Discriminative Detection of Volatile Sulfur Compound Mixtures with a

We demonstrated the discrimination of volatile sulfur compound mixtures with different mixing ratios by using an array of the plasma-polymerized film ...
0 downloads 0 Views 319KB Size
Anal. Chem. 2005, 77, 4228-4234

Discriminative Detection of Volatile Sulfur Compound Mixtures with a Plasma-Polymerized Film-Based Sensor Array Installed in a Humidity-Control System Michiko Seyama,*,† Yuzuru Iwasaki,† Shigeki Ogawa,‡ Iwao Sugimoto,§ Akiyuki Tate,† and Osamu Niwa†,|

NTT Microsystem Integration Laboratories, NTT Corporation, 3-1 Morinosato-wakamiya, Atsugi, Kanagawa 243-0198 Japan, NTT Advanced Technology, 3-35-1 Shimorenjyaku, Mitaka, Tokyo 181-0013 Japan, and School of Bionics, Tokyo University of Technology, 1404-1 Katakura, Hachioji, Tokyo 192-0982 Japan

We demonstrated the discrimination of volatile sulfur compound mixtures with different mixing ratios by using an array of the plasma-polymerized film (PPF)-coated quartz crystal resonators. The PPF sensor array, which contains PPFs prepared from amino acids and synthetic polymers, exhibited different response patterns to mono or mixed volatile sulfur compounds (VSCs) (hydrogen sulfide and methanethiol) under a dry environment. The sensor array was installed in a desktop-size relative humidity controller. The relative humidity and temperature conditions of the sample flow to the sensor cell were equalized to those of the inner atmosphere of the sensor cell based on the concept of the two-separate-temperatures method. In this way, the baseline drift of PPF sensor response caused by the introduction of a highly humid sample was successfully suppressed. We compared the sensor array responses under the controlled humidity conditions. Presorption of water molecules by PPFs caused a decrease of sensor sensitivity, but the films still had the ability to discriminate sub-ppmv VSC mixtures having 6:1, 1:1, and 1:6 mixture ratios of hydrogen sulfide and methanethiol. Gas sensor arrays represent a versatile integrated technology that is a promising alternative to costly time-consuming analytical methods for online monitoring and alarm systems1,2 in many fields, such as food and beverage manufacturing,3,4 environmental assessment,5 and personal health care.6 Though the developed data processing procedures, which are known as chemometrics, * To whon correspondence should be addressed. E-mail: [email protected]. Tel: +81 46 240 3046. Fax: +81 46 240 4728. † NTT Corp. ‡ NTT Advanced Technology. § Tokyo University of Technology. | Present address: National Institute of Advanced Industrial Science and Technology, Tsukuba, Ibaraki, 305-8566 Japan. (1) Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325-332. (2) Go ¨pel, W. Sens. Actuators, B 1998, 52, 125-142. (3) Hivert, B.; Hoummady, M.; Mielle, P.; Mauvais, G.; Henrioud, J. M.; Hauden, D. Sens. Actuators, B 1995, 27, 242-245. (4) Shurmer, H. V.; Gardner, J. W.; Chan, H. T. Sens. Actuators 1989, 18, 361371.

4228 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

have contributed much to improving the discrimination ability of an array,7-9 the molecular selectivity of each sensor is still a most important factor that determines the gas discrimination ability. We have investigated plasma-polymerized film (PPF) prepared by radio frequency (rf) sputtering of a solid organic target as a sorptive material for gas sensor arrays. With the plasma (sputtering) process, films with a molecularly dense structure that possesses a large adsorption area can be fabricated from various organic solid materials.10-15 The chemical structure of the sputtered film is related to the starting organic solid material. We have already prepared a sensor array comprising quartz crystal resonators (QCRs) coated with PPFs (PPF sensors) that can discriminate mono10-12 or mixed volatile organic compounds (VOCs) in aromas13 or indoor air,14 including even a chiral odorous compound,15 and can also detect several-ppbv-level VOCs.12 Gas sensor arrays can discriminate an anomalous condition of the gaseous mixture samples from a normal one without qualitative or quantitative determination of each chemical in the mixture samples, which has been very useful for alarm systems in many applications. Recently, there is much interest in developing sensor array technology that can obtain more specific information about each component in the gaseous mixture (5) Grate, J. W.; Rose-Phersson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H. Anal. Chem. 1993, 65, 1868-1881. (6) Natale, C. D.; Macagnano, A.; Martinelli, E.; Paolesse, R.; D’Arcangelo, G.; Roscioni, C.; Finazzi-Agro`, A.; D’Amico, A. Biosens. Bioelectron. 2003, 18, 1209-1218. (7) Grate, J. W.; Wise, B. M.; Abraham, M. H. Anal. Chem. 1999, 71, 45444553. (8) Abraham, M. H.; Andonian-Haftvan, J.; Whitig, G. S.; Leo, A.; Taft, R. S. J. Chem. Soc., Perkin Trans. 1994, 2, 1777-1791. (9) Grate, J. W.; Wise, B. M. Anal. Chem. 2001, 73, 2239-2244. (10) Sugimoto, I.; Nakamrua, M.; Kuwano, H. Sens. Actuators, B 1996, 37, 163168. (11) Sugimoto, I. Analyst 1998, 123, 1849-1854. (12) Sugimoto, I.; Seyama, M.; Nakamura, M. J. Environ. Monit. 1999, 1, 135142. (13) Kasai, N.; Sugimoto, I.; Nakamura, M.; Katoh, T. Biosens. Bioelectron. 1999, 14, 533-539. (14) Seyama, M.; Sugimoto, I.; Nakamura, M. Biosens. Bioelectron. 2004, 20, 814-824. (15) Sugimoto, I.; Nakamura, M.; Seyama, M.; Ogawa, S.; Katoh, T., Analyst 2000, 125, 169-174. 10.1021/ac0484833 CCC: $30.25

© 2005 American Chemical Society Published on Web 06/01/2005

Figure 1. Concept of the relative humidity control based on the two-separate-temperatures method. T1 and T2 are temperature and es(T1) and es(T2) are the saturated vapor pressure at T1 and T2 over liquid water (T1 < T2). U (%) represents the RH of air.

because there is a wide range of potential applications.16-18 In the field of environmental monitoring, regulations are gradually becoming stricter and ways of controlling the concentration level of specific chemicals is required.16 For precise control of the fermentation process, the ability to monitor varying specific components in a generated gaseous mixture is desired. In the field of personal health care, determination of mixing ratios of specific volatile chemicals known as indicators of diseases (lung cancer,6 breast cancer,17 periodontitis,18,19 etc.) emitted with biogases including the breath are expected as a fast-screening method. Not only the discrimination ability of volatile chemicals but also the effect from water vapor or humidity should be considered since humidity is a serious noise source for sorptive polymer-based sensor arrays when monitoring real samples. Therefore, in this work, we investigated the ability of our PPF sensor array to discriminate the mixture ratio of volatile compounds under dry and humid environments. We prepared model mixture samples by referring to the existing real samples of halitosis volatile sulfur compounds (VSCs). VSCs from human breath is a promising new indicator of periodontitis.18,19 Periodontitis is a major cause of tooth loss and sometimes indicative of a weakening of the immune system due to other diseases, such as diabetes and cardiovascular problems.20,21 From a healthy human breath, a common halitosis VSC, hydrogen sulfide, can be detected at sub-ppm levels. When periodontitis progresses with the formation of a so-called biofilm consisting of aggregated bacteria, methanethiol increases in the breath. According to recent clinical research, a person whose breath includes hydrogen sulfide and methanethiol with a ratio of 6:1, 1:1, or 1:6 suffers light, moderate, or serious periodontitis, respectively.21 And VSCs are volatile, adsorptive, and thermally unstable compounds,22 so that the procedure for on-site monitoring is required. In this study, we introduced the newly developed relative humidity (RH) control system based on the two-separate-temperatures method (illustrated in Figure 1) that allows us to control the RH and the temperature of analytes without heating to high (16) Penza, M.; Cassano, G. Sens. Actuators, B 2003, 89, 269-284. (17) Phillips, M.; Cataneo, R. N.; Ditkoff, B. A.; Fisher P.; Breenberg, J.; Gunawardena, R.; Kwon, C. S.; Rahbari-Oskoui, F.; Wong, C. Breast J. 2003, 8, 184-191. (18) Beck, J.; Garcia, R.; Heiss, G.; Vokonas, P. S.; Offenbacher, S. J. Periodontol. 1996, 67, 1123-1137. (19) Lee, C. H.; Chung, S. C.; Lee, S. W.; Kim, Y. K. J. Periodontol. 2003, 74, 32-37. (20) Oho, T.; Yoshida, Y.; Shimazaki Y.; Yamashita Y.; Koga, T. Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endod. 2001, 91, 531-534. (21) Soskolne, W. A.; Klinger, A. Ann. Periodontol. 2001, 6, 91-98. (22) Ochiai, N.; Takino, M.; Daishima, S.; Cardin, D. B. J. Chromatogr., B 2001, 762, 67-75.

temperature to eliminate the effect from high humidity. The main components of the system were two containers for a water reservoir and a sensor cell with Peltier devices, an electrical circuit board for temperature control, and a pump, and the parts were integrated in a desktop-size box (27 L in volume). We prepared the sensor array comprising PPFs prepared from different amino acids (phenylalanine and histidine) by referring to the good affinity of hydrogen sulfide and methanethiol with amines that is utilized in the industrial sulfur removal process.23 We also put a PPF prepared from lipophilic polyethylene in the array because we expected the PPF to have higher affinity with methanethiol than hydrogen sulfide.24 We prepared the model mixture sample of hydrogen sulfide and methanethiol mixture with a ratio of 6:1, 1:1, or 1:6 at sub-ppm level, which is a concentration level similar to human breath. The basic detection of and discrimination ability to VSC mixtures containing sub-ppm-level hydrogen sulfide and methanethiol was first investigated under a dry environment. And then, using the RH control system, we investigated the VSC discrimination ability of the PPF sensor array under the regulated humidity condition. EXPERIMENTAL SECTION PPF Sensor Preparation and Sensor Responses. PPFs were prepared by rf-sputtering using a vacuum instrument with a parallel-electrode configuration. The sputtering targets were polyethylene (PE), poly(chlorotrifluoroethylene) (PCTFE), Dphenylalanine (Phe), D-histidine (His), and QCRs with Au electrodes whose fundamental resonant frequency was 9 MHz were mounted on the substrate holder. During sputtering, He or Kr gas was introduced to the vacuum chamber and power of ∼100 W was applied. PPFs were deposited to a thickness of 500 nm on both sides of the QCR. PPF thickness was monitored by the resonant frequency change of the QCR. The resonant frequency of the PPF-coated QCR sensors is correlated with the increase in mass of the PPFs by adsorption as indicated by Sauerbrey’s equation.25 The resonant frequency change (∆f) of QCR is related to the mass increase (∆m) as

∆f ) (-) 2fo2∆mA-1(Fqµq)-1/2

(1)

where fo is the fundamental resonance frequency, A the electrode area, Fq the density of quartz (2.65 g cm-3), and µq the shear modulus of quartz (2.95 × 1011 dyn cm-2). We used an 8-mm-i.d. QCR with a 0.13-cm2 Au electrode whose fundamental resonance frequency was 9 MHz. The correlation between the resonant frequency change ∆f and the mass load therefore becomes

∆f (Hz) ) (-)1.05∆m (ng)

(2)

The negative sign in the above equations indicates that the resonant frequency of the QCR decreases as the mass load of the QCR increases. We denote the resonant frequency shift caused by the increase of mass loading due to gas sorption simply as ∆f in this paper. (23) Jou, F.-Y.; Mather, A. E. J. Chem. Eng. Data 2000, 45, 1096-1099. (24) Toda, K.; Ohira, S. I.; Tanaka, T.; Nishimura, T. Environ. Sci. Technol. 2004, 38, 1529-1536. (25) Sauerbrey, G. Z. Z. Phys. 1959, 155, 206-222.

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4229

Reflective Fourier Transform Infrared (FT-IR) Spectroscopy. Reflective FT-IR measurement was performed with a spectrometer with a microscopic attachment equipped with a mercury-cadmium-tellurium detector that was cooled with liquid nitrogen (Multiscope and System2000, PerkinElmer, Ltd.). We measured the PPFs deposited on a QCR with Au electrodes and used a reference spectrum obtained with a bare QCR with Au electrodes. The measurement was performed with the resolution of 2 cm-1. RH Control System. The RH control concept is based on the two-separate-temperatures method (Figure 1). When the temperature of a half-filled water reservoir is T1, the RH of air that goes through the water vapor-saturated overhead space and enters the connected room (T2) is determined by

U (%) ) es(T1)/es(T2) × 100

4230 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

under dry condition Ymax (mS) under humid condition ∆f (Hz)

Phe

PE

His

81.9 75.4 6360

88.9 83.8 3062.5

117.1 117.5 5832.5

(3)

where es(T1) and es(T2) are the saturated vapor pressure at T1 and T2 (°C) over liquid water (T1 < T2). We prepared two stainless steel containers whose temperature was controlled by Peltier devices using feedback from platinum thermopiles embedded in each container. A water reservoir and flow-through aluminum sensor cell that houses the PPF sensor array were installed in each container. All sample flows were connected by Teflon tubes, and a flow-control pump system (MFC100, DKK-TOA Co.) was installed in the system. The PPF sensors were connected to a circuit board comprising an oscillation circuit, a resonant frequency measurement circuit, and the interface for data transfer. The circuit board was also installed in the system. The board can measure the resonant frequency with the precision of 0.1 Hz, and the frequency change data were stored in a personal computer.12 In this work, the RH of the sensor cell was controlled by setting T1 and T2 to 5 and 20 °C, respectively. This system can control the temperature with a precision of (0.1 °C; therefore, theoretically, the RH can be controlled to the 0.1% level. The es at a certain temperature is determined by the Goff-Gratch equation26 by applying the coefficients from the international temperature scale of 1990. The es(5) and es(20) were calculated as 871.96 and 2338.7 Pa, respectively, and the RH of the inner atmosphere of the sensor cell was 37.2% RH at 20 °C. We monitored this condition using a RH and temperature sensor (Hygromer115C, Rotronics) placed near the outlet of the cylindrical sensor cell. No hydrogen sulfide and methanethiol were lost by passing through our RH controller, which was confirmed by gas chromatographic measurement with the human breath samples. Both samples, which were collected in the Tedlar bags directly from a human panel (not periodontitis patients) or through our RH controller, contained 90 ppbv hydrogen sulfide and several-ppbvlevel methanethiol (smaller than the quantitative detection limit of 2 ppbv). This gas chromatographic measurement was performed according to the procedures outlined in Table 2, Article 5 of the Offensive Odor Control Law of the Japanese Ministry of the Environment, by using a gas chromatograph equipped with a liquid oxygen-cooled preconcentrator and a flame photometric detector (GC14A/FLS-1, Shimadzu Co.). (26) Plate, E. J. Engineering Meteorology; Elsevier: New York, 1982.

Table 1. Admittance and Resonant Frequency Changes of PPF Sensors under Dry and Humid (RH-Controlled) Conditions

Figure 2. (a) Design of the RH control system and (b) measurement setup for VSC samples with the novel humidity control system.

VSC Gas-Sorption Measurement. The VSC source was a 10-L cylinder that contained 5 ppmv hydrogen sulfide or methanethiol with nitrogen-based gas. The cylinders were purchased from Takachiho Trading, Co. Ltd.. We used two mass-flow meters (MFMs) with an accuracy of 1 mL min-1 in order to adjust the concentrations of VSC sample gases. MFM1 (calibrated with nitrogen) was set at the position just after the VSC flow from the canister was mixed with nitrogen, and MFM2 (calibrated with air) was set just in front of the sensor cell as shown in Figure 2. The VSC gases were mixed with nitrogen gas by using the mass flow controllers with an accuracy of 0.1 mL min-1, and the total amount of nitrogen-based VSC gas flow was ∼40 mL min-1 (measured with the MFM1). An air flow of ∼160 mL min-1 was mixed with the nitrogen-based VSC gas flow through the MFM1. With this system, the VSC source in the canisters can be diluted by 2000, and considering the error of the mass flow controller (1% of maximum flow rate), the error in concentration was (2.5 ppbv. To confirm the baseline level of the PPF sensors before the gas sorption test, the reference air, a mixture of only nitrogen and air, was introduced to the sensor cell. At that time, VSC sample flow was set to zero. When the flow from VSC gas sources was increased to obtain the VSC samples at the desired concentration, the amount of nitrogen flow was decreased by the same amount to make the total sample gas flow to the sensor cell equal throughout the gas sorption test. The total flow rate of a gas sample to the sensor cell was regulated as 200 mL min-1. The mixture ratio of VSCs was defined as the ratio of the common divider of the concentration of hydrogen sulfide gas (CH2S) and methanethiol gas (CCH3SH). The concentration of a mixture sample is the sum of CH2S and CCH3SH in ppbv. After the gas sorption test, we cleared the sensor cell and all lines using the reference gas. The flow of mass-flow-controlled gas sample to the RH control system was monitored with a flow meter placed at the outlet of our system to confirm that there was no counter flow. RESULTS AND DISCUSSION RH Control at Introduction of Humid Air. A humid air sample (97% RH at 24 °C) that had a RH level similar to that of

Figure 3. Sensor signals to humid air samples with (dotted lines) or without (solid lines) humidity control with the PPF sensor array (Phe sensor, PE sensor, and PCTFE sensor). Room air was applied to the sensor cell with a pump during the first 5 min in order to confirm the background level.

human breath was prepared by moisturizing air in a sealed glass container. To evaluate the effect of RH control at the introduction of highly humid air, we compared the PPF sensor responses with activation of the RH controller (indicated as “with control”) and without activation (indicated as “without control”) in which the temperature of the second container (T2) was kept as 20 °C but the water reservoir was empty without temperature control (Figure 3). Without RH control, a sudden change in the sample RH (at 5 min in Figure 3) resulted in large resonant frequency shifts of the PE and Phe sensors, which have a high affinity to many VOCs.12 Such baseline drift of those sensors with maximum shifts around 70-80 Hz decreased after several minutes, but the decreasing rate was slow because of the high affinity of the Phe and PE films to water vapor. As a reference, we tested a PCTFE sensor (QCR coated with PPF prepared from fluoropolymer, PCTFE), which has very low affinity with water vapor and with VOCs as well. The PCTFE sensor showed small shifts in the resonant frequency (∼3 Hz) even without RH control. When the RH control system was activated, large baseline drifts of both Phe and PE sensors were successfully suppressed with deviation less than 1 Hz as shown by solid lines in Figure 3. VSC Discrimination under Dry Condition. We examined the basic detection and discrimination ability of the PPF sensor array for VSCs under a dry condition prepared by keeping the water reservoir in the humidity control system empty without temperature control and regulating the temperature of the second container (T2) to 20 °C. We used synthetic gases as the sample source and diluting gas to keep the inner atmosphere of the sensor cell dry. A RH sensor placed at the outlet of the sensor cell indicated RH lower than 1%. Based on the data from our chromatographic survey of VSC concentrations from a nonpatient’s breath (in the Experimental Section), we prepared the VSC gas of ∼90 ppbv. Phe, PE, and His sensors had enough sensitivity to detect sub-ppmv-level hydrogen sulfide and methanethiol as shown in Figure 4. The ∆f

Figure 4. Response curves of the Phe sensor, PE sensor, and His sensor under the dry condition. Comparison of ∆f of the three PPF sensors to 84 ppbv hydrogen sulfide and 84 ppbv methanethiol.

of the three PPF sensors began to increase when VSC gas was introduced to the sensor cell (at 10 min in Figure 4). The PCTFE sensor showed no response to sub-ppmv-level hydrogen sulfide and methanethiol. It is known that hydrogen sulfide has an affinity to amine groups23 that may exist in the amino acid-based PPF.27 As we expected, the ∆f of the amino acid-based Phe and His sensors was larger than that of the PE sensor during 84 ppbv hydrogen sulfide sorption, and the ∆f of the PE sensor was larger than that of the Phe sensor for a 84 ppbv methanethiol gas sample. However, the His film, which was expected to have more free amino acid structures than the Phe film (Supporting Information) had lower affinity to both hydrogen sulfide and methanethiol than the Phe film. After 60 min, we changed the gas flow from the VSC gas to the reference air (200 mL min-1). But the ∆f of the three PPF sensors did not immediately decrease. Instead, the ∆f stayed at the same level before the flow change or increased for a while between 30 min and 2 h. We consider that this was because VSCs still remained in the measurement line. VSC molecules, which had adsorbed on the walls of our measurement line and sensor cell, began to be released into the gas flow because VSC concentration decreased after the sample flow was changed to the reference air. The ∆f of the PPF sensors began to decrease due to desorption of VSCs from PPFs after an interval whose length depended on the measured VSC concentration. Prior to subsequent testing, we confirmed that the measurement line was completely clean and the VSC totally desorbed by checking whether the resonant frequency of PPF sensors recovered to the level it was before the VSC sorption test. Next, we compared the PPF sensor responses to 84-ppbv VSC samples that included VSCs with different mixture ratios. Figure 5 shows the ∆f of Phe, PE, and His sensors after 60-min sorption, described as ∆f60(Phe), ∆f60(PE), and ∆f60(His), respectively, to (27) Sugimoto, I.; Nakamura, M.; Kuwano, H. Anal. Chem. 1994, 66, 43164323.

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4231

Figure 5. Comparison of ∆f’s of the Phe, PE, and His sensors after 60-min gas sorption test under dry condition. Total concentrations of all test gases were 84 ppbv. The mixture ratio of hydrogen sulfide and methanethiol is indicated at the bottom of the graph. The ratios 1:0 and 0:1 indicate only hydrogen sulfide and methanethiol gas, respectively.

the VSC mixture samples whose total VSC concentrations were equally 84 ppbv. We prepared VSC mixture samples that included hydrogen sulfide and methanethiol at the mixture ratio of 6:1, 3:1, 1:1, 1:3, or 1:6. Responses to 84 ppbv hydrogen sulfide or methanethiol were labeled 1:0 or 0:1 and are also shown in Figure 5. There was a relation between the patterns of Phe sensor and PE sensor responses and the concentration ratios of the VSC samples. When there was more hydrogen sulfide in the VSC sample than methanethiol (1:0, 6:1, 3:1), the ∆f60(Phe) was larger than ∆f60(PE). For the binary sample with the same concentration of hydrogen sulfide and methanethiol, ∆f60(Phe) and ∆f60(PE) were similar. On the other hand, when there was more methanethiol than hydrogen sulfide in the VSC sample (1:3, 1:6, 0:1), ∆f60(Phe) was smaller than ∆f60(PE). The intensities of PPF sensor responses to mixture samples are not a simple sum of sensor response to each corresponding monogas concentration in the mixture. A small concentration of methanethiol with hydrogen sulfide (6:1) seemed to enhance VSC sorption by the Phe and PE sensors. But in other cases, the ∆f60(Phe) to sample gases containing methanethiol was smaller than that to only hydrogen sulfide gas (labeled as 1:0). The difference in intensities of PE sensor responses to the mixture samples containing larger concentrations of methanethiol (1:3, 1:6, 0:1) was within 1 Hz, so that methanethiol sorption by PE film was not affected by the coexistence of the smaller concentration of hydrogen sulfide as much as the Phe and His films. The ∆f values for VSC mixture samples with the same mixture ratio increased with increasing total concentration (200 ppbv); however, the concentration range for good discrimination was regulated. Therefore, the response pattern including the relative intensities between the PPF sensors in the array should be considered to fabricate the sensor array for discrimination of mixing ratio of the gas mixture. Average VSC concentration from breath samples of periodontitis patients is nearly the twice that of the healthy people,20 so we consider that the PPF array sensor presented here still has the basic ability for screening of periodontitis from a highly humid sample.

(30) Sugimoto, I.; Nakamura, M.; Kasai, N.; Katoh, T. Polymer 2000, 41, 511522 (31) Pu ¨ttmer, A.; Hauptman, P.; Lucklum, R.; Krause, O.; Henning, B. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 1997, 44, 60-66.

CONCLUSION A plasma-polymerized organic film-coated quartz crystal resonator sensor (PPF sensor) array was applied to detect a VSC

Figure 8. PCA score plots based on data collected under the humidity control condition. The mixture ratios of hydrogen sulfide and methanethiol are 6:1 (circles), 1:1 (rectangles), and 1:6 (lozenges). Numbers indicate total concentration of mixture gas in ppbv.

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4233

mixture, which may be an indicator for the progress of periodontitis when detected from human breath. Although the chemical structure of hydrogen sulfide and methanethiol is similar, the sensor array composed of sensory films prepared from polyethylene, D-phenylalanine, and D-histidine had a basic ability to discriminate the mixtures with the mixing ratios of 6:1, 1:1, and 1:6 under a dry environment. Baseline drifts of the PPF sensor array caused by changes of sample RH were successfully suppressed by using a newly developed desktop-size RH controller based on the two-separatetemperatures method. This RH controller enables us to measure the real samples around room temperature, so it is applicable for detection of thermally unstable molecules. The sensing ability of PPF sensors under a dry environment differed from that under the humid environment prepared with the RH controller because of presorption of water molecules by PPFs. However, our PPF sensor array installed in the proposed RH controller has the ability to discriminate halitosis VSCs at the sub-ppmv level. Therefore, the RH-controlled sensor array is applicable for training with data

4234

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

sets based on chemometrics to develop a system tuned for special targets where the sample has a high or unstable humidity like human breath. ACKNOWLEDGMENT The authors thank Kensuke Sasaki of NTT Afty for his technical assistant and Dr. Toshiaki Maeda of AIST for support in the development of the humidity control system. They also thank Isao Shinozuka of NTT Advanced Technology for his help with chromatographic measurements. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review October 14, 2004. Accepted March 30, 2005. AC0484833