Nanoassembled Thin Film Gas Sensors. III ... - ACS Publications

Feb 19, 2010 - Graduate School of Environmental Engineering, The University of Kitakyushu, 1-1 Hibikino, Kitakyushu 808-0135,. Japan, and Graduate ...
0 downloads 0 Views 3MB Size
Download PDF
Anal. Chem. 2010, 82, 2228–2236

Nanoassembled Thin Film Gas Sensors. III. Sensitive Detection of Amine Odors Using TiO2/ Poly(acrylic acid) Ultrathin Film Quartz Crystal Microbalance Sensors Seung-Woo Lee,*,† Naoki Takahara,† Sergiy Korposh,† Do-Hyeon Yang,† Kiyoshi Toko,‡ and Toyoki Kunitake† Graduate School of Environmental Engineering, The University of Kitakyushu, 1-1 Hibikino, Kitakyushu 808-0135, Japan, and Graduate School of Information Science and Electrical Engineering, Kyushu University, 6-10-1 Hakozaki, Fukuoka 812-8581, Japan Quartz crystal microbalance (QCM) gas sensors based on the alternate adsorption of TiO2 and polyacrilic acid (PAA) were developed for the sensitive detection of amine odors. Individual TiO2 gel layers could be regularly assembled with a thickness of ∼0.3 nm by the gas-phase surface sol-gel process (GSSG). The thickness of the poly(acrylic acid) (PAA) layer is dependent on its molecular weight, showing different thicknesses of ∼0.4 nm for PAA25 (Mw 250 000) and 0.6-0.8 nm for PAA400 (Mw 4 000 000). The QCM sensors showed a linear response to ammonia in the concentration range 0.3-15 ppm, depending on the deposition cycle of the alternate TiO2/PAA layer. The ammonia binding is based on the acid-base interaction to the free carboxylic acid groups of PAA and the limit of detection (LOD) of the 20-cycle TiO2/PAA400 film was estimated to be 0.1 ppm when exposed to ammonia. The sensor response was very fast and stable in a wide relative humidity (rH) range of 30-70%, showing almost the same frequency changes at a given concentration of ammonia. Sensitivity to n-butylamine and ammonia was higher than to pyridine, which is owing to the difference of molecular weight and basicity of the amine analytes. The alternate TiO2/PAA400 films have a highly effective ability to capture amine odors, and the ambient ammonia concentration of 15 ppm could be condensed up to ∼20 000 ppm inside the films. Detection of odorous amine compounds is a target of primer importance in many areas of human activities.1,2 Amine compounds, especially, have been used as an indicator of the food * Corresponding author. Phone: +81 93 695 3293. Fax: +81 93 695 3384. E-mail: [email protected]. † The University of Kitakyushu. ‡ Kyushu University. (1) Timmer, B.; Olthuis, W.; Berg, A. Sens. Actuators, B 2005, 107, 666–677. (2) Pacquit, A.; Tong Lau, K.; McLaughlin, H.; Frisby, J.; Quilty, B.; Diamond, D. Talanta 2006, 69, 515–520.

2228

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

quality3 and are also a challenging target of environment pollutant detection.4 Among others, ammonia could be regarded as a “biomarker” for the diagnosis of several diseases and therefore the device for the ammonia measurement with the detection limit of 50-2000 ppb and fast response time are highly desired.1 Recently, organic polymers were deposited on different transducers as the sensitive element for the ammonia detection,5 such as potentiometric,6 optical,7 and mass sensitive ones.8 Among these techniques, quartz crystal microbalance (QCM) is extremely sensitive and a powerful tool for monitoring mass changes in the nanogram range.9,10 The basic principle of QCM sensors is the measurement of the frequency shift as a result of the mass adsorbed on the QCM resonator. How to modify the resonator surface properties becomes an essential issue for enhancing the range of applications of the QCM sensor, and deposition of thin films on the QCM resonator surface allows sensitive and selective determination of particular chemical analytes to be achieved.11-13 The layer-by-layer (LbL) process is a versatile technique for the deposition of multilayered nanothin films onto different substrates.14,15 Recently, we have employed the LbL process for the preparation of multilayered organic films as sensitive (3) Olafsdottir, G.; Nesvadba, P.; Di Natale, C.; Careche, M.; Oehlenschlager, J.; Tryggvadottir, S. V.; Schubring, R.; Kroeger, M.; Heia, K.; Esaiassen, M.; Macagnano, A.; Jørgensen, B. M. Trends Food Sci. Technol. 2004, 15, 86–93. (4) WHO. Air Quality Guidelines for Europe, No. 23, European Series, WHO Regional Publications: Copenhagen, Denmark, 1987. (5) Harsanyi, G. Mater. Chem. Phys. 1996, 43, 199–203. (6) Meyerhoff, M. E. Anal. Chem. 1980, 52, 1532–1534. (7) Korposh, S. O.; Takahara, N.; Ramsden, J. J.; Lee, S.-W.; Kunitake, T. J. Biol. Phys. Chem. 2006, 6, 125–133. (8) Ding, B.; Yamazakia, M.; Shiratori, S. Sens. Actuators, B 2005, 106, 477– 483. (9) Lu, C.; Czanderna, A. W. Applications of Piezoelectric Quartz Crystal Microbalances; Elsevier: Amsterdam, The Netherlands, 1984. (10) Kanazawa, K.; Cho, N.-J. J. Sens. 2009, article ID 824947. (11) Paolesse, R.; Di Natale, C.; Macagnano, A.; Davide, F.; Boschi, T.; D’Amico, A. Sens. Actuators, B 1998, 47, 70–76. (12) Dickert, F. L.; Balumler, U. P. A.; Stathopulos, H. Anal. Chem. 1997, 69, 1000–1005. (13) Ji, Q.; Yoon, S. B.; Hill, J. P.; Vinu, A.; Yu, J.-S.; Ariga, K. J. Am. Chem. Soc. 2009, 131, 4220–4221. (14) Ariga, K.; Hill, J. P.; Ji, Q. Phys. Chem. Chem. Phys. 2007, 9, 2319–2340. (15) Ariga, K.; Hill, J. P.; Lee, M. V.; Vinu, A.; Charvet, R.; Acharya, S. Sci. Technol. Adv. Mater. 2008, 9, 1–96. 10.1021/ac901813q  2010 American Chemical Society Published on Web 02/19/2010

elements on glass substrate7 and optical fiber16 for ammonia gas detection. Poly(acrylic acid) (PAA) is a promising candidate for the creation of ammonia sensors due to the presence of free carboxylic functional groups, which leads to the high sensitivity and selectivity toward amine compounds and allows the use of PAA with different transducers.8,17,18 A main drawback of this material, when used with QCM transducers, is the undesired sensitivity to humidity and long recovery time at higher (>1 ppm) ammonia concentrations.19 Incorporation of PAA into inorganic matrixes may solve this problem because the nanocomposite film may retain unique properties, i.e., functionality and flexibility, of the organic compound while the inorganic part provides robustness and stability.20,21 Inorganic-organic hybrid materials have a high potential for sensor applications, for instance, MoO3 hybrid films with conducting polymers improved the sensitivity and selectivity toward aldehyde compounds.22 Over the last several years, we have studied the surface sol-gel process as a means for the preparation of ultrathin metal oxide films.23 The individual sol-gel procedure of this process is independent of each other, and organic, polymeric, biological, and metallic materials are readily incorporated as second components and as unit layers, if they are reactive with the amorphous metal oxide layer.24 Very recently, we examined a novel surface sol-gel process, a gasphase surface sol-gel process. An important feature of this approach is that ultrathin metal oxide films of controlled thickness can be fabricated from metal alkoxides in the gas-phase.25,26 This unique process is used in this study for the preparation of TiO2/PAA composite films, which were deposited on QCM electrodes in order to examine their response toward odorous ammonia and amine compounds. The employment of the TiO2 matrix is advantageous to incorporate the organic PAA polymer owing to the strong complex formation between Ti and COOH moieties, which will give a higher stability of PAA inside the film. Additionally, it is hypothesized that the presence of the TiO2 matrix would suppress the mobility of the PAA, thus reducing the influence of water on the sensor response. Sensor parameters, such as selectivity, sensitivity, and stability, are investigated, and the effect of the thickness and chemical composition of the film on the sensor response is discussed. EXPERIMENTAL SECTION Materials. Titanium n-butoxide, Ti(O-nBu)4, was purchased from Kishida Chem. Poly(acrylic acid)s, PAA400 (Mw 4 000 000) (16) Korposh, S.; Kodaira, S.; Batty, W. J.; James, S. W.; Lee, S.-W. Sens. Mater. 2009, 21, 179–189 (no. 4). (17) Oprea, A.; Weimar, U. Sens. Actuators, B 2005, 111-112, 572–576. (18) Oprea, A.; Barsan, N.; Weimar, U. Sens. Actuators, B 2005, 111, 577–581. (19) Sahm, M.; Oprea, A.; Baˆrsan, N.; Weimar, U. Sens. Actuators, B 2007, 127 (1), 204–209. (20) Huang, J.; Ichinose, I.; Kunitake, T.; Nakao, A. Langmuir 2002, 18, 9048– 9053. (21) Huang, J.; Ichinose, I.; Kunitake, T.; Nakao, A. Nano Lett. 2002, 2, 669– 672. (22) Hosono, K.; Matsubara, I.; Murayama, N.; Woosuck, S.; Izu, N. Chem. Mater. 2005, 17, 349–354. (23) Lee, S.-W.; Ichinose, I.; Kunitake, T. Langmuir 1998, 14, 2857–2863. (24) Ichinose, I.; Kawakami, T.; Kunitake, T. Adv. Mater. 1998, 10, 535–539. (25) Yang, D.-H.; Takahara, N.; Mizutani, N.; Lee, S.-W.; Kunitake, T. Chem. Lett. 2006, 35, 990–991. (26) Takahara, N.; Yang, D.-H.; Ju, M.-J.; Hayashi, K.; Toko, K.; Lee, S.-W.; Kunitake, T. Chem. Lett. 2006, 35, 1340–1341.

Figure 1. Schematic illustration of the film preparation by the gasphase surface sol-gel process.

and PAA25 (Mw 250 000, 35 wt % in H2O), were purchased from Sigma-Aldrich. Ammonia (30 wt % in H2O), n-butylamine, pyridine, ethanol, toluene, and chloroform were used to produce the analyte gas and were purchased from Wako Pure Chem., Japan. All of these chemicals were reagents of analytical grade and used without further purification. Deionized pure water (18.3 MΩ cm) was obtained by reverse osmosis followed by ion exchange and filtration (Nanopure Diamond, Barnstead, Japan). Standard ammonia gas of 100 ppm in dry air was purchased in the cylinder from Japan Air Gases Corp. TiO2/PAA Alternate Films. The gas-phase surface sol-gel (GSSG) process was used for the deposition of the (TiO2/PAA)n (n ) 5, 10, and 20) films on a gold-coated quartz crystal microbalance electrode, as described in our previous reports.25,26 Prior to the film deposition, the gold-coated QCM electrode was modified with 2-mercaptoethanol. The PAA/TiO2 film was deposited on the hydroxyl terminated QCM electrode, as shown in Figure 1. For the deposition of TiO2 layers, the previously reported procedure was employed.25,27 Afterward, the QCM electrode coated with a TiO2 thin layer was immersed in aqueous solutions of 0.05 wt % of PAA400 or 0.1 wt % of PAA25 for 20 min at 25 °C. Consequently, alternate films of 5, 10, and 20 cycles (one cycle is a TiO2/PAA bilayer) were prepared by alternate adsorption of Ti(O-nBu)4 and PAA on the QCM electrode with intermediate water washing for the hydrolysis of the metal alkoxide precursor, rinsing with deionized water, and drying by flushing with N2 gas. Pure TiO2 Films. In order to study the response of pure TiO2 films to amine odors as reference, the gas-phase sol-gel process was used for the deposition of pure TiO2 films of 5, 10, and 20 cycles without adsorption of PAA. Dip-Coated TiO2 and TiO2/PAA Alternate Films. The dipcoating method was employed to study the effect of the film thickness on the sensor parameter, and thicker films of about 50-180 nm were fabricated. A mercaptoethanol-modified QCM electrode was immersed at 6.0 × 10-4 m/s into a solution of Ti(On Bu)4 in toluene/ethanol (1, 5, and 10 mM, v/v ) 1:1) and withdrawn at 1.6 × 10-4 m/s under the nitrogen atmosphere (27) See the Supporting Information.

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2229

Figure 2. Schematic diagram of the measurement setup.

in order to fabricate a thick TiO2 film. For the alternate deposition of TiO2 and PAA, the QCM electrode was immersed into a 0.5 wt % aqueous solution of PAA400 for 20 min at 25 °C, as described earlier in TiO2/PAA Alternate Films. QCM Measurements. Frequency change measurements of film assembly and analyte detection were conducted using quartz crystal microbalance (9 MHz) devices manufactured by USI System, Fukuoka, Japan. The mass increase due to adsorption can be estimated from the QCM frequency shift by using the Sauerbrey equation.23 In our system, a frequency decrease of 1 Hz corresponds to a mass increase of ∼0.9 ng. The thickness (d, in Angstroms) of an adsorbed film on one side of a resonator is given by 2d )

∆F 1.832F

(1)

where F is the film density (in grams per cubic centimeter) and ∆F is the frequency shift of the QCM (in hertz). Gas Preparation. Desired gas concentrations are produced using a two-arm flow system, as shown in Figure 2.16 The final analyte concentration (volume fraction) c in the measurement chamber was calculated using the following equation: c)

L1z L1(1 + z) + L2

(2)

where z is the mole fraction of the analyte in the headspace (i.e., the ratio of the partial pressure of the solution ps ) (W0 - W1)/V at a given temperature to the atmospheric pressure P, i.e., z = ps/P), and L1 and L2 are the flow rates of the air that passed and bypassed through the bottle, respectively. The baseline of each experiment was recorded by flowing dry air through the QCM chamber until the frequency reached equilibrium. Gas concentrations of the analytes (ammonia, nbutylamine, pyridine, ethanol, toluene, and chloroform) were adjusted from 0 to 100 ppm. In the case of ammonia gas, the gas concentration generated using this system was in good agreement with that of the standard ammonia gas in cylinder (see Figure S1 in the Supporting Information). SEM, AFM, and XPS Measurements. Atomic force microscopy (AFM) measurements were carried out using a scanning probe microscope JSPM-5200 at room temperature in the noncontact mode (silicon cantilevers NSC12/Ti-Pt/15, MicroMasch, Spain). Scanning electron microscopy (SEM) measurements were 2230

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

undertaken with the help of a Hitachi S-5200 at 10-15 kV acceleration voltages. A 2 nm thick platinum layer was deposited on all samples using a Hitachi E-1030 ion sputter at 15 mA and 10 Pa prior to measurements in order to prevent the charge up of the samples surface. The X-ray photoelectron spectroscopy (XPS) measurements were carried out on an ESCA/XPS (Shimadzu-Kratos, AXIS-HSi) using Al KR (1486.6 eV) radiation (at applied power of a 15 kV and 20 mA). The base pressure in the analysis chamber was less than 10-9 Pa. Smoothing, background removal, and peak fitting were carried out using the Vision 2 Processing System analysis software package. All the peaks were corrected with C(1s) at 285 eV as the reference. For sample preparation, the gold plate was polished with alumina polishing suspension of particle size 0.05 mm on a microcloth pad. To remove the trace alumina on the gold electrode surface, the gold plate was sonicated in deionized water for about 10 min, rinsed with deionized water, and dried by N2 gas. Prior to film deposition, the gold plate was treated with piranha solution, rinsed with deionized water, and dried by N2 gas. RESULTS AND DISCUSSION Deposition of PAA and TiO2 Layers. The frequency linearly decreases with the adsorption cycles due to the alternate adsorption of Ti(O-nBu)4 and PAA, indicating that each component of the TiO2/PAA film is regularly deposited in each sequential cycle (see Figure S2 in the Supporting Information). Average frequency changes of the (TiO2/PAA400)20 alternate film were 17 ± 8 and 40 ± 19 Hz for the TiO2 and PAA400 layers, respectively. In the current setup, a QCM frequency decrease of 1 Hz corresponds to a thickness increase of 0.273/F Å, where F is the density of the adsorbed film, and we can readily estimate the thicknesses of the PAA and TiO2 layers from their ∆F values. With the use of the density of PAA (1.4 g/cm3),28 the ∆F value of 40 ± 9 Hz corresponds to a thickness increase of 0.8 ± 0.2 nm for the PAA400 layer. Similarly, the thickness of the TiO2 layer was calculated to be 0.3 ± 0.1 nm from the ∆F value (17 ± 8 Hz) and the bulk density (1.7 g/cm3) of the TiO2 gel.29 Therefore, the thickness of the alternate TiO2/PAA film is estimated to be 1.1 nm per one bilayer adsorption. However, the surface coverage of each component, in practice, appears not to be fully reached 100% every time when we considered the standard deviations of the averages frequency shifts: in some cases the whole electrode surface is covered and in some cases not. Although the bilayer is represented as an idealized model of the film structure in this study, each layer of the TiO2/ PAA film perhaps is considered not to be continuous because of exceptionally thin TiO2 layers. On the other hand, the frequency shift for PAA25 was about 21 Hz, which is 2 times smaller than that for PAA400. The averaged film thickness of PAA25 is estimated to be 0.4 ± 0.1 nm. Perhaps, this difference in thickness between PAA25 and PAA400 is attributed to their molecular weights and structural features in water. Similarly, the thickness of the film deposited by the dip-coating method was calculated (Figure S3 in the Supporting Information). Table 1 shows the frequency shift and thickness of the individual layer coated on the QCM electrode by the GSSG process and the (28) Ariga, K.; Lvov, Y.; Kunitake, T. J. Am. Chem. Soc. 1997, 119, 2224–2231. (29) Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. Langmuir 1996, 12, 3038– 3044.

Table 1. Frequency Shift and Thickness of the Individual Layer Deposited by the GSSG Process and Dip-Coating Method frequency shift/Hz sample (TiO2/PAA400)5 (TiO2/PAA400)10 (TiO2/PAA400)20 (TiO2/PAA25)5 (TiO2/PAA25)10 (TiO2/PAA25)20 (TiO2)5 (TiO2)10 (TiO2)20

PAA

Ti(O-nBu)4

thickness/nm PAA

Gas-Phase Surface Sol-Gel Process 28 ± 8 19 ± 4 0.6 ± 0.2 39 ± 9 18 ± 8 0.8 ± 0.2 40 ± 9 17 ± 8 0.8 ± 0.2 18 ± 6 18 ± 8 0.4 ± 0.1 21 ± 9 18 ± 9 0.4 ± 0.2 20 ± 7 17 ± 9 0.4 ± 0.1 21 ± 6 19 ± 6 20 ± 8

TiO2

total

0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1 0.3 ± 0.1

4.5 11.0 22.0 3.5 7.0 14.0 1.5 3.0 6.0

Dip-Coating Method (TiO2/PAA400)20 62 ± 41 106 ± 29 1.2 ± 0.8 1.7 ± 0.5 58.0 Ti(O-nBu)4 1 mM (TiO2/PAA400)20 110 ± 39 398 ± 51 2.1 ± 0.8 6.4 ± 0.8 170.0 Ti(O-nBu)4 10 mM

dip-coating methods. The thicknesses of the films prepared by the dip-coating method are 58 and 170 nm when 1 and 10 mM of Ti(O-nBu)4 were used, respectively. These values are 2 and 9 times thicker as compared to that of the (TiO2/PAA400)20 alternate film prepared by the GSSG process. Film Morphology. Sensor parameters such as sensitivity, response time, and recovery time are strongly dependent on the techniques used for the QCM coating, and the film morphology may be helpful to understand such a relation. Parts a and b of Figure 3 show scanning electron micrographs of the cross section of the (TiO2/PAA400)20 films deposited using the GSSG process and the dip-coating method on gold-coated QCM resonators. The dip-coated film of Ti(O-nBu)4 (1 mM) and PAA400 is 2 times thicker compared to the corresponding film prepared by the

Figure 3. SEM cross-sectional views of (TiO2/PAA400)20 films deposited by the (a) GSSG process and (b) dip-coating method. AFM images of the surface morphology of (c) PAA25 and (d) PAA400 deposited on TiO2 gel-immobilized mica. The scale bar is 100 nm.

Figure 4. XPS spectra of (TiO2/PAA)3 GSSG films on a ld plate: (a) TiO2/PAA400 film and (b) TiO2/PAA25 film. The inset shows the XPS spectra in the energy region of C(1s).

GSSG process: 54 ± 3 and 27 ± 2 nm, respectively. These results are in a good agreement with the values of 58 ± 4 and 22 ± 2 nm calculated from the QCM frequency shifts (see Table 1). The film density in each case was estimated to be 1.70 and 1.18 g/cm3 from the observed film thickness and the total frequency shift (3360 and 1140 Hz), respectively. Interestingly, the film density is ∼1.5 times lower in the case of the GSSG film. This is because the practical thickness of the GSSG film was observed thicker than that (22 nm) calculated from the average frequency shift and bulk density of each component, 1.4 g/cm3 for PAA and 1.7 g/cm3 for the TiO2 gel (see Table 1). This suggests that the GSSG process does not influence the adsorption structure of PAA, and analytes can be more diffusible in the TiO2/ PAA400 film deposited using the GSSG process, as compared with the corresponding dip-coating film. AFM measurements offer useful information to evaluate the surface morphology of the adsorbed PAA polymers. The PAA layer was prepared on mica with a TiO2 layer (∼0.52 nm as a value of root-mean-square, rms) deposited by the GSSG process.25 As can be seen in parts c and d of Figure 3, a substantial difference in the size of the adsorbed PAA polymers was observed. The morphology of the two samples differed, corresponding to their molecular weights. The PAA25 layer revealed an extremely smooth surface and its roughness was determined to be only 0.49 nm (Figure 3c). On the other hand, the surface roughness of the PAA400 layer increased to 8.0 nm (Figure 3d). This film morphology, which depends on the molecular weight of the PAAs used, will influence the internal structure of PAA in the TiO2/ PAA film. The chemical composition of the (TiO2/PAA)3 films deposited using the GSSG process was inferred from the XPS data (see Figure 4 and Table 2). The composition ratio of Ti/COOH in the TiO2/PAA400 film was estimated to be 1.0:23.4, as calculated from the peak areas after correction of relative sensitivity factors, where CCOOH indicates C(1s) corresponding to COOH (288 eV, relative to standard C(1s) at 285.0 eV). This elemental ratio is about 3 times bigger than that (Ti/COOH ) 1.0:7.2) of the TiO2/PAA25 film. However, the amount of the carboxyl groups calculated with an acrylic acid unit (AA, 72.06 g/mol) Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2231

Table 2. XPS Elemental Analysis (mol % Element) of (TiO2/PAA)3 GSSG Films C(1s) sample

COOH

CH2-CH

O(1s)

Ti(2p3/2)

TiO2/PAA400 TiO2/PAA25

11.9 (23.4) 9.3 (7.2)

59.7 (117) 61.8 (47.9)

27.8 (54.6) 27.5 (21.4)

0.51 (1.0) 1.29 (1.0)

is estimated to be 0.25 and 0.50 nmol for PAA25 and PAA400 from the adsorbed masses of PAA25 and PAA400 per cycle (18 and 36 ng, see Table 2), respectively. In practice, they are just 2 times different. Similarly, the elemental ratio of Ti/COOH can be calculated from the averaged QCM frequency shift of each component. The approximate molar ratio is estimated to be 1.0:3.7 and 1.0:7.5 for the PAA25 and PAA400 films, respectively, when the detailed composition of the hydrolyzed TiO2 component was assumed as TiO(OH)(OR)2 (227.09 g/mol), where R means a normal butyl group. This difference between XPS and QCM measurements may be related to the surface sensitivity of XPS and that the PAA layer was an outermost one. Nevertheless, it is evident that higher molecular weight PAA polymers can offer much more free carboxylic groups as binding sites for amine analytes. Sensitivity of TiO2/PAA Films. Figure 5a shows the typical dynamic responses of the QCM electrode coated with (TiO2/ PAA400)20 film to the ammonia concentration. When the ammonia gas was admitted into the measurement chamber, the

Figure 5. (a) Dynamic responses of the QCM electrode coated with a (TiO2/PAA400)20 film to ammonia at different concentrations. The inset shows a comparison of the calibration curves with data taken at (9) 5 s; (2) 60 s; and the (b) saturation point. (b) Calibration curves for (TiO2 /PAA)n (n ) 5, 10, and 20) films: (b) 5-cycle film, (2) 10cycle film, and (9) 20-cycle film. The frequency shift at a given concentration was taken at 20 s after ammonia exposure. 2232

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

Figure 6. Comparison of dynamic responses of (TiO2/PAA)20 films exposed to 15 ppm of ammonia: (a) GSSG film and (b and c) dipcoating films deposited with 1 and 10 mM Ti(O-nBu)4 solutions, respectively.

frequency decrement was observed. The response was fast and saturated within a few seconds at high concentrations of the analyte, whereas long time responses for saturation (Langmuirtype saturation curve) are observed at low concentrations. When the 100 ppm concentration of ammonia was applied, the isotherm of the ammonia is saturated very rapidly (data not shown). The inset of Figure 5a shows a comparison of the calibration curves with QCM data taken at 5 s, 20 s, and the saturation point after ammonia exposure. The obtained QCM frequency changes at 20 s (curve b) shows a linear sensitivity in the concentration range of 0-15 ppm. This linearity can be extended up to 40 ppm when shorter exposure time (5 s, curve a) was applied, whereas saturated exposure time (curve c) shows a much narrower linear range. This indicates that the film/analyte interaction is based on a chemically specific one between ammonia molecules and the COOH groups of PAA because the sensor response will not be relevant to the exposure time if the binding of the analyte is based on physical adsorption. The calibration curve was plotted from these results, demonstrating a linear response of the QCM electrode to the ammonia concentration (see Figure 5b), where the sensor response values were taken 20 s after the beginning of the ammonia exposure. The thicker film, (TiO2/PAA400)20, has a higher sensitivity toward ammonia than the corresponding 5- and 10-cycle films. The frequency decrease is attributed to the adsorption of ammonia to free carboxyl groups of PAA, and the molecular weights of the PAA polymers used for film preparation may be influencing the sensor response (see Figure S4 in the Supporting Information). Thus, the higher sensor response is observed when PAA400 is used. Figure 6 compares the ammonia sensitivity of the (TiO2/ PAA)20 film as prepared by using different techniques. It is evident that the QCM electrode coated with thin films by the GSSG process (curve a) has a higher sensitivity and a faster response time compared to the other two electrodes with films deposited by the dip-coating technique (curves b and c). Moreover, the thicker film (Ti(O-nBu)4 10 mM, curve c) showed a much slower response and sensitivity than the thinner film (Ti(O-nBu)4 1 mM, curve b). It appears that the

Table 3. Sensor Parameters to Ammonia for Different Sensor Films sample

thickness/nm

sensitivity/slope

linear rangea/ppm

response timeb/s

(TiO2/PAA400)5 (TiO2/PAA400)10 (TiO2/PAA400)20 (TiO2/PAA25)20

4.5 11.0 22.0 7.0

Gas-Phase Surface Sol-Gel Process 0.33 ± 0.02 0.1-15 0.52 ± 0.03 0.1-15 0.98 ± 0.03 0.1-15 0.56 ± 0.03 0.1-15

(TiO2/PAA400)20 Ti(O-nBu)4 1 mM (TiO2/PAA400)20 Ti(O-nBu)4 10 mM

58.0

0.2 ± 0.01

170.0

0.46 ± 0.02

Dip-Coating Method 0.1-8 0.1-8

recovery time/s

LODc/ppm

120 s, where the concentration range is from 0.3 to 15 ppm and from 0.3 to 5 ppm, respectively. It is important to note that the recovery of the sensor signal was complete in rather short time of 10-200 s for concentrations 10 ppm, respectively. Humidity Effect. The effect of relative humidity (rH) on the sensor parameter is crucial in the practical application. In particular, the humidity is a significant interference factor for the QCM measurement because the QCM device is, in principle, highly dependent on mass changes. In cases of micrometer-thick PAA films, the response to a target gas may be influenced by humidity with 1 to 2 orders of magnitude.8 In contrast, our system based on the TiO2/ PAA ultrathin film is quite stable in an intermediate humidity range. Figure 9 shows QCM sensor responses of the (TiO2/ PAA400)20 film to the exposure of 6 ppm ammonia at different rH. Interestingly, the QCM sensor response to ammonia is constant and stable in the wide rH range (8-70%), regardless of the baseline change due to the increment of the rH: ∼80 Hz decrease in the range of 8-30% rH and minor changes in frequency in the range of 30-70% rH. From the above results, it is evident that the current system has a strong merit of absence of significant influences from moisture. This most plausibly is attributed to the structure of the nanocomposite film, in which as mentioned above the TiO2 gel is composed of TiO(OH)(OCH2CH2CH2CH3)2. The PAA polymer complexed with Ti would be additionally surrounded by the hydrophobic domains of the TiO2 gel, such as Ti-O-Ti of the titanium-oxygen network and Ti-OCH2CH2CH2CH3 of the unhydrolyzed alkoxide moiety. On the other hand, the hydrophilic Ti-OH domain would be internally hidden by making a hydrogen bonding with the COOH moiety of PAA. Therefore, the internal property of the TiO2/PAA film may be quite hydrophobic and will suppress water penetration. This film structure would prevent an excessive adsorption of water into the film, especially at high humidity and cause rapid transport of analytes. Consequently, in our system, the diffusion of ammonia into the TiO2/PAA layers is not greatly influenced by water molecules, which was observed experimentally (see Figure 9), although at high humidity levels a small amount of adsorbed water molecules are helpful to destruct hydrogen bonds between the carboxylic groups of PAA, causing the improved response time of ammonia.19,29-31 This process is schematically illustrated in Scheme 1. (30) Hoerter, M.; Oprea, A.; Baˆrsan, N.; Weimar, U. Sens. Actuators, B 2008, 134, 743–749. (31) Sahm, M.; Oprea, A.; Baˆrsan, N.; Weimar, U. Sens. Actuators, B 2008, 130, 502–507. (32) Swartz, M. E.; Krull, I. S. Analytical Method Development and Validation; Marcel Dekker, Inc.: New York, 1997.

Binding Mechanism of Amine Odors. The binding of amine compounds to the carboxyl moiety of PAA will be dominated by their acid-base equilibrium that is closely related to the basicity (pKB) of the amine compounds. The order of basicity of the amine compounds used in this study is n-butylamine > ammonia > pyridine, corresponding to 10.77, 9.36, and 5.17 as the pKBH+ values, respectively. In addition, this acid-base equilibrium is improved by the aid of water (eqs 4 and 5), and the sensor response, in practice, is accelerated at higher rH (see Figure S5 in the Supporting Information).

An additional evidence of the binding of ammonia to the free carboxylic acid groups of PAA was obtained from Fourier transform-infrared (FT-IR) measurements. As shown in Figure S6 in the Supporting Information, the peak at 1716 cm-1, which corresponds to free carboxylic moieties, disappears after exposure of the TiO2/PAA film to ammonia (Figure S6, solid line in the Supporting Information). Instead of that, a new peak at 1551 cm-1 related to the formation of the carboxylate species appears, indicating the binding of ammonia to the available carboxyl groups (Figure S6, dashed line in the Supporting Information). From the above results, the sensing mechanism for the two different film structures could be described using Scheme 2: (i) the films deposited using the dip-coating method have a lower accessibility for amine analytes, and the interaction between amine analytes and PAA is possible mainly in the area close to the surface of the PAA film because the access of the analyte molecules is suppressed by the thick TiO2 layer (Scheme 2a). (ii) In contrast to that, the GSSG process offers highly improved gas diffusion of amine analytes, which are permeable in the entire film (Scheme 2b,c). Consequently, the lower frequency shift and slower sensor response are observed when thick TiO2 films are employed, as compared with ultrathin films prepared by the GSSG process (see Figure 6). CONCLUSIONS In the present study, we developed a nanoassembled thin film approach for the sensitive detection of amine odors. The TiO2/PAA inorganic-organic hybrid ultrathin films prepared using the gasphase surface sol-gel process enabled the rapid penetration of amine gases into the entire film. The concentration of the ammonia analyte condensed inside the alternate TiO2/PAA film was 1370 times higher as compared with the ambient exposure concentration. Incorporation of nanoassembled PAA layers into TiO2 ultrathin layers is an example for the improvement of intrinsic problems of QCM sensors. The general applicability of the present approach is noteworthy. The combinations of a variety of functional organic polymers and the gas-phase surface sol-gel process Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2235

are numerous and will expand the application area of the proposed sensor to various analytes.

ACKNOWLEDGMENT This work was supported by MEXT via Kitakyushu Knowledgebased Cluster Project and Promotion Program for Fire and Disaster Prevention Technologies.

2236

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

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 August 11, 2009. Accepted January 22, 2010. AC901813Q