Anal. Chem. 2008, 80, 2957-2962
Hydrogel-Based Piezoresistive pH Sensors: Investigations Using FT-IR Attenuated Total Reflection Spectroscopic Imaging Joerg Sorber,*,† Gerald Steiner,*,‡,⊥ Volker Schulz,† Margarita Guenther,† Gerald Gerlach,† Reiner Salzer,‡ and Karl-Friedrich Arndt§
Institute for Solid-State Electronics, Institute for Analytical Chemistry, and Institute for Physical Chemistry and Electrochemistry, Dresden University of Technology, 01062 Dresden, Germany, and Clinical Sensoring and Monitoring, Medical Faculty “Carl Gustav Carus”, Dresden University of Technology, 01307 Dresden, Germany
The strong swelling ability of the pH-responsive poly(acrylic acid)/poly(vinyl alcohol) (PAA/PVA) hydrogel makes the development of a new type of sensor possible, which combines piezoresistive-responsive elements as mechanoelectrical transducers and the phase transition behavior of hydrogels as a chemomechanical transducer. The sensor consists of a pH-responsive PAA/PVA hydrogel and a standard pressure sensor chip. However, a timedependent sensor output voltage mirrors only the physical swelling process of the hydrogel but not the corresponding chemical reactions. Therefore, an investigation of the swelling behavior of this hydrogel is essential for the optimization of sensor design. In this work, Fourier transform infrared (FT-IR) spectroscopic imaging was used to study the swelling of the hydrogel under in situ conditions. In particular, laterally and time-resolved FTIR images were obtained in the attenuated total reflection mode and the entire data set of more than 80 000 FT-IR spectra was evaluated by principal component analysis (PCA). The first and third principal components (PCs) indicate the swelling process. Molecular changes within the carboxyl groups were observed in the second and fourth PC and identified as key processes for the swelling behavior. It was found that time-dependent molecular changes are similar to the electrical sensor output signal. The results of the FT-IR spectroscopic images render an improved chemical sensor possible and demonstrate that in situ FT-IR imaging is a powerful method for the characterization of molecular processes within chemicalsensitive materials. Hydrogels are a special class of polymer networks which absorb large amounts of water into their structures. Such materials are used in biomedical applications like diagnostic, therapeutic, and implantation devices.1,2 Recently, hydrogels have received great attention for being chemical sensors. In particular, hydrogels * To whom correspondence should be addressed. E-mail: Joerg.Sorber@ tu-dresden.de (J.S.),
[email protected] (G.S.). † Institute for Solid-State Electronics. ‡ Institute for Analytical Chemistry. § Institute for Physical Chemistry and Electrochemistry. ⊥ Clinical Sensoring and Monitoring, Medical Faculty “Carl Gustav Carus”. 10.1021/ac702598n CCC: $40.75 Published on Web 02/28/2008
© 2008 American Chemical Society
that are sensitive to pH have become increasingly important for use in chemical microsensor technology. Poly(acrylic acid)/poly(vinyl alcohol) (PAA/PVA) hydrogel has been one of the most studied polymer networks during the past 10 years due to its promising applications in biomedical engineering. PAA/PVA hydrogel swells and shrinks in response to a change in pH value of the surrounding solution.3 The basic condition for swelling is the dissociation of the carboxylic acid (COOH) of PAA to the negatively charged carboxylate ion (COO-). The chain-bound, negatively charged carboxylate groups repel each other. The resulting electrostatic force causes swelling. The swelling capacity can be enhanced both by osmotic pressure and by chain repulsion.4 If the chains of an ionized network carry charges of the same sign, the chains drift apart because of electrostatic repulsion. Furthermore, the free counterions inside the hydrogel create an osmotic pressure between solvent and gel, which drives the diffusion from the solvent into the network. PAA/PVA is able to absorb up to several hundred times its own weight of salt-free solvent. A combination of the PAA/PVA hydrogel and a mechanoelectrical transducer offers a lot of promising applications.4,5 Piezoresistive silicon pressure sensor chips are very advantageous to be used as such a mechanoelectrical transducer because they are low-cost, well-investigated, and show high reliability as well as long-term stability.6 Due to the comparatively excellent behavior of piezoresistive sensor technology, long-term stability of hydrogelbased sensors is determined by the hydrogel’s behavior. A model to correlate the swelling equilibrium of nonionic hydrogels in water and aqueous nonelectrolyte solutions was presented in ref 7. Due to its complexity, the time-dependent swelling ability of pH-sensitive polyelectrolyte hydrogels is not (1) Gehrke, S. H.; Lee, P. I. In Specialized Drug Delivery Systems; Tyle, P., Ed.; Marcel Dekker: New York, 1990. (2) Kim, S. W.; Bae, Y. H.; Okano, T. Pharm. Res. 1992, 3, 283-290. (3) Arndt, K.-F.; Richter, A.; Ludwig, S.; Zimmermann, J.; Kressler, J.; Kuckling, D.; Adler, H.-J. Acta Polym. 1999, 50, 383-390. (4) Gerlach, G.; Guenther, M.; Sorber, J.; Suchaneck, G.; Arndt, K.-F.; Richter, A. Sens. Actuators, B 2005, 111-112, 555-561. (5) Sorber, J.; Gerlach, G.; Arndt, K.-F.; Richter, A. German Patent DE 101 29 986 C2, Dec 6, 2001. (6) Gerlach, G.; Werthschu ¨ tzky, R. Tech. Mess. 2005, 72, 53-76. (7) Hu ¨ ther, A.; Scha¨fer, B.; Xu, X.; Maurer, G. Phys. Chem. Chem. Phys. 2002, 4, 835-844.
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 2957
Figure 1. Functional principle of hydrogel-based piezoresistive pH sensor: (1) deflected bending plate; (2) mechanoelectrical transducer, comprising piezoresistors connected to a Wheatstone bridge; (3) swollen PAA/PVA hydrogel; (4) standard pressure sensor chip; (5) TO8 package; (6) channels for solution inlet and outlet; (7) bonding wire; (8) solution; (9) pedestal.
well-understood yet. This concerns the physical swelling process as well as the accompanying structural chemical changes. Monitoring the sensor output voltage allows the investigation of the physical swelling process. However, it does not recognize chemical changes. A reliable assessment of the swelling process of the hydrogel layer requires molecular-sensitive imaging techniques. Fourier transform infrared (FT-IR) imaging spectroscopy combines the high sensitivity for molecular structure analysis with a spatial resolution down to few micrometers.8 Additionally, Fourier transform infrared attenuated total reflection (FT-IR-ATR) spectroscopic imaging is a novel and excellent tool for a time- and space-resolved analysis of molecular processes.9 In this study, FTIR-ATR spectroscopic imaging has been used for the first time for in situ examination of the swelling process. This technique enables the characterization of the PAA/PVA hydrogel with the corresponding chemical processes running in the background. The knowledge of temporally and spatially resolved processes occurring chemically in the hydrogel during the swelling process supports the development of sensors with even better properties. Thin hydrogel layers with a thickness of 50-90 µm directly deposited on a flat stainless steel substrate have been used to investigate the chemical process within the hydrogel. This allows the observation of the chemical mechanisms at the hydrogel layer surface. EXPERIMENTAL SECTION pH-Sensitive Hydrogel Sensor. For the pH sensor design, a sensor chip with a distortable thin silicon bending plate was used. Commercially available pressure sensor chips (Aktiv Sensor GmbH, Stahnsdorf, Germany) were used as mechanoelectrical transducers for the transformation of bending plate deflections into an appropriate electrical output voltage. In this case, it is correct to use the term bending plate10 because the bending state is dominant compared to the tensile stress over the whole plate length. Like the air pressure in a pressure sensor, the swelling hydrogel leads to a deflection of the silicon bending plate locally thinned-down within the sensor chip. The chip itself is mounted on a modified TO8 package (Figure 1). The analyte is pumped through the sensor with a flow rate of 25 µL/min. The circular hydrogel layer (diameter 1.5 mm) has (8) Levin, I. W.; Bhargava, R. Annu. Rev. Phys. Chem. 2005, 56, 429-474. (9) Chan, K. L. A.; Kazarian, S. G. Appl. Spectrosc. 2003, 57, 381-389. (10) Timoshenko, S. P.; Woinowsky-Kreiger, S. Theory of Plates and Shells, 2nd ed.; McGraw-Hill: New York, 1964.
2958 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
Figure 2. Measuring setup: (1) stainless steel plate; (2) PAA/PVA hydrogel; (3) ZnSe ATR crystal; (4) flow-through measuring cell; (5) infrared light; (6) MCT 64 × 64 array detector, (R) angle of incidence, (F) force applied.
an initial thickness of 80 µm, and it touches the square surface of the bending plate (width 3 mm, thickness 20 µm). A Wheatstone bridge of four integrated piezoresistors delivers the sensor output voltage which corresponds to the deformation of the bending plate and hence to the swelling. Gel Preparation. The polymer components were dissolved separately in distilled water (80 °C, 15 wt % PVA and 7.5 wt % PAA). The stock solutions were mixed with a resulting composition of 80 wt % PVA and 20 wt % PAA. PVA serves as the scaffolding and PAA as the pH-responsive component. The hydrogel layer was deposited on a stainless steel disc by spincoating, drying, and thermal cross-linking (20 min at 130 °C). A 1 h conditioning was performed in deionized water. FT-IR Imaging. FT-IR spectroscopic images were collected using a Bruker FT-IR imaging spectrometer Hyperion (Bruker Optik GmbH, Ettlingen, Germany) coupled to a macro ATR unit (Specac Inc., Woodstock, GA) with a ZnSe crystal as the optical element. To perform in situ measurements under continuous flow condition a dedicated accessory was developed, which is shown in Figure 2. A stainless steel plate covered with a thin hydrogel layer is placed onto the surface of the ATR crystal. The stainless steel plate has a central hole in order to permit the continuous flow of a NaOH solution with a constant pH of 10 to the hydrogel. At the same time the steel plate ensures good contact of the hydrogel layer to ATR crystal surface. The PAA/PVA hydrogel layer was pressed toward the ATR crystal by a constant force of F ) 10 N. The angle of incidence, R, of 45° ensures total internal reflection of the infrared light. The reflected light is captured by the imaging detector array of the spectrometer. It imaged a sample area of approximately 4 × 4 mm2. In order to eliminate radiation of unwanted wavelengths and prevent Fourier fold-over perturbations a low-pass filter was inserted into the beam. After an automatic camera gain and offset optimization, 11 interferograms were coadded for each of the 4096 image pixels at a resolution of 4 cm-1. With the use of Happ-Genzel apodization and a zero filling factor of 1, the spectra were acquired in the internal reflection mode in the spectral range of 950-3600 cm-1. The frame rate of the camera was 252 Hz, yielding a total measurement time of 1.5 min per spectroscopic image. A series of 20 FT-IR spectroscopic
Figure 3. Sensor output voltage upon pH changes between pH 2 and pH 12.
Figure 5. FT-IR-ATR spectroscopic image of the PAA/PVA hydrogel layer. The framed area is used for subsequent evaluations. Figure 4. FT-IR-ATR spectrum of a PAA/PVA hydrogel layer.
images was captured over a time range of 420 min. Finally, the MatLab (version 5.6, MathWorks Inc., Natric, MA) package served as mean for data preprocessing, image processing, and feature extraction. Spectra of the sample image were ratioed against a neat surface of the ATR crystal, transferred to absorbance values, and normalized to the largest band, the δ(O-H) mode (see below). RESULTS AND DISCUSSION Swelling Behavior after pH Changes. Figure 3 shows measured data of a sensor as sketched in Figure 1. In this case, 0.1 M HCl and 0.1 M NaOH solutions were pumped alternately through the sensor. The measurements show directly the physical effects of swelling and shrinking. Typically, spikes occur systematically at the beginning of the swelling process, at defined stages during swelling, and before the shrinking starts. To explain this very typical behavior, ATR spectra were analyzed with respect to temporal and spatial changes. Spectral Absorption in PAA/PVA. PAA/PVA hydrogel has a number of dominant absorption bands in the infrared fingerprint region between 1000 and 1800 cm-1. Figure 4 shows an FT-IRATR spectrum taken from a 500 µm thick PAA/PVA hydrogel layer after admission of NaOH solution. Bands between 1000 and 1300 cm-1 arise mainly from stretching modes of the C-O bond, partly in combination with the
deformation mode of the C-H groups.11,12 The bands between 1380 and 1500 cm-1 are assigned to the deformation (δ) (C-H) modes.13,14 The strong band between 1500 and 1750 cm-1 originates from the δ(O-H) mode of water12 (1640 cm-1) as well as from the carbonyl stretching mode of the COOH groups (above 1700 cm-1) and the COO- groups (below 1600 cm-1). Spatially Resolved Evaluation. Figure 5 displays the infrared spectroscopic image of a PAA/PVA hydrogel layer. The graycoded image is calculated from the integrated intensities of the ν(C-O) bands in the spectral range from 1000 to 1300 cm-1. Herein, dark pixels indicate smallest absorption. They occur at the inner hole with no abundant PAA/PVA hydrogel and outside the PAA/PVA ring. The hydrogel ring is clearly discernible; its strong absorption is indicated by areas of bright pixels. However, the absorption of the PAA/PVA hydrogel is not uniform across the ring area. This indicates either fluctuations in the density of the molecular chains or variations in the optical contact between the polymer and the optical crystal. For the following data evaluation, the area within the white dashed lines was used. This area exhibits typical fluctuations in the PAA/PVA absorbance (11) Griffiths, P. R.; De Haseth, J. A. Fourier Transform Infrared Spectrometry, 2nd ed.; Wiley: Hoboken, NJ, 2007. (12) Zaman, A. A.; Tsuchiya, R.; Moudgil, B. M. J. Colloid Interface Sci. 2002, 256, 73-78. (13) Sokrates, G. Infrared and Raman Characteristic Group Frequencies: Tables and Charts; Wiley: Chichester, U.K., 2004. (14) Kirwan, L. J.; Fawell, P. D.; van Bronswijk, W. Langmuir 2003, 19, 58025807.
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
2959
Figure 6. Normalized integrated absorbance in the spectral range between 1000 and 1150 cm-1 of the PAA/PVA hydrogel vs time.
values, and it covers parts of all sample regions (hole, hydrogel, outer area). The swelling process starts immediately after admission of the NaOH solution. This is indicated by the increase of the ν(C-O) absorbance bands over time (Figure 6). All values are normalized to the integrated absorbance at t ) 0 min. The plot is similar to one swelling cycle of the PAA/PVA hydrogel sensor shown in Figure 3. The absorbance increase during swelling (Figure 6) indicates chemical as well as physical effects since the amount of molecular groups in the observed hydrogel volume cannot grow. These measurements result from a definite area which is determined by the optical layout and by the detector geometry. The investigated layer thickness is determined by the evanescent field of the IR radiation, which falls off exponentially from the ATR surface over a distance of several hundred nanometers. What changes within this volume is the density of PAA/PVA hydrogel becoming smaller during swelling. That implies the number of hydrogel groups reached by IR radiation gets smaller as well and the measured absorbance should lessen but not increase as in Figure 6. Under the experimental conditions used here, the optical contact between the PAA/PVA hydrogel and the ATR crystal is very critical. The pressure within the hydrogel increases during swelling. This improves the optical contact between the hydrogel and the ATR crystal. Molecules that have direct contact to the ATR crystal experience a much higher strength of the evanescent field and surface absorb more infrared light than molecules in a certain distance from the surface. When the conditioned PAA/ PVA hydrogel layer is placed onto the ATR crystal surface, a thin film of water is formed between the ATR crystal surface and the PAA/PVA hydrogel network.15 As a consequence, regions of high strength of the evanescent field are initially occupied by water molecules. Hydrogel molecular groups gradually replace water molecules close to the ATR crystal surface as the pressure increases. This replacement causes the increase of the ν(C-O) absorbance in Figure 6. The swelling of the PAA/PVA hydrogel in Figure 6 reaches equilibrium after 250 min. Similar to Figure 3, the absorbance change in Figure 6 reveals two regions of reduced swelling rates, in the initial phase and between 90 and 120 min. A similar behavior (15) Ikada, Y.; Uyama, Y. Lubricating Polymer Surfaces; Technomic Publishing Company: Lancester, PA, 1993.
2960 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
was observed by Tamura et al.16 They found that volume first decreases before it starts to increase. The decrease was attributed to the sudden rise in the ionic strength at the very beginning of the experiment. Slight variations in the occurrence of the features in Figure 3 and in Figure 6 may originate from the different thicknesses of the PAA/PVA hydrogel layers and from the influence of the external force of F ) 10 N in case of the spectroscopic experiment. Dissociation of Carboxyl Groups. Dissociation of carboxylic acid was already mentioned as key condition for hydrogel swelling. As the swelling process is coupled to the dissociation of carboxyl groups, the evaluation was focused on this molecular entity. The multivariate method of choice for the evaluation of the FT-IR spectra proved to be principal component analysis (PCA). PCA provides information about spectral regions with the highest variance during a measurement series orsas in case of this studysacross a spectroscopic image. Principal components (PCs) of a spectroscopic image can be visualized by loading plots and score maps. The loading plots resemble spectra with features both in positive and negative directions. Loading plots provide insight in the molecular processes during the experiment. The score maps reveal the weight of the loading plot for each pixel in the image. This permits localization of the molecular processes in particular regions of the swelling zone. The PCA in Figure 7 was performed with the entire data set which consists of 20 individual FT-IR spectroscopic images of the FT-IR spectra taken from the area highlighted in Figure 5. The four PCs cover 98.6% of the total variance across the investigated area (first PC, 87.1%; second PC, 4.2%; third PC, 3.8%; fourth PC, 1.6%). The shape of the loading plot of the first PC (Figure 7A) clearly resembles the experimental spectrum in Figure 4. The first PC is similar to average spectrum and indicates the δ(O-H) mode of water plus contributions from the carbonyl stretching modes of -COO and -COO- (cf., the Spectral Absorption in PAA/PVA section). Apparently, the first PC comprises all components of the hydrogel without chemical changes. The shape of the first PC and its dominating share in variance suggests it mainly has a physical background like reflection at the boundary. The loading plot of the second PC exhibits a distinct maximum at the characteristic frequency of carboxylic acid. The third PC has its maximum at the same position as the first PC, but it is narrower than the latter. Moreover, its small contribution to the variance in the measurement series may be related to the behavior of a single constituent in the sample. According to the location of its maximum, the third PC indicates the influence of water on the measurement series. The contribution of this PC is negative and the scores slightly increase over time; this indicates fading participation of water. The loading plot of the fourth PC clearly indicates carboxylate groups by the presence of the antisymmetric stretching vibration at about 1550 cm-1 and the symmetric stretch at 1430 cm-1. The molecular background of the swelling process is revealed by the loading plots, whereas the extent of the changes is indicated by the corresponding score maps (Figure 7B). One has to bear in mind that the shape of a loading stays constant over the measurement series, i.e., any particular PC is related to a particular (16) Tamura, T.; Yoshida, S.; Miyamoto, Y.; Kawauchi, S.; Satoh, M.; Komiyama, J. Polym. Int. 2000, 49, 147-152.
Figure 7. First four PCs of the FT-IR-ATR spectroscopic images in the spectral range between 1300 and 1800 cm-1: (A) loading plots and (B) score maps. The area between the dashed lines in first score image of the fourth PC will be used for subsequent analysis.
(constant) chemical composition. The amount of this particular entity (pure or mixed composition) may change over the measurement series. This is revealed by the intensity of the pixels in the score maps. Brighter pixels display a stronger participation at a particular position. Thus the first PC describes the behavior of a constant composition of H2O/-COOH/-COO-. The increasing contributions of the first PC over time (raising brightness of the pixels in the score maps) can merely be caused by physical changes s the disappearing water film between the ATR crystal and the hydrogel layer during swelling. Initially the water film occupied the region of highest strengths of the evanescent field. In the score map of the second PC the number of bright pixels increases over time. Brighter pixels indicate formation of nondissociated carboxylic acid. As the water is squeezed out, the hydrogel extends toward higher field strengths and gains intensity in its ATR spectra. This finding is supported by the behavior of the third PC. As described above, the third PC merely indicates water. The pixels in the score maps of the third PC get brighter
over time, but the loading plot has a negative sign. The third PC describes the disappearance of the water film during swelling. In case of the fourth PC, bright pixels are already observed at the very beginning of the experiment (Figure 7B, 0 min). As the NaOH solution is added, carboxylic acid is dissociated into carboxylate within a few seconds. Above pH ) 6, the carboxylic acid groups become fully ionized.17 In order to assess the development of the score values of the fourth PC for the course of the swelling process, we averaged the score values of all pixels in the PAA/PVA hydrogel area (indicated by the dashed lines in Figure 7B, fourth PC, 0 min). Figure 8 shows the averaged score values of the fourth PC of the PAA/PVA hydrogel area versus time. The trace shows three regions of significantly lower scores (corresponding to carboxylate absorbance), located around 20, 90, and 250 min. These minima appear much more pronounced than the minima in Figure 6, (17) Elliott, J. E.; Macdonald, M.; Nie, J.; Bowmann, C. N. Polymer 2004, 45, 1503-1510.
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
2961
Figure 8. Averaged score values of the fourth PC.
which showed the integral absorbance for the complete sample instead of the behavior of a single molecular entity. The more general parameter in Figure 6 constitutes the overlay of several processes, e.g., dissociation and pressure changes. These overlaying effects could well be separated by PCA; hence, the minima in Figure 8 are more distinct than those in Figure 6 and slightly shifted. The minima indicate time intervals of reduced concentration of carboxylatesin other words time periods of reduced dissociation of carboxylic acid. Every carboxylate group carries one negative charge; at time periods of reduced dissociation the net amount of charges in the hydrogel framework drops. This in turn results in a reduced driving force for swelling. The spectroscopic results clearly reveal the molecular background of the observed spikes in the sensor behavior of the PAA/PVA hydrogel layer (Figure 3). These findings are essential for any dedicated optimization of this kind of sensor. CONCLUSIONS pH-sensitive sensors based on swelling of a thin layer of PAA/ PVA hydrogel exhibit discontinuities in their response. The
2962
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
swelling of the PAA/PVA hydrogel is initiated by an increase in pH upon admission of NaOH, whereas during the swelling process the pH remains constant. The molecular background for the discontinuities was investigated by in situ FT-IR-ATR spectroscopic imaging, which yielded laterally and time-resolved molecular spectra for the hydrogel samples. Principal component analysis of the spectroscopic images revealed the different behavior of the various structural groups, and carboxylate groups turned out to be of key importance. The score value of the fourth PCs corresponding to the absorbance of the carboxylate groupss exhibits significant deviations from its mean value during the swelling process. Significantly lower score values, i.e., lower carboxylate concentrations, were observed at 20, 90, and 250 min. The results correspond to the sensor output signal and explain the observed discontinuous sensor response. Similar changes were also found for the integral absorbance of the FT-IR spectra in the carboxyl region. These changes indicate molecular processes as well, but they are related to the overall behavior of the polymer rather than to the behavior of a particular structural group. Effects in the FT-IR spectra due to overall behavior are less pronounced than the structure-group-specific results obtained by PCA. The structure-group-specific results relate the swelling process to changes in the molecular structure of the hydrogel. ACKNOWLEDGMENT The authors owe thanks to the Deutsche Forschungsgemeinschaft (Collaborative Research Center (SFB) 287, Project C11) for funding this work.
Received for review November 22, 2007. Accepted January 18, 2008. AC702598N
