Unformatted Document Excerpt
Course Hero has millions of student submitted documents similar to the one
below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.
Course Hero has millions of student submitted documents similar to the one below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.
sensors 4.2.2 for temperatures and humidity Temperature sensors There are a number of ways to measure temperatures microscopically, such as thermocouples, resistance thermometers, and thermistors. Among these, thermocouple technology is one of the mostt accurate measurement methods. Mechanism of thermocouples is that they use two different materials that generate electrical voltages in accordance with temperature differences. This phenomenon is called "Seebeck effect" The Seebeck coefficient is defined as generated voltages per unit temperature, so its unit is V/C or V/C. Other temperature measurement material is resistance thermometers where the resistivity change of a material for temperature change is utilized. The most popular thermoresistence material is platinum. Self-heating is major source of error in measure temperatures by use of patinum, which is generated during measuring temperatures due to applied voltages and currents. Another temperature measurement methos is semi-conductor based devices which have large resistance changes with temperatures, but their properties are non-linear and vary from sample to sample. So, in order to obtain high spatial resolution, thermocouple technology combined with micro electro mechanical systems (MEMS) technology is considered the most common and reliable method. Before statement of MEMS based thin film thermocouple(TFTC) , we shall review the basic principle of Seebeck effect. A material with Seebeck coefficient S is subjected to temperature difference T can exhibit thermoelectric voltage(often call thermoelectric power) , E, is given by E = S T (1)
If two materials are connected as shown in Fig. 2.1 (a) to form a thermocouple(TC), the two ends of such TC is then subjected to two different temperatures, T1(low) and T2(high). Due to the Seebeck effect generated thermoelectric voltage, E a , is proportional to the temperature differences. By use of Eq.(1) , we can obtain the he following equation. Ea = (S1-S2)(T2-T1) (2)
where, S1and S2 denote the Seebeck coefficient of material 1 and 2, respectively, and T1 and T2 are temperatures at point 1 and point 2, respectively, and Ea = E1 E2.
T1 material 2 T3 material 2
Figure 2-1 Schematic diagram of thermocouples; (a) 2 different materials, and (b) 3 different materials
When three different materials(their Seebeck coefficients are S1, S2 and S3) are used to form a new thermocouples as shown in Fig. 2-1(b), the generated thermoelectric power, E b , is expressed as the following equation; Eb = (S1-S2)(T3-T1) +(S3-S2)(T2-T3) (2)
where T1, T2 and T3 are the temperature at left end, middle point and righ end of the threematerial TC, see Fig. 2-1(b). Designing thin film thermocouple (TFTC) for measurements of temperature distibution accurately with high spatial resolution, one needs to start with the fundamental analysis which includes the consideration of model shown in Fig. 2-1.
Noble Metal Thermocouple Combinations 1 2 3 Type S Type R Type B BP BN (Pt2) + 10% Rh3)) (Pt + 13% Rh) (Pt + 30% Rh) (Pt + 30% Rh) (Pt + 6% Rh) (Pt) (Pt) (Pt + 6% Rh) (Pt-67) (Pt-67) -50 to 1767 -50 to 1767 0 to 1820 812.9 818.6 56.1 789.4 733.3 7.64 7.84 1.15 7.84 6.69
Base Metal Thermocouple Combinations 4 Type E EP EN 5 Type J JP JN 6 Type K KP KN 7 Type T TP TN (Ni-Cr) (Ni-Cr) 4) (Pt-67) (iron) (iron) 4) (Pt-67) (Ni-Cr) (Ni-Cr) 4) (Pt-67) (copper) (copper) 4) (Pt-67) (Cu-Ni) (Pt-67) (Ni-Cr) 5) (Cu-Ni) (Pt-67) (Cu-Ni) 5) (Ni-Al) (Pt-67) (Ni-Al) 5) (Cu-Ni) (Pt-67) (Cu-Ni) 5) -270 to 400 -270 to 1372 -210 to 1200 6488.6 2157.5 4331.2 5016.0 3494.7 1521.3 5341.4 989.4 4352.0 54.79 16.69 38.10 40.91 30.84 10.08 48.34 9.95 38.39 -270 to 1000 7846.8 89.23
Table 2.1 Several standard thermocouples 1) Generated voltages(V) from ice point - Benzoic TP(0.00 ~ 122.37 C) 2) Pt : Platinum, 3) Rh: Rhodium 4) positive thermoelement, 5) negative thermoelement
Park and Taya (2006) designed a 2D array of TFTC based on copper and constantant, see Fig.2-2. 10mm (1,10) sensing point (10,10) sensing point
(1,1) sensing (10,1) sensing point
constantan pad ; A constantan wire bonds with conductive 100m epoxy.
Fig. 2.2 Two dimensional temperature sensor film based on copper and constantan lines where the sites of temperatures to be measures are the junctions of copper and constantan lines(Park and Taya, 2006). The key materials used in the 2D TFTC are two metal lines with high Seebeck coefficients, silicon substrate, SiO2 and AlN, whose properties are listed in Table 2.2. It is noted that use of AlN is to provide electrically insulator and thermal conductor material. For the measurements of 4
higher temperature distributions, one needs to use those TC materials with higher temperature capability, which can be selected from Table 2.1 under the condition that those TC materials are easy for silicon/lithography technology.
Thermal Conductivity (W/mK) at Room Temperature 99.5% Alumina Aluminum Nitride (AlN) Silicon Silicon dioxide (SiO2) Copper Aluminum Air 25.1 ~ 35.6 100 ~ 115 125 1.5 395 247 0.024
Volumetric Resistivity (-m) at Room Temperature 11012 11011 640 11014 1.5510-8 2.65 x 10-8 -
Table 2.2 Thermal conductivity and volumetric resistivity Humidity sensors Humidity sensors are very popular and important for inspection of the environment for food processing and medical and health products and others. Humidity sensors thus range from large scale to miniaturized devices. Here we will discuss briefly on miniaturized humidity sensors. The principles of sensing humidity are basically cast to two groups, (1) sensing the change in electrical properties( capacitance and conductivity) and (2) those in optical properties(such as refraction index change). Yeo et al (2008) made a comprehensive review on the humidity sensors based on optics. Before discussing the humidity sensors, we should define "humidity" in terms of "relative humidity(RH)" which is given by RH =( Pw/Psw) x 100% (3)
Where Pw is partial pressure of water vapor and Psw is the saturated water vapor pressure, both at given temperature. Thus, the range of RH is typically 15-85%, and it is very rare that RH would reach less than 5% and more than 95%.
The sensor based on capacitance change is composed of sensing polymer sandwiched by a pair of electrode where the water vapor can penetrate into the sensing polymer under two mechanisms, one is based on Henry's law applied to surface and the other is Fick's law applied to volume of the sensing polymer. The space-temporal behavior of water vapour can be characterized by solving both equations. Telelin and Pellet(2006) used the above model to predict the dynamic behavior of water vapour distribution as a function of time and depth of the polymer sensing film made of divinyl siloxane benzocyclobutene. The response time of this capacitative humidity sensor is very fast in order of a few seconds due to very thin film of the sensing polymer. The humidity sensors based on the capacitance and resistance change , however suffer from temperature dependency and cross-sensitivity to some chemical species. Nevertheless, the majority of the humidity sensor market is dominant by the above capacitance and resistance based sensors. Fiber optic(FO) methods for detection of humidity are increasingly popular due to its easiness of operation. There are four mechanisms associated with FO humidity sensors, (i) spectroscopy,(ii) evanescent wave, (iii)in-fiber grating and (iv) interferometrics. All of the FO humidity sensors are based on similar set up, a part of fiber optic surface is covered with sensing chemical to which water vapour is absorbed , which is detected by the above four principles.
4.2.3 Chemical sensors The mechanisms based on a number of chemical sensors are designed and used can be categorized into the following three groups: electrochemical signals optical signals mass change The chemical sensors based on each mechanism will be described in the following. Sensors based on electrochemical signals This group of chemical sensors has been the main stream of chemical sensors, which can be further categorized into: electrical conductivity 6
interfacial potencial change via potentiometry electrochemical reaction current via amperometric analysis Sensors based on electrical conductivity The principle of the sensor operation based on the change of the electrical conductivity () is straight-forward, based on the conductance (S) equation : S = A/d (4) Where A is the area of the interface between the conductive material and the electrode, d is the distance of the electrode. If digitized electrode is used, see Fig. 3.1 ,the interface area A can be increased and given by A= P h Where P is the total length of the periphery of the digitized electrode and h is the average thickness of the conductive film. Then, Eq. (4) is rewritten as S = Ph/d (6) It is noted that the ratio of P/d is fixed for a given geometry of the digitized electrode, which can be made large number, several of 10,000(Wohltjen et al, 1985). Fig. 3.1 shows simple circuit of the chemical sensor based on conductive film composed of bias potential (Ebias) of order of 1 volt, which is needed for measurements of low conductive polymer film. Even then, the order of the electric current in the circuit is as small as picoamperes. To compensate such small current, Wohltjen et al designed ratiometeric measurement system, shown in Fig. 3.2(a) where sample made of conductive polymer film and reference made of paraffin are used so that the output potential (Eout) is obtained as Eout = - Ssample Ebias/Sref (7) (5)
d Microelectrode Selective coating Array Insulating Substrate
Fig. 3.1 (a)Typical chemical sensor design made of conducting polymer, (b) close up the digitized electrode with its gap of d, covered by conducting polymer with thickness t(Wohltjen et al, 1985).
Relative conductance change [%]
10 80 60 40 20 0 -
Fig.3.2 the circuit of the chemical sensor based on ratiometric measurements, (b) typical signals of detection of NH3 as a result of the relative conductance change base on the circuit(a). For a given conductance of the reference film (paraffin), and bians potential, one can calculate the conductance of the sample (conductive polymer). The examples of such signals of detecting NH3 by the chemical sensor with the ratiometric circuit of Fig. 3.1(a)(Wohltjen et al, 1985). This ratiometric measurement method alleviates the temperature calibration which would be often required for the chemical sensors of other designs. For both sample and reference sensors are subjected to the same temperature if they are mounted on the same substrate, say silicon which is true if standard silicon lithography is used. By using Cu-based phthalocyanine conductive polymer and Langmuir-Blodgett film coating method and also the digitized electrode along with the ratiometric method, Wohltjen et al measured 2 ppm anmonia(NH3) vapour successfully, as shown in Fig. 3. 2(b). It is noted here that the above method can be repeated as far as dry air is purged to clean the sensor surface. The high sensitivity of the above chemical sensor design rests in the high ratio of P/d(= over 20,000) by the digitized electrode. IF one can use nanofiber mat made of conducting polymer 8
along with similar digitized electrode, where the nanofiber mat provide a much larger surface to volume ratio, therefore the surface on which analyte gas is absorbed , is very large, thus, the sensitive of such a chemical sensor is expected to be even better than the chemical sensor of Fig. 3.1. For example, working with Profesor Alan MacDiarmid of University of Pennsylvania, Prof. Ko's group prepared conducting nanofibers from a mixture of polyaniline doped with camphorsulfonic acid (PAn.HCSA) and polyethylene oxide (PEO) in chloroform (Norris et al, 2000). The chemosensor set up is shown in Fig. 5(a) and its photo of the as-spun nanofibers is shown in Fig. 5(b). De-doping of the electrospun PAn.HCSA/PEO fibers (11-50 wt% PAn.HCSA) was achieved by suspending the non-woven fibrous mat above the vapor of concentrated aqueous ammonium hydroxide solution. Within 3 seconds of exposing the non-woven mat to the ammonia vapor, the green non-woven fiber mat turned to blue indicating that the emeraldine salt in the blend fibers was converted to emeraldine base, Fig. 5(c). After the non-woven mat was removed from the ammonia source, the polyaniline in the non-woven mat spontaneous turned back to the original green color of the as-spun mat, see Fig. 5(c). This de-doping and spontaneous re-doping occurs at an extremely fast rate relative to films cast from the same solutions.
(c) Fig. 3.3
The enhanced performance in a high surface to volume ratio material is clearly demonstrated in the rate of de-doping of these polyaniline blend nanofibers with NH3 vapor as compared to a cast film,Fig.5(c)(Huang et al, 2003; Virji et al, 2004). The enhanced performance in a high surface to volume ratio material is clearly demonstrated in the rate of de-doping of these polyaniline blend nanofibers with NH3 vapor as compared to a cast film,Fig.5(c). It is noted here that there are several processing routes for nanofibers made of conducting polymers, (1) electrospinnig(Norris, et al, 2000:Aussawasathien et al, 2005; Liu et al, 2005;Pinto et al, 2008) and aqueous/organic interface polymerization(Huang et al, 2002; Janata and Josowicz, 2002) The above evidences emphasized the advantages of using nanofibers made of conducting polymers for use as a fast sensitive sensor element, and flashing the sensor surfaces by N2 gas or other gases, will make these sensor surfaces refreshed, thus reusable for a good number of times. Metal oxide semiconductor(MOS) sensors Metal oxide semiconductors(MOS) have been used extensively as the key sensing materials for detection of analytes by measuring the change in the resistance(R) or impedance of the MOS sensor due to absorbed analytes. Popular MOSs are grouped to two types, n-type (SnO2, ZnO, In2O3, WO3, Fe2O3, Ga2O3, TiO2) and p-type(oxides of Ni , Cu and Co). Applications of these MOS sensors are to detect leakage of gases , particularly toxic gases(NO2 , CO and CO2), but they include those sensors for food stuffs, coffee (Gardner et al, 1992), milk (Sberveglier et al, 1998), strawberry (Hirshfelder et al, 1998), meat(Braggines, et al, 1999) and edible oils (Bazzo et al , 1998). The resistance change, R is a non-linear function of the concentration of analyte, C, for example(Kohl, et. al, 2000). R/R0 = k C1/2 (8)
where R0 is the initial resistance before the absorption of analyte, k is sensor-dependent constant. It is noted that Eq.(8) was obtained empirically by curv-fitting by Kohl et al(2000), similar empirical relation was proposed by Mielle et al (2000). The recognition of a specific odor from a mixture of different gases is normally based on arrayed sensors, the raw data of which provide the basis for a series of analytical models. Wilson et al 10
(2000) conducted such analyses based on the raw data taken from 30 arrayed sensor of SnO2 system for a mixture of 7 different odors. They used four different models, principal component analysis(PCA), controid proximity metric(CPM), multi-layer perceptron neural network, and radial basis function neural network. They found that use of CPM is very effective in quantifying the degree of clustering in all 30 dimensions of the data set and observed a good relation between CPM and neural network results. Another set of metal oxide semiconductor sensors are based on field emission transitor(FET). Among these, SiC based Schottky diode sensor made of p-type silicon of subtrate, n-type silcon, source and drain metals and Pt catalytic gate, is best suited for sensing various gases(H2, CxHy, NH3, O2, H2O, NxOy, CO2) at high temperatures of up to 1000C(Spetz et al, 2000). Sensors based on potentiometry The principal of this sensor is the potential differences that exist across the sensor membrane that separates two solutions. The origin of the membrane potential lies on the binding of a charged species at the membrane surface and /or transport of such species through the membrane. If the binding and transport are selective to a given charged species, then, the membrane potential of the cell is given by Ecell = (RT/nF) ln(a1/a2) (9)
where a1 and a2 are the concentration of on the sample(analyte) side and reference side of the membrane, respectively. IF the membrane is not entirely selective, Eq.(9) is generalized as Ecell = constant + (0.059/zi) ln ( ai + Kij aj zi/zj ) (10)
where ai is the activity of the analyte ion with charge zi, aj is the activity of j-th interferent ion with charge zj and Kij is the potentiometric selectivity coefficient between the analyte ion(i) and the j-th interferent ion, the repeated index j in the equation is to be summed over j=1, N where N is the total number of interferent ions. Sensor design based on ion selective membrane is clearly aimed at minimizing the selectivity coefficient, Kij. Subtaintial research has been done on cations (H+, Ca2+, K+, Na+, alkali metal ions) but less on anions and more complex molecular ions. In the case of anion sensing , use of various metal-ligand complexes , including metalloporphyrins has been much progressed. Typical designs of gas sensors for CO2, NH3, NO2 and SO2 based on 11
ion-selective membrane of such metal-ligand polymers are illustrated in Fig. 3.3 where (a) and (b) denote static dynamic sensing for gas containing the analyte gases with these four species are shown.
Ion-Selective Electrode (ISE) H2O + CO2 H2O + NH3 H2O +2NO2 Gas or Liquid Sample Polymer Membrane Aqueous Recipient Solution
HCO3 + H + NH4 + OH NO2- + NO3- + 2H+
CO2 NH3 NO2 SO2 Gas-permeable membrane Polymer
CO2 NH3 NO2 SO2
Water H O + CO 2 2 Flow
H2O + NH3 H2O +2NO2 H2O + SO2
HCO3 + H
NH4 + OH
NO2- + NO3- + 2H+ HSO3- + H+
Fig. 3.3 Typical design of chemical sensors based on ion-selective-electrode (ISE) for sensing four different gases with CO2, NH3, NO2 and SO2, (a) four analyte gases are diluted in solution then analyzed by the ISE , (b) four analyte gases permetate through gas permeable membrane into flowing water channel, then ionized which will be analyzed at the end of the channel by polymer membrane with ISE. Chemical sensors based on reactions The mechanism inherect in the chemical reactions of analyte and the reacted species with and without intermediate medium such as enzyme is rather complicated. But the signal in terms of electric current reflects on the amount (or concentration) of the analyte reacted through the enzyme. The chemical sensors based on reactions with enzyme are well-studied for sensing glucose where glucose oxidase is used as the sensing enzyme and the reaction product is hydrogen peroxide which provides electrical current at the electrode. The amount of the electrical current measured at the electrode is propotional to the concentration of glucose. It is noted that this chemical sensors based on such enzymatic reactions will consume the amount of such enzyme thus limiting the life of the chemical sensor, or requiring the periodic replacement of the sensor electrode with the enzyme. In the case of the glucose sensor based on the enzymatic 12
reaction which samples human serum, would react other interferences in serum. Therefore, the technical challenge in designing such chemical sensors is to minimize the reactions with such various interferent species. One approach is use of polymer film modified electrode which would block off unwanted interferences by the polymer modified electrode which contains glucose oxidase is immobilized by cross-linking with glutaraldehyde,see Fig. 3.4(a) (Geise and Yacynych, 1989). It is noted that the reaction kinetics provides a non-linear relation between the concentration of analyte and the sensor signal of electric current which is proportional to the reaction product of H2O2, see Fig. 3.4(b).
Interferences Polymer film Glucose H2O2
GO GOX GOX GOX Pt Pt GOX
surface Glutaraldehyd e crosslinking Cur
3 2 1 0 0 2 4 6 8 100
Fig. 3.4 Sensor for detection of glucose in blood serum with fluidic channel, (b) typical current signals as a function of concentration of glucose( Geise and Yacynch, 1989).
Chemical sensors based on mass change in the piezo-electric mechanical motions This mechism is mainly due to the mechanical motions (vibrations and wave propagations) in piezoelectric sensor , on which analyte or analytes with given mass. Two methods are popular, (i) Acoustic wave based sensors, (ii) cantilever-beam in resonance mode being changed due to mass attachment. The success of this kind of mechanical sensors rests on the miniaturization of the piezoelectric sensor elements as this would increase the resolution of detectable mass amount into very small level. Acoustic Wave Based Sensors
Acoustic wave based sensors is based on the mechanical oscillations propagation between two metal electrodes sensing change in resonance frequency due to mass attachment on the piezosensor surface. There are two types, surface acoustic wave(SAW) and bulk acoustic wave(BAW) types where both types, the surface of piezoelectric material is covered by chemically interactive membrane(CIM) on which an analyte will be absorbed, thus, its physical property is change. This change in the physical properties( such elastic stifinness) would change the acoustic wave propagation speed, and also shift of the resonance frequency. The piezoelectric materials that have been used for this sensor are quartz, ZnO, ALN, Bi12GeO20, -SiO2, LiTaO3, LiNbO3, PZT). The advantages of the acoustic wave based sensors are high sensitivity, short response time, low power consumption , small size and overall robustness while the disadvantages are temperature and humidity dependant, difficulty of replacing the sensor material, relatively large noise level. Canti-lever type sensors Canti-lever type sensors are increasing popular now for use in various applications, particularly as biosensors, owing to its very high sensitivity for small amount of analyte. Recently, Goeders et al reviewed the current status of all types of cantilever sensors (2008). The smaller the size of cantilever sensor , the smaller amount of analyte can be detected. Therefore, design of micronsized cantilever sensor is mainstream of research in this field. There are two principles by which the cantilever sensor works, (i) static motion, and (ii) dynamic motion of resonance frequency. The first type is to detect the amount(mass) of analyte adsorbed on the surface of the cantilver by measuring the deflection by optical method. The deflection () of a cantilever of length L with distributed weight (w0) is given by a well-known equation: = w0L4/(8EI) where E is Young's modulus of the cantilever and I is the second moment of inertia of rectangular cross section( wt3/12) and where w and t are the width and thickness of the cantilever, Fig. 3.5(a). The measurement of is made by using the reflection of a beam light off the cantilever onto a segmented photodiode where the light source is laser of light emitting diode(LED). Often many cantilevers with specific chemical interactive film are used to detect a good number of analytes. (11)
Uniform load (wo/unit length)
(a) stressed layer on top (1) and on bottom surface (2).
Fig. 3.5 A cantilever beam loaded with (a) uniformly distributed weight, w0, and (b) with added
If the analytes are bonded on top and bottom surfaces of the cantilever, they induce the stresses , Fig. 3.5(b), then, the Radius of curvature(1/R) is given by (Stoney, 1909): 1/R = 6 (1-) (1 2)/(Et2) where and E are Poisson's ratio and Young's modulus of the cantilever material. Normally the signals of the chemical sensors in static mode are measured optically for their deflections based on the above equations. Instead of measuring the deflections of such a cantilever beam, one can measure the stress induced within the cantilever if it is made of piezoresistive material, such as boron-doped polysilicon where "piezo-resistivity" of the polysilicon is utilized. Recently, Loui et al (2008) proposed a new chemical sensor based on arrayed mcirocantilevers made of polysilicon doped with boron for detection of a set of volatile chemical species which include water vapor, methanol, isopropanol, 1, 4-Dioxane, toluene, hexane, benzene, methylene chrloride, ethyl acetate, acetone, acetonitrile, chemical warfare agents;VX and sulfur mustard(ED). They used a set of polymer films coated on the polysilicone, each will react with a chemical gas by swelling, thus inducing the stress for which Eq.(12) is valid. The dimension of each cantilver is 120 m (length), 50m (width) and 0.5 m (thickness). 6 cantilever beams with two reference cantilever beams constitute one set chemical sensor. The voltage changes due to the piezo-resistive polysilicon cantilevers are taken as a function of time, which are converted by AD converter to a set of digital data. The peaked voltage data set can be considered points in a six-dimensional(6-D) Euclidian space, which can be converted to principal component values in the 6-D space. By using this principal component analysis(PCA), they are able to discriminate all different chemical gases. (12)
Fig. 3.6 Schematic of chemical gas sensor based on arrayed piezo-resistive silicon cantilevers, (a) gas mixing system, (b) analytical algorithm based on principal component analysis(PCA) method (Loui et al, 2008). The sensitivity can be enhanced if we use the dynamic method based on a simple mass-springdashpot model and the shift of its resonance frequency as a result of unknown analyte to be absorbed on the surfaces of the cantilever. The equation of motion of such a mass-spring-dashpot model , Fig. 3.7(a), is given by
d 2x dx m 2 + c + kx = F (t ) dt dt
Where m, c, k are mass, dashpot constant, and spring constant, respectively, and F(t) is generalized force. The natural frequency (lowest frequency) is obtained by solving the homogeneous equation with c=0 and F(t) =0 as
nat = k / m
When damping (with non-zero c) exists, Eq.(13) is modified to
d 2x dx 2 + 2 nat + nat x = 0 2 dt dt
Where parameter is related to
c 2 nat m
c 2 km
To quantify the sensitivity of a cantilever sensor with resonance frequency peak, one can often use quality factor, Q which is defined for a lightly damped one-degree-of freedom system (Ginsgerg, 2001),
i fi 1 = = 2 FWHM
where i = 2fi and =2 FWHM and where FWHM is defined as full width at at half magnitude at i-th resonance peak, see Fig. 3.7(b). Q Higher value is desired as this would increase frequency resolution, i.e lower the minimum frequency. Fig. 3. 7(b) illustrates the case of higher Q at 50Hz making it more detectable, which is shited to higher frequency with the same Q.
fi=50kHz, Q=10 fi=50kHz, Q=100 fi=75kHz, Q=100
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
Fig. 3.7 (a) Mass-spring-dashpot model for use in cantilever biosensor, (b) frequncy response curve, normalized amplitude vs frequency where definition of quality factor Q is defined. A unloaded cantilver beam in a vacuum will have a number of natural frequencies and its-ith frequency fi is given by (Mafarland et al, 2005)
i2 k fi = 2 3 M b
where Mb is the mass of the cantilever beam and i is solution of the following algebraic equation,
cosh i cos i + 1 = 0
by M, then , the i-th frequency due to this increase mass , fi+m is changed to
If a layer of molecules or other species are added on the surface of a beam, the mass is increased
i2 k = 2 3 M b + M
Added mass normally increases the second moment of inertia, I, making the beam stiffer , i.e k+ effect, then, the i-th frequency due to this stiffness change, f+k, is given by
3E ads I i2 k = + 3 ads L Mb 2 3 M b
where the absorbed layer has elastic modulus, Eads, and Iads is given by
t wt = ads + wt ads ads - t b - y cm 12 2
E t + E ads (2t ads t b + t ads ) = bb 2 E ads t ads + 2 Eb t b
And where the centroid of the cross section with absorbed layer, ycm is given by
and tb and tabs are the thick of the beam and added layer, respectively. The effect of the surface stress () is given by
where i is defined by
2L3 = i 1 + 2 3 EI
If one can include both the effects of added layer(mass and stiffness) and surface stress, then resulting i-th frequency, fi+m, +k,+ is given by
+ m , + k , +
3E I k + 3 ads ads M b + M L ( M b + M )
Applications The chemical sensors based on the above micro-cantilver design have many applications, which can be roughly grouped into (1) sensing volatile organics(alcohol(Lang et al, 1998)), (2) Chemical warfare agents(dimethylmethylphosphonate)( Pinnaduwage, et al,2003),(2) explosives (Pamula and Fair, 2000), (3) toxic metals(Cr2+, Ca2+, Cs+, CrO42- )(Li, et al, 2001), (4) biological applications which have been progressed very rapidly in several subbranches, (i) cells, (ii) virus,
(iii) antigen-antibody interactions, (iv) DNA hybridization and (v) enzymes. Some notable biological applications are stated briefly below: Llic et al (2001) reported the first successful application of micro-cantilever biosensor to detection of E. coli bacteria by immobilizing anti-bodies on the surface of the cantilver. When a number of cells are bound to the surface, the resonance frequency is shifted , which is found to be proportional to the amount of bound cells. Llic et al (2004) reported first the use of microcantilever sensor in detecting immunospecific binding of viruses in liquid. Cambell and Mustharasan (2006) reported successful detection of pathogen Bacillus anthracis spores in liquid under both stagnant and flow conditions. Recently, many studied are reported on the detection of antibody-antigen bondings using known pairs which demonstated the effectiveness of using the micro-cantilver sensors. Among these, the work of Wu and et al (2001) is noteworthy, as they demonstrated the new method of detection of early stage of prostate cancer using prostratespecific-antigen(PSA) over the range of 0.2 ng/mL to 60 mg/mL. Clinically, the concentration range of PCA in blood serum 0.01 10 ng/mL or more is considered to be first srage. If this small concentration of PCA is realized in clinical testing, this would be quite beneficial for patients with first stage of prostate cancer. For the lower limit of the current clinical testing is 4ng/mL, while the micro-cantilver sensor by Wu et al can detect even smaller concentration as shown in Fig. 3. 8. If this sensor is extended to multi-cantilever sensor system for detection of multiple cancer-specific antigens, this seems very promising direction as the biosensors based on cantilever method does not require any labeling , yet detection limit is very small. Fritz and his co-worker (2000) pioneered the use of the micro-cantilever sensor for detection of nucleic acid hybridization and optical measurement system of static deflection of the cantilever beam,Fig. 3. 9. KcKendry et al (2002) reported multiple label-free biosensor system based on arrayed microcantilever beams like Fig. 3.9, for detection of ligand-receptor interactions like DNA hybridization where the deflection was measured optically.
Fig. 3.8 Microcantilever biosensor for detection of prostate cancer antigen(PSA) concentration in term of steady state deflection. Use of smaller sized cantilever can successfully detect the PSA with smaller concentration than the amount detectable by standard clinical testing(4 ng/ml)(Wu et al, 2001).
Fig. 3.9 Typical micro-cantilever beams for use in biosensors(Fritz, et al, 2000)
Chemical sensors based on optics Chemical sensors based on optics may be grouped to two types, (i) fiber optics based sensors and (ii) surface plasmon resonance sensors. Both types will be reviewed below. Fiber optics based chemical sensors
Walt et al (1989) reviewed FO based sensors. A typical optical fiber is continuous glass fiber of diameter ranging 100- 1000 m. It is flexible and can perform continuous spectroscopy at remote sites which could not be accessible by other sensors.The fiber optics(FO) based sensors have a number of advantages over the electrochemical sensors, simple design and accessible to analyte site in harsh environment, immune to electromagnetic interference, and requires no reference electrode. Due to these advantages, examples of applications based on FO sensors are in situ toxic waste monitoring, in vivo blood gas monitoring and in situ processing control monitoring. A popular FO sensor is composed of optical fiber with its end modified with reagent, light source (laser, LED) , photodiode and date acquisition system. Transduction mechanisms used in FO sensors are absorption, fluorescence or reflectance). Among these, use of fluorescence as the reagent attached to tip of an optical fiber is most effectively used in vivo and situ sensing applications. Walt et al (1989) designed a optical fiber tip made of copolymerized fluorenscent dye or biomolecures in the polymer attached at the fiber tip where the base polymer is to increase surface area, thus shortening the sensor response time. Before the bonding the base polymer, one needs to active the surface of glass optical fiber by using silanizing agent After attachment of such polymer base on the optical fiber tip, the next key step is to prepare for indicator reagents such as fluoresceinisothiocyanate(FITC) and fluoresceinamine. If the immobilized sensing reagent contains also a bioreceptor such as an enzyme or antibody, such FO sensor will become an effective biosensor. Walt et al discussed several FO sensors targeted for sensing, pH, CO2 gas, gasoline, Penicillin. Sepaniak et al (1989) designed a new biosensor based on antibody- based heterogeneous fluoroimmunoassay(FIA) where the dye used is FITC. The measured optical signals are proportional to the amount of antigens bound on the FIA sensing surface. They employed three different termini, type A , B and C as shown in Fig. 3.10. For type A design, the antibodies are directly bonded on the tip surface, but the sensing signals are poor due to small amount of antibodies attached to the tip surface and their denaturing upon bonding to the fiber tip. Terminus design B provided larger amount of antibodies bonded to the tip surface area without denaturing, but its repeated sensing capability is limited. Terminus design C overcomes this difficulty of repeatability by using six column bundle, Fig. 3.10(c) where the sensing FO sensor is surrounded by rinsing liquid tubes and anti-BPT solution tube. This FO sensor with terminus design C provides successful regenerable immunosensor.
FIBER OPTIC /
6 COLUMN BUNDLE
chamber wall heat shrink sensing chamber
10,000 m.w. cutoff membrane
Anti-BPT inlet rinse inlet outlet fiber optic
Fig. 3.10 Three different sensing termini used in FO sensor ( Sepaniak, et al, 1989) FBG sensors Fiber Bragg grating sensors have been extensively applied to many application field, particularly measurement of strain field which surrounds fiber with Bragg grating. First fiber optic is grated by using UV lights so as to create a periodic grating for localized area or entire length.
3.11 Schematics of reflected / transmitted wavelengths and related Bragg relations for: a) uniform fabricated grating, b) uniform deformed grating, c) non-uniformly
Surface plasmon resonance (SPR) biosensors An incident light can be excited at the interface of three-laminate composite resulting in surface plasmon wave(SPW) at a particularly incident angle. Based on this concept, a new biosensor is designed, called as surface plasmon resonance (SPR) biosensor. Homola et al (1999) made an extensive review on SPR biosensors. Three different concepts have been proposed and put into practical SPR sensors, (i) prism complex with attenuated total reflection (ATR), (ii) grating complex and (iii) optical wave guide. These three concepts are illustrated in Fig. 3.12.
analyte Metal layer Flow cell analyte Metal layer Metal layer with grating analyte Wave guide
Fig. 3.12 Surface plasmon resonance sensors, (a) ATR-prism, (b) grating, (c) optical fiber (Homola et al, 1999).
The ATR prism-based SPR system has been extensively used since Kretschmann and Raether (1968) and Otto (1968) designed this in 1968, where optical wave incident to the prism with a specific angle (i) is reflected totally and surface plasmon wave propagates at the interface between metal layer and analyte layer if the propagation constant of such a surface plasmon wave (kspr) is given by
k spr = k o
m + na
where ko is the free space wave number and equal to 2/o (o is the free space wave length), na is the refractive index of the analyte and m is complex-valued dielectric constant, m = mr + j mi . The wave vector corresponding to the incident wave along the interface (x-axis) is given by
k x = k o n glass sin ( in )
When kx is equal to kspr, then a surface plasmon wave is generated, resulting in the reduction in the intensity of the reflected wave. There are two methods of analysis based on ATR prism SPR system, angle modulation and wave length modulation modes. In the former design, a single wave such as collimated laser wave, Fig 3.13(a) is used as an incident wave and the reflected
intensity is measured as a function of incident angle i, Fig. 3.13(b). If a sensor chip is mounted on an unknown analyte layer, the reflected intensity- i relation is shifted as shown in dashed line in Fig. 3.13(b). This shift can be measured in terms of the change in i, i or the change in the /intensity, I which is better quantity to measure than i. In the case of wave length modulation with light which has infinite number of wavelength, Fig. 3.13(c), the intensities of the reflected multiple waves exhibit the lowest intensity at a certain wave length of incident waves, solid curve in Fig. 3.13(d). The details of SPR analysis based on Maxwell's equations and transmission line theory is given in a book (Taya, 2005).
I i Wave length Incident angle i Wave length Wave length
Fig. 3. 13 Two types of SPR design, (a) single wave incident (b) resulting in the shift of the SPR peak, (c) multiple-incident waves , (d) resulting in the peak shift (Taya,2005).
Upon unknown analyte mounted on the surface of SPR sensor chip, the intensity- wavelength curve is shifted to dashed one in Fig. 3.13(d), providing the change in the wave length, . Fig. 3.14(a) illustrates an SPR model of a five layer laminate model. We shall calculate the reflection coefficient (reflectivity), of the SPR in this multi-layer system. We are interested in the effect of polymer layer (its thickness 10nm and n=1.4). It is noted in Fig. 3.14(a) that the Crlayer is needed as adhesive bonding layer between prism (BK7, n=1.5) and Au. Use of water and polymer layer is to simulate a realistic SPR characterization of unknown molecules in a flow cell (mostly water).
Water n=1.33 Polymer (10[nm], n=1.4) Au (50[nm])
with polymer layer without polymer layer
24 Ii Ir
spr Fig. 3.14 (a) Fiver-layer SPR model , (b) Simulation of reflectivity vs incident angle() relation of the five-layer SPR model of (a)(Taya,2005). Fig. 3.14(b) shows the results of the numerical simulation of the reflectivity-incident angle (i) relation for the multi-layer model of Fig. 3.14(a) with (solid curve) and without polymer layer (dashed curve). Fig. 3.14(b) demonstrates the sensitivity of SPR sensor sensing the existence of polymer layer. To quantify the sensitivity, one can use the change in SPR angle () or the change in the reflectivity () at a specific incident angle (=750 was used in the figure). If the polymer layer is to simulate unknown molecules in more realistic set of SPR characterization, the thickness of the unknown molecules can be as small as a few nm or so. Even for such a extremely thin layer, the SPR sensor system based on either or i, can detect the existence of the unknown molecules of very small amount. The SPR sensor, however, can not distinguish the in-plane resolution (i.e. x-y plane) as the unknown layer is assumed to be uniform in the x-y plane (sensor surface). Cahill et al (1997) designed a new SPR sensor based on wave length modulation mode and use of a prism with retro-reflection, Fig. 3.15. Advantage of using this retro-reflector SPR is its portability of the probe portion which can be attached to any analyte medium. Akimoto et al (2000) examined the sensitivity of this retro-reflection type SPR.
Input li ht inc
Sensor 1 Side view
Reflection 2taper 0.5 cm surface
Output li ht
Fig. 3.15 Retro-reflector SPR system (Cahill et al, 1997).
Akimoto et al (2003) proposed a new SPR design based on step-like sensor surface which provides integrated reference surface. Rich and Myseka (2002) made an extensive literature survey (700 papers published in 2001) on biosensors based on optical sensing methods, most of which are based on commercial SPR systems, Biocore AB (Uppsala, Sweden), Affinity Sensors (Franklin, MA, USA), Nippon Laser and Electronics (Nagoya, Japan), Texas Instruments (Dallas, TX, USA), IBIS Technologies (Enschede, The Netherlands) and Analytical -Systems (Woodinville, WA, USA). SPR biosensors
Diagnosis at an early stage and prognosis for successful therapeutic intervention are essential for the recovery of a cancer patient. Monitoring cancer biomarkers in blood, urine and other body fluids is an important method for early detection, since some of the biomarkers allow for identification of a disease at its very early stage, even before its symptoms can be recognized by a patient (Landman et al, 1998). Carcinoembryonic antigen (CEA) is a widely used tumor marker for diagnostic and therapeutic purposes in gastrointestinal, breast and lung cancer (Aquino et al, 2004). The normal range for CEA in an adult non-smoker is less than 2.5 ng/mL and for a smoker less than 5.0 ng/mL in human serum, but its level exceeds 100 ng/mL upon development of certain cancers (Fernandez, 2004). A rising CEA level indicates progression or recurrence of the cancer, which indicates detection of CEA in human sera samples can be used to diagnose and monitor cancer at its early stage. Furukawa et al (2008) recently analyzed the glycoforms on human CEA purified from liver metastases of colon carcinoma and pleural and ascites fluids, and quantitatively detected more than 120 glycans on human CEA. Substantially different glycosylation profiles between CEAs of liver metastases of colon carcinoma and pleural and ascites fluids were observed. In this regard, it will be very important not only quantify CEA level in serum but also quantitatively evaluate glycoforms attached for the improved sensitivity and specificity of diagnosis and prognosis. Indeed, although prostate-specific antigen (PSA) tests often suffer from lack of specificity in distinguishing benign prostate hyperplasia from prostate cancer, recent studies indicate that N-glycans of PSA found in prostate cancer differ significantly from those seen in benign prostate hyperplasia and therefore could be a potential indicator leading to improved sensitivity in diagnosing prostate cancer (Ohyama, et al, 2004).
The majority of techniques currently employed to detect interactions of antigens and antibodies require a fluorescent- or enzymatic-labeling to report binding event (Cooper, 2003) which is time-consuming, and may not be a suitable platform for the successive interaction analysis to reveal both the CEA level and the glycoform attached. Since surface plasmon resonance (SPR) can provide label-free and real-time detection, it has attracted strong attention as a biosensor in the past decade (Karlsson et al, 1994;Liedberg et al, 1995; Homola et al, 1999). Biosensors based on SPR have been extensively used to monitor molecular interactions for its outstanding sensitivity, reliability, reproducibility as well as its capability to monitor multiple interactions successively(Rich and Myszka, 2000; Huang et al, 2005). However, its application to the screening of disease markers in human body fluids (e.g. serum) is so far limited. The detection of the cancer marker CEA in human serum using a SPR biosensor has not reported yet, except for recent work by Su et al(2008). Su et al (2008) first studied the effects of three different sensor chip, CM5, C1 and EthyleneGlycol terminated alkanethiol Self Assembly Monolayer(EGSAM) in their performance of detection of CEA in buffer solutions. These three different design of sensor chip surface and their SPR signals in relative unit according to Biacore SPR equipment are shown in Fig. 3.16.
HS O O O O O COOH (1-mercapto-11-undecyl)hexa(ethylene glycol) carboxylic acid
(1-mercapto-11-undecyl)tetra(ethylene glycol) HSC11EG4
Fig. 3.16 Three different designs of sensor chip , C5, C1 and EG-SAM and their SPR signals based on C1 and EG-SAM as a function of time for different concentrations of CEA(Su et al, 2008).
RU 300 250 200 150 100 50 0 0 200 400 600 800 1000
CM5 C1 SAM
Fig. 3.17 Kinetic analysis of CEA antibody and antigen interactions, detection of CEA antigens with a series of concentrations (800 ng/mL to 12.5 ng/mL) by using BIACORE C5 sensor chip: color lines are experimental sensorgrams, and black lines are simulated sensorgrams by using BIAevaluation 4.1(Su et al, 2008). Kinetic analysis should provide important parameters in designing the experiment conditions to improve the detection sensitivity. Therefore, the interaction of CEA with immobilized anti-CEA antibody was analyzed under a series of concentration (12.5 ng/mL 800 ng/mL), see Fig. 3.17. The kinetic parameters were calculated based on the global analysis of the experimental data using a 1: 1 binding model accounting for mass transport. The association rate constant (ka), dissociation rate constant (kd) and equilibrium constant (KD) were calculated to be 4.22 x 105, 3.0 x 10-5, and 7.11 x 10-11, respectively. Based on the kinetic parameters, the change of resonance unit with the interaction time to the analyte solutions could be simulated by using the BIA evaluation software. As shown in Figure 3.17, the experimental data and the simulated data of the interactions of immobilized anti-CEA antibodies with antigens of a series of concentrations agree very well.
By using C5 type sensor chip and sandwich assay, Su et al (2008) studied the effective of using SPR biosensor in detection of CEA in serum and found that the SPR can dectect as small as 25 ng/mL , as shown in Fig. 3.18.
2nd anti-CEA MAb 3rd anti-CEA MAb (100 ug/ml) (100 ug/ml)
800 ng/ml 400 ng/ml 200 ng/ml 100 ng/ml 50 ng/ml 25 ng/ml
100 Response 0
400 500 Time (sec)
Fig. 3. 18 Sandwich assay of a series of CEA-spiked serum sample(Su et al, 2008). 4.2.4 Infrared sensors The chemical sensors based on infra-red(IR) waves are composed of IR fibers with protective thin or thick coating where there are two types of IR fiber material, chalcogenide glass fibers(As2Se3-xTex) and polycrystalline silver halide fibers(AgClxBrx-1). The former IR fiber is targeted at wave lengths of 2 to 12 m with attenuation losses of 0.1 dB/m, while the latter at broader wave lengths up to 20 m with attenuation losses of 0.2 dB /m(MIzaikoff et al, 1995. The principle of IR optical fiber sensor is based on either attenuation of total reflection elements based on Fresnel's and Snell's equations, or evanescent wave that appears at the interface between the IR optical fiber surface and adjacent optically thinner medium containing unknown analyte with capability of detection of such analyte within a very thin region , dev which is given by Dev = (29)
Where n1 and n2 are refraction index of higher (IR fiber) and lower medium(analyte containing medium) and is the IR wave.Mizaifoff et al (1995) developed two types of IR fiber chemical sensors, thick and thin coating on the surface of the IR optical fiber. For thick coated IR fiber
sensor made of silver halide, the surface of the IR fiber of length of 10 cm is coated with 20 mm thick coating of low density polyethylene or polyisobutylene. This IR optical fiber with thick coating subjected to a aqueous solution contaminated with chlorinated hydrocarbon. The IR optical fiber with thin coating was used for detection of glucose in a complex aqueous solution in the physiological range where the thin coating is made of enzyme glucose oxidase which would be reacted to provide glucose acide and hydrogen peroxide. However the speed of the reaction of glucose was found lower than expected. The slow speed of the reaction is partially attributed to the fact that the surface surface is limited. If the sensor surface is increased to make it more 3-D nanostructure surface so as to increase the surface to volume ratio, then the IR optical sensor diction speed may be much more enhanced.
References Akimoto, T, Ikebukura, K, and Karube, I, 2003, Biosensors and Bioelectronics, 18, 1447-1453. Aussawasathien, D, Dong, JH and Dai, L, 2005, Synthetic Metals, 154,37-40. Bazzo, S, Loubet, F, Labreche, S, Tan, TT, 1999, in Electronic Noses and Sensor Array based System, Design and Applicatinos, WJ Hurst Ed, TEchnomic Publishing, Lancaster, p.225-234. Braggins, TJ, Frost, DA, Agnew MP, Farouk, MM, 1999, in Electronic Noses and Sensor Array based System, Design and Applicatinos, WJ Hurst Ed, Technomic Publishing, Lancaster, p.51-82. Cahill, CP, Johnston, KS and Yee, SS, 1997, Sensors and Actuators, B45, 161-166. Cambell, GA and Mutharasan, R, 2006, Biosensors and Bioelectronics, 22, 78. Janata, J and Josowicz, M, 2002, Nature, vol.2, 19-22. Fritz, J, Baller, MK, Lang HP, Rothuizen HVP, Meyer, E, Guntherodt, HJ, Gerber, C and Gimzewski, JK, 2000, Science, 288, 316. Gardner, JW, Craven, M, Dow, C. andHines, EL, 1998, Meas. Scie. Tech., 9, 120-127 Geise, R J, and Yacynych, A, 1989,in Chemical sensors and microinstrumentation, eds Murray RW et al, Amer. Chem. Soc. Washington DC, ACS Symp. Ser. 403, pp. 65-77. 30
Ginsberg, 2001, Mechanical Structural Vibrations, John Wiley and Sons, New York. Goeders, KM, Colton, JS and Bottomley, LA, 2008, Chem. Rev., 108, 522-542. Hirschfelder, M, Hanrieder, D, Ulrich, D, 1998, Semin. Food Anal., 3, 27-36. Homola, J, Yee, SS, Ganglitz, G, 1999, Sensors and Actuators, B54(1-2),3-15. Huang, J, Virji, S, Weiller ,B and Kaner, RB, 2003, J. Amer. Chem. Soc., 125, 314-515 Kohl D, Heinert, L, Bock, J, Hofmann, Th, Schieberle, P, 2000, Sensors and Actuators B 70, 4350. Kretschmann, E and Raether, H, 1968, Z. Natureforsch, 23A, 2135-2136. Lang, HP, Berger R, Battiston, F, Ramseyer, JP, Meyer, E, Andreoli, C, Brugger, J, Vettiger, P, Despont, M, Mezzacasa, T, Scandella, L, Guentherodt, HJ, Gerber, C and Gimzewski, JKK, 1998, Appl. Phys. A66, S61. Llic, B, Czaplewski, D, Craighead, HG, Neuzil, P, Campagnolo, C and Batt, C, 2001, J. Appl. Phys. Letts, 77,450. Llic, B, Yang, Y, Craighead, HG, 2004, Appl. Phys. Letts, 85, 2604. Li, HF, Thundat, T, Dabestani, R, Brown GM, Britt, PF and Bonnesen, PV, 2001, Anal. Chem., 73, 1572. Liu, H, Kameoka, J, Czaplewski, DA and Graighead, HG, 2004, 4(4), 671-675. McFarland AW, POggi, MA, Doyle, MJ, Bottomley, LA and Colton , JS, 2005, Appl. Phys. Letts., 87, 053505. Loui, A, Ratto, TV, Wilson, TS, McCall, SK, Mukerjee, EV, Love, AH and Hart, BR, 2008, Analyst, 133, 608-615. McKendry, R, Zhang, J, Arntz, Y, Strunz, T, Hegner, M, Lang HP, Baller, MK, Certa, U, Meyer, E, Guntherodt, HJ, and Gerber , C., 2002,Proc. National Academy of Science, 99,(15), 97839788. Mizaikoff, B, Goebel, R, Krska, R, Taga, K , Kellner, R, Tacke, M and Katzir, A, 1995, Sensors and Actuators B29, 58-63. Murray RW, 1989, in Chemical sensors and microinstrumentation, eds Murray RW et al, Amer. Chem. Soc. ACS Symp. Ser. 403, pp.1-19. Otto, A, 1968, Z. Physik,216,398-410. Pamula, VK and Fair, RB, 2000, Proc. SPIE-Int. Soc. Opt. Eng., 44038, 547. Pinnaduwage, LA, Gehl, A, Hedden, DL, Muralidharan, G, Thundat, T, Lareau, RT, Sulchek, T, Manning, L, Rogers, B, Jones, M, Adams, JD, 2003 , Nature, 425,474 Pinto, NJ, Ramos, I, Rojas, R, Wang, PC and Johnson, AT, Jr, 2008, Sensors and Actuators B 129, 621-627. 31
Sberveglieri, G, Comini, E, Faglia, G, Niederjaufner, G Benussi, G, Contarini, G and Povolo, M, 1998, Semin. Food Anal., 3, 67-76. Spetz, AL, Tobias, P, Uneus, L, Svenningstorp, H, Ekedahl, LG and Lundrom, I, 2000, Sensors and actuators B 70, 67-76. Su, FY, Xu, C, Taya, M, Murayama, K, Shinohara, Y and Nishimura, S, 2009, "Detection of Carcinoembryonic Antigens Using a Surface Plasmon Resonance Biosensor", Sensor, 8, 42824295. Taya, M, 2005, Electronic Composites, Cambridge University Press. Virji, A, Huang, J, Kaner RB, and Weiller BH, 2004, Nano Letts, 4(3), 491-496. Walt, DR, Munkholm, C, Yuan, P, Luo, S and Barnard, S, in Chemical sensors and microinstrumentation, edited by RW Murray, et al, Americal Chemical Society Series Vol. 403, 252-272. Wilson, DM, Dunman, K, Roppel T and Kalim, R, 2000, Sensors and actuators B62, 199-210. Wu, G, Datar, RH, Hansen KM, Thundat, T, Cote, RJ and Majundar, A, 2001, Nature Biotechnology, 19, 856-860.