This chapter will start with a short introduction into immune reactions. Then optical evanescent field immunosensors will be discussed and we will demonstrate an SPR immunosensor, based on a differential detection of the SPR angle. We will also show how SPR can be used for multichannel detection of layer growth on one and two-dimensional arrays of sensor surfaces.
Many studies have been devoted to the adsorption of proteins onto solid surfaces and the immunological reaction between antibody and antigen. Such studies are needed for the development of diagnostic assays, biomaterials and purification methods for proteins.
The presence of certain substances in blood or urine can be detected using diagnostic assays, based on the very specific immune reaction between antigen and antibody. These assays can be manufactured by first adsorbing a layer of antibodies on a solid phase. When a blood or urine sample, containing the antigen against which the antibody is directed, is brought in contact with the coated solid phase, the specific immune reaction occurs and the antigen will bind to the antibody. This binding, resulting in a layer growth, can be detected directly by optical methods, or indirectly by adding a labelled antibody that will bind to the antigens bound to the first layer of antibodies.
In the development of biomaterials to be used for implants one frequently aims at the precaution of blood clotting and protein adsorption on the surface of those materials. The assay technique may then be used to study which proteins adsorb onto such surfaces.
For protein purification purposes, a ligand can be bound covalently to a hydrophilic surface. The binding of specific proteins to this ligand should then be reversible, and leave the protein intact also upon release from the ligand.
Cells of a host animal will start to produce antibodies when a compound foreign to the body (an antigen) is introduced. These antibodies will react specifically with the introduced antigen, eventually destroying it. The parts of the antigens that evoke this immune response are known as antigenic determinants. After some time, a heterogeneous population of antibodies will be created, each of which will only recognize one specific antigenic determinant. However, the different antibodies (polyclonal antibodies) are specific for different sites at the antigen, and many antigenic determinants will be recognized. Monoclonal antibodies (all specific to the same antigenic determinant) can be produced by selecting and cloning individual antibody cells. This makes it possible to produce antibodies with a constant quality, and to improve their specificity in diagnostic assays.
A frequently applied method to immobilize proteins at a solid surface is physical adsorption. The increase in entropy by the release of ordered water molecules at the surface of the protein and substrate make adsorption energetically favorable. Most proteins adsorb very rapidly onto hydrophobic surfaces and almost no desorption occurs when the surface is washed with a buffer solution. Another immobilization method is the formation of covalent bonds between the protein and the surface. For this work however, only physical adsorption was used for the immobilization of proteins.
Proteins have compact, tightly folded structures, made up out of one or more polypeptide chains. Under physiological conditions, globular proteins spontaneously adopt their native conformation. Because many proteins have an elliptical shape their orientation at the interface is important as well. It determines the thickness of the monolayer and whether the binding sites for the antigens are exposed to the aqueous medium.
The reversible reaction between an antigen binding site at the antibody (Ab) and a certain site at the antigen (Ag), can be described by
with ka and kd as the association and dissociation rate constants, and AbAg indicating the antibody-antigen complex. The affinity constant Ka (l/mol) is defined by
The nature of the formation of the immune complex obviously depends on whether the antibodies are monoclonal or polyclonal. In the former case, only one antigen can react with every antibody, while in the latter more antigens can bind to one antibody.
For diagnostic assays either the antibody or the antigen is immobilized on the solid phase, and therefore a number of factors influence the formation of the immune complex: (i) Diffusion of the immune compound in the solution can limit the reaction, especially at low concentrations. (ii) The immune reactivity of the adsorbed protein might be affected by conformational changes that can take place during the immobilization. (iii) The orientation of the protein molecule after immobilization might make the binding sites inaccessible. (iv) If the surface concentration of the immobilized protein is too high, the binding sites might not always be accessible for the complementary immune compound, because of steric hindrance.
In immunosensing the formation of the immune complex is detected via an appropriate transduction event. Transducing systems for direct detection of antibody-antigen interactions include gravimetric, potentiometric, acoustic and optical devices.1,2
Sensitivity is obviously an important requirement for these sensors, but also speed, the ability to do measurements on low volume samples, and the possibility to regenerate the sensor after use are important. Preferably it should be possible to measure in more than one channel for reference, difference or multiparameter measurements. For successful commercialization the instrument should also be small, cheap and easy to use.
In the last years, progress has been made particularly in the field of optical evanescent field immunosensors. With these sensors, biomolecular interactions within the evanescent field are detected via a change in the refractive index or layer thickness at the sensor surface, and labelling of antigens is not required. A number of different configurations has been used.
In grating coupler waveguide sensors3 the effective refractive index of a guided mode is changed by the adsorption of molecules at the waveguide surface. The change is detected as a shift of the incoupling angle for the excitation of the guided mode. For an optimal excitation the output at the end face of the waveguide is maximized.
The refractive index changes within the evanescent field of waveguides can also be detected interferometrically. In the Mach-Zehnder waveguide interferometer4 a light beam is split and made to propagate through the waveguide in a reference arm and a signal arm. After that, the light is recombined and the interference signal results.
In the waveguide SPR sensor5 instead of a prism, coupling between a dielectric monomode waveguide and a metal cover layer is used. This makes the design of integrated sensors easier.
The resonant mirror (RM)6,7 is aimed at combining the enhanced sensitivity of the waveguiding devices (due to the longer interaction length as compared to SPR sensors) with the ease of construction and operation of SPR sensors. In a configuration similar to that of Kretschmann, the metal layer is replaced by a dielectric layer structure. The waveguide is interrogated by a combination of p and s polarized light. If p and s polarized light excite corresponding modes in the waveguide, the phase difference between the reflected p and s beams is p,6 which can be easily measured by using an appropriately placed polarizer in the reflected beam. A sharp peak is thus observed in the reflectance as a function of the angle of incidence. The position of this peak is sensitive to changes at the sensing interface.
Because in waveguides in most cases p and s polarized modes can be excited, more information can be obtained and the values for layer thickness and refractive index of the adsorbed layer can be separately determined, unlike with common SPR systems.8 Waveguide sensors in general also have the advantage of a higher sensitivity due to a propagation length which is much longer than in SPR sensors.8 The interaction length in SPR devices can, however, be adjusted by choosing an appropriate wavelength and metal layer. By doing this, the interaction length (~ propagation length of the SPs) can be varied from less than 1 mm in the UV to several mm in the IR.9 Moreover, SPR sensors are more surface specific, because their evanescent field is confined to a smaller distance from the interface. The much shorter propagation length makes SPs less sensitive than guided modes,8 but at the same time this makes SPR a better option for 2D array multisensors. The smaller sensitive area also implies that smaller sample volumes are needed for the same surface density. Finally, high light intensities can easily be coupled in, and increase the signal to noise ratio.
SPR sensors have also been used as gas sensors for halothane,10,11 ammonia12 and NO2,13 after coating the metal layer with an appropriate chemical sensing layer, and even as a temperature sensor.14 Most often however, they are used as immunosensors for the label-free detection of biomolecular interactions in real time.15,16 Applications for immunosensors can be found in medical science (diagnostic assays, drug monitoring), environmental monitoring (pesticide detection) and industrial processes (concentration determination, in-line process monitoring). The biomolecular interactions that have been studied include the kinetics of association and dissociation of antigens and antibodies, the activity of enzymes and hormone-receptor interactions. SPR sensors have also been used to assay for specific nucleic acid sequences.17
Although evanescent field optical sensors have a high sensitivity, their specificity as biosensors has to be provided by biochemistry. When molecules are studied that are not directly detectable (i.e. the refractive index or thickness change is too small), the small molecules can be immobilized and detected using biomolecular recognition by larger molecules. The sandwich assay is another method that can be used when a layer growth is too small to be detected directly. In this case the antigens are 'labeled' with a secondary antibody, before or after binding to the first antibody at the surface. Other possibilities include displacement assays, replacement assays and inhibition assays.18 Very often however, direct monitoring of the binding of the antibodies to the antigens immobilized on the sensor surface is possible.
Different detection principles have been used to detect the resonant excitation of surface plasmons and the shift in the condition for this excitation caused by refractive index changes within the evanescent field. The most straightforward method is probably to keep the incident angle fixed at a value for which the reflectance is half way down its minimum and monitor changes in the reflectance.9,10,13,16,19 However, the linear region is rather small and the incline (or width) of the resonance curve might change appreciably during the monitored process. Angular scans using a rotation table offer the possibility of using Fresnel theory to obtain the layer parameters from a fit of the reflected intensity as a function of the angle of incidence over a large range of angles. A single scan however usually takes several minutes, and thus even slow changes cannot be monitored in real time.14,20-22 Angular scans can be speeded up by using a scanning mirror.23
It is also possible to measure the reflectance curve fast by using a convergent light beam covering a suitable interval of incidence angles reflected at the prism base and imaged onto a photodiode array. The resonance angle can then be determined from the position of the minimum on the detector array (using a fitting algorithm).24 Using the second dimension of a CCD detector a one-dimensional array of sensing areas could in principle be used simultaneously.20,25-28 This however, has not yet been demonstrated in a sensor measurement.
Another possibility is to use the wavelength dependence of SPR. The angle of incidence is kept fixed and a broadband lightsource is used. Only one wavelength will couple to the SP, and a spectral analysis of the reflected light will show an absorption dip at this wavelength.29 This detection principle has also been used to detect SPR in a miniaturized setup where the plasmons are excited in the metal-cladded tip of a multimode fiber. At the fiber end face the light is reflected by a mirror and half of the light is guided back to a spectrometer where the resonance wavelength is measured.30
Using an acousto-optic tunable filter (AOTF) the differential reflectivity as a function of the wavelength has been measured. To do this, a frequency modulation is added to the drive signal of the AOTF to modulate the wavelength of the output. The wavelength modulated reflectance is measured by a lock-in amplifier which results in a signal which is proportional to the differential of the reflectance with respect to the wavelength.31
With an acousto-optic deflector (AOD) the differential reflectance with respect to the incident angle has been measured as well. This has been published as a patent only.32 We will use a smaller, cheaper and simpler way to modulate the incident angle and present results of immunosensor measurements using this system.
We will use a beam with a modulated angle of incidence in a SPR sensor device. Thereby, only angle dependent signals are measured, using a lock-in amplifier. Apart from the advantage for the signal to noise ratio this also implies that we can measure the first derivative of the dip in the SPR curve. This makes feedback to the incident angle possible, keeping it at the SPR angle (null detection), and results in a measurement with a very high dynamic range. With a device (see Fig. 2.1) that can be very small the SPR angle can be measured as a function of time, monitoring an adsorption or an immunoreaction, as will be demonstrated. The angular measurement can be made into an absolute one, if the critical angle is included in a scan.
By measuring the DC reflectance as well as the AC component and second harmonic, the three main features of the SPR reflectance curve can be measured simultaneously: dip depth, position and width. In general, changes in the refractive index profile in the evanescent field will result in a change of any of these three parameters of the SPR reflectance minimum. These values can give a quick Lorentz approximation of the SPR reflectance curve around the SPR minimum (see Fig. 2.2):
with xc and d as the position and depth, and w as a measure for the width. The second derivative in the minimum is directly related to the width:
or for \(d\ll w\) :
In the case of a bare metal layer the position, depth and width values could even be used to directly calculate approximate values for the complex refractive index and the thickness of the metal layer.33
A spread in the thickness of a measured layer (due to roughness) is known to change the shape of the resonance minimum.21 An antibody-antigen / aqueous interface is an example of such a rough surface. Theoretically and experimentally this leads to an increase in the width of the resonance curve as compared to the case of a smooth layer with the same average thickness. The position of the resonance angle however, is not influenced.14 Therefore, the measurement of resonance depth and width will give more information than a measurement of the dip position only.
If feedback on the incident angle with the rotation table is not used, the sensor output is still linear around the resonance minimum. The shape of the reflectance curve around the resonance minimum is similar to a parabola and therefore, the first derivative is almost linear. With a high frequency, low amplitude modulation of the incident beam the SPR measurement is fast and the beam probes one fixed spot on the surface. This makes this detection method a very good option for scanning a two-dimensional array of sensor surfaces. In this case the second derivative would give the (change of the) sensitivity during a measurement; something that normal fixed angle measurements cannot do.
A standard HeNe-laser (2 mW) was used as a light source. The angle of incidence was modulated by placing a piezo-electric strip with a small mirror in the light path. By applying a sine voltage (255 Hz) to the piezo actuator the strip will bend slightly and the angle of incidence will be modulated in an approximately sinusoidal way (typical amplitude: <0.1 deg.). The modulated lightbeam is used to excite surface plasmons in the Kretschmann configuration (Fig. 2.1). A simple cuvet is pressed against the 46 nm gold layer evaporated (1 nm/s at 10-6 mbar) on the surface of a 60° prism (BK7 glass), allowing measurements to be made in liquid media. The prism / cuvet combination can be rotated with a rotation table (MicroControle Systems; accuracy: 1 mdeg.), to feedback the AC signal to the angle of incidence or to make an angular scan. A photodiode connected to a preamplifier and a lock-in amplifier (Stanford Systems Inc.) is used to detect the modulated reflectance (DC, AC and second harmonic).
To render the gold layer hydrophobic, a thiol layer (mercaptopentadecane) was deposited on top of it by self-assembly from a solution in ethanol. Proteins were obtained from Sigma (St. Louis, USA) and used without further purification. All solutions were made using phosphate buffered saline (PBS pH 7.3).
Figure 2.3 shows the result of an angular reflectance curve measurement for the bare gold layer in air. Using the setup described the reflectance can be measured as a DC, AC and second harmonic signal with the lock-in amplifier. These signals correspond to the reflectance and its first and second derivative with respect to the angle of incidence. In the first derivative we can see a peak caused by the sudden change of the slope of the reflectance curve at the critical angle. Since the critical angle only depends on the media on either side of the layer system, it can be used as a reference to turn the relative angular measurement into an absolute one. At the SPR angle we have a zero crossing for the first derivative. In a sensor measurement we can use the AC signal as a feedback to follow the SPR resonance angle. By interpolation, using the error signal, the angular resolution can be further improved, not limited by the angular resolution of the rotation table but by the signal to noise ratio (see Fig. 2.4(a)).
To get an idea of the physical stability, a 12 hour sensor measurement with feedback was performed in air, in which case no significant shift of the SPR angle is expected (see Fig. 2.4(b)). The slope of the zero crossing was used to estimate an appropriate threshold value that the feedback signal should exceed before the incident angle was incremented by 1 mdeg. using the rotation table.
Using Fresnel theory we can calculate the sensitivity of the SPR angle to a protein layer growth (approximate refractive index: 1.5) in an aqueous medium (refractive index: 1.33). This results in an almost linear relation with a slope of about 0.26 deg/nm, which means that an angular shift of 1 mdeg. (externally measured) corresponds to a layer growth of about 4 pm. Using literature values for the refractive index of water as a function of temperature,34 the dependence of the resonance angle on the temperature of the aqueous bulk medium can be determined using Fresnel theory as well. Experimentally the influence of the temperature is higher than expected theoretically (see Fig. 2.5). Most probably this means that the temperature influences the dielectric constant of the metal layer and maybe also that the temperature of the water near the interface is higher (the slope of the experimental curve in Fig. 2.5 is comparable to the theoretical slope around 50°C).
Figure 2.6 shows the result of an experiment where an immunoreaction was monitored with the setup. The hydrophobized gold layer was first coated with human serum albumin (2·10-6 M hSA). After a washing step the content of the cuvet was exchanged for 2·10-7 M \(\alpha\)hTRF (anti human transferrin) to test for specificity. No growth was observed until (after several washing steps) 10-7 M \(\alpha\)hSA was used. As can be seen in Figure 2.6(b) the accuracy of the angular measurement is better than 1 mdeg. The slight increase in the signal after the washing steps can probably be explained by temperature changes. The use of a flow cell would reduce the measuring time greatly, thereby reducing drift problems. When a flow cell is used one can also completely exchange one solution for another, which is not possible with a static cuvet, without drying the interface.
The result of measuring DC, AC and second harmonic reflectance signals during an immunoreaction is presented in Figure 2.7. The position of the SPR dip shifts as it did in the last experiment, the dip depth however hardly changes. The second derivative of the reflectance at the resonance angle depends on temperature (measured in contact with the metal cuvet) as well as surface structure, in a way that is not clear from this experiment alone. The resonance width is expected to decrease with an increase of temperature (a lower refractive index leads to a lower SP damping), and increase with increasing surface roughness (the roughness leads to a higher damping).
The linear range of the differential measurement without feedback is about 2 deg. as can be seen in Figure 2.8. This range is generally sufficient for immunosensor measurements. It simplifies the setup, and makes fast multichannel measurements possible, by scanning the modulated laser beam over a number of sensing areas on one sensor surface. Because the modulation amplitude is small, the movement of the spot is much smaller than the spot itself, which therefore is essentially stationary on one sensing area. Using a glass cube rotated by a stepping motor the beam with a fixed angle of incidence (except for the modulation) was scanned past three sensing areas, without feedback. A multichannel flow cell was used to lead different solutions past the sensing areas. The result of this experiment is shown in Figure 2.9. By rotating the glass cube around another axis two-dimensional arrays could be scanned as well.
The SPR angle could be measured accurately (resolution better than 1 mdeg.) and fast (several ms) using a differential measuring technique where the angle of incidence was modulated. Using a lock-in amplifier gave the usual advantages of an improved signal to noise ratio and the elimination of the influence of ambient light (the measurement can continue during sample manipulation). A rotation table was used to do measurements with feedback and a high dynamic range. When the rotation table is left out, the dynamic range is still sufficient for immunosensing. This allows the modulated beam to be scanned over an array of sensing areas without feedback, as was demonstrated. The light spot is stationary and the measurement is fast; these are essential advantages for this option. The fact that almost no moving parts are used add to the simplicity and the mechanical stability of the setup.
In this section we will demonstrate how SPR can be used for multichannel immunosensing measurements. For the first time, four separate immunoreactions have been monitored simultaneously in real time. To do this, a plasmon carrying gold layer against which a four channel flowcell was pressed, was imaged at a fixed angle of incidence. Taking the method a step further we have first coated the four channels with antibodies then turned the flowcell by 90° in such a way that the flow channels overlapped the areas coated in the first step. In a second step antigens were applied to the different antibodies on the surface. Thus all antibody-antigen combinations can be measured in a two-dimensional array of sensor surfaces in real time.
There are a number of reasons why multisensing is desirable: (i) simultaneous measurements save time; (ii) leading a solution past a number of sensor surfaces is preferred when only a low sample volume is available; (iii) multi-analyte mixtures can be measured; (iv) reference and duplo measurements can easily be included.
Because SPs have a short propagation length (for gold and in the visible part of the spectrum typically in the mm range), they are particularly well suited for imaging a two-dimensional array of small sensing areas. All sensing areas can be present at the same sensor surface, and the imaging system allows measurements to be made on different spots of the sample simultaneously and independently. The fact that the sensor surfaces are visually observed in real time (and thus problems such as passing air bubbles can be detected) is a practical advantage of this method. By recording the images on video tape the size and position of the sensor surfaces can also be chosen after the measurements. Other advantages are the inherent sensitivity of surface plasmons and the instrumentational simplicity. Finally, because small sensor surfaces can be defined relatively few molecules are needed to obtain a certain surface density.
The experimental setup is shown in Fig. 2.10. The sample delivery system consisted of a four channel flowcell and peristaltic pump. The flowrate used was 4.9 ml/hr. Layers of 2 nm Ti and 47.8 nm Au were evaporated on a glass slide which was brought in contact with a prism (BK7 glass; n=1.515) using a suitable matching oil. A light beam from a 2 mW HeNe laser (l=632.8 nm) was expanded to a 2 cm diameter and attenuated by a factor of 300. The angle of incidence was chosen slightly to the left (see Fig. 1.8) of the minimum in the reflectance curve, therefore the increase in reflectance will be approximately linear with the increase in layer thickness. After internal reflection in the prism the light was imaged by a large diameter lens (f=90 mm) on a CCD video camera with a linear response (VCM 3250; Philips). Images were recorded on videotape for later analysis. A video digitizer (VisionPlus AT OFG; Imaging Technology, Inc., Woburn, MA) was used to calculate the
average intensity of the defined sensor areas in the image as a function of time when the tape was played back. All chemicals were kindly provided by Organon Teknika (Boxtel, The Netherlands). Solutions were prepared in PBS (pH=7.3). Human chorionic gonadotrophin (hCG) and luteinizing hormone (LH) were used as antigens. Three different monoclonals of the ahCG antibody ([1C],[7B],[3A]) were used as well as an aLH antibody. Bovine serum albumin (bSA) was used to block unoccupied sites after the initial adsorption of antibodies to avoid unspecific adsorption of antigens.
One-dimensional array of sensor surfaces
As an initial experiment the adsorption of bSA was measured with three different concentrations and a reference channel without bSA. Fig. 2.11 shows the result of this experiment which agrees with the expectation that for a lower concentration the adsorption will proceed slower.
Having demonstrated the feasibility of a multichannel sensor based on SPR, we proceeded with real immunosensor measurements. First, the four channels were coated with 10-6 M ahCG [7B] by adsorption to the gold surface (see Fig. 2.12). To avoid non-specific binding of the corresponding antigen to unoccupied sites at the surface, these were blocked by bSA (10-5 M). Three different concentrations of hCG were used for the immunoreaction (2·10-7 M, 4·10-8 M and 8·10-9 M). In the fourth channel 8·10-9 M LH was added, which is known to have a high crossreactivity with ahCG because of its structural similarity with hCG. In the right part of Fig. 2.12 we see that the occurrence of the immunoreaction can be measured very well. The influence of the concentration on the kinetics of the immunoreaction is also apparent. The LH molecules indeed show a very high crossreactivity with the ahCG, but the measurement also shows that a part of them are washed away when the solution is exchanged for the buffer solution.
Two-dimensional array of sensor surfaces
Even more measurements can be performed simultaneously if a two-dimensional array of sensor surfaces is used for measurements. Figure 2.13 explains the approach that was chosen. In a first step the channels are coated with antibodies A,B,C,D. Then the multichannel flowcell is removed from the surface, turned 90° and pressed to the surface again. In a second step the antigens 1,2,3,4 are applied, flowing past the sensor areas with the different coatings. In this way all specific and crossreactivities can be measured simultaneously. Of course it is also possible to include duplo and reference measurements.
Figure 2.14 shows the result of a two-dimensional multichannel measurement. The four channels were coated with 3·10-7 M solutions of the ahCG monoclonals ([1C],[7B],[3A]) and aLH. After the blocking step with bSA the flowcell was turned 90°. In the second step the first two channels were used for a duplo measurement with 2·10-7 M hCG. In the third channel 10-8 M LH was used, while the fourth channel was used as a reference channel. The A4 area was used to correct for disturbances due to ambient light etc. Rows 1 and 2 show similar results, except for the A1 area. The results in column B are very similar to those in Fig. 2.12. The sensor areas for the 2D measurement were four times smaller (about 1 mm2) than those used during the immobilization of the antibodies, explaining the difference in the noise level. The drift in the measurements is probably mainly of a chemical nature, because mechanical drift would have the same effect on all sensor surfaces.
The feasibility of one and two-dimensional multichannel immunosensing was demonstrated. The problem of immobilization of a number of different analytes on the multisensor surface was tackled by using a multichannel flowcell. In Ref. 35 it was demonstrated that small quantities can also be deposited in specific locations on a sensor surface by using an ink jet nozzle. In our experiments the immobilization by adsorption as well as the immunoreaction were measured.
(1) Lowe, C. R. Phil. Trans. R. Soc. London 1989, B324, 487.
(2) Thompson, M.; Krull, U. J. Anal. Chem. 1991, 63, 393.
(3) Tiefenthaler, K. Biosensors Bioelectron. 1993, 8, xxxv.
(4) Heideman, R. G.; Kooyman, R. P. H.; Greve, J. Sensors and Actuators B 1993, 10, 209.
(5) Harris, R. D.; Wilkinson, J. S. Sensors and Actuators B 1995, 29, 261.
(6) Cush, R.; Cronin, J. M.; Stewart, W. J.; Maule, C. H.; Molloy, J.; Goddard, N. J. Biosensors Bioelectron. 1993, 8, 347.
(7) Buckle, P. E.; Davies, R. J.; Kinning, T.; Yeung, D.; Edwards, P. R.; Pollard-Knight, D.; Lowe, C. R. Biosensors Bioelectron. 1993, 8, 355.
(8) Lukosz, W. Biosensors Bioelectron. 1991, 6, 215.
(9) Knoll, W. MRS Bulletin 1991, 16, 29.
(10) Liedberg, B.; Nylander, C.; Lundström, I. Sensors and Actuators 1983, 4, 299.
(11) Nylander, C.; Liedberg, B.; Lind, T. Sensors and Actuators 1982/83, 3, 79.
(12) Van Gent, J.; Lambeck, P. V.; Bakker, R. J.; Popma, Th. J. A.; Südholter, E. J. R; Reinhoudt, D. N. Sensors and Actuators A 1991, 25-27, 449.
(13) Jory, M. J.; Cann, P. S.; Sambles, J. R. J. Phys. D: Appl. Phys. 1994, 27, 169.
(14) Chadwick, B.; Gal, M. Jpn. J. Appl. Phys. 1993, 32, 2716.
(15) Flanagan, M. T.; Pantell, R. H. Electron. Lett. 1984, 20, 968.
(16) Daniels, P. B.; Deacon, J. K.; Eddowes, M. J.; Pedley, D. G. Sensors and Actuators 1988, 15, 11.
(17) Pollard-Knight, D.; Hawkins, E.; Yeung, D.; Pashby, D. P.; Simpson, M.; McDougall, A.; Buckle, P.; Charles, S. A. Ann. Biol. Clin. 1990, 48, 642.
(18) Biosensors; Cass, A. E. G., Ed.; The Practical Approach Series; Oxford University Press: Oxford, 1990.
(19) Severs, A. H.; Schasfoort, R. B. M. Biosensors Bioelectron. 1993, 8, 365.
(20) Hickel, W.; Knoll, W. Thin Solid Films 1991, 199, 367.
(21) Pollard, J. D.; Sambles J. R. Optics Comm. 1987, 64, 529.
(22) Brink, G.; Sigl, H.; Sackmann, E. Sensors and Actuators B 1995, 24-25, 756.
(23) Lenferink, A. T. M.; Kooyman, R. P. H.; Greve, J. Sensors and Actuators B 1991, 3, 261.
(24) Ivarsson, B.; Jönsson, U.; Karlsson, R.; Liedberg, B.; Renck, B.; Roos, H.; Sjödin, H.; Stenberg, E.; Ståhlberg, R.; Urbaniczky, C.; Lundström, I. In Proc. 3rd Int. Meet. on Chemical Sensors, Sept. 24-26, 1990, pp 157.
(25) Jorgenson, R. C.; Jung, C.; Yee, S. S.; Burgess, L. W. Sensors and Actuators B 1993, 13-14, 721.
(26) Dougherty, G. Meas. Sci. Technol. 1993, 4, 697.
(27) Matsubara, K.; Kawata, S.; Minami, S. Appl. Opt. 1988, 27, 1160.
(28) Löfås, S.; Malmqvist, M.; Rönnberg, I.; Stenberg, E.; Liedberg, B.; Lundström, I. Sensors and Actuators B 1991, 5, 79.
(29) Zhang, L.-M.; Uttamchandani, D. Electron. Lett. 1988, 24, 1469.
(30) Katerkamp, A.; Bolsmann, P.; Niggemann, M.; Pellmann, M.; Cammann, K. Mikrochim. Acta 1995, 119, 63.
(31) Jory, M. J.; Bradberry, G. W.; Cann, P. S.; Sambles, J. R. Meas. Sci. Technol. 1995, 6, 1193.
(32) Gass, P. A.; Sambles, J. R. 1992 Methods of and apparatus for measuring using acousto-optic devices, UK patent GB2254693B.
(33) De Bruijn, H. E.; Kooyman, R. P. H.; Greve, J. Appl. Opt. 1990, 29, 1974.
(34) Weast, R. C., Ed.; Handbook of Chemistry and Physics, 56th ed.; CRC Press: Cleveland, 1975.
(35) Kimura, J.; Kawana, Y.; Kuriyama, T. Biosensors 1988, 4, 41.