Functionalized Ga2O3 Nanowires as Active Material in Room

Jul 22, 2010 - ... vapor pressure of P0) was diluted with N2 before entering a miniature bell jar gas ... Because of the experimental setup limitation...
16 downloads 0 Views 2MB Size
Download PDF
pubs.acs.org/Langmuir This article not subject to U.S. Copyright. Published 2010 by the American Chemical Society

Functionalized Ga2O3 Nanowires as Active Material in Room Temperature Capacitance-Based Gas Sensors Lena Mazeina,* F. Keith Perkins, Victor M. Bermudez, Stephen P. Arnold, and S. M. Prokes* Naval Research Laboratory, 4555 Overlook Avenue S. W., Washington, DC 20375 Received May 3, 2010. Revised Manuscript Received June 30, 2010 We report the first evidence for functionalization of Ga2O3 nanowires (NWs), which have been incorporated as the active material in room temperature capacitance gas-sensing devices. An adsorbed layer of pyruvic acid (PA) was successfully formed on Ga2O3 NWs by simple room temperature vapor transport, which was confirmed by Fourier transform infrared spectroscopy. The effect of the adsorbed PA on the surface properties was demonstrated by the change in the response of the NW gas-sensing devices. Results indicate that the adsorption of PA reduced the sensitivity of the Ga2O3 NW device to common hydrocarbons such as nitromethane and acetone while improving the response to triethylamine by an order of magnitude. Taking into account the simplicity of this functionalization together with the ease of producing these capacitance-based gas-sensing devices, this approach represents a viable technique for sensor development.

1. Introduction One of the most important advantages of the recently developed capacitance-based sensors1 using Ga2O3 nanowires (NWs) is that such devices can operate at room temperature (RT) in contrast to the high temperatures required for conductance-based Ga2O3 gas sensors.2-9 Other advantages are that these capacitance-based sensors have a rapid and reversible response to different hydrocarbons,1,10 they do not require a heat source to improve the response or device recovery, and they use little power. In addition, the fabrication is inexpensive, and no contacts to single NWs are required. However, these devices are not chemically selective and show strong responses to such common hydrocarbons as acetone, which is a serious disadvantage for practical applications under realistic conditions. Thus, this work explores the functionalization of Ga2O3 NWs with the ultimate goal of achieving higher selectivity and decreasing the cross-sensitivity of these devices. A common way to functionalize oxide surfaces is by reacting surface OH groups with the desired functional molecule. Most oxide surfaces are terminated in a layer of OH groups resulting from the dissociative adsorption of ambient H2O. In this case, functionalization is readily achieved by reacting the OH groups with species such as alkoxysilanes,11 chlorosilanes,12 or phospho-

nic acids.13,14 However, the pristine (100) surface of β-Ga2O3, which is the most stable surface of this material, does not dissociate H2O.15,16 In principle, hydroxylation could be achieved by damaging the surface by, for example, ion bombardment, with the goal of increasing the reactivity with H2O.17 The same effect might be produced by plasma treatment in a humid environment.14 However, such approaches are quite likely to damage or destroy a NW substrate or a device in which NWs are incorporated. Hence, we seek a less aggressive functionalization chemistry, which is applicable to these nanostructures. As recently reported,15 Ga2O3 nanoribbons adsorb carboxylic acids, in particular acetic and valeric acids, forming strong bridgelike bonds with carboxyl groups. In this paper, we report a simple and reproducible method for functionalizing of Ga2O3 NWs using pyruvic acid (PA) (a keto-carboxylic acid, CH3COCOOH, also known as 2-oxopropanoic acid), which is terminated by a functional keto group. Functionalization is demonstrated by changing the relative chemical reactivity of the Ga2O3 NWs to different hydrocarbons. This approach can be applied to different oxide systems for a variety of applications including chemical and biological sensing, solar cells, drug delivery, etc.

*To whom correspondence should be addressed. E-mail: lena_mazeina@ yahoo.com (L.M.) or [email protected] (S.M.P.).

The sensors were fabricated on Si (100) (0.001 Ω/cm) substrates with a 250 nm thick thermal oxide using a design described in our previous work.1 For each sensor, an interdigitated array of 250 nm Pt electrodes (with a 5 nm Ti adhesion layer) was first deposited on Si by a standard photolithography and lift-off process. The whole device was then covered with a 5 nm electron beam-evaporated gold layer used as a catalyst for the Ga2O3 NW growth. The Ga2O3 NWs were incorporated into the devices by the vapor-liquid-solid (VLS) growth method18 at 900 °C in a horizontal furnace using the method described earlier.19 A

(1) Arnold, S. P.; Prokes, S. M.; Perkins, F. K.; Zaghloul, M. E. Appl. Phys. Lett. 2009, 95, 103102. (2) Fleischer, M.; Meixner, H. Sens. Actuators, B 1991, 4, 437. (3) Fleischer, M.; Seth, M.; Kohl, C.-D.; Meixner, H. Sens. Actuators, B 1996, 36, 297. (4) Li, Y.; Trinchi, A.; Wlodarski, W.; Galatsis, K.; Kalantar-zadeh, K. Sens. Actuators, B 2003, 93, 431. (5) Schwebel, T.; Fleischer, M.; Meixner, H.; Kohl, C.-D. Sens. Actuators, B 1998, 49, 46. (6) Fleischer, M.; Meixner, H. Sens. Actuators, B 1995, 26, 81. (7) Pohle, R.; Fleischer, M.; Meixner, H. Sens. Actuators, B 2000, 68, 151. (8) Cuong, N. D.; Park, Y. W.; Yoon, S. G. Sens. Actuators, B 2009, 140, 240. (9) Liu, Z.; Yamazaki, T.; Shen, Y.; Kikuta, T.; Nakatani, N.; Li, Y. Sens. Actuators, B 2008, 129, 666. (10) Snow, E. S.; Perkins, F. K.; Houser, E. J.; Badescu, S. C.; Reinecke, T. L. Science 2005, 307, 1942. (11) Lee, J.; Jung, B.-J.; Lee, J.-I.; Chu, H. Y.; Do, L.-M.; Shim, H.-K. J. Mater. Chem. 2002, 12, 3494. (12) Malinsky, J. E.; Jabbour, G. E.; Shaheen, S. E.; Anderson, J. D.; Richter, A. G.; Marks, T. J.; Armstrong, N. R.; Kippelen, B.; Dutta, P.; Peyghambarian, N. Adv. Mater. 1999, 11, 227.

13722 DOI: 10.1021/la101760k

2. Materials and Methods

(13) Bardecker, J. A.; Ma, H.; Kim, T.; Huang, F.; Liu, M. S.; Cheng, Y.-J.; Ting, G.; Jen, A. K.-Y. Adv. Funct. Mater. 2008, 18, 3964. (14) Giza, M.; Thissen, P.; Grundmeier, G. Langmuir 2008, 24, 5551. (15) Bermudez, V. M. Langmuir 2008, 24, 12943. (16) Pan, Y.; Liu, C.; Mei, D.; Ge, Q. Langmuir 2010, 26, in press. (17) Purvis, K. L.; Lu, G.; Schwartz, J.; Bernasek, S. L. J. Am. Chem. Soc. 2000, 122, 1808. (18) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (19) Mazeina, L.; Picard, Y. N.; Prokes, S. M. Cryst. Growth Des. 2009, 9, 1164.

Published on Web 07/22/2010

Langmuir 2010, 26(16), 13722–13726

Mazeina et al.

Article Table 1. Capacitance Response to Acetone, Nitromethane, and TEA from Untreated and Functionalized Devicesa response ΔC/C  103

analyte

P0 (mbar, at 25 °C)

ppm at P = 0.1% P0

pure Ga2O3 NWs

Ga2O3 NWs-PA

acetone 308 304 0.063 (0.096 at P/P0 = 0.4%) nitromethane 48.1 47.5 0.038 (P/P0 = 0.2%) TEA 76 75 0.058 a Listed are the measured values of ΔC/C corresponding to P/P0 = 0.1% (or 0.2 and 0.4% as marked).

specially designed boat was used,1 where several devices were placed at the hottest zone of the furnace downstream from the source. Growth was performed for 1 min using metallic Ga (99.999%) as a source and Ar (∼800 mL/min) as a carrier gas. The oxygen source was residual O2 in the background. For functionalization, we used as-received concentrated PA (Alfa Aesar, 98%). Because Si wafers with a thermal oxide had stronger interference fringes and strong Si-O peaks (see below), all Fourier transform infrared (FTIR) measurements were performed using double side-polished Si substrates with no intentional oxidation. Ga2O3 NWs were grown on one side of the substrate and then placed in a desiccator next to a beaker with PA for 2 h at RT. The RT vapor pressure of PA is ∼1.6 Torr.20 Only the side with NWs was exposed to PA. No surface area measurements were performed to estimate NW surface coverage during the adsorption because of the insufficient amount of grown NWs on each substrate piece and because NWs are not free-standing but in contact with the substrate. To check the stability of the resulting adsorbate, after the exposure, the samples were (a) flushed in a constant N2 flow, (b) pumped in vacuum (50 mtorr) for several hours, (c) heated at different temperatures, and (d) washed in different solvents (water, CCl4, and CHCl3). Control samples (bare Si substrates with and without thermally grown SiO2) underwent the same procedure as described above. To confirm the presence of adsorbed PA on the NW surface, FTIR measurements were performed at 4 cm-1 resolution using a “wide-band” HgxCd1-xTe (MCT-B) detector. To avoid interference fringes resulting from the Si substrate, data were collected by transmission in p-polarization with the angle of incidence set close to the Brewster angle (θB ≈ 74° for Si in the mid-IR).21 Spectra of bare as well as functionalized NWs were recorded on both Si and Si/SiO2 substrates and showed identical results. However, we only report spectra recorded using Si for the reasons mentioned above. The spectrum of pure PA in CCl4 (10 μL in 5 mL of solvent) was collected using a KBr liquid cell. The RT gas-sensing setup and measurements were performed as reported earlier.1,10 Briefly, N2 saturated with the vapor of interest (at a vapor pressure of P0) was diluted with N2 before entering a miniature bell jar gas chamber at the desired analyte concentration (P/P0). Typically, a device was exposed to 10 pulses with analyte concentrations from 0.1% P0 to 1% P0 in 0.1% P0 increments. Because of the experimental setup limitations, 0.1% P0 is the lowest possible dilution, which corresponds to 304, 47.5, and 75 ppm for acetone, nitromethane, and triethylamine (TEA), respectively (see Table 1). A change in the capacitance was measured by a balanced ac bridge circuit.1 Gas-sensing experiments were performed on several devices to check reproducibility. Devices without NWs were exposed to the same analytes as devices with NWs, and no response was observed.

3. Results and Discussion The fabricated devices showed a high yield of Ga2O3 NWs (Figure 1) with minimal NW growth on the Pt pads and fingers (Figure 1c) and a resistivity of 10-14 Ω/cm. The NWs are kinked (20) Plath, K. L.; Takahashi, K.; Skodje, R. T.; Vaida, V. J. Phys. Chem. A 2009, 113, 7294. (21) Faggin, M. F.; Hines, M. A. Rev. Sci. Instrum. 2004, 75, 4547. (22) Prokes, S. M.; Carlos, W. E.; Glembocki, O. J. Proc. SPIE Int. Soc. Opt. Eng. 2005, 6008, 60080C/1.

Langmuir 2010, 26(16), 13722–13726

0.014 (P/P0 = 0.4%) no response 0.84

Figure 1. SEM images of a gas sensor device with Ga2O3 NWs as sensing materials. (a) Low and (b) high magnification view of the device. (c) SEM images of Pt pads and (d) Ga2O3 NWs.

Figure 2. FTIR spectrum of (a) Ga2O3 NWs exposed to PA for 2 h in comparison with (b) the spectrum of pure PA in CCl4. The blank region in panel (b) is obscured by an adsorption band of the solvent, CCl4. Inserts show the PA molecule (c) and two possible geometries of PA adsorption on the Ga2O3 surface: bridging (d) and monodentate (e). Intensities for the spectrum of PA in CCl4 were reduced by a factor of 10 for display purposes.

with many bends resulting from the growth under oxygen-deficient conditions,22 with diameters of 30-100 nm and lengths of tens of μm (Figure 1d). The FTIR spectrum of the NWs confirmed the monoclinic β-Ga2O3 structure (see the Supporting Information, Figure S1). The FTIR spectrum of Ga2O3 NWs exposed to PA is shown in Figure 2. All bands were reproducible in several experiments, although the strong 1739 cm-1 peak varied over a range of 1739-1752 cm-1 from sample to sample. The spectrum persisted after flushing in an N2 flow and after pumping for several hours, indicating that the functionalized surface is stable. The spectra observed after exposure of the unprocessed control piece of Si with thermal oxide showed no PA adsorption (Figure S2a in the Supporting Information). The spectra of Si substrates exposed to PA showed a shoulder in the region 1780-1740 cm-1 (Figure S2b in the Supporting Information), which could correspond to the CdO vibration bands of adsorbed PA; however, its intensity is too weak as compared to the one observed for functionalized NWs; thus, adsorption of PA on Si can be neglected. In addition, a dense meshlike network of Ga2O3 NWs on the Si substrate prevents adsorption on the underlying Si substrate. DOI: 10.1021/la101760k

13723

Article

The vibrational spectrum of PA in CCl 4 solution is in good agreement with recent vapor-phase data20 except for some differences in relative intensities. Major changes occur when the molecule interacts with the NWs, a full analysis of which is beyond the scope of the present work. Briefly, the free-molecule structure in the ∼1000-1500 cm-1 range (which is assigned20,23 to various δ(CH3) deformation and ν(C-C) and ν(C-O) stretching modes) appears broadened and shifted but otherwise intact in the adsorbed state. In all forms of PA (vapor, liquid, and solid), the keto ν(CdO) appears in the 1728-1740 cm-1 range; therefore, we assign the strong adsorbate peak at 1739 cm-1 to a “free” keto ν(CdO). For the pyruvate anion (in Naþ or Kþ salts), this mode is shifted slightly lower,24 to 1709 cm-1; thus, PA does not appear to form an ionic bond with Ga2O3 NWs. The carboxyl ν(CdO), which appears at 1788 cm-1 in CCl4 solution, is conspicuously absent for the adsorbate. This mode is sensitive to the molecular environment. For example, it is not readily apparent in liquid PA due to the shift and broadening, which result from formation of H-bonded dimers.23 Modes are seen in the 1600-1700 cm-1 range and near 1450 cm-1, which are consistent with, respectively, the νas(OCO) and νs(OCO) modes of a bridging carboxylate25 or a carboxylate anion,24 but these are uncharacteristically weak relative to the keto ν(CdO). We see two modes, at 1621 and 1660 cm-1, both of which can be assigned to νas(COO) of a carboxylate, thus suggesting two possible carboxylate geometries. The peak at 1660 cm-1 can also be assigned to the CdC stretch from the enol tautomer.26 The separation between the symmetric and the asymmetric modes (Δ = 170-210) suggests a bridgelike geometry25,27 (Figure 2d). We also observe a broad peak with a maximum at ∼3200-3300 cm-1 (not shown), which might correspond to hydrogen-bonded OH from PA, as in the structure shown in Figure 2e. It could also be due to OH groups formed on the Ga2O3 NW surface28,29 when H is released from the carboxyl group during formation of the bridging structure (Figure 2d) or to the OH group of the possible enol tautomer of PA.26 The possible presence of a molecular OH group together with only weak asymmetric and symmetric COO vibrations suggests that the second geometry may involve a dative bond between the carboxyl CdO and a coordinatively unsaturated Ga atom together with a CO-H---O bond to an unsaturated surface O atom (Figure 2e). Qualitatively, the carboxyl ν(CdO) would be red-shifted by this interaction and might not be resolvable from the keto ν(CdO), which would be relatively unaffected. A structure of this type has been predicted theoretically,15 but not observed experimentally, for formic acid (HCOOH) on the more stable of the β-Ga2O3 (100) surfaces. Figure 2 shows only two tentative models for PA adsorption. We do not exclude that other geometries are possible as well since the interaction of PA with Ga2O3 NWs is a very complex process. However, we believe that both geometries are the most plausible ones since both of them have a free keto group, the existence of which is suggested by the presence of the strong peak at 1739 cm-1. This peak is observable upon heating the NWs and disappears only upon heating at 310 °C (see the Supporting Information, Figure S3). It is still present although weaker in intensity when the devices are soaked in CCl4 and CH3Cl for 30 min but disappears (23) Ray, W. J.; Katon, J. E.; Phillips, D. B. J. Mol. Struct. 1981, 74, 75. (24) Hanai, K.; Kuwae, A.; Sugawa, Y.; Kunimoto, K.-K.; Maeda, S. J. Mol. Struct. 2007, 837, 101. (25) Tackett, J. E. Appl. Spectrosc. 1989, 43, 483. (26) Kakkar, R.; Pathak, M.; Gahlot, P. J. Phys. Org. Chem. 2008, 21, 23. (27) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th ed.; Wiley: New York, 1986; p 232. (28) Collins, S. E.; Baltanas, M. A.; Bonivardi, A. L. Langmuir 2005, 21, 962. (29) Meriaudeau, P.; Primet, M. J. Mol. Catal. 1990, 61, 227.

13724 DOI: 10.1021/la101760k

Mazeina et al.

Figure 3. Response to acetone (insert) by devices with (a) as grown Ga2O3 NWs and (b) NWs functionalized with PA.

Figure 4. Responses to nitromethane: (a) from devices with pure Ga2O3 NWs, (b) from devices functionalized with PA, and (c) from devices where PA was removed by heating.

Figure 5. Reponses to TEA (insert) by pure Ga2O3 NWs (a) and by Ga2O3 NWs functionalized with PA (b).

after soaking in water for 30 min (Figure S4 in the Supporting Information). The vibrational mode of the keto group shifts by 20 cm-1 after treatment in CCl4 and CHCl3, which might be another confirmation of the presence of two different geometries. To check the effect of the functionalization, we exposed the devices with pure and functionalized NWs to acetone (Figure 3), nitromethane (Figure 4), and TEA (Figure 5). The three compounds were chosen first because they represent common classes of organic compounds: ketones, nitro-alkanes, and amines. They also represent different groups of strong and weak electron acceptors and donors. The detection of these materials by vapor sensors is also interesting because they have many industrial applications and yet are toxic and need to be monitored. In addition, acetone and nitromethane can be used in explosive materials. TEA also has a number of commercial applications and is considered to be a carcinogen.30 (30) Wang, C.-C.; Lu, K.-L.; Chen, X.-Y. Int. Biodeterior. Biodegrad. 2007, 59, 202.

Langmuir 2010, 26(16), 13722–13726

Mazeina et al.

Figure 6. Change in the sensor response to acetone (circles) and nitromethane (squares) as a function of the pulse dose (0.1-1% P0) for pure (open symbols) and functionalized devices (filled symbols). Numbers indicate ppm for the acetone and nitromethane doses, respectively, at the first, fifth, and tenth pulses. Note that the response of the PA-functionalized device to nitromethane represents the sensor noise limit. The actual difference between pure and functionalized devices responses is even bigger.

The response to acetone was first reported earlier,1 and the response to nitromethane and TEA is reported for the first time in this work. The response of devices with pure Ga2O3 NWs to acetone (Figure 3) is significantly higher than that to nitromethane (Figure 4), which has a lower P0 (Table 1). However, both responses are reversible and completely recover after the analyte vapor is removed, which suggests weak physical adsorption. Devices showed sensitivity to 304 ppm of acetone and 47.5 ppm of nitromethane (first dose, see Figure 6). However, we would like to emphasize that this is not the limit of the sensitivity of our devices, and responses to lower concentrations are expected as well. However, higher dilutions are currently not available due to the limitations of our gas-sensing apparatus. The magnitude of the response increases with the dose (Figure 6) following the Freundlich adsorption isotherm.1 However, the responses to acetone and nitromethane are not distinguishable. Thus, the pure Ga2O3 NW devices not only lack selectivity but actually can be contaminated by significant amounts of either acetone or nitromethane, which is not optimal for practical applications. Devices showed sensitivity to 74 ppm of TEA (see Table 1). However, in contrast to the responses to nitromethane and acetone, the response of the untreated NWs to TEA is equal for each dose increment and shows poor recovery (Figure 5a). This indicates that the device response at RT consists of a stronger irreversible chemisorption component and a weaker fast-recovering physisorption component. These devices fully recover after heating to 100-130 °C. The response to TEA parallels that observed for pyridine on the β-Ga2O3 (100) surface.15 Here, the weak and reversible physisorption is associated with coordinatively unsaturated (CU) octahedral Ga sites, which are weak Lewis acids. The stronger chemisorption bond is ascribed to CU tetrahedral sites, which are exposed on the (100) surface when O vacancies form. The effect of functionalization on the device response is shown in Figures 3-5. It can be seen that adsorbing PA significantly decreased the response to acetone (Figure 3) by more than a factor of 5 (see Figure 6). No response is seen below 0.3% P0, and only a weak response, equivalent to 14% of that of the untreated sensor, is observed for the fourth dose (see Table 1). For bigger doses, the response becomes stronger. This can result from interaction of acetone either with NW sites free from PA or with adsorbed PA itself. The formation of weak hydrogen bonds between acetone Langmuir 2010, 26(16), 13722–13726

Article

and adsorbed PA, taking into account the presence of acidic OH groups, is also possible. The negative baseline trend might suggest that acetone removes adsorbed PA. No response to nitromethane is observed for concentrations below P/P0 = 1% (Figure 4). Thus, by adsorbing PA, we have passivated the surface of the Ga2O3 NWs against physical adsorption of acetone and nitromethane. As we mentioned earlier, the IR spectrum shows that PA is completely removed at 310 °C. However, even heating to 170 °C for several minutes eliminates a significant amount of PA, and after heating, the devices show a nitromethane response comparable to that of pure Ga2O3 NWs (see Figure 4). Desorption of the PA evidently occurs without leaving a significant residue or altering the NW surface since the bare NW response to acetone or nitromethane is fully recoverable. In addition, re-exposure of the NW device to PA results in a sensor response similar to that seen after the initial exposure to PA (i.e., no response to nitromethane, not shown). Nitromethane could form hydrogen bonds with the acidic OH from PA analogous to what we suggested for the acetone interaction with adsorbed PA; however, nitromethane forms only very weak hydrogen bonds;31,32 hence, there is an insufficient coverage of nitromethane to register a sensor response. The functionalization of Ga2O3 NWs with PA increased the device response to TEA by 1 order of magnitude (see Figure 5 and Table 1). There is again a physi- and a chemisorption component in the response. The former recovers more slowly for the NWs treated with PA (compare the “peaks” in Figure 5a with the “saw teeth” in Figure 5b, which indicates a slower recovery). Heating to 100-150 °C does not recover the device, confirming the strong chemical interaction. After all high-binding energy sites have been filled, only physical adsorption occurs, and the last doses show equal intensity of the response. A mechanism can now be proposed for the effects of PA functionalization on the sensor response. On the bare Ga2O3 surface, octahedral and tetrahedral CU Ga sites act as Lewis acids and are the active sites in adsorption.15 On a pristine (100) surface, the only active adsorption sites are the weakly acidic CU octahedral Ga atoms. Even a strong electron donor such as pyridine interacts only weakly with these sites.15 Any plausible model for PA adsorption involves bonding to unsaturated Ga sites, which are then passivated (i.e., blocked) against adsorption of analyte vapors. The potential adsorption sites are then determined by the functional groups of the PA. None of these are strong electron donors or acceptors, and the adsorption of acetone and nitromethane is thus suppressed. We believe that adsorbed PA forms at least a monolayer on the Ga2O3 surface and blocks a majority of all active Ga adsorption sites since we did not see any response for nitromethane from PA-functionalized devices. The carboxylic OH, which is present in the structure shown in Figure 2e only, however, is a strong Bronsted acid and can react with TEA to give the observed chemisorption effect. The C atom of the keto group is weakly electrophilic and can thus contribute to the reversible physisorption observed for TEA and also, at higher P/P0, for acetone, which is a weak electron donor via the carbonyl O atom. Thus, PA functionalization replaces the nonspecific adsorption sites of the bare Ga2O3 surface with sites that are targeted for a particular class of molecules (in this case, amines). The sensor results in Figure 5 show that both the chemisorption and the physisorption sites provided by PA are more effective in bonding to TEA than are those found on the bare NW surface. A detailed (31) Nagakura, S.; Gouterman, M. J. Chem. Phys. 1957, 26, 881. (32) Ungnade, H. E.; Roberts, E. M.; Kissinger, L. W. J. Phys. Chem. 1964, 68, 3225.

DOI: 10.1021/la101760k

13725

Article

Mazeina et al.

study of the IR spectra and adsorbate geometry, both before and after interaction with acetone, nitromethane, and TEA, will be published in a subsequent paper.

4. Conclusions We have examined the effect of functionalization on the sensing responses of capacitance-based devices with Ga2O3 NWs as the active material. Devices with pure Ga2O3 NWs showed strong responses to acetone and nitromethane. However, strong responses to common hydrocarbons might be a significant disadvantage for practical applications since the sensor can register a false positive or be saturated by ambient species. To avoid this, we have shown that passivation and a decrease in sensitivity to these common species (by 5 times and at least by 10 times to acetone and nitromethane, respectively) can be achieved by functionalizing the (33) Devdas, S.; Mallik, R. R.; Coast, R.; Henriksen, P. N. Surf. Sci. 1995, 326, 327. (34) Devdas, S.; Mallik, R. R. Int. J. Adhes. Adhes. 2000, 20, 349. (35) Dobson, K. D.; McQuillan, A. J. Spectrochim. Acta, Part A 2000, 56, 557. (36) Dobson, K. D.; McQuillan, A. J. Spectrochim. Acta, Part A 1999, 55, 1395. (37) Gotovac, S.; Yang, C.-M.; Hattori, Y.; Takahashi, K.; Kanoh, H.; Kaneko, K. J. Colloid Interface Sci. 2007, 314, 18. (38) Yang, X.; He, Z. H.; Zhou, X. J.; Xu, S. H.; Leung, K. T. Appl. Surf. Sci. 2006, 252, 3647. (39) Delgado, J. M.; Berna, A.; Orts, J. M.; Rodes, A.; Feliu, J. M. J. Phys. Chem. C 2007, 111, 9943.

13726 DOI: 10.1021/la101760k

NW surface with PA. At the same time, chemical specificity of a sensor is important as well. We showed that functionalization with PA enhances the response to TEA by 1 order of magnitude. In general, functionalization with carboxylic acids, especially those that are liquids under ambient conditions, is simple and does not require expensive equipment. It is reversible, and if necessary, the NW surface can be completely recovered by heating at a reasonable temperature or by extensive washing in water. A similar approach can be used for functionalization of other materials that adsorb carboxylic acids (e.g., TiO2, Al2O3, and metals33-39, etc.) for a variety of applications including dye-sensitization, bio- and chemical sensing, drug delivery, etc. Acknowledgment. This work was supported by the Office of Naval Research. L.M. thanks the National Research Council for administrative support. Art Snow is thanked for providing distilled TEA. Supporting Information Available: FTIR spectrum of Ga2O3 NWs (Figure S1), FTIR spectra of silicon substrates after exposure to PA (Figure S2), evolution of FTIR spectra of adsorbed PA upon heating (Figure S3), and FTIR spectra of adsorbed PA after treatment with different solvents (Figure S4). This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2010, 26(16), 13722–13726