- Email: [email protected]
A surface plasmon resonance system for the measurement of glucose in aqueous solution
Sensors and Actuators B 105 (2005) 138–143
A surface plasmon resonance system for the measurement of glucose in aqueous solution W.W. Lam a,∗ , L.H. Chu a , C.L. Wong b , Y.T. Zhang a a
Department of Electronic Engineering, Joint Research Center for Biomedical Engineering, The Chinese University of Hong Kong, Shatin, NT, Hong Kong b Department of Physics and Materials Science, The City University of Hong Kong, Kowloon, Hong Kong Received 12 July 2003; received in revised form 16 April 2004; accepted 26 April 2004 Available online 22 December 2004
Abstract In this paper, a surface plasmon resonance (SPR) system is proposed as an alternative approach for glucose measurement. The developed SPR system, which is based on wavelength interrogation technique, has a high detection resolution of 8.67 × 10−6 refractive index unit (RIU), equivalent to 6.23 mg/dL of glucose in water. The experimental results show that the proposed SPR system is capable of measuring the glucose at the concentration levels in physiological ranges, potentially leading to its further development towards a minimal-invasive glucose measurement of interstitial fluid. © 2004 Published by Elsevier B.V. Keywords: Surface plasmon resonance (SPR); Glucose measurement; Refractive index; Wavelength modulation
1. Introduction Diabetes mellitus is a chronic disease and is a leading cause of death in many countries. The World Health Organization (WHO) forecasts the population of diabetes will reach 150 million worldwide by total 2025 [1]. A well-designed treatment for patient depends on frequent measurement of blood glucose level [2]. Therefore, they need to prick their finger to take the blood sample for conventional blood glucose measurement, which is a painful process and can raise other complications. The development of minimum or non-invasive blood glucose measurement techniques can avoid distressing blood harvest and improve living style for diabetic’s patients. The paper proposed an optical configuration for measuring glucose level in water, which results are comparable to physiological glucose range in blood. The results encourage researchers to investigate non-pure mixture, interstitial fluid, measurement with SPR. Surface plasmon resonance (SPR) technique based on an optical measurement approach is highly sensitive to the refractive index unit (RIU) of the sample on its surface. It has been widely adopted in analytical chemistry for detecting the presence and the concentration of chemical substances.
∗ Corresponding author. Tel.: +852-92890369; fax: +852-27885236. E-mail addresses: iou [email protected] (W.W. Lam), [email protected] (Y.T. Zhang).
0925-4005/$ – see front matter. © 2004 Published by Elsevier B.V. doi:10.1016/j.snb.2004.04.088
Some researchers have demonstrated the feasibility of using SPR effect for glucose solution measurement [3,4]. However, the concentration of glucose studied cannot be compared with physiological blood glucose concentration. In this work, a SPR system based on the wavelength interrogation technique is proposed, for measuring refractive index (RI) changes associated with varying the constituent concentrations of glucose–water mixture corresponding to the physiological blood glucose ranging from 30 to 600 mg/dL. This SPR system is constructed by simple optics and electronic components, and it can be miniaturized for future portable monitoring device.
2. Theory Surface plasmon resonance is a phenomenon that the charge-density oscillation induced by a p-polarized optical beam, which undergoes total internal reflection inside a metal-coated glass prism (Fig. 1). The metallic film is usually made of gold or silver and the optimum thickness suggested by the literature [5] is about 50 nm in order to provide better SPR coupling efficiency. During total internal reflection, electric fields of the incident photons leak into the metallic layer and the fields interact with the electrons confined in the metal film. The electron constellation of the metallic film will be excited by the incident photon energy
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 1. Surface plasmon resonance is excited by totally internally reflected p-polarized light at a metal-coated interface when the momentum vector of the incident light in the plane of the surface (kx ) matches with the momentum vector of the surface plasmons in the metallic film (ksp ). The exponential behavior of the enhanced evanescent fields is also indicated on the right.
[5–8], if the condition for surface plasmon excitation is satisfied (kx = ksp ), that is, εmr εa (1) np sin(θ) = εmr + εa 1/2 εp ),
where np is the refractive index of the prism (np = θ the incident angle of the light beam to the interface of prism and metal film, and εmr and εa are the wavelength-dependent complex permittivity of the metallic film and dielectric sample, respectively. From the Drude formula [9], the complex dielectric constant of metal εm and the resonant wavelength λ can be interrelated and written as εm (λ) = εmr + iεmi = 1 −
λ 2 λc λ2p (λc + iλ)
(2)
where εmr and εmi denote the real part and imaginary part of the dielectric constant of metal, respectively, λp denotes the plasma wavelength, λc denotes the collision wavelength, and λ denotes the resonant wavelength. Under the excitation conditions, the energy of the incident light is transferred to charge-density waves (surface plasmons) and the intensity of the light reflected from the surface is reduced, which can be registered as a strong attenuation dip in the SPR curve. According to Eqs. (1) and (2), a small change in refractive index of sample solution on the metallic film will deviate the resonance conditions of SPR. The variation of the resonance condition corresponds to the shift of the light wavelength which causes the SPR effect providing that the incident angle of the light, θ is fixed. As refractive indices of materials are wavelength-dependent, so by taking differentiation of wavelength, λ, with respect to dielectric refractive index, na , the spectral sensitivity can be obtained. spλ =
ε2mr dλ dna (n3a /2) |dεmr /dλ| + (εmr + n2a )εmr (dnp /dλ)(na /np )
(3)
139
Fig. 2. Experimental scheme.
From Eq. (3), Homola and co-workers [4,9] reported that longer resonant wavelength gives higher spectral sensitivity and that the spectral sensitivity of SPR system should thus be improved at longer resonant wavelength.
3. Experimental setup The configuration of the SPR system built for glucose level measurement is shown in Fig. 2. A commercial halogen bulb (manufactured by Philips, Type 7748XHP) serves as the broadband optical light source for exciting the surface plasmon wave. The emission spectrum, which is shown in Fig. 3, spans from about 400 to 1200 and 970 nm is the peak emission wavelength of the light source for a given silicon photo-detector. The emitted light is collimated through sets of compound lenses and an adjustable iris (up to 6 mm diameter). The collimated light is then impinged onto the bottom of a coupling prism at an angle greater than critical angle. The coupling prism is a 60◦ SF-18 prism coated with a ∼50 nm thick gold film via DC sputtering. The reflected light is polarized by a polarizer and then focus into the entrance slit of the monochromator (Acton Research Corporation, SpectraPro-150). The decomposed signal light will be picked up by a silicon photo-detector (manufactured by Integrated Photomatrix, IPL10050) or germanium photo-detector which covers the wavelength region from 800 to 1800 nm (manufactured by Newport, 818IR), amplified by a current pre-amplifier (Standford Research System, SR570), and stored in a personal computer for further analysis.
4. Results and discussions Two experiments of glucose measurements were carried out using the proposed system. In the first experiment, the initial resonant wavelength was tuned to 928 nm by adjusting the incident angle. Nine samples of glucose solution ranging from 30 to 600 mg/dL were prepared and kept at room temperature. The corresponding range of changes in RI is
140
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 3. Emission spectrum of the halogen bulb used in our experiment.
Fig. 4. The normalized SPR curves of some typical glucose–water samples.
from 4.17 × 10−5 to 8.33 × 10−4 RIU with a conversion factor of 7.19 × 105 mg/dL RIU [10,11]. For the sample preparation, a weighting balance with resolution of 0.1 mg was used (AdventurerTM Balances, AR2140). The required glucose concentrations were obtained by calculating the ratios of glucose and de-ionized water in weight percentage. The sample solutions were then fed into a Teflon made sample chamber in turns and the spectra of the reflected light were recorded by a computer-controlled data logger. In order to avoid contamination, the sample chamber was washed by de-ionized water thoughtfully after each sample measurement. After retrieving the spectra of the glucose solutions, an off-line signal processing procedure was used to process the data and to deduce calibration model. Firstly, the obtained spectra of the sample solutions were normalized with respect to the spectra of the air (as shown in Fig. 4) in order to remove the color effects of the light source and the variations
caused by other optical components. Then, an eighth-order polynomial curve-fitting algorithm was employed to determine the resonance wavelength of each sample. Finally, the estimated resonance wavelength is plotted versus the concentration of the samples in Fig. 5, which results were calculated by averaging eight sets of data. It can be seen from Fig. 4 that the resonance wavelength shifts to right as the concentration of glucose increases. Fig. 5 shows a linear regression line fitted with the experimental data. Quantitatively, the correlation between the regression line and the experimental data is 0.9976. These results show the potential of an SPR system for measuring glucose concentrations at the physiological blood glucose levels. The variations as expressed by the standard deviations for each concentration can be attributed to many factors such as the temperature fluctuation, inaccurate sample preparation and noises in the system.
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
141
Fig. 5. The relationship between resonant wavelengths and glucose concentrations with initial λspr ∼ 928 nm.
Fig. 6. Resonant wavelength drift vs. time for our system. The analytical cell was filled with sample water.
The resolution of the system can be deduced by finding the derivative of the calibration curve, and then multiplied by the maximum resolution of the instruments. In the proposed system, the resolution of the monochromator is 1 nm, and thus, the detection resolution can be estimated as 8.61×10−5 RI units (equivalent to 61.94 mg/dL glucose solution). The stability and the maximum resolution of the overall system (with the off-line signal processing algorithm) were studied by monitoring the variation of the spectral dip over a period of 1 h. The result of the stability test is plotted in Fig. 6. The temperature variation during this period was 22.40 ± 0.10 ◦ C. The drift is assumed to be mainly due to temperature variation and random fluctuation associated with the compo-
nents in the system. The standard deviation of the sensor signal indicates the instruments have a maximum resolution of 0.31 nm. This corresponds to the maximum detection resolution of 2.67 × 10−5 RI units (equivalent to 19.21 mg/dL of glucose in water). In order to improve the sensitivity, the second experiment was performed with the system at a higher resonant wavelength. Fig. 7 obtained at the resonant wavelength of about 1200 nm gives the results that were calculated by averaging nine sets of data ranging from 0 to 140 mg/dL. The correlation between the linear regression line and the experimental data is 0.9969. The system successfully resolved 20 mg/dL of glucose in water and its maximum resolution is further enhanced to 8.67 × 10−6 RI units, which corresponds to
142
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 7. The relationship between resonant wavelengths and glucose concentrations with initial λspr ∼ 1200 nm.
6.23 mg/dL of glucose in water. Comparing with other spectral SPR sensing techniques at shorter resonant wavelengths [4,12–15], this result is a very encouraging. It is worth noticing that one of the major noise sources in the system is the temperature variation of the medium above the metal layer. The temperature coefficient of the refractive index of water between 20 and 30 ◦ C, on average, is 1.0 × 10−4 ◦ C−1 [16], which is corresponding to the wavelength shift of about 3.58 nm (or 71.94 mg/dL of glucose in water). As the medium covers the whole of the evanescent field, its variations produce a maximum noise effect on the SPR measurement. Thus temperature must be controlled or monitored to better than a fraction of degree (◦ C). Therefore, the threshold sensitivity of our experimental setup could be further increased if the system is better engineered and a temperature control system is employed. Since light interacts with glucose which will undergo absorptions obeying the Beer–Lambert’s law, chemical noise may be introduced in the measurement. For the wavelengths we used in the SPR experiment are around 930 and 1200 nm which are corresponding to second overtone absorption of glucose [17]. Fortunately, the effect of the absorption is very small as it is not in the obvious absorption band, first overtone, and the path distance is 2 mm only. A literature [18] also claimed that the change of 5 mmol/L (90 mg/dL) glucose concentration costs ∼10−5 absorbance unit changes in finger. Besides, the calibration model is based on the shift of resonance wavelength with respect to different concentrations rather than the absorbance of a particular wavelength. So, the whole measurement phenomenon is dominated by SPR effect. Nevertheless, the SPR setup with high sensitivity explored the capability to detect the concentration of glucose in physiological range. It is the first step for measuring glucose in
interstitial fluid and in blood. Blood glucose can be measured traditionally by fluorescence sensing method which adopted boronic acid to react with glucose to form fluorescent substance [19]. Therefore, we would like to suggest diphenylbornonic acid, a weak acid which does not react with gold in low concentration, as the binding agent coated on the gold film and the measurement could be optimized if suggested chemical is adopted in the reported SPR system.
5. Conclusion A surface plasmon resonance system based on wavelength interrogation for glucose measurements has been presented. The system has a high resolution of 8.67 × 10−6 RI units, which corresponds to the measurement error of 6.23 mg/dL of glucose in aqueous water. A small glucose concentration difference of 20 mg/dL is successfully resolved. The results of these experiments suggest an alternative way for glucose measurement via an optical approach and a possibility of developing a non-invasive blood glucose device based on SPR effect.
Acknowledgements The support of this project by ITF grant and co-sponsorship by IDT and STL are greatly appreciated. The authors would like to express their gratitude to Professor Ho Ho Pui and Mr. Pun Sio Hang for their aggressive discussion. In addition, equipment support from Mr. Wu Shu Yuen at the Department of Physics and Materials Science, City University of Hong Kong is also appreciated.
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
References [1] World Health Organization, Fact sheet of diabetes mellitus, Fact Sheet No. 138, last accessed April 2002, http://www.who.int/inf-fs/ en/fact138.html [2] National Institute of Health, Diabetes overviews, last accessed April 2003, http://www.niddk.nih.gov/health/diabetes/pubs/dmover/ dmover.htm [3] A. Ymetic, J.S. Kanger, R. Wijn, P.V. Lambeck, J. Greve, Development of a multichannel integrated interferometer immunosensor, Sens. Actuat. B 83 (2002) 1–7. [4] Z. Wang, Y. Chen, Analysis of mono- and oligosaccharides by multiwavelength surface plasmon surface (SPR) spectroscopy, Carbohydr. Res. 332 (2001) 209–213. [5] J. Homola, S.S. Yee, G. Gauglitz, Surface plasmon resonance sensors: review, Sens. Actuat. B 54 (1999) 3–15. [6] H.P. Ho, W.W. Lam, S.Y. Wu, Surface plasmon resonance sensor based on the measurement of differential phase, Rev. Scientific Instr. 73 (10) (2002) 3534–3539. [7] F. Markey, What is SPR anyway, Bia J. 1 (1999) 14–17. [8] F. Markey, Principles of Surface Plasmon Resonance, Real-Time Analysis of Biomolecular Interactions, Springer-Verlag, Tokyo, 2000, pp. 13–22. [9] J. Homola, I. Koudela, S.S. Yee, Surface plasmon resonance sensors based on diffraction gratings and prism couplers: sensitivity comparison, Sens. Actuat. B 54 (1999) 16–24. [10] J.S. Marier, S.A. Walker, S. Fantini, M.A. Franceschini, E. Gratton, Possible correlation between blood glucose concentration and the
[11]
[12]
[13]
[14]
[15]
[16] [17] [18] [19]
143
reduced scattering coefficient of tissues in the near infrared, Opt. Lett. 19 (24) (1994) 2062–2064. M. Kohl, M. Cope, Influence of glucose concentration on light scattering in tissue-simulating phantoms, Opt. Lett. 19 (24) (1994) 2170–2172. H.P. Ho, S.Y. Wu, M. Yang, A.C. Cheung, Application of white light-emitting diode to surface plasmon resonance sensors, Sens. Actuat. B 80 (2001) 89–94. K.H. Chen, C.C. Hsu, D.C. Su, Measurement of wavelength shift by using surface plasmon resonance heterodyne interferometry, Opt. Commun. 209 (2002) 167–172. C.P. Cahill, K.S. Johnston, S.S. Yee, A surface plasmon resonance sensor probe based on retro-reflection, Sens. Actuat. B 45 (1997) 161–166. G.G. Nenninger, J.B. Clendenning, C.E. Furlong, S.S. Yee, Reference-compensated biosensing using a dual-channel surface plasmon resonance sensor system based on a planar lightpipe configuration, Sens. Actuat. B 51 (1998) 38–45. R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics, 68th ed., CRC Press, Boca Raton, FL, 1978. R.A. Shaw, H.H. Mantsch, Infrared Spectroscopy in Clinical and Diagnostic Analysis, Encyclopedia of Analytical Chemistry. O.S. Khalil, Spectroscopic and clinical aspects of noninvasive glucose measurements, Clin. Chem. 45 (2) (1999) 165–177. V.V. Karnati, X. Gao, S. Gao, W. Yang, W. Ni, S. Sankar, B. Wang, A glucose-selective fluorescence sensor based on boronicaciddiol recognition, Bioorg. Med. Chem. Lett. 12 (23) P3373– P3377.
A surface plasmon resonance system for the measurement of glucose in aqueous solution W.W. Lam a,∗ , L.H. Chu a , C.L. Wong b , Y.T. Zhang a a
Department of Electronic Engineering, Joint Research Center for Biomedical Engineering, The Chinese University of Hong Kong, Shatin, NT, Hong Kong b Department of Physics and Materials Science, The City University of Hong Kong, Kowloon, Hong Kong Received 12 July 2003; received in revised form 16 April 2004; accepted 26 April 2004 Available online 22 December 2004
Abstract In this paper, a surface plasmon resonance (SPR) system is proposed as an alternative approach for glucose measurement. The developed SPR system, which is based on wavelength interrogation technique, has a high detection resolution of 8.67 × 10−6 refractive index unit (RIU), equivalent to 6.23 mg/dL of glucose in water. The experimental results show that the proposed SPR system is capable of measuring the glucose at the concentration levels in physiological ranges, potentially leading to its further development towards a minimal-invasive glucose measurement of interstitial fluid. © 2004 Published by Elsevier B.V. Keywords: Surface plasmon resonance (SPR); Glucose measurement; Refractive index; Wavelength modulation
1. Introduction Diabetes mellitus is a chronic disease and is a leading cause of death in many countries. The World Health Organization (WHO) forecasts the population of diabetes will reach 150 million worldwide by total 2025 [1]. A well-designed treatment for patient depends on frequent measurement of blood glucose level [2]. Therefore, they need to prick their finger to take the blood sample for conventional blood glucose measurement, which is a painful process and can raise other complications. The development of minimum or non-invasive blood glucose measurement techniques can avoid distressing blood harvest and improve living style for diabetic’s patients. The paper proposed an optical configuration for measuring glucose level in water, which results are comparable to physiological glucose range in blood. The results encourage researchers to investigate non-pure mixture, interstitial fluid, measurement with SPR. Surface plasmon resonance (SPR) technique based on an optical measurement approach is highly sensitive to the refractive index unit (RIU) of the sample on its surface. It has been widely adopted in analytical chemistry for detecting the presence and the concentration of chemical substances.
∗ Corresponding author. Tel.: +852-92890369; fax: +852-27885236. E-mail addresses: iou [email protected] (W.W. Lam), [email protected] (Y.T. Zhang).
0925-4005/$ – see front matter. © 2004 Published by Elsevier B.V. doi:10.1016/j.snb.2004.04.088
Some researchers have demonstrated the feasibility of using SPR effect for glucose solution measurement [3,4]. However, the concentration of glucose studied cannot be compared with physiological blood glucose concentration. In this work, a SPR system based on the wavelength interrogation technique is proposed, for measuring refractive index (RI) changes associated with varying the constituent concentrations of glucose–water mixture corresponding to the physiological blood glucose ranging from 30 to 600 mg/dL. This SPR system is constructed by simple optics and electronic components, and it can be miniaturized for future portable monitoring device.
2. Theory Surface plasmon resonance is a phenomenon that the charge-density oscillation induced by a p-polarized optical beam, which undergoes total internal reflection inside a metal-coated glass prism (Fig. 1). The metallic film is usually made of gold or silver and the optimum thickness suggested by the literature [5] is about 50 nm in order to provide better SPR coupling efficiency. During total internal reflection, electric fields of the incident photons leak into the metallic layer and the fields interact with the electrons confined in the metal film. The electron constellation of the metallic film will be excited by the incident photon energy
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 1. Surface plasmon resonance is excited by totally internally reflected p-polarized light at a metal-coated interface when the momentum vector of the incident light in the plane of the surface (kx ) matches with the momentum vector of the surface plasmons in the metallic film (ksp ). The exponential behavior of the enhanced evanescent fields is also indicated on the right.
[5–8], if the condition for surface plasmon excitation is satisfied (kx = ksp ), that is, εmr εa (1) np sin(θ) = εmr + εa 1/2 εp ),
where np is the refractive index of the prism (np = θ the incident angle of the light beam to the interface of prism and metal film, and εmr and εa are the wavelength-dependent complex permittivity of the metallic film and dielectric sample, respectively. From the Drude formula [9], the complex dielectric constant of metal εm and the resonant wavelength λ can be interrelated and written as εm (λ) = εmr + iεmi = 1 −
λ 2 λc λ2p (λc + iλ)
(2)
where εmr and εmi denote the real part and imaginary part of the dielectric constant of metal, respectively, λp denotes the plasma wavelength, λc denotes the collision wavelength, and λ denotes the resonant wavelength. Under the excitation conditions, the energy of the incident light is transferred to charge-density waves (surface plasmons) and the intensity of the light reflected from the surface is reduced, which can be registered as a strong attenuation dip in the SPR curve. According to Eqs. (1) and (2), a small change in refractive index of sample solution on the metallic film will deviate the resonance conditions of SPR. The variation of the resonance condition corresponds to the shift of the light wavelength which causes the SPR effect providing that the incident angle of the light, θ is fixed. As refractive indices of materials are wavelength-dependent, so by taking differentiation of wavelength, λ, with respect to dielectric refractive index, na , the spectral sensitivity can be obtained. spλ =
ε2mr dλ dna (n3a /2) |dεmr /dλ| + (εmr + n2a )εmr (dnp /dλ)(na /np )
(3)
139
Fig. 2. Experimental scheme.
From Eq. (3), Homola and co-workers [4,9] reported that longer resonant wavelength gives higher spectral sensitivity and that the spectral sensitivity of SPR system should thus be improved at longer resonant wavelength.
3. Experimental setup The configuration of the SPR system built for glucose level measurement is shown in Fig. 2. A commercial halogen bulb (manufactured by Philips, Type 7748XHP) serves as the broadband optical light source for exciting the surface plasmon wave. The emission spectrum, which is shown in Fig. 3, spans from about 400 to 1200 and 970 nm is the peak emission wavelength of the light source for a given silicon photo-detector. The emitted light is collimated through sets of compound lenses and an adjustable iris (up to 6 mm diameter). The collimated light is then impinged onto the bottom of a coupling prism at an angle greater than critical angle. The coupling prism is a 60◦ SF-18 prism coated with a ∼50 nm thick gold film via DC sputtering. The reflected light is polarized by a polarizer and then focus into the entrance slit of the monochromator (Acton Research Corporation, SpectraPro-150). The decomposed signal light will be picked up by a silicon photo-detector (manufactured by Integrated Photomatrix, IPL10050) or germanium photo-detector which covers the wavelength region from 800 to 1800 nm (manufactured by Newport, 818IR), amplified by a current pre-amplifier (Standford Research System, SR570), and stored in a personal computer for further analysis.
4. Results and discussions Two experiments of glucose measurements were carried out using the proposed system. In the first experiment, the initial resonant wavelength was tuned to 928 nm by adjusting the incident angle. Nine samples of glucose solution ranging from 30 to 600 mg/dL were prepared and kept at room temperature. The corresponding range of changes in RI is
140
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 3. Emission spectrum of the halogen bulb used in our experiment.
Fig. 4. The normalized SPR curves of some typical glucose–water samples.
from 4.17 × 10−5 to 8.33 × 10−4 RIU with a conversion factor of 7.19 × 105 mg/dL RIU [10,11]. For the sample preparation, a weighting balance with resolution of 0.1 mg was used (AdventurerTM Balances, AR2140). The required glucose concentrations were obtained by calculating the ratios of glucose and de-ionized water in weight percentage. The sample solutions were then fed into a Teflon made sample chamber in turns and the spectra of the reflected light were recorded by a computer-controlled data logger. In order to avoid contamination, the sample chamber was washed by de-ionized water thoughtfully after each sample measurement. After retrieving the spectra of the glucose solutions, an off-line signal processing procedure was used to process the data and to deduce calibration model. Firstly, the obtained spectra of the sample solutions were normalized with respect to the spectra of the air (as shown in Fig. 4) in order to remove the color effects of the light source and the variations
caused by other optical components. Then, an eighth-order polynomial curve-fitting algorithm was employed to determine the resonance wavelength of each sample. Finally, the estimated resonance wavelength is plotted versus the concentration of the samples in Fig. 5, which results were calculated by averaging eight sets of data. It can be seen from Fig. 4 that the resonance wavelength shifts to right as the concentration of glucose increases. Fig. 5 shows a linear regression line fitted with the experimental data. Quantitatively, the correlation between the regression line and the experimental data is 0.9976. These results show the potential of an SPR system for measuring glucose concentrations at the physiological blood glucose levels. The variations as expressed by the standard deviations for each concentration can be attributed to many factors such as the temperature fluctuation, inaccurate sample preparation and noises in the system.
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
141
Fig. 5. The relationship between resonant wavelengths and glucose concentrations with initial λspr ∼ 928 nm.
Fig. 6. Resonant wavelength drift vs. time for our system. The analytical cell was filled with sample water.
The resolution of the system can be deduced by finding the derivative of the calibration curve, and then multiplied by the maximum resolution of the instruments. In the proposed system, the resolution of the monochromator is 1 nm, and thus, the detection resolution can be estimated as 8.61×10−5 RI units (equivalent to 61.94 mg/dL glucose solution). The stability and the maximum resolution of the overall system (with the off-line signal processing algorithm) were studied by monitoring the variation of the spectral dip over a period of 1 h. The result of the stability test is plotted in Fig. 6. The temperature variation during this period was 22.40 ± 0.10 ◦ C. The drift is assumed to be mainly due to temperature variation and random fluctuation associated with the compo-
nents in the system. The standard deviation of the sensor signal indicates the instruments have a maximum resolution of 0.31 nm. This corresponds to the maximum detection resolution of 2.67 × 10−5 RI units (equivalent to 19.21 mg/dL of glucose in water). In order to improve the sensitivity, the second experiment was performed with the system at a higher resonant wavelength. Fig. 7 obtained at the resonant wavelength of about 1200 nm gives the results that were calculated by averaging nine sets of data ranging from 0 to 140 mg/dL. The correlation between the linear regression line and the experimental data is 0.9969. The system successfully resolved 20 mg/dL of glucose in water and its maximum resolution is further enhanced to 8.67 × 10−6 RI units, which corresponds to
142
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
Fig. 7. The relationship between resonant wavelengths and glucose concentrations with initial λspr ∼ 1200 nm.
6.23 mg/dL of glucose in water. Comparing with other spectral SPR sensing techniques at shorter resonant wavelengths [4,12–15], this result is a very encouraging. It is worth noticing that one of the major noise sources in the system is the temperature variation of the medium above the metal layer. The temperature coefficient of the refractive index of water between 20 and 30 ◦ C, on average, is 1.0 × 10−4 ◦ C−1 [16], which is corresponding to the wavelength shift of about 3.58 nm (or 71.94 mg/dL of glucose in water). As the medium covers the whole of the evanescent field, its variations produce a maximum noise effect on the SPR measurement. Thus temperature must be controlled or monitored to better than a fraction of degree (◦ C). Therefore, the threshold sensitivity of our experimental setup could be further increased if the system is better engineered and a temperature control system is employed. Since light interacts with glucose which will undergo absorptions obeying the Beer–Lambert’s law, chemical noise may be introduced in the measurement. For the wavelengths we used in the SPR experiment are around 930 and 1200 nm which are corresponding to second overtone absorption of glucose [17]. Fortunately, the effect of the absorption is very small as it is not in the obvious absorption band, first overtone, and the path distance is 2 mm only. A literature [18] also claimed that the change of 5 mmol/L (90 mg/dL) glucose concentration costs ∼10−5 absorbance unit changes in finger. Besides, the calibration model is based on the shift of resonance wavelength with respect to different concentrations rather than the absorbance of a particular wavelength. So, the whole measurement phenomenon is dominated by SPR effect. Nevertheless, the SPR setup with high sensitivity explored the capability to detect the concentration of glucose in physiological range. It is the first step for measuring glucose in
interstitial fluid and in blood. Blood glucose can be measured traditionally by fluorescence sensing method which adopted boronic acid to react with glucose to form fluorescent substance [19]. Therefore, we would like to suggest diphenylbornonic acid, a weak acid which does not react with gold in low concentration, as the binding agent coated on the gold film and the measurement could be optimized if suggested chemical is adopted in the reported SPR system.
5. Conclusion A surface plasmon resonance system based on wavelength interrogation for glucose measurements has been presented. The system has a high resolution of 8.67 × 10−6 RI units, which corresponds to the measurement error of 6.23 mg/dL of glucose in aqueous water. A small glucose concentration difference of 20 mg/dL is successfully resolved. The results of these experiments suggest an alternative way for glucose measurement via an optical approach and a possibility of developing a non-invasive blood glucose device based on SPR effect.
Acknowledgements The support of this project by ITF grant and co-sponsorship by IDT and STL are greatly appreciated. The authors would like to express their gratitude to Professor Ho Ho Pui and Mr. Pun Sio Hang for their aggressive discussion. In addition, equipment support from Mr. Wu Shu Yuen at the Department of Physics and Materials Science, City University of Hong Kong is also appreciated.
W.W. Lam et al. / Sensors and Actuators B 105 (2005) 138–143
References [1] World Health Organization, Fact sheet of diabetes mellitus, Fact Sheet No. 138, last accessed April 2002, http://www.who.int/inf-fs/ en/fact138.html [2] National Institute of Health, Diabetes overviews, last accessed April 2003, http://www.niddk.nih.gov/health/diabetes/pubs/dmover/ dmover.htm [3] A. Ymetic, J.S. Kanger, R. Wijn, P.V. Lambeck, J. Greve, Development of a multichannel integrated interferometer immunosensor, Sens. Actuat. B 83 (2002) 1–7. [4] Z. Wang, Y. Chen, Analysis of mono- and oligosaccharides by multiwavelength surface plasmon surface (SPR) spectroscopy, Carbohydr. Res. 332 (2001) 209–213. [5] J. Homola, S.S. Yee, G. Gauglitz, Surface plasmon resonance sensors: review, Sens. Actuat. B 54 (1999) 3–15. [6] H.P. Ho, W.W. Lam, S.Y. Wu, Surface plasmon resonance sensor based on the measurement of differential phase, Rev. Scientific Instr. 73 (10) (2002) 3534–3539. [7] F. Markey, What is SPR anyway, Bia J. 1 (1999) 14–17. [8] F. Markey, Principles of Surface Plasmon Resonance, Real-Time Analysis of Biomolecular Interactions, Springer-Verlag, Tokyo, 2000, pp. 13–22. [9] J. Homola, I. Koudela, S.S. Yee, Surface plasmon resonance sensors based on diffraction gratings and prism couplers: sensitivity comparison, Sens. Actuat. B 54 (1999) 16–24. [10] J.S. Marier, S.A. Walker, S. Fantini, M.A. Franceschini, E. Gratton, Possible correlation between blood glucose concentration and the
[11]
[12]
[13]
[14]
[15]
[16] [17] [18] [19]
143
reduced scattering coefficient of tissues in the near infrared, Opt. Lett. 19 (24) (1994) 2062–2064. M. Kohl, M. Cope, Influence of glucose concentration on light scattering in tissue-simulating phantoms, Opt. Lett. 19 (24) (1994) 2170–2172. H.P. Ho, S.Y. Wu, M. Yang, A.C. Cheung, Application of white light-emitting diode to surface plasmon resonance sensors, Sens. Actuat. B 80 (2001) 89–94. K.H. Chen, C.C. Hsu, D.C. Su, Measurement of wavelength shift by using surface plasmon resonance heterodyne interferometry, Opt. Commun. 209 (2002) 167–172. C.P. Cahill, K.S. Johnston, S.S. Yee, A surface plasmon resonance sensor probe based on retro-reflection, Sens. Actuat. B 45 (1997) 161–166. G.G. Nenninger, J.B. Clendenning, C.E. Furlong, S.S. Yee, Reference-compensated biosensing using a dual-channel surface plasmon resonance sensor system based on a planar lightpipe configuration, Sens. Actuat. B 51 (1998) 38–45. R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics, 68th ed., CRC Press, Boca Raton, FL, 1978. R.A. Shaw, H.H. Mantsch, Infrared Spectroscopy in Clinical and Diagnostic Analysis, Encyclopedia of Analytical Chemistry. O.S. Khalil, Spectroscopic and clinical aspects of noninvasive glucose measurements, Clin. Chem. 45 (2) (1999) 165–177. V.V. Karnati, X. Gao, S. Gao, W. Yang, W. Ni, S. Sankar, B. Wang, A glucose-selective fluorescence sensor based on boronicaciddiol recognition, Bioorg. Med. Chem. Lett. 12 (23) P3373– P3377.