the table are therefore given for base polymers/most common polymer grades. Most favorable material ◮ ◮ ◮ ◮ least favorable material.
COC PC PDMS PEEK PET High-density PE Low-density PE PI PMMA PS Polytetrafluoroethylene Polyvinylchloride PVDC Polyvinylidene fluoride
Material
Water vapor permeability
Properties
Hot embossing
Table 1. Suitability of selected polymer materials for biochip manufacturing
POLYMER-BASED MICROSYSTEM TECHNIQUES 5
6
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
a temporary surface activation not produced by molding techniques).25
4.1
Laser Fabrication
A wide range of lasers able to cut and pattern an even wider range of materials suitable for polymer biosensor packaging is now available.26 Laser micromachining systems have the advantage of being noncontact tools that can be rapidly reprogrammed to produce varied patterns, making them particularly suitable for the design and development phase of the microfluidic biosensor. Lasers are available that produce either pulsed or continuous radiation and are characterized by the wavelength of the light they produce. For micromachining applications pulsed lasers are essential since they allow greater control of the molten, or heat-affected zone in comparison to continuous systems. Generally the shorter the pulse length, the smaller the heat-affected zone. Table 2 shows a range of the popular pulsed lasers used for microfabrication and their operating characteristics. It should be obvious that for any material to be able to be laser machined it should absorb light at the wavelength of the laser to be used. For this reason, many infrared laser sources (e.g., carbon dioxide lasers (10.6 µm) or fiber lasers (1.09–1.55 µm depending upon type)) will produce wildly varying results depending on the nature of the polymer used or the precise detail of
additives in the polymer. Unfortunately the buyer of the bulk polymer often is unable to know what these additives are and in what concentration they are present, therefore much laser work tends to be by trial and error. The most popular pulsed visible lasers are specially modified Nd:YAG (neodymium–yttrium–aluminum–garnet) lasers that can produce powerful pulses of green (533 nm) or ultraviolet (256 nm) light at high repetition rates. The alternate wavelengths are produced by placing wavelength doubling or tripling crystals in the beam, a technique that can be applied to some other lasers, for example the infrared fiber laser or the green copper vapor laser (CVL), enabling each to produce ultraviolet pulses. Ultraviolet pulses are the most useful wavelength for micromachining applications since they produce the least thermal damage and, since most polymers strongly absorb ultraviolet radiation, can provide control of the depth of laser machining. Excimer lasers are pulsed gas lasers that produce ultraviolet light without the aid of doubling crystals. Another major difference from other laser sources is the large rectangular beam produced by excimer lasers. Hence rather than being used as a focused spot, excimer lasers are generally used as an illumination source for a stencil or mask that is imaged onto the polymer workpiece. If an image-reducing lens is used, the features produced at the workpiece can be of submicron size, and for most polymer materials the machined depth is typically less than a micron per laser pulse.
Table 2. Pulsed laser types and their operating characteristics
Pulsed laser source
Type
Frequency multiplied
Wavelength
Carbon dioxide (CO2 ) Ti:Sapphire Nd:YAG Nd:YAG 2 ω Nd:YAG 3 ω Nd:YAG 4 ω Copper Vapor (CVL)
Infrared Infrared Infrared Visible Ultraviolet Ultraviolet Visible Visible Ultraviolet Ultraviolet Ultraviolet Ultraviolet Ultraviolet Ultraviolet
Fundamental Fundamental (tunable) Fundamental Doubled Tripled Quadrupled Fundamental Fundamental Doubled Doubled Fundamental Fundamental Fundamental Fundamental
9.24–10.64 µm 700–1080 nm 1.064 µm 532 nm 355 nm 266 nm 511 nm 578 nm 255 nm 271 nm 308 nm 248 nm 193 nm 157 nm
Copper Vapor 2 ω (CVL 2 ω) Excimer Excimer Excimer Excimer (a)
(XeCl) (KrF) (ArF) (F2)
Typical specifications. Values will vary depending upon configuration and operating conditions.
Pulse width(a)
Repetition rate(a)
25 µs–1 ms 20–100 fs 10–300 ns
20 kHz 75–120 MHz 2–100 kHz
20 ns
10–20 kHz
20 ns
100 Hz–6 kHz
POLYMER-BASED MICROSYSTEM TECHNIQUES
Hence by computer-controlled manipulation of the number of pulses and the mask shape, complex three-dimensional shapes such as channels, ports, weirs, wells, mixers, and bifurcators can be rapidly fabricated in polymers.27 It is necessary to implement some form of polymer replication process once increasing manufacturing volumes are required. A range of microreplication processes for thermoplastic polymers are available that are generally smaller-scale implementations of their macroworld counterparts. These include hot embossing, injection molding, reaction injection molding, injection compression molding, and thermoforming.28
4.2
Hot Embossing
Hot embossing is a popular replication process since it is relatively easy to tool-up for and is a comparatively easy process to execute. It is able to achieve excellent replication of high-aspect-ratio microstructures, for example 8-µm-wide beams 150 µm tall in PMMA,8 but has the disadvantage of a slow cycle time that can be up to 20–30 min. In the hot-embossing process a mold tool is created that has the inverse features of the desired shape. This can be done by direct precision machining of metals or can be a nickel electroform grown from a previously microfabricated master. This master could be made in a number of ways including laser ablation of polymers, wet or dry lithographic etching in silicon,29 UV lithography in thick SU-8 photoresist, or synchrotron exposure using the LIGA process (a German acronym for lithography, electroplating and replication). In some cases the polymer or silicon master can itself be used as the embossing tool.29–31 The tool is mounted into a press and heated to a temperature slightly above the glass transition temperature, Tg of the polymer to be embossed (PMMA 106 ◦ C, PC 150 ◦ C). Polymer sheet is introduced into the press, which is closed, and a force of between 20–30 kN over a 4-in.-diameter area is applied under a vacuum of around 10−1 mbar. After a hold time of a few minutes and with the force still applied, the tool is then cooled to below Tg to stabilize the polymer before opening and demolding. Optimization of the process will reduce the thermal stresses in the part realizing improved replication, but usually at the cost of increased cycle time.
7
Nanosized features are readily reproduced by hot embossing. While this may be useful it also means that imperfections and roughness in the tool are also readily reproduced.
4.3
Injection Molding
Injection molding is a highly developed process for macroreplication and is now increasingly available for microscale thermoplastic replication.28 The process has the advantage of extremely fast cycle times, of the order of a few seconds per cycle, but at the cost of a considerably more complex molding tools. In this process, a microstructured mold insert is placed within a specially formed mold cavity within the injection-molding machine. Polymer beads are heated above Tg and forced to flow into the mold cavity at high pressure where they rapidly cool to form a solid component that is ejected from the tool. This cyclic temperature control is called variotherm (variothermal ). The resulting parts can have high degrees of internal stress and variable rates of shrinkage due to the rapid cooling of the polymer in the tool. Minimizing these effects as well as creating an effective ejection system for removing the part from the tool becomes part of the skill in designing good injection-molding systems. The ability to produce many millions of parts per year at relatively low cost makes this an important part of the industrial manufacturing process. Table 3 compares selected manufacturing technologies for polymer biochips and rates them according to their suitability for tool making, prototyping, and volume production. Having created the polymer components of the microfluidic system they must then be assembled and bonded together to form complete units. 5 SURFACE MODIFICATION OF POLYMER MATERIALS
Surface modification techniques change the surface characteristics of a material for a specific application without severely affecting the bulk properties of the polymer substrate. A range of biological, physical, and chemical methods are employed to modify surface properties such as wettability, permeability, biocompatibility,
8
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
LIGA Silicon bulk micromachining Laser micromachining (CO2 , λ = 10 µm) Laser machining (Nd:YAG, λ = 355 nm) Laser machining (Excimer, λ = 248 nm) Micromilling µEDM Plasma etching Hot embossing Injection molding Roto-cutting blanking Stereolithography
Investment costs/ operational costs
Process time/throughput
Process flexibility
3D capabilities
Precision
Technology
Minimum structure width
Table 3. Selected manufacturing technologies for polymer biochips
/ / / / / / / / / / / /
Application Toolmaking Toolmaking Prototyping production Prototyping toolmaking Prototyping toolmaking Prototyping toolmaking Toolmaking Prototyping production Toolmaking production Production Production Prototyping toolmaking
µEDM: microelectrodischarge machining. Most favorable process ◮ ◮ ◮ ◮ least favorable process.
chemical inertness, bondability, electrical characteristics, or optical properties.32 The majority of polymer packaging technologies use surface modification techniques in order to condition the polymer surfaces for the bonding process. Most polymer surfaces are hydrophobic, leading to poor wetting of the surface and therefore a poor spreading of adhesives or poor adhesion during bonding. One commonly used surface modification method is the use of a corona discharge to oxidize the polymer surface by ionized particles.33 Although corona treatment is very cost effective compared with other surface modification methods, its short shelf-life and limitations in the treatable thickness of polymer sheets (up to 250 µm) restrict its applicability. An alternative to corona discharge is gas plasma treatment. Typically, gas plasma treatment of polymers are utilized to ablate surface contamination, introduce chemically functional groups to the surface, and/or to introduce cross-linking.34 Common gases utilized in this process include oxygen, nitrogen, and argon. One of the most widely used applications of gas plasma treatment is the oxygen plasma modification of PDMS utilized to convert hydrophobic Si–C siloxane groups to hydrophilic
SiOx groups.35 Similar to the corona discharge process, the surface modification can be short lived due to polymer chain mobility. Another major disadvantage of gas plasma treatment is the requirement for an evacuated environment. In an atmospheric plasma treatment process a polymer film can be passed through the plasma beam without the need for a vacuum, allowing for continuous in-line processing.36 Many surface modification methods combine a chemical surface treatment with physical changes to surface properties. Polymer surfaces can be grafted with chemicals that provide excellent adhesion to a large range of materials. This process starts with a surface activation step, followed by the deposition of chemicals dissolved in a water solution (e.g., silanes) which bond to the activated polymer.37 In another surface modification approach, PET surfaces have been modified using a saponification reaction, in which polymer substrates were immersed in a bath of highly concentrated NaOH to etch the surface immediately before bonding.21 The effect of photodegradation has been utilized to modify the surface of PMMA by exposing the polymer film to UV light to soften the top surface of the substrate.38
POLYMER-BASED MICROSYSTEM TECHNIQUES
The nonspecific adsorption of proteins to polymer surfaces (and surfaces of other materials) is a significant problem encountered in a variety of biotechnological and bioanalytical applications. The behavior of hydrophilic polymer surfaces can vary considerably from nonfouling to selectively binding to high binding, depending on their respective physical and chemical surface characteristics. Proteins also adsorb to hydrophobic surfaces, but tend to denature, resulting in a thin, denatured, but persistently attached protein film preventing subsequent protein adsorption. This behavior may lead to a depletion of available target proteins and may significantly distort the measurements. For processes such as PCR, the use of different additives, for example, polyethylene glycol (PEG) or bovine serum albumin (BSA) in the PCR buffer has been shown to reduce nonspecific binding of key assay components to polymer surfaces and to improve process yield.15,39–41 The grafting of polyethylene oxide (PEO) has also been demonstrated to reduce nonspecific binding.42 Permeability of the substrate material is another essential property to control. Microsized bioassays
9
only handle small amount of fluids and reagents. It is therefore essential to maintain the volume of these fluids. A small fluid reduction caused by permeable materials affects the integrity of the process. Modifying the surface to become less permeable assists in avoiding this problem. Barrier layers on flexible polymer substrates are usually formed by depositing a thin layer of inorganic material like aluminum or silicon oxide on commodity polymers, such as PE or PET.43 A major disadvantage of this method is the loss of optical transparency preventing the use of optical detection techniques. In another approach, high barrier polymers are formed by co-extruding commodity polymers with polymer barrier layers, such as polyvinylidene chloride (PVDC). A very promising approach to surface modification is surface coating with parylene, a conformable, transparent coating based on polymerized para-xylylene. Parylene is deposited via chemical vapor deposition. In addition to preventing fluid losses through substrate materials, parylene reduces protein adsorption and cell adhesion.44 Since the process requires vacuum
Investment/operational costs
Process compatibility
Option for selective treatment
Surface treatment
Sustainability of effect
Strength of effect
Table 4. Surface modification of polymer materials for biochip manufacturing
Surface modification techniques for adhesion promotion Saponification Photodegradiation (UV) Corona treatment Plasma treatment Polymer grafting Primer deposition
/ / / / / /
Surface modification techniques for biocompatibility PEO PEG BSA Parylene deposition
/ / / /
Surface modification techniques for water vapor permeability reduction Barrier layer deposition Parylene deposition Barrier layer co-extrusion Most favorable process
◮ ◮
/ / / ◮ ◮
least favorable process.
10
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
conditions, it is difficult to integrate into continuous productions systems. Surface modification techniques are usually applied before bonding the prefabricated polymer layers or parts together. Unfortunately, surface requirements for bondability, biocompatibility, and permeability are usually contradictory and potentially conflicting. Surface modifications to condition the polymer surfaces for the bonding process often leave reactive functional chemical groups on the surface, which may inhibit bioanalytical processes. The same inhibition can occur when treating a surface to minimize loss of reagent. Again, the consequences of these contradictions are amplified by the large surface-tovolume ratio typical for miniaturized microfluidic devices. Each surface modification addressing one requirement therefore needs cross-checking to ensure compatibility with other requirements. Table 4 shows a qualitative assessment of selected surface modification techniques of polymer materials for biochip manufacturing.
6 ASSEMBLY AND PACKAGING OF POLYMER-BASED MICRODEVICES
Almost all microfluidic devices are based at least partly on fully enclosed and sealed microfluidic structures (i.e., channels, reservoirs, process chambers). Polymer microfabrication techniques, however, are usually only capable of generating open fluidic structures, and rely on bonding and sealing technology for the completion of the microfluidic device. The simplest way to bond and seal a microfluidic structure is to cap a single planar microstructured polymer substrate on one or both sides with an unstructured cover layer. More sophisticated devices may be assembled from several stacked layers of microstructured polymer films or substrates, creating true three-dimensional microfluidic systems. The number of layers that can be bonded together is only limited by the applied bonding technique, the complexity of the design, and the feature size of the microfluidic structures. As an example, Figure 2 shows a PCR cartridge, driven by pneumatically actuated peristaltic on-chip pumps and controlled by pneumatically actuated on-chip valves. The device is fabricated from seven vertically assembled polymer layers.
The actual microfluidic reactor consists of three microstructured layers. The pneumatic control circuit is realized by another three microstructured layers, and an elastomeric membrane layer joins and separates the two three-layer prefabricates. Primary functions for assembly, bonding, and sealing are to realize fully functional microfluidic devices by joining microstructured prefabricates, to prevent leakage from microfluidic features and to provide sufficient structural integrity within the assembled device. Several bonding technologies can be used for assembly, bonding, and sealing of prestructured polymer layers (see Table 5). Most of these are adapted from standard polymer manufacturing technologies. For microfluidic circuits, bonding technologies that enable a selective bonding and sealing only at preselected areas (e.g., around the channel walls) are especially interesting. Although these technologies are usually more costly than bulk bonding techniques, many of them reduce the risk of involuntarily blocking channels and microfluidic structures and/or avoid accidental exposure of biological fluids to potentially nonbiocompatible auxiliary materials (e.g., adhesives, solvents). The joining of polymer substrates using adhesives is widely used during prototyping of polymer microfluidic devices. Adhesives are capable of bonding between dissimilar polymers as well as bonding polymers to metal layers or polymer prefabricates with large metallization areas. Additionally, adhesive bonding processes do not require extensive capital investment. The direct application of an adhesive layer onto the surface of the polymer substrate carries a high risk of channel blocking. More commonly, adhesives (i.e., ultraviolet curable adhesives) are selectively applied to the bond surface using screen-printing techniques.45 However, adhesives spread after application and clamping, potentially entering microsized features and clogging channels, mixers, or fluidic junctions. A voidless adhesive joint with liquid or thixotropic adhesives and screen printing is difficult to realize around complex or densely packed microstructures. The use of adhesive films instead of liquid adhesives prevents the undesired flow of adhesives into microfeatures, but necessitates prestructuring of adhesive film to create the required microfluidic vias between the polymer layers and careful alignment of the
POLYMER-BASED MICROSYSTEM TECHNIQUES
11
(a)
(b) Figure 2. (a) Design for a polymer biochip fabricated from vertically assembled microstructured polymer layers. (b) Polymer biochip fabricated from vertically assembled microstructured polymer layers—manufactured layers and completed device.
Table 5. Systematization of polymer bonding techniques
Bulk bonding
Direct bond between structural layers
Bond mediated by auxiliary materials
Lamination Thermal diffusion bonding UV-assisted thermal diffusion bonding Plasma-assisted thermal diffusion bonding
Adhesive bonding Adhesive tape bonding Solvent bonding Chemical etching–assisted thermal diffusion bonding Screen-printed adhesives and adhesive bonding Prestructured adhesive tape and adhesive bonding Light-absorbing dyes laser welding Microwave absorber Microwave welding
Ultrasonic welding Selective bonding Transmission laser welding Reverse conductive laser welding
adhesive film relative to the structures on the polymer substrates, thus increasing manufacturing costs significantly. Additionally, adhesives have to be selected carefully as there is a high risk of undesired molecular interaction of the biological assay with the surface chemistry of the adhesive layer,
which may influence or even inhibit the biochemical processes on the device.12 Thermal lamination is a simple, effective method to cap single planar layers that contain microfluidic structures. A very common laminate consists of a thin PE/PET film thermally bonded
12
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
to PET substrates using a hot laminator.22,46,47 Thermal lamination is especially suited to seal simple devices consisting of only one microstructured layer, but it has considerable limitations in sealing large, shallow features (reservoirs, reaction chambers) without any additional structural support, as the laminating film sags into these structures during lamination and potentially interferes with their intended function. Additionally, multilayer devices usually cannot as easily be sealed with thermal lamination since the laminating film isolates microfluidic layers from each other by blocking microfluidic vias between layers. One approach to realize vertically integrated microfluidic multilayer devices via lamination is to microstructure (relatively thick) PE/PET/PE films (i.e., by laser manufacturing or roto-cutting) and laminate them together (i.e., in a reel-to-reel system). For prototype or small series production, thermal diffusion bonding, realized by applying heat and pressure over a given time to preassembled polymer slides, is a suitable method to bond prestructured polymer layers to each other. As no auxiliary materials such as adhesives are required, potential channel blocking is avoided, and biocompatibility is maintained. The bonding success depends heavily on the mobility of molecular chains in the polymer. Thus, only very similar materials with identical glass transition temperature can be bonded together. Successful thermal diffusion bonding has been reported for PC, PET, PMMA, and COC.12,48,49 Bonding temperature and applied pressure are critical as an unsuitable parameter combination will deform the material and collapse the channels. For the device in Figure 2, two three-layer PC prefabricates were manufactured separately by thermal diffusion bonding. The laser-machined PC layers, which had alignment features incorporated into the design, were placed into an in-house developed hot embossing tool. A temperature of 135 ◦ C and a pressure of 4.2–4.5 MPa were applied for 20 min.12 A major disadvantage of thermal diffusion bonding is the required process time of up to 30 min for a bond with sufficient strength. UV treatment prior to bonding allows for bonds to form at a significantly faster rate compared to thermal diffusion bonding of untreated surfaces.38 The number of layers that can be bonded together by this technology is limited by the complexity of the design
and the feature size of the cut-out structures. The pressure distribution to any bond area situated above or below a void in the structure (e.g., a channel or a process chamber) is very uneven and might locally prevent successful bonding. Strict observation of design rules is required. Solvent bonding involves the exposure of a polymer surface to a suitable solvent. Upon joining two solvent-exposed surfaces, the interfaces of both substrates diffuse in one another and form a bond after the solvent evaporates from the assembly. Although solvent bonding is a common joining method for the assembly of polymer parts with many material/solvent combinations being available, it has rarely been used for bioanalytical devices. One approach is to deposit a thin layer of COC with a lower glass transition temperature on a thick layer of COC with a higher glass transition temperature by dissolving it in toluene and spin-coating it on the thicker substrate. The solvent-bonded parts were subsequently thermal diffusion bonded to each other.50 The optical transparency of many polymer materials has been used for a number of selective bonding techniques. A common bonding technique is through-transmission laser welding used to bond two polymer parts with different optical transmission characteristics. A laser transmits energy through the transparent layer. The laser energy is absorbed by the subjacent opaque polymer, causing the material to heat past its melting temperature. As a result, the two substrates will locally join. Scanning a focused laser beam around microsized features enables selective sealing and bonding. In reverse conductive welding, the energy-absorbing layer is not part of the device, but forms a workbench that heats up during energy absorption. From there, the thermal energy is conducted back to the interface between the polymer layers. This process creates a large heat-affected zone, which leads to distortion of microsized features in the vicinity of the bond. Another way to weld polymer layers together is based on an energy-absorbing dye. The dye is deposited onto at least one of the surfaces to be bonded and subsequently heated by a laser source with a wavelength corresponding to the absorption wavelength of the dye.51 In ultrasonic welding, high frequency mechanical energy is applied via an acoustic horn to
POLYMER-BASED MICROSYSTEM TECHNIQUES
13
Cost
Process time and throughput
Technical flexibility
Biocompatibility
Transparency
Compatibility with metallization
Channel clogging / distortion
Structural complexity
Geometrical resolution
Technical maturity
Table 6. Process performance of selected polymer bonding techniques
Adhesive bonding (bulk) Adhesive bonding (screen-printed adhesive) Adhesive tape bonding (bulk) Adhesive tape bonding (prestructured tape) Lamination Thermal diffusion bonding Thermal diffusion bonding (UV-assisted) Thermal diffusion bonding (O2 plasma-assisted) Thermal diffusion bonding (chemical etch–assisted) Solvent bonding Laser welding (transmission) Laser welding (reverse conductive) Laser welding (absorbent dye) Ultrasonic welding Microwave welding (microwave absorber) Most favorable process
◮ ◮
◮ ◮
least favorable process.
the polymer assembly creating frictional heat between molecules and causing the polymer to melt and join.52 Another bonding technique utilizes microwave technology to join polymer layers in microfluidic devices. Most polymers are transparent to microwave radiation. If a microwave absorber such as a conductive polymer or a metal film is added to the joint interface between two polymer layers, it will selectively absorb the microwave energy. This interaction results in a local heat generation at the interface, leading to bulk polymer flow across the joint and formation of a weld.12,53 Table 6 qualitatively assesses the process performance of selected polymer bonding techniques. The analysis shows that again, owing to the strengths and weaknesses of each of the discussed bonding techniques, a single preferred bonding technique cannot be identified. The selection of a suitable bonding technique depends heavily on the complexity of the design, the utilized materials, and the fabrication technology most suitable for the respective production volumes. A
more detailed discussion of polymer bonding techniques is available in the literature.54–56
7 SUMMARY
Polymer microfabrication offers a wide variety of techniques for fabricating microdevices. Some of the techniques are relatively new and are borrowed from other microfabrication processes while the vast majority are adaptations of techniques already established in the macromanufacturing environment. We can expect to see even more innovations arising from the combination of other traditional manufacturing processes, for example printing processes, that when combined with polymer fabrication will produce increasingly integrated microsystems. The development of new polymer materials will further accelerate this development. The low material cost and great structural resolution possible with polymers makes for a highly cost-effective approach to designing and fabricating complex devices. As with all
14
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
product development, careful attention must be paid to the way the manufacturing process is scaled in volume. However, low cost, great design flexibility, and the ability to cost-effectively achieve high production volumes mean that we are seeing an increasing introduction of innovative and commercially successful disposable biosensor systems into the market.
REFERENCES 1. D. R. Reyes, D. Iossifidis, P.-A. Auroux, and A. Manz, Micro total analysis systems. 1. Introduction, theory, and technology. Analytical Chemistry, 2002, 74, 2623–2636. 2. P.-A. Auroux, D. Iossifidis, D. R. Reyes, and A. Manz, Micro total analysis systems. 2. Analytical standard operations and applications. Analytical Chemistry, 2002, 74, 2637–2652. 3. L. J. Kricka, Miniaturization of analytical systems. Clinical Chemistry, 1998, 44, 2008–2014. 4. M. A. Northrup, M. T. Ching, R. M. White, and R. T. Watson, DNA Amplification in a Microfabricated Reaction Chamber, In: Digest of Technical Papers: Transducers ‘93 , Yokohama, 1993, pp. 924–926. 5. M. A. Northrup, B. Bennett, D. Hadley, P. Landre, S. Lehew, J. Richards, and P. Stratton, A miniature analytical instrument for nucleic acid based on micromachined silicon reaction chambers. Analytical Chemistry, 1998, 70, 918–922. 6. E. T. Lagally, P. C. Simpson, and R. A. Mathies, Monolithic integrated microfluidic DNA amplification and capillary electrophoresis analysis system. Sensors and Actuators. B, 2000, 63, 138–146. 7. E. T. Lagally, C. A. Emrich, and R. A. Mathies, Fully integrated PCR-capillary electrophoresis microsystem for DNA analysis. Lab on a Chip, 2001, 1, 102–107. 8. H. Becker and U. Heim, Hot embossing as a method for the fabrication of polymer high aspect ratio structures. Sensors and Actuators. A, 2000, 83, 130–135. 9. W. Schomburg, Review of micro moulding of thermoplastic polymers. Journal of Micromechanics and Microengineering, 2004, 14, R1–R14. 10. M. L. Hupert, M. A. Witek, Y. Wang, M. W. Mitchell, Y. Liu, Y. Bejat, D. E. Nikitopoulos, J. Goettert, M. C. Murphy, and S. A. Soper, Polymer-Based Microfluidic Devices for Biomedical Applications, in Microfluidics, BioMEMS, and Medical Microsystems, H. Becker and P. Woias (eds), SPIE, Bellingham, 2003, SPIE Vol. 4982, pp. 52–64. 11. M. Chanda and S. K. Roy, Plastics Technology Handbook, 4th Edn, CRC Press, Boca Raton, 2007. 12. M. Schuenemann, D. Thomson, M. Atkin, S. Garst, A. Yussuf, M. Solomon, J. Hayes, and E. Harvey, Packaging of Disposable Chips for Bioanalytical Applications, in Proceedings of the IEEE 54th Electrical Components and Technology Conference, P. Thompson (ed), IEEE Press, Piscataway, 2004, pp. 853–861.
13. M. A. Burns, B. N. Johnson, S. N. Brahmasandra, K. Handique, J. R. Webster, K. Madhavi, T. S. Sammarco, P. M. Man, D. Jones, D. Heldsinger, C. H. Mastrangelo, and D. T. Burke, An integrated nanoliter DNA analysis device. Science, 1998, 282, 484–487. 14. K. R. Hawkins and P. Yager, Nonlinear decrease of background fluorescence in polymer thin-films—a survey of materials and how they can complicate fluorescence detection in µTAS. Lab on a Chip, 2003, 3, 248–252. 15. J. Yang, Y. Liu, C. B. Rauch, R. L. Stevens, R. H. Liu, R. Lenigk, and P. Grodzinski, High sensitive PCR assay in plastic micro reactors. Lab on a Chip, 2002, 2, 179–187. 16. X. Yu, D. Zhang, T. Li, L. Hao, and X. Li, 3D microarrays biochip for DNA amplification in polydimethylsiloxane (PDMS) elastomer. Sensors and Actuators. A, 2003, 108, 103–107. 17. J. S. Rossier, M. A. Roberts, R. Ferrigno, and H. H. Girault, Electrochemical detection in polymer microchannels. Analytical Chemistry, 1999, 71, 4294–4299. 18. S. Qi, X. Liu, S. Ford, J. Barrows, G. Thomas, K. Kelly, A. McCandless, K. Lian, J. Goettert, and S. A. Soper, Microfluidic devices fabricated in poly (methyl methacrylate) using hot-embossing with integrated sampling capillary and fiber optics for fluorescence detection. Lab on a Chip, 2002, 2, 88–95. 19. J. Steigert, S. Haeberle, T. Brenner, C. M¨uller, C. P. Steinert, P. Koltay, N. Gottschlich, H. Reinecke, J. R¨uhe, R. Zengerle, and J. Ducr´ee, Rapid prototyping of microfluidic chips in COC. Journal of Micromechanics and Microengineering, 2007, 17, 333–341. 20. S. Metz, R. Holzer, and P. Renaud, Polyimide-based microfluidic devices. Lab on a Chip, 2001, 1, 29–34. 21. M. Atkin, J. P. Hayes, N. Brack, K. Poetter, R. Cattrall, and E. Harvey, Disposable Microchip Fabrication for DNA Diagnostics, in Biomedical Applications of Micro- and Nanoengineering, D. V. Nicolau (ed), SPIE, Bellingham, 2002, SPIE Vol. 4937, pp. 125–135. 22. J. Rossier, F. Reymond, and P. E. Michel, Polymer microfluidic chips for electrochemical and biochemical analyses. Electrophoresis, 2002, 23, 858–867. 23. C. S. Effenhauser, G. J. M. Bruin, A. Paulus, and M. Ehrat, Integrated capillary electrophoresis on flexible silicone microdevices: analysis of DNA restriction fragments and detection of single DNA molecules on microchips. Analytical Chemistry, 1997, 69, 3451–3457. 24. C. H. Ahn, J. W. Choi, G. Beaucage, J. H. Nevin, J. B. Lee, A. Puntambekar, and J. Y. Lee, Disposable smart lab on a chip for point-of-care clinical diagnostics. Proceedings of the IEEE, 2004, 92, 154–173. 25. M. A. Roberts, J. S. Rossier, P. Bercier, and H. Girault, UV laser machined polymer substrates for the development of microdiagnostic systems. Analytical Chemistry, 1997, 69, 2035–2042. 26. C. Khan Malek, Laser processing for bio-microfluidics applications (Part 1). Analytical and Bioanalytical Chemistry, 2006, 385, 1351–1361. 27. E. C. Harvey, P. T. Rumsby, M. C. Gower, and J. L. Remnant, Microstructuring by Excimer Laser, in Micromachining and Microfabrication Technology, K. W. Markus (ed), SPIE, Bellingham, 1995, SPIE Vol. 2639, pp. 266–277.
POLYMER-BASED MICROSYSTEM TECHNIQUES 28. M. Heckele and W. K. Schomburg, Review on micro moulding of thermoplastic polymers. Journal of Micromechanics and Microengineering, 2004, 14, R1–R14. 29. J. Elders, H. V. Jansen, M. Elwenspoek, and W. Ehrfeld, DEEMO: A New Technology for the Fabrication of Microstructures, in Proceedings of the IEEE Micro Electro Mechanical Systems, IEEE Press, Piscataway, 1995, pp. 238–243. 30. J. Narashimhan and I. Papautsky, Polymer embossing tools for rapid prototyping of plastic microfluidic devices. Journal of Micromechanics and Microengineering, 2004, 14, 96–103. 31. C. Khan Malek and R. Duffait, Packaging using hotembossing with a polymeric intermediate mould. International Journal of Advanced Manufacturing Technology, 2006. DOI 10.1007/s00170-006-0595-2. 32. S. L. McArthur and K. M. McLean, Surface Modification, in Encyclopedia of Biomaterials and Biomedical Engineering, G. E. Wnek and G. L. Bowlin (eds), Marcel Dekker, New York, 2004. 33. C. M. Chan, T. M. Ko, and H. Hiraoka, Polymer surface modification by plasmas and photons. Surface Science Reports, 1996, 24, 3–54. 34. P. K. Chu, J. Y. Chen, L. P. Wang, and N. Huang, Plasmasurface modification of biomaterials. Materials Science and Engineering R, 2002, 36, 143–206. 35. D. C. Duffy, J. C. McDonald, O. J. A. Schueller, and G. M. Whitesides, Rapid prototyping of microfluidic systems in poly(dimethylsiloxane). Analytical Chemistry, 1998, 70, 4974–4984. 36. M. J. Shenton and G. G. Stevens, Surface modification of polymer surfaces: atmospheric plasma versus vacuum plasma treatments. Journal of Physics D: Applied Physics, 2001, 34, 2761–2768. 37. W. S. Gutowski, S. Li, C. Filippou, P. Hoobin, and S. Petinakis, Interface-interphase engineering of polymers for adhesion enhancement: Part II. Theoretical and technological aspects of surface-engineered interphaseinterface systems for adhesion enhancement. Journal of Adhesion, 2003, 79, 483–519. 38. R. Truckenm¨uller, P. Henzi, D. Herrmann, V. Saile, and W. K. Schomburg, Bonding of polymer microstructures by UV irradiation and subsequent welding at low temperatures. Microsystem Technologies, 2003, 10, 372–374. 39. E. Kiss, J. Samu, A. Toth, and I. Bertoti, Novel ways of covalent attachment of poly(ethylene oxide) onto polyethylene: surface modification and characterization by XPS and contact angle measurements. Langmuir, 1996, 12, 1651–1657. 40. M. Zhang, T. Desai, and M. Ferrari, Proteins and cells on PEG immobilized silicon surface. Biomaterials, 1998, 19, 953–960. 41. B. C. Giordano, E. R. Copeland, and J. P. Landers, Towards dynamic coating of glass microchip chambers for amplifying DNA via the polymerase chain reaction. Electrophoresis, 2001, 22, 334–340. 42. P. Kingshott and H. J. Griesser, Surfaces that resist bioadhesion. Current Opinion in Solid State and Material Science, 1999, 4, 403–412. 43. M. Hanika, H.-C. Langowski, U. Moosheimer, and W. Peukert, Inorganic layers on polymeric films—
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
15
influence of defects and morphology on barrier properties. Chemical Engineering and Technology, 2003, 26, 605–614. Y. S. Shin, K. Cho, S. H. Lim, S. Chung, S. Park, C. Chung, D. Han, and J. K. Chang, PDMSbased micro PCR chip with Parylene coating. Journal of Micromechanics and Microengineering, 2003, 13, 768–774. J. Han, S. Lee, A. Puntambekar, S. Murugesan, J.-W. Choi, G. Beaucage, and C. H. Ahn, UV Adhesive Bonding Techniques in Room Temperature for Plastic Lab-on-aChips, in Proceedings of Micro Total Analysis Systems 2003, M. A. Northrup, K. F. Jensen, and D. J. Harrison (eds), Transducers Research Foundation, San Diego, 2003, pp. 1113–1116. M. A. Roberts, J. S. Rossier, P. Bercier, and H. H. Girault, UV laser machined polymer substrates for the development of microdiagnostic systems. Analytical Chemical, 1997, 69, 2035–2042. J. S. Rossier, G. Gokulrangan, S. Svojanovsky, G. S. Wilson, and H. H. Girault, Characterization of protein adsorption and immunosorption kinetics in photoablated polymer microchannels. Langmuir, 2000, 16, 8489–8494. J. Yang, Y. Liu, C. B. Rauch, R. L. Stevens, R. H. Liu, R. Lenigk, and P. Grodzinski, High sensitivity PCR assay in plastic micro reactors. Lab on a Chip, 2002, 2, 179–187. X. Zhu, G. Liu, Y. Guo, and Y. Tian, Study of PMMA thermal bonding. Microsystem Technologies, 2007, 13, 403–407. F. Bundgaard, T. Nielsen, D. Nilsson, P. Shi, and G. Perozziello, Cyclic Olefin Copolymer (COC/Topas )— an Exceptional Material for Exceptional Lab-on-a-chip Systems, in Proceedings of Micro Total Analysis Systems 2004, T. Laurell, J. Nilsson, K. Jensen, D. J. Harrison, and J. P. Kutter (eds), Royal Society of Chemistry, Cambridge, 2004, Vol. 2, pp. 372–377. L. Dosser, K. Hix, K. Hartke, R. Vaia, and M. Li, Transmission Welding of Carbon Nanocomposites with Direct-diode and Nd:YAG Solid State Lasers, in Photon Processing in Microelectronics and Photonics III, P. R. Herman, J. Fieret, A. Pique, T. Okada, F. G. Bachmann, W. Hoving, K. Washio, X. Xu, J. J. Dubowski, D. B. Geohegan, and F. Traege (eds), SPIE, Bellingham, 2004, SPIE Vol. 5339, pp. 465–474. R. Truckenm¨uller, Y. Cheng, R. Ahrens, H. Bahrs, G. Fischer, and J. Lehmann, Micro ultrasonic welding: joining of chemically inert polymer microparts for single material fluidic components and systems. Microsystem Technologies, 2007, 12, 1027–1029. A. A. Yussuf, I. Sbarski, J. P. Hayes, M. Solomon, and N. Tran, Microwave welding of polymeric microfluidic devices. Journal of Micromechanics and Microengineering, 2005, 15, 1692–1699. C. A. Harper, Plastics Joining, in Handbook of Plastics, Elastomers, and Composites, 4th Edn, C. A. Harper (ed), McGraw-Hill, New York, 2002, pp. 507–560. T. Velten, H. H. Ruf, D. Barrow, N. Aspragathos, P. Lazarou, E. Jung, C. Khan Malek, M. Richter,
16
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
J. Kruckow, and M. W¨ackerle, Packaging of BioMEMS: strategies, technologies and applications. IEEE Transactions on Advanced Packaging, 2005, 28, 533–546. 56. S. Garst, M. Schuenemann, M. Solomon, M. Atkin, and E. Harvey, Fabrication of Multilayered Microfluidic Packages, in Proceedings of the IEEE 55th Electrical Components and Technology Conference, P. Thompson (ed), IEEE Press, Piscataway, 2005, pp. 853–861.
FURTHER READING H. Becker and C. Gartner, Polymer microfabrication methods for microfluidic analytical applications. Electrophoresis, 2000, 21, 12–26. D. Thomson, J. P. Hayes, and H. Thissen, Protein Patterning in Polycarbonate Microfluidic Channels, in BioMEMS and Nanotechnology, D. V. Nicolau (ed), SPIE, Bellingham, 2004, SPIE Vol. 5275, pp. 161–167.
44 Microelectrochemical Systems Stuart A. G. Evans and Lindy J. Murphy Oxford Biosensors Ltd., Yarnton, UK
1 INTRODUCTION
The use of microelectrodes in the field of biosensors has led to increasingly lower detection limits and sample volumes, due to their small dimensions and high sensitivity of measurement. Detection limits as low as femtomolar concentrations of DNA or zeptomolar concentrations of analytes, and sample volumes as low as picoliters have been reported. In addition, advances in microfabrication techniques have resulted in increasing numbers of lab-on-a-chip-type devices with inbuilt electrochemical detection being reported, some of which are commercially available. Microelectrodes have also been fundamental to the development of the technique of scanning electrochemical microscopy (SECM), which allows investigation of redox processes at electrode surfaces with high resolution. This article describes the electrochemical response and methods of fabrication of microelectrodes, and outlines some of the recent applications of microelectrodes in the field of bioelectrochemistry.
2 MICROELECTRODES: DEFINITION AND PROPERTIES
Microelectrodes, as their name suggests, differ from conventional electrodes (macroelectrodes) with respect to their size. Macroelectrodes typically have dimensions in the meters to millimeters scale, depending on their application,
whereas microelectrodes (which are also known as ultramicroelectrodes or UME s) are regarded as having at least one dimension in the micrometer range. The question of how small an electrode must be in order to be defined as a microelectrode has been discussed in great detail, but with no clear resolution. Part of the reason being that the term microelectrode was initially used in the 1940s for electrodes with dimensions in the millimeter range, but in the late 1970s it was used for smaller electrodes with dimensions in the micrometer range. It is generally accepted that for an electrode to be considered a microelectrode it must have at least one dimension, the critical dimension, smaller than the diffusion layer thickness, under the experimental conditions employed.1 For the purpose of this review, a microelectrode is defined as an electrode having at least one dimension smaller than 25 µm but greater than 10 nm. It will therefore not include the so-called nanodes,2 with critical dimensions that reside in the nanometer range. For clarity, the critical dimension can be the thickness of the electrode for microring, microband, or tubular microband electrodes, or the radius of the electrode for microdisc, hemisphere, or spherical microelectrodes. When one electrode dimension is below the critical size, the electrode response has been shown to deviate from the standard theory for macroelectrodes and to exhibit some unique properties. Under appropriate experimental conditions, for example during slow scan voltammetry, the voltammetric response of microelectrodes is very
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
different to that observed at electrodes of conventional size because the diffusion layer thickness can greatly exceed the dimensions of the microelectrode (see Figure 1).3 When this occurs, the microelectrode attains a time-independent steadystate response, characterized by a sigmoidalshaped voltammogram. This is similar to the polarograms obtained with a dropping mercury electrode or the current-voltage curves obtained with a rotating disc electrode, but in this case it is due to high diffusion rates under quiescent conditions. To explain this phenomenon fully, a simple model will be described where a microelectrode is immersed in a solution of an oxidizable redox species, with the microelectrode poised at a potential sufficient to oxidize the redox species at a diffusion-controlled rate. Initially, after application of the potential, the electrode perturbs the solution and causes the formation of a diffusion layer that moves out from the electrode into solution. At short times, the diffusion layer thickness, δ, is very thin and so the electrode is much larger than the diffusion layer thickness. Consequently, the nonuniform current distribution resulting from high mass transport of redox species to the edge of the microelectrode (the edge effect) has little contribution to the measured current and the electrode response is described by that of an infinitely large planar electrode (see Figure 2a and b). As time progresses, the diffusion layer thickness increases and eventually exceeds the dimensions of the microelectrode. Under these conditions, the edge effect becomes dominant and results in the 20
i (mA)
10
(a)
(b)
(c)
Figure 2. Illustration depicting the diffusion fields to (a) a macroelectrode (planar diffusion), (b) a microelectrode at short time after application of a potential step (planar diffusion), and (c) a microelectrode at long times after a potential step (radial diffusion).
formation of a spherical diffusion field and the attainment of a steady-state response with high current density (Figure 2c). Conversely, for short timescale experiments, for example with cyclic voltammetry recorded at high scan rate, the diffusion layer thickness is smaller than the size of the microelectrode, semi-infinite planar diffusion is dominant and the voltammetry reverts to the peak-shaped behavior seen at electrodes of conventional size. Figure 3 compares cyclic voltammograms recorded using a 14-µm thick carbon microband electrode in a 10-mM Ru(NH3 )6 Cl3 solution recorded with fast and slow scan rates. The slow scan rate voltammogram displays the characteristic sigmoidal shape consistent with radial diffusion to a microelectrode, while the fast scan voltammetry (FSV) has peaks consistent with planar diffusion.
0 −0.5
−0.25
0
0.25
−10 −20 −30 E vs Ag/AgCl (V) Figure 1. Cyclic voltammograms for a 14-µm-thick screen printed carbon microband electrode in 10 mM Ru(NH3 )6 Cl3 , recorded with scan rates of 10 mV s−1 (black line) and 100 mV s−1 (gray line).
3 ADVANTAGES OF MICROELECTRODES
Microelectrodes, because of their small size, have several advantages compared with macroelectrodes and, as a result they have been the focus of several comprehensive review articles.4–6 As the electrolysis currents recorded during microelectrode experiments are typically small, the impact of ohmic phenomena (iR drop) is greatly reduced making microelectrodes amenable to undertaking electrochemistry in a wide variety
MICROELECTROCHEMICAL SYSTEMS
3
4 TYPES OF MICROELECTRODES AND THEIR FABRICATION
(a)
(b)
(d)
(e)
(g)
(h)
(c)
(f)
(i)
Figure 3. Showing the most commonly used geometries of microelectrodes and microelectrode arrays; (a) microdisc, (b) microring, (c) microband, (d) microcylinder, (e) microsphere, (f) microhemisphere, (g) interdigitated array, (h) microdisc array, (i) random array of microdiscs.
of chemical media including nonaqueous solvents, gas, ice, polymer films, and in low-conductivity aqueous solutions with little or no supporting electrolyte.7 The minimal distortion from iR drop also enables the use of microelectrodes for fast scan rate voltammetry with scan rates of over 1 × 106 V s−1 .8 In addition to this, the response time is reduced since the capacitive (nonfaradaic) response of an electrode decreases with electrode radius, so more information can be gained in the early part of chronoamperometric transient responses and in fast scan rate voltammetry for the investigation of high-speed electron transfer reactions that were previously inaccessible with macroelectrodes. The small physical size also makes them ideally suitable for experimental conditions where either space or sample volume is at a premium, for example during single cell studies in nanoliter volumes.9 Finally, the steady-state response obtained with microelectrodes makes them ideally suitable for electroanalytical applications, since the limiting current is directly proportional to the analyte concentration giving an excellent signalto-noise ratio and as low as zeptomole detection limits.10
Microelectrodes fall into one of two main categories; single or array microelectrodes. Figure 3 shows the most common geometries of microelectrode, which have been fabricated and utilized in the literature. Of these, the most commonly used microelectrode geometry, accounting for approximately 50% of all microelectrode studies,11 is the microdisc electrode (also known as inlaid microdisc). Of the other common geometries, the microcylinder accounts for 20%, microarray electrodes (both random and uniform) account for a further 20%, while the remaining 10% is split mainly between the microband and microring, with a small percentage attributable to the microsphere and microhemisphere electrodes. Fabrication of the single microelectrodes is typically undertaken by sealing a microwire, thin foil, or fine fiber into an insulating material such as glass or epoxy resin. The reviews by Zoski and Murray and coworkers provide highly detailed and comprehensive discussion of the topic including the design, fabrication, and characterization of microelectrodes.12,13 Briefly, microdisc electrodes are fabricated by either heat sealing a microwire into a glass capillary (under vacuum) and then polishing the end of the capillary to yield a microdisc electrode or by inserting the microwire into the capillary and then pulling the metal/glass assembly using a pipette puller. To prepare microcylinder electrodes, the microwire is again inserted into the glass capillary, but in this case a small length of microwire ( fc . (d) At the extreme values of analyte concentration, S(0) (black line) is minimal; for analyte concentrations [A] ≈ K, it rises to a maximal value. The characteristic relaxation time for the interaction between the analyte and the nanopore, τ = 1/2πfc , decreases monotonically with increasing analyte concentration (dashed line). In the limit of zero analyte concentration, τ ∼ 1/koff .
The plot in Figure 5(d) also illustrates that at low analyte concentration, the characteristic lifetime τ (dashed line) equals 1/koff because the dominant timescale is the mean time the analyte is bound to a site on the channel. As the analyte concentration increases, a second timescale, the time it takes the analyte to bind and react with the site, contributes and τ decreases monotonically to zero. The variation of S(0) with analyte concentration depends on both the pK, the product of the number of binding sites, N, and koff . Thus, if the pK and koff are determined from a calibration of the mean current (Figure 5b) and τ (Figure 5d), the number of binding sites can be estimated from the values of S(0).15 This method discriminates particularly well between different analytes that bind to the nanopore
because it makes use of both the thermodynamic (pK) and kinetic (kon and koff ) information. For example, it was demonstrated that a single αHL channel can distinguish between isotopic ion species (i.e., aqueous hydronium and deuterium ions).15,16 The spectral method was subsequently applied to genetically engineered versions of the αHL channel tuned to bind divalent cations.17 3 NEUTRAL POLYMER AND DNA TRANSPORT IN A SINGLE NANOPORE 3.1
Neutral Polymer Probes of Channel Structure Interact with the Pore
Krasilnikov and colleagues developed a method to use nonelectrolyte polymers of poly(ethylene
DETECTION AND CHARACTERIZATION OF IONS, DNA, AND PROTEINS
glycol) (PEG) to estimate the diameter of ion channels.27,28 It is well known that PEG decreases the bulk conductivity of ionic solutions. Thus, PEGs that are sufficiently small enter the pore and decrease the channel conductance. Polymers larger than the diameters of the two channel mouths rarely partition into the pore and therefore have little or no effect on the conductance. The dependence of channel conductance on the PEG molecular weight determines the pore’s PEG molecular mass cutoff, and by inference, the pore diameter. The single-channel recordings in Figure 6(a) illustrate the effect of different molecular mass PEGs on the αHL channel conductance. Note that the relatively large PEG 8000 rarely decreases the conductance. In contrast, PEG 200 significantly reduces the mean current. The ratio of the conductance in the presence of PEG to that in the absence of the polymer demonstrate that PEGs with molecular mass less than 3000 partition into the channel (Figure 6b). The diameter of the αHL channel is estimated from these data and the measured values of PEG hydrodynamic radii29 are indicated on the plot. The single-channel current recordings in the presence of PEG 2000 (Figure 6a) are noisy. Ion current fluctuations should indeed occur when the
polymer randomly partitions into and out of the pore. However, the observed noise is orders of magnitude greater than expected based on the calculated residence time for the polymer diffusion inside the pore. Specifically, the one-dimensional diffusion equation x2 = 2DτD 30 suggests the polymer should diffuse the length of the channel in a time τD ∼ 10 ns. However, the current recordings shown in Figure 6(a), which were filtered to 1-kHz bandwidth, indicate otherwise. Indeed, the mean residence time for PEG in the αHL pore, deduced from the excess current noise, was ∼100 µs.18
3.2
Detecting Individual Polynucleotides that Thread through a Single αHL Channel
The previous result (Figure 6a) indicated that the αHL nanopore can interrogate a polymer for a time much greater than the time taken for the polymer to diffuse through the channel. Because of this property and because the αHL channel can remain fully open for long times, an opportunity is provided to study the details of DNA transport in a highly confined space. Specifically, we demonstrated that individual molecules of single-stranded DNA are
Channel conductance ratio
Hydrodynamic radius (nm) 0.5 1 2
30 pA
1s
(a)
PEG 200
PEG 2000
PEG 8000
1.0
0.8
0.6
Hard spheres Scaling theory
100 (b)
4
1.2
w 3 law
No PEG
7
Ratio of bulk conductivities
10 000 1000 PEG molecular weight
Figure 6. Estimating the size of the αHL channel with nonelectrolyte polymers. Polymers of poly(ethylene glycol), PEG, reduce the bulk conductivity of an electrolyte solution. (a) Sufficiently small PEGs partition into the solitary channel and reduce the current of spontaneously forming channels. (b) The dependence of the single-channel conductance on the polymer Molecular Weight (MW) is used to estimate the PEG MW cutoff, and thus provides an estimate for the diameter of the aqueous-filled channel pore.
8
MINIATURIZED, MICRO AND PARTICLE SYSTEMS +poly[U] cis
Lifetime (µs)
3000
50 pA
−120 mV
300 µs
0
1300 µs
(a)
ssDNA
1500
(b)
+
+++++ + + +
+ ++
400 200 Mean poly[U] length (nt)
dsDNA trans (c)
cis
Figure 7. Polynucleotides are driven into a single αHL channel by a transmembrane potential difference. (a) Single-channel recordings in the absence and presence of single-stranded RNA show transient blockades. (b) Polymer-induced blockade lifetimes (inset: histogram of blockade lifetimes for a given length poly[U] RNA show 3 characteristic lifetime values). The polynucleotide-induced lifetimes for the two slowest blockade types are proportional to the polymer length, which suggests the polynucleotide threads completely through the nanopore. (c) PCR demonstrates that single-, but not double-stranded DNA is transported through the αHL channel from the cis to the trans side.
driven electrophoretically into and through a single αHL ion channel.19,31 Because the mobility of negatively charged polynucleotides is less than that of monovalent ions and the polymer occupies space that small mobile ions normally would, polynucleotides decrease the channel conductance when they are inside the pore (Figure 7a). The lifetime of the polymer-induced current blockades is proportional to the contour length for polynucleotides comparable to or longer than the pore length, (Figure 7b). Polymerase chain reaction technology (PCR) confirmed that singlestranded, but not blunt-ended double-stranded, DNA was transported through the pore. The latter two results strongly suggest that the polymer threads through the pore as a linear rod.19 In that report, we hypothesized that a single nanopore could rapidly sequence DNA if each base caused a unique current blockade level. We subsequently demonstrated that different homopolymers cause distinctly different ion current blockade signatures.32,33 Because DNA sequencing with a single nanopore is yet to be realized, intense investigation continues.34–41 3.3
Sensor Technologies Based on DNA – Nanopore Interactions
For relatively short polynucleotides that interact with the αHL channel, the time-averaged
t Blockade rate ~[polymer]free
Complex cannot enter pore
Complex blocks pore for t ~1/koff
Figure 8. Sensor models based on the interaction between a single nanopore and polymers. It is assumed that polymers with binding sites for analytes have unfettered access to the nanopore. The entry or transport of individual polymers causes a transient decrease in the ionic current (left). Addition of analytes that bind to the polymer change the ability of the latter to partition into (center) and/or transport through the pore (right). In the latter model, the analyte : polymer complex blocks the pore for a time that corresponds to the mean lifetime of the complex (i.e., τ ∼ 1/koff ). [Adapted from Kasianowicz, et al.,42 2001.]
blockade rate increases linearly with the polymer concentration.31 This, in part, permitted simultaneous multiple analyte detection with polymers and a single nanopore (Figure 8a–c).42 Briefly, a binding site for a specific analyte is attached to an αHL channel–permeant polymer. Analyte
DETECTION AND CHARACTERIZATION OF IONS, DNA, AND PROTEINS
binding to this site alters the ability of the polymer to thread through the pore. In the first case, the analyte concentration is deduced from the decrease in the mean number of blockades per unit time (Figure 8b). For the second detection scheme (Figure 8c), the analyte concentration is estimated from the mean time for nanopore occlusion by the analyte/polymer complex after the electric field is applied. Because different polymer types cause different current blockade patterns (see below), a single nanopore can be used to simultaneously detect different analytes. This method is particularly useful because it does not require the analytes to be labeled. In addition, changing the applied potential permits the force on the bond between the analyte and polymer to be varied, which may help identify a particular analyte in the presence of molecules with similar, but not identical properties.
4 ADVANCED SIGNAL PROCESSING METHODS: READING INFORMATION ENCODED IN POLYMERS 4.1
Polynucleotide-induced Current Blockades Characteristic of Polymer Type
As shown in Figure 7, the transport of individual polynucleotides through a single ion channel is easily observed electronically because the polymers occlude the channel and thereby reduce the flow of ions through the nanopore. The length of a polymer can be estimated from the mean current blockade lifetime (Figure 7b). Can information encoded in the polymer be read from the electronic signals? The current recordings in Figure 9(a) illustrate the blockades caused by identical length poly[dT], poly[dA], and poly[dC] molecules. Note that the lifetime and substate patterns for a given homopolynucleotide are clearly distinguishable from those caused by the others. Nevertheless, the variation in individual blockade patterns for a given polymer type requires a stochastic analysis of the signals. The current blockades depend on the characteristics of the polynucleotide (e.g., base composition, secondary structure, and interactions between polymer subunits) and the nanopore (e.g., pore geometry and local electric field in the lumen). The
Fully open
I=0
poly[dT]
poly[dC]
9 poly[dA]
25 pA 2 ms
(a) Open
I=0 Open 50 pA
I=0 5 ms
(b)
Figure 9. Polynucleotides threading through a single αHL channel cause transient ionic current blockades that are characteristic of the polymer. (a) Blockades caused by individual 100-nt long poly[thymine] (poly[dT]), poly[cytosine] (poly[dC]), and poly(adenosine) (poly[dA]). (b) The blockade patterns for poly[dT] depend upon the side to which the polymer is added.
signals caused by 100-base-long homopolymers of thymine (i.e., poly[dT]100 ) are discussed in greater detail because their rich state structure reveals characteristics of the pore geometry. Additionally, the signals provide an example of how information encoded in a polymer might be read electronically via a nanopore. The current blockades shown in Figure 9(b) for poly[dT]100 entering the cis entrance of the αHL channel, as indicated by the cartoon inset, often show a characteristic shallow-then-deep blockade pattern. In contrast, when the polymer is driven from the opposite direction (i.e., trans side), the blockade pattern is reversed (i.e., a deep-thenshallow pattern prevails). Figure 10 illustrates many blockades induced by poly[dT]100 . This “forest from the trees” representation obscures the details of any individual blockades (Figure 9b) but it aids the visual identification of the most probable occluded current states. For example, the three darkest bands in the current recording (Figure 10a, left) and the largest peaks in the current amplitude histogram for ∼104 blockades (Figure 10a, right) correspond to the fully open state and two most probable occluded current
10
MINIATURIZED, MICRO AND PARTICLE SYSTEMS 150
State Open
I (pA)
100
1
50
2 2a 3
0 0 (a)
1000 Time (ms)
2000 Current amplitude histogram
State 100
I (pA)
Open 50
1′
2′ 0
3′ 0
(b)
300
600
Time (ms)
Current amplitude histogram
Figure 10. Ionic current time series for poly[dT]100 transit events with lifetimes between 40 µs and 2 ms (left) and current amplitude distribution (right) for polymers entering the pore from either the (a) cis or (b) trans pore entrances for applied potentials of V = −120 or +120 mV, respectively. The two time series depict the open channel current band and bands corresponding to the three most probable occluded states. The limited number of occluded states may represent the negotiation of the polymer through the various-diameter segments of the pore.
states (states 1 and 3). The speckle between the lower two dark bands (Figure 10b, left) represents a less frequently occurring, but statistically significant state (state 2) and an even less probable state (state 2a). We ignore the latter state in subsequent discussion. The technique for determining the state sequences of individual events relies on Viterbi decoding43 of the dwell times within states of Gaussian mixtures fit to the observed amplitude densities. For the range of event lifetimes under consideration (≤ 2 ms), three blockade states and one open channel state are statistically adequate to describe the amplitude distribution. Using these population distributions, the state sequences in current flow levels from individual molecules as they entered the ∼1.5-nm diameter
opening in the channel were decoded. Under these experimental conditions only hundreds of ions per microsecond flow through the pore when it is partially occluded by a polynucleotide. Thus, statistical techniques are fundamental to characterizing the blockade states. For relatively short lifetime current blockades, the states and state sequences suggest that the polymer–pore interaction is simple. Moreover, it can be demonstrated that the state parameters and their sequence depend on the direction of transit, and how long the polymer and pore interact with each other. A statistical analysis of the current amplitude histogram in Figure 10(a, right), shows that three occluded states result in a good fit to the data. The cis open channel state mean current
DETECTION AND CHARACTERIZATION OF IONS, DNA, AND PROTEINS
4.2
Fully open
1 2 3
I=0
.
.
.
Figure 11. A cartoon that illustrates a possible mechanism for the three most probable transient current blockades caused by poly[dT]100 .
(∼120 pA), and the occluded state mean current values (∼70, ∼46, and ∼17 pA), have probability weights of 0.62, 0.2, 0.03, and 0.15 respectively. Similar results are obtained for events caused by poly[dT]100 molecules entering the trans entrance (Figure 10b). In that case, the trans fully open single-channel current average (∼88 pA), and the three most probable occluded states (∼53, ∼26, and ∼8 pA) occur with probability weights 0.53, 0.04, 0.19, and 0.24, respectively. Interestingly, the ratios of mean current values for each of the three most probable occluded states to the respective mean open channel current for poly[dT]100 entering the pore from the cis side (Figure 10a) do not appear to correspond to those for the three most likely occluded states for polymers threading the pore in the opposite direction (Figure 10b). One interpretation of the latter result, illustrated in Figure 11, suggests that the degree of ionic current blockade correlates with the amount of poly[dT] mass in either the pore vestibule (shallow blockade), or the smallest channel aperture (deeper blockade). This simple hypothesis is consistent with the blockade patterns caused by poly[dT] transport in either direction and the lifetime distributions of poly[dT]-induced blockades. These results, and others shown here, demonstrate that DNA can be used to probe the geometry of αHL channel. By inference, this technique may prove useful for probing the structures of other nanometer-scale pores, including those made in solid-state materials.
11
Blockade State Sequences Evolve with the Event Lifetime
Because the physical length of the homopolymer is constant for the poly[dT]100 experiment, differences in the blockade event lifetime and the state sequences allow us to characterize the physical properties of the nanopore. Figure 12 illustrates how the morphology of poly[dT]100 -induced current blockades evolves with increasing event lifetime over the range from 60 ≤ τcis ≤ 600 µs (i.e., for polymer added to the cis side). Here, the single-channel current time series for ensembles of events at three representative lifetimes (i.e., τcis = 60, 140, and 600 µs) are aligned at the onset of each channel blockade. The colors indicate the frequency of current values; that is, the closer to red end of the spectrum, the greater the number of occurrences. Because few 60-µs events exhibit the deep blockade state (state 3), the polymer most likely only entered the pore vestibule and did not thread completely through the pore. The 140-µs ensemble shows a bifurcation of the event set into shallow blockades, which are qualitatively similar to the 60-µs events, and deep blockades in which the homopolymer most likely was driven past the narrowest diameter of the channel and
∆tcis 60 µs
States 1
140 µs
600 µs
1,3
1,3,1→3
Figure 12. Event time-amplitude histograms showing characteristic ionic current signatures for increasing blockade duration. Poly[dT]100 -induced current blockades stratified by duration show that the event structure evolves in a simple manner. Sixty microsecond long events show a state 1 conductance level; 140 µs long events show a bimodal conductance morphology (state 1 and state 3); 600 µs long events show state 1 and increasingly frequent state 3 blockades. The 600 µs blockades also show the emergence of state 1 to state 3 transitions within the events. These signatures most likely represent the progress of the polymer through the various limiting apertures of the nanopore. In these experiments, the polynucleotide entered the αHL channel from the cis pore mouth. The color scale is adjusted to the declining frequency of longer events to best visualize the common event morphologies.
12
MINIATURIZED, MICRO AND PARTICLE SYSTEMS
threaded through the pore (see Figure 11 center). The 600-µs event ensemble shows a third event morphology, a shallow-then-deep blockade, that is composed of three event types: shallow blockades, deep blockades, and shallow-then-full blockades. Figure 9(a, left) shows three such individual events. 4.3
Extracting Information Encoded in Polymers
Figure 13 illustrates typical single-channel current blockades for poly[dT]100 entering from the cis entrance of the channel, and the Maximum A Posteriori probability estimate of the state sequence (black), subject to the constraint imposed by a persistent Hidden Markov Model (HMM), which applies a penalty for state transitions, superimposed on the actual time series data (gray). For polymers entering the pore from the cis side (Figure 13a), the blockades show the three common patterns of Figure 12; i.e., state 1 only, state 3 only, and the intra-event transition from state 1 to state 3. Also included is an example of a relatively rare event comprised of a transition from state 2 to state 3. A corresponding class of events is observed when the polymer enters the trans side (Figure 13b). However, note that the transitions are from a more occluded state to a less occluded one, with the opposite two-step pattern observed when the polymer enters the pore from the cis entrance. The relatively small number of current blockade states and patterns also
suggests that a simple description of the blockade mechanism at these short event lifetimes may be valid. Figure 13(b) shows individual events illustrating shallow blockades, deep-then-shallow blockades, and events proceeding directly to a deep blockade. A mirror symmetrical morphology is observed in trans-to-cis event ensembles of this duration range: the deep-then-partial blockade. These results most likely reflect the fact that the channel cross section is asymmetric: the narrower segments are closer to the trans pore entrance. Analysis of a large number (∼2 × 105 ) of poly[dT]100 -induced blockade events, from both the cis and trans directions suggest an interesting and potentially important use of polymers as molecular rulers for the ion channel. The wide range of blockade durations, from 20 µs to over 2 full seconds, also suggests that the very long events may represent ssDNA that are folded in ways requiring substantial time periods to unfold and thereby access states that allow transit. The state structure of the amplitude distribution for longer events, characterized by components of a Gaussian mixture, evolves substantially over logarithmic increments of event duration beyond the 60–600-µs range just discussed, and is shown in Figure 14. Progressively greater proportions of long-duration events are spent in deep blockade states and require many more Gaussian mixture components to model adequately than do the deep blockade portions of the short-duration events. We have resolved up to 39 states with the aid of digital
150 cis
trans
100
1
I (pA)
I (pA)
100 1'
50
2
50
3
2' 3'
0
0 0
6 Time (ms)
12
0
6 Time (ms)
12
Figure 13. Typical single-channel current blockades for poly[dT]100 entering from the cis entrance (a) or trans entrance (b) and the maximum-likelihood estimate of the amplitude state sequence (black) superimposed on the data (gray). The technique for determining the state sequences from the data relies on Viterbi decoding of the dwell times.
DETECTION AND CHARACTERIZATION OF IONS, DNA, AND PROTEINS 40 µs–2 ms
0
50
100
2.02–20 ms
150
0
50
100
20.02–200 ms
0 150 I (pA)
50
100
13
200.02 ms–2 s
150
0
50
100
150
Figure 14. Blockade event lifetime histograms for poly[dT]100 molecules that enter the cis side of a single αHL nanopore. For blockade lifetimes over a wide range (40 µs–2 s), virtually all of the events (99%) have lifetimes 0.995), and little protein-to-protein variation was observed as shown in Figure 9(d). The detection limit of the proteins using this staining method was 7.0 ng/band of BSA (signal-to-noise ratio was 3.0) when washing the gel, which is as sensitive as the general staining methods using Dye 1, and it was noted that the sensitivity for proteins was maintained regardless of reducing the staining procedures and time. Other proteins at various concentrations (chymotrypsinogen A, transferrin, IgG) were stained by Dye 1 under the same experimental conditions, and their calibration graphs were constructed as shown in Figure 9(d). As a result, this protein staining using Dye 1 indicated only a slight protein-to-protein variation similar to the result of the general staining method. Under the same experimental conditions, 2D SDS-PAGE was carried out for the separation of 40 µg of mouse brain lysates, and well-resolved gel images were observed. These gel images are shown in Figure 10. The fluorescent image of proteins with some nonspecific spots and small background was observed despite the SDS removal and excess dyes in the gel as shown in Figure 10(a). By washing the gel, the background and nonspecific spots in the gel disappeared, and a clearer gel image was observed as shown in Figure 10(b). As a result, the binding of Dye 1 to proteins during the
Low MW (a)
pH 3
pH gradient
pH 3
pH gradient
pH 9
High MW
Electrophoresis
10
Low MW (b)
pH 9
Figure 10. Rapid staining with Dye 1 for 2D PAGE of mouse brain lysate (40 µg) before (a) and after (b) washing gel.
SDS-PAGE experiment was successful regardless of the 1D or 2D SDS-PAGE gel separation. SYPRO Orange is a commercially available fluorescent protein staining dye which makes it possible to stain proteins in the presence of SDS and is a simple and rapid method.8,9 Although this stain works well with the 1D SDS-PAGE gel, its performance with 2D gels was inconsistent and failed to achieve the sensitivity levels obtained by the 1D SDS-PAGE.10 Moreover, no proteins were stained by SYPRO Orange under the same experimental conditions, because the movement of SYPRO Orange into the gel was prevented by the 1D strip gel. Both Dye 1 and SYPRO Orange could bind to proteins using a simple and rapid method, whereas Dye 1 succeeded in the staining of proteins not only in 1D SDS-PAGE gels, but also in 2D gels with a high sensitivity. From this viewpoint, it was demonstrated that Dye 1 had significantly improved characteristics than the commercially available staining dyes. Our procedures significantly reduced protein staining times for the SDS-PAGE compared to the general method for the SYPRO Ruby staining of 18 h and for CBB at 105 min. For this study, it took 15 min (or 45 min in the case of washing of the gel after SDS-PAGE together with staining). Moreover, the gel image analysis to detect the protein bands in this experiment could be
PROTEIN CHIPS AND DETECTION TOOLS
directly performed after the SDS-PAGE experiment without the labor-intensive treatments such as fixation, washing, and lengthy staining of the general staining protocol using SYPRO Ruby and CBB. Although the detection limit of the protein for Dye 1 (7 ng/band) was slightly lower than that for SYPRO Ruby (1–2 ng/band), the detection limit of the proteins in this study was as sensitive as the short-protocol silver staining methods (7 ng/band), and was much higher than that of CBB (64 ng/band). The SDS-PAGE experiment using Dye 1 made it possible to carry out the highthroughput protein analysis and highly sensitive detection of proteins, which satisfies the requirements in the rapidly growing field of proteomics. The present study demonstrated the highperformance staining for 1D and 2D SDS-PAGE using the novel protein-binding fluorophore, Dye 1. The proteins in the gel could be stained by Dye 1 during the SDS-PAGE experiment by preparation of the electrophoresis buffer solution containing Dye 1 under optimum conditions and by the binding to proteins in the gel during the SDS-PAGE experiment. This method for SDS-PAGE significantly simplified the staining protocols without any loss of the protein-to-protein variation and sensitivity. Recently, proteomic analysis has become an important field and it is increasingly important to refine the techniques generating proteomic data. This highly sensitive, rapid, and easy handling staining method using Dye 1 should be widely applicable and convenient for multiple scientific disciplines including biochemistry, medicine, and pharmacology.
5 PROSPECTS FOR THE FUTURE
The fully automated 2D electrophoresis system developed by the authors is believed to be an epoch-making system that allows anyone to conduct a simple, rapid 2D electrophoretic analysis conventionally limited to a few highly trained researchers. This makes protein analysis more familiar to us; it may be used for laboratory tests in the near future. The global market scale of products related to the fully automated 2D electrophoresis, including instruments and consumable items, reached US$3.13 million in 2003.11 If a chip is developed that allows post-translational
11
modifications including glycosylation and phosphorylation, the market scale will greatly expand in the future. ACKNOWLEDGMENTS
The development of the automated 2D electrophoresis system was financially supported as the High-throughput Biomolecule Analysis System Project by the New Energy and Industrial Technology Development Organization (NEDO), Japan. Authors thank cooperative researchers: Y. Unuma, Y. Maruo, T. Matsushima, K. Takahashi, M. Mieda, M. Nakamura from Sharp corporation, K. Sakairi, C. Hayashida, M. Kano, K. Ueyama from Toppan Printing, I. Namatame, K. Yodoya, Y. Ishii, T. Shibata, H. Inamochi, Y. Nakada, Y. Ogawa, H. Marusawa, T. Komatsu, Y. Saito from Astellas Pharma, K. Yano, S. Akutsu, and I. Karube from Tokyo University of Technology. REFERENCES 1. G. A. Scheele, Two-dimensional gel analysis of soluble proteins. Charaterization of guinea pig exocrine pancreatic proteins. Journal of Biological Chemistry, 1975, 250, 5375. 2. P. H. O’Farrell, High resolution two-dimensional electrophoresis of proteins. Journal of Biological Chemistry, 1975, 250, 4007. 3. D. A. Wolters, M. P. Washburn, and J. R. Yates III, An Automated Multidimensional Protein Identification Technology for Shotgun Proteomics. Analytical Chemistry, 2001, 73, 5683. 4. N. Ramachandran, E. Hainsworth, B. Bhullar, S. Eisenstein, B. Rosen, A. Y. Lau, J. C. Walter, and J. LaBaer, Self-Assembling Protein Microarrays. Science, 2004, 305, 86. 5. K. Usui, A. Hiratsuka, K. Shiseki, Y. Maruo, T. Matsushima, K. Takahashi, Y. Unuma, K. Sakairi, I. Namatame, Y. Ogawa, and K. Yokoyama, A selfcontained polymeric 2-DE chip system for rapid and easy analysis. Electrophoresis, 2006, 27, 3635. 6. Y. Suzuki and K. Yokoyama, Design and Synthesis of Intramolecular Charge Transfer-Based Fluorescent Reagents for the Highly-Sensitive Detection of Proteins. Journal of the American Chemical Society, 2005, 127, 17799. 7. F. Gharahdaghi, C. R. Weinberg, D. A. Meagher, B. S. Imai, and S. M. Mische, Mass spectrometric identification of proteins from silver-stained polyacrylamide gel: a method for the removal of silver ions to enhance sensitivity. Electrophoresis, 1999, 20, 601. 8. T. Steinberg, L. Jones, R. P. Haugland, and V. Singer, SYPRO orange and SYPRO red protein gel stains:
12
ARRAY TECHNOLOGIES
one-step fluorescent staining of denaturing gels for detection of nanogram levels of protein. Analytical Biochemistry, 1996, 239, 223. 9. L. Steinberg, R. P. Haugland, and V. Singer, Applications of SYPRO orange and SYPRO red protein gel stains. Analytical Biochemistry, 1996, 239, 238.
10. J. P. Malone, M. R. Radabaugh, R. M. Leimgruber, and G. S. Gerstenecker, Practical aspects of fluorescent staining for proteomic applications. Electrophoresis, 2001, 22, 919. 11. World 2 D Gel Electrophoresis Market, Frost & Sullivan, Palo Alto, 2004.
55 Surface-Enhanced Laser Desorption/ Ionization (SELDI) Technology Lee O. Lomas1 and Scot R. Weinberger2 1
Ciphergen Biosystems Inc., Fremont, CA, USA and 2 GenNext Technologies Inc., Montara, CA, USA
1 OVERVIEW OF SELDI BIOCHIP TECHNOLOGY
Surface enhanced laser desorption/ionization mass spectrometry (SELDI-MS) was first conceptualized in the early 1990s when Hutchens and Yip demonstrated that chromatographic affinity probes used to specifically enrich fragments of lactoferrin could then be directly presented to a laser desorption/ionization source for mass spectrometric detection.1 The capability of the probe to actively participate in the extraction of the analyte and subsequent removal of sample components that interfere with or suppress ionization was a logical improvement over the classical matrix-assisted laser desorption/ionization (MALDI) applications where the sample probe surface plays a passive role in the analytical scheme and merely presents the sample to the mass spectrometer for analysis; in order to produce usable MALDI-MS signal, crude samples must first be fractionated and purified of any ionization suppressants such as salts, chaotropic agents, detergents, and so on. SELDI, as originally defined by Hutchens and Yip, consists of two subsets of technology: surface enhanced affinity capture (SEAC) and surface enhanced neat desorption (SEND).1 By far the SELDI array technology showing the most utility to date is SEAC and as such is generally
referred to as SELDI in the published domain. In this format, the probe surface plays an active role in the extraction, concentration, and presentation of the analyte and elimination of ionization suppressants. Figure 1 depicts the elements of a SELDI biochip and associated variety of chemical and biochemical SELDI array surfaces used in differential protein expression applications. Chemical surface arrays are derivatized with classic chromatographic separation ligands such as reverse phase, ion exchange, immobilized metal affinity capture (IMAC), and normal phase media. Such surfaces, with broad binding properties, are typically used for general protein profiling and de novo biomarker discovery, where large populations of proteins are compared (e.g., from diseased vs normal samples) with the goal of elucidating differentially expressed elements. Biomolecules bind to these surfaces through hydrophobic, electrostatic, coordinate covalent bond or Lewis acid–base interaction, the strengths of which can be directly modulated by modifying the binding and/or washing buffer compositions. Biochemical arrays are created by immobilizing bait molecules upon the surface of preactivated SELDI surfaces via covalent attachment using either primary amines or hydroxyl groups. In this way, specific protein-interaction arrays of virtually any content may be created, including antibodies,
Handbook of Biosensors and Biochips. Edited by Robert S. Marks, David C. Cullen, Isao Karube, Christopher R. Lowe and Howard H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.
2
ARRAY TECHNOLOGIES
Functionalized hydrogel
Hydrophobic barrier
Chemical surface Reverse phase Quaternary a mine (anionic) Carboxymethyl (cationic) Nitrilotriacetic acid (IMAC) Silica (normal phase) Biochemical surfaces Carboimidazol reactive Epoxy reactive
SELDI biochip Figure 1. Elements of a SELDI biochip. The array consists of a glass-coated aluminum strip that displays discrete affinity locations or spots. Each spot incorporates a hydrogel that is functionalized with classical chromatographic ligands, such as C9–C12 aliphatic chains (reverse phase), quaternary amines (strong anionic), carboxymethyl (weak cationic), nitrilotriacetic acid (IMAC) and silicon oxide (normal phase). The biochips also incorporate a hydrophobic barrier that surrounds the spot locations and prevents sample movement between spots. Dimensions and spot locations are such that 12 strips side-by side provide the standard 96-well microtiter footprint and sample processing can be achieved using standard robotics.
receptors, enzymes, DNA, small molecules, ligands, and lectins. In contrast with standard chromatographic media, these biochemical surfaces provide a greater degree of enrichment of captured analytes, because of the high specificity of biomolecular interactions. Because specific biochemical interaction motifs demonstrate high affinity and low equilibrium disassociation constants, biochemical surfaces facilitate a vast array of microscale experiments that facilitate the analysis of very low sample volumes. Such experiments include SELDI immuno assays, targeted protein identification and/or purification, ligand binding domain analysis, epitope-mapping experiments, post–translational modification detection, as well as reliable quantitative studies, even when fishing for target proteins within a complex biological sample. When compared to MALDI, SELDI-based arrays have demonstrated not only a simplified workflow, but also improved analytical sensitivity and associated mass detection limit. The latter is attributed to a marked reduction in sample loss
inherent in combining and miniaturizing sample processing workflow. 2 USE OF SELDI IN PROTEOMICS 2.1
Differential Protein Display and Biomarker Discovery
Over the course of the last decade, SELDI technology has mostly been applied to challenges of proteomics research. Among today’s most popular proteomic research activities is differential protein display or expression monitoring. Differential protein display is a comparative technique that contrasts protein profiles between different organisms, individuals, pathogenic and/or metabolic conditions, and phenotypic response to environmental or chemical challenges. Unlike differential studies of transcription, differential protein display studies are not easily enabled by amplification strategies such as reverse polymerase chain reaction (PCR). In this manner, differential protein
SELDI TECHNOLOGY
studies require approaches that isolate and enrich both major as well as minor protein constituents from a complex biological mixture, with specific attention to preanalytical and analytical biases that may compromise conclusional integrity. The proteomic community performs differential protein display for biomarker discovery in two formats; the so-called top down and bottom up approaches. SELDI has been used most extensively in the top down approach whereby protein mass signatures within a biological sample are detected and compared without any further reduction to peptide fragments by post sample collection methods such as global tryptic digestion. The implication of this is that proteins and protein modifications due to the biology of the system are preserved through to detection as they may be indicative of the biological question to be answered. For such experiments, the workflow is performed in two phases; a scouting phase whereby a large number of chromatographic separations are performed using multiple array chemistries and binding/washing conditions, and a validation phase whereby select differentially expressed protein candidates are validated using only the chromatographic surface and binding/washing condition that gave the initial differential profile. As an example, the process of SELDI-based sample preparation using four different chemical surfaces is demonstrated in Figure 2. A series of orthogonal SELDI surfaces including reverse phase, anionic, cationic, and IMAC loaded with a transitional metal such as Cu2+ , Ni2+ , or Zn2+ are arranged in plate format. A complex biological sample is deposited upon every chemically active “spot” of each array. After binding, the spots on each array are washed with appropriate buffers in gradient manner. Within a given array, subsequent spots experience a greater degree of stringency, removing analytes with comparatively weaker surface interaction potential and enriching for those of strong surface affinity. In some cases, particularly when using specific biomolecular interactions, purification to almost-complete homogeneity is possible, without substantial loss of analyte. Benefits of this workflow also reveal insight into the physical–chemical properties of the analyte retained on the surfaces such as hydropathicity, charge, pI, and post–translational modifications such as phosphorylation and/or glycosylation. Such de facto knowledge can be further exploited
3
particularly in analyte purification whereby such knowledge can be used to design efficient scaleup purification strategies by matching the array chemistry and binding/washing conditions used for discovery with that of a more conventional chromatographic bead chemistry in a column format. This is particularly useful when additional purified material is needed for the purpose of protein identification and further characterization. In contrast to research-based proteomics, clinical proteomic studies endeavor to follow the progress of disease within an individual or a small population with the ultimate goal of finding biomarkers potentially useful as diagnostic agents or new drug targets; this topic is now known best as “translational medicine”. Under such circumstances, sample or tissue availability is limited and the dependence upon highly efficient, small-scale techniques is becoming more and more essential. Typically protein populations between groups are compared using univariate and/or multivariate statistical analysis schemes with the ultimate goal of elucidating a protein or groups of proteins whose expression levels correlate with a given clinical condition.2 Automation requirements here are primarily focused upon running many samples in a massively parallel manner and only proteins of interest are further characterized to provide insight into identification and post–translational modifications, or to provide insight into the disease mechanism or host response to disease. Because of its reproducibility, throughput, and starting material requirements, SELDI-MS has gained acceptance as a tool of choice for clinical proteomic studies.
3 SELDI CLINICAL PROTEOMIC STUDIES
Clinical proteomics aims to scan the realm of expressed proteins to identify biomarkers that can answer specific clinical questions. The most obvious are markers that can be used for diagnosis or prognosis. Another important issue that clinical proteomics promises to help resolve, is the ability to predict a patient’s response to a specific drug. For example, diagnostic markers can themselves be candidates for drug targets and pharmaceutical companies pursue clinical proteomics to identify markers that predict toxicity of candidate drugs. The overriding determinant of the success of a clinical proteomics program is the choice
4
ARRAY TECHNOLOGIES ∆ Salt
Cu2+ Ni2+ Ga2+
∆ Organic
∆ Imidazole
∆ pH
Q
CM
Chaotroph
RP
IMAC
IMAC-Cu2+ 20 Condition 1
10
Peak intensity
0 20 Condition 2 10 0 60 40
Condition 3
20 0 2000
4000
6000
8000
10 000
Molecular mass (M/z)
Figure 2. Typical workflow for a discovery phase using SELDI biochips. A series of arrays of different chromatographic characteristics are screened, such as anionic (Q), cationic (CM), IMAC, and reverse phase. Samples are incubated on chemistries under a number of different binding/washing conditions to modulate the specific population of proteins retained by the surface. After final wash to remove any buffer components that may interfere with MS detection, MS spectra are generated that represent a protein fingerprint of that sample under the specific chemistry and binding conditions; which can then be compared across samples such as control versus disease to identify differentially expressed proteins.
of clinical question followed by careful study design and implementation.3 The underlying clinical question will drive the decision on the choice of proteomics technique, the success criteria, how many samples to examine, how to analyze the data, and ultimately, whether the clinical proteomics
program is a success. Therefore, the initial task is to balance the advantages of simplifying the clinical study with the practical utility of the outcome. Obviously the more limited the population on which an outcome is based, the more limited the population the outcome can be applied to,
SELDI TECHNOLOGY
unless additional work is performed to demonstrate the validity of the biomarkers for a more general population. The importance of proper study design cannot be more exemplified than by a study published by Petricoin et al.4 that described the grail of clinical proteomics, the capability of a panel of biomarkers to distinguish early-stage ovarian cancer with 100% sensitivity, 95% specificity and a positive predictive value of 94%. Although this study energized the research community, subsequent scrutiny indicated a number of study design flaws that ultimately provided a conclusory bias that is now generally acknowledged.5 Although this study polarized the proteomic community with respect to the utility of biomarkers, it did solidify for all the importance of all aspects of study design including sampling methods, sample collection, preparation and data generation, candidate biomarker selection based on appropriate statistical analysis, biomarker validation, identification, and finally quantitative assay generation. Today, clinical proteomics is becoming a concerted effort that involves biostatisticians, clinicians, and the bio-analytical core facilities. During the course of the last two years, researchers have used SELDI array technology to perform biomarker discovery research in a variety of diseases including infectious disease,6–8 Alzheimer’s disease,9–13 and cancer.14–17 Further information regarding SELDI array technology and clinical proteomics research is found in the following recent reviews.18–23 As with research proteomics, once the samples are defined and appropriately collected, a typical SELDI clinical proteomic study begins with a discovery phase, in which assay conditions are tested on a relatively small number of samples. Usually, profiling proceeds with at least 30 samples in each classification group (e.g., disease vs healthy or treated vs untreated). This number of samples is usually enough to yield greater than 90% statistical confidence in single markers with p values