Vapor Sensitivity of Networked Gold Nanoparticle Chemiresistors

Aug 4, 2007 - 1,12-Dodecanedithiol was used as a hydrophobic, flexible linker ... water in the concentration range 100−5000 ppm at 0% relative humid...
0 downloads 0 Views 194KB Size
J. Phys. Chem. C 2007, 111, 12855-12859

12855

Vapor Sensitivity of Networked Gold Nanoparticle Chemiresistors: Importance of Flexibility and Resistivity of the Interlinkage Yvonne Joseph,*,† Antun Peic´ ,†,§ Xudong Chen,‡ Josef Michl,‡ Tobias Vossmeyer,†,| and Akio Yasuda† Materials Science Laboratory, SONY Deutschland GmbH, Hedelfinger Strasse 61, 70327 Stuttgart, Germany, and Department of Chemistry and Biochemistry, UniVersity of Colorado, Boulder, Colorado 80309-0215 ReceiVed: March 14, 2007; In Final Form: June 21, 2007

In this study we investigated the response behavior of chemiresistors made from differently interlinked networks of gold nanoparticles. Our results show that the degree of flexibility and conductivity of the interlinkage between the nanoparticles has a profound impact on the response characteristics of such sensors. 1,12-Dodecanedithiol was used as a hydrophobic, flexible linker compound. For comparison, [4]-staffane3,3′′′-dithiol provided a rigid, rodlike linkage, and 4,4′-terphenyldithiol was used as a rigid linker with enhanced conductivity due to its delocalized aromatic moieties. As determined by AFM, all three sensor coatings had similar thicknesses (∼30 nm), but the degree of interlinkage, as measured by XPS, was significantly higher in the case of the flexible network. All three materials showed linear current-voltage characteristics. The vapor sensitivity was tested by dosing the sensors with toluene, 1-propanol, 4-methyl-2-pentanone, and water in the concentration range 100-5000 ppm at 0% relative humidity. The flexible 1,12-dodecanedithiol interlinked film responded with an increase in resistance to these analytes and with the highest sensitivity to toluene. In striking contrast, the rigid staffane interlinked film responded with a decrease in resistance to all four analytes and with the highest sensitivity to 4-methyl-2-pentanone. The rigid but more conductive 4,4′-terphenyldithiol interlinked coating gave hardly any response, although microgravimetric measurements showed that similar amounts of analytes were absorbed as in the case of the other two sensor films. The different response characteristics are discussed in terms of film swelling and changes in permittivity.

Introduction Over the past few years the use of gold nanoparticles for the development of novel biochemical and chemical sensors has attracted considerable attention.1 Such sensors are fabricated by depositing thin films of ligand-capped or interlinked gold nanoparticles onto appropriate transducers, which are then used for liquid or gas phase operation. For gas phase applications mass-sensitive sensors2,3 and especially chemiresistors,4-12 as depicted schematically in Figure 1, are under investigation. In the latter sensor type the gold particles provide the sensor coating with sufficient electrical conductivity and enable a simple electrical signal transduction. The organic film component stabilizes the particle structure in the film and provides sites for selective analyte sorption. As indicated by earlier investigations,6 the kind and amount of analyte sorbed in the film is usually determined by its solubility in the organic component. Although notable progress toward applications of these sensors has been made,13,14 the underlying molecular mechanisms controlling their sensitivity are still only incompletely understood. In several publications the sensing mechanism has been discussed in the context of an activated tunneling model,4,9-12 which describes the conductivity of nanoparticle/ * Corresponding author. Phone: +49 711 5858 836. Fax: +49 711 5858 484. E-mail: [email protected]. † SONY Deutschland GmbH. ‡ University of Colorado. § Current address: Max Planck Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany. | Current address: Institut fu ¨ r Physikalische Chemie, University Hamburg, Grindelallee 117, 20146 Hamburg, Germany.

Figure 1. Chemiresistor device with sensitive coating comprising Au nanoparticles interlinked with organic dithiols. In this study three different dithiols were used, depicted on the right of this figure. All three molecules have similar lengths (C12-chain; AuDT, 1.70 nm; AuSF4, 1.57 nm; AuTPT, 1.47 nm; calculated with DMol3 (GGA/ BLYP) embedded in Acelrys, Materials Studio 4.0).

organic composites as a product of two terms. The first term describes the tunneling current between neighboring particles, which decays exponentially with increasing interparticle distance. The second term takes into account the thermal activation of charge carriers. The activation energy in this Arrhenius term is inversely proportional to the permittivity of the organic matrix. Details are given in the Supporting Information. It is apparent that any process changing either the particle separation (e.g., by swelling) or the permittivity of the organic matrix (e.g., by void filling with high dielectric constant analytes) will be detectable by changes in the conductivity. According to the model it was suggested that in the case of analytes with low dielectric constants the typically observed resistance increase

10.1021/jp072053+ CCC: $37.00 © 2007 American Chemical Society Published on Web 08/04/2007

12856 J. Phys. Chem. C, Vol. 111, No. 34, 2007 is related to film swelling as the overriding component of the sensing mechanism.2,4,9-12,15,16 Further, in case of some analytes with high dielectric constants, such as methanol or water, the observed decrease in resistance was ascribed to an increase in permittivity of the organic matrix as the dominating component of the sensing mechanism.9-12 Although the qualitative explanations given for the observed sensor responses are quite plausible, it is very difficult to give direct evidence for the hypothesis that analyte-induced swelling plays a major role in the sensor response mechanism. This is mainly due to the fact that even only marginal swelling should result in a significant increase in resistance due to the exponential relationship between the tunneling current and the average interparticle distance. Ellipsometric measurements indicated that swelling of films from ligand-capped nanoparticles may indeed occur upon dosing with high vapor concentrations.10 On the other hand, neutron reflectivity measurements16 showed that even at saturated vapor concentrations swelling of Au nanoparticle/dendrimer films was negligible within the resolution of the method. It is noted that the accuracy of those measurements is limited by the inherent roughness of the films. An alternative, experimental approach to investigating the importance of film swelling vs changes in permittivity is to characterize sensor coatings with a controlled ability to swell. Networked nanoparticle films comprising flexible or rigid linker molecules provide a very useful model system for such investigations. Ideally, the linker molecules must not differ in properties other than flexibility, such as length, permittivity, electronic delocalization, and solubility. Choosing 1,12-dodecanedithiol (DT) as the linker for a flexible network, the closest existing saturated hydrophobic rigid rodlike hydrocarbon molecule with a similar length is [4]-staffane-3,3′′′-dithiol (SF4, Figure 1). Besides the question of how the flexibility of the linker molecules affects the response characteristics, we were interested in investigating how the conductivity of the linker molecules affects the sensitivity of the sensors. To address this question, we tested sensor coatings comprising the rigid and partially delocalized linker 4,4′-terphenyldithiol (TPT, Figure 1), for which conductivity can be expected to be higher than that of [4]-staffane-3,3′′′-dithiol. The three Au nanoparticle/linker composites investigated in this study were deposited onto substrates with interdigitated electrodes and quartz microbalances (QCMs) via layer-by-layer self-assembly. The resulting coatings were characterized by atomic force microscopy (AFM), UV/vis and X-ray photoelectron (XP) spectroscopies, and resistance measurements. The sensor response characteristics of chemiresistors and QCMs were measured by dosing the films with diluted vapors from toluene, 4-methyl-2-pentanone, 1-propanol, and water. Experimental Section The synthesis of the (4 ( 1.3) nm dodecylamine protected Au nanoparticles,4 of 1,12-dodecanedithiol,4 and of [4]-staffane3,3′′′-dithiol17 as well as the film assembly4 and the UV/vis spectrometer4 have been described earlier. The synthesis of 4,4′terphenyldithiol is given in the Supporting Information. Details on the QCM measurements2 and the photoelectron spectrometer used to determine the film compositions as well as the XP spectra data evaluation procedures18 have been described in the literature. For characterization of the film morphology and thickness, atomic force microscopic techniques were used. The microscope and measurement equipment as well as procedures for the vapor dosing experiments have been given elsewhere.4

Joseph et al. Results and Discussion Thin composite films were deposited onto glass substrates with interdigitated gold electrodes and quartz microbalances (QCMs) by interlinking Au nanoparticles (∼4 nm core diameter) with 1,12-dodecanedithiol (DT), 4,4′-terphenyldithiol (TPT), or [4]-staffane-3,3′′′-dithiol (SF4) via layer-by-layer self-assembly. The stepwise film formation was clearly visible by an intensity increase of a bluish-purple color, which is due to plasmon absorption by the nanoparticles. We noted that compared with the other two materials the AuTPT deposition was accompanied by the most rapid increase in color intensity, indicating that the TPT linker most efficiently networked the nanoparticles. Possibly, the deposition process was promoted by π-π interactions between neighboring TPT molecules. The thickness of the films was investigated with AFM by imaging the edges of several scratches produced by moving a needle across the film while applying a gentle pressure. These measurements revealed thicknesses of 33.6 ( 2.7 nm (31.9 ( 2.0 nm), 43.7 ( 2.5 nm (39.7 ( 6.4 nm), and 42.6 ( 1.3 nm (54.7 ( 4.2 nm) for AuDT, AuSF4, and AuTPT, respectively (values for QCM devices in parentheses). Interestingly, the thickness-normalized UV/vis spectra (Supporting Information) indicated that the density of Au nanoparticles in the AuSF4 film was significantly lower than in the AuDT material. All three composite films showed linear current-voltage characteristics in the bias range (1 V. The room-temperature conductivities σ(RT) of the films were calculated by taking into account their resistances R, thicknesses d, and the geometry of the electrode structures (σ(RT) ) g/[(2n - 1)ldR], with n ) 50, number of electrode finger pairs; g ) 10 µm, gap between electrodes; l ) 1800 µm, overlap length of electrode fingers). The resulting values are 7.7 × 10-4, 6.8 × 10-6, and 6.9 × 10-3 S/cm for AuDT, AuSF4, and AuTPT, respectively. The conductivity of AuDT is in good agreement with earlier investigations.4 As expected, the more conductive molecular linker TPT gave the most conductive network. Here, we assume that the charge transport of the nonconjugated molecule (DT) is dominated by nonresonant tunneling whereas the conjugation of the TPT molecule gives rise to states in the gap used for resonant tunneling.19 However, it is not distinguishable by our measurements whether the charge propagation occurs along the molecular backbone or even in between neighboring TPT molecules or both. The low conductivity of AuSF4 is most likely due to the nonconjugated, rigid structure of the SF4 molecules, which provide larger spacing between the nanoparticles than the flexible DT linker and more efficient electrical isolation of the particles than the TPT linker. XPS investigation of the S 2p levels revealed the presence of different sulfur species in the films: sulfur bound to Au nanoparticles (S-Au, ∼162.2 eV), free thiol groups (S-H, ∼163.5 eV), radiation-damaged sulfur (SRD, ∼164.4 eV), and oxidized sulfur (SOx, >166 eV). The corresponding XP spectra are given in Figure 2. The peak assignment for S-Au and S-H is in agreement with the literature.20-24 The relative intensities of these peaks reveal that films assembled with the rigid linker molecules (AuTPT and AuSF4) contained a much larger fraction of free thiols than the AuDT film. This result can be explained by taking into consideration the higher degree of motional freedom in the case of the flexible linker molecule DT, which promotes successful approach and binding of the thiol groups to the nanoparticles. Molecular dynamics studies25 revealed that dodecanethiol molecules protecting gold nanoparticles start chain

Networked Gold Nanoparticle Chemiresistors

J. Phys. Chem. C, Vol. 111, No. 34, 2007 12857

Figure 2. X-ray photoelectron spectra of S 2p regions of AuDT, AuTPT, and AuSF4. The pie diagrams give the relative amounts of different sulfur species.

melting at 280 K and at around room temperature 85% trans configurations of the chains exist. Thus, assuming that these results can be transferred to the dodecanedithiol molecule, backbinding is unlikely. If we neglect back-binding of DT, the observed differences in the S-Au/S-H ratios imply that the degree of interlinkage is significantly higher in films with flexible linker compounds (DT) than in films with rigid linker compounds (SF4 and TPT). As explained above, the SRD peak was attributed to radiationinduced damage. However, we note that this peak may also be explained by a slightly asymmetric peak shape, which was not taken into account in the data evaluation procedure. The small amount of detected oxidized sulfur is most likely due to oxidation reactions occurring during storage of the samples for a few weeks at ambient conditions. The chemical sensitivity of the films was characterized by dosing them with vapors of toluene, 4-methyl-2-pentanone, 1-propanol, and water while monitoring their resistance. These analytes were used as test vapors because they have comparable vapor pressures (29, 21, 20, and 23 mbar at 20 °C, respectively). Thus, their partitioning in the sensor films is mainly determined by their chemical nature and not by major differences in vapor pressure. Further, the dielectric constants of the analytes vary significantly: 2.4 (toluene), 13.11 (4-methyl-2-pentanone), 20.8 (1-propanol), and 80.1 (water) at 20 °C.26 Taken together, these analytes are well-suited to study the chemical selectivity of the sensors including the influence of differences in their dielectric constants on the response characteristics of the chemiresistors. Typical response transients of the chemiresistors AuDT, AuSF4, and AuTPT, and the corresponding QCM sensors, are shown in Figure 3. These transients were recorded by dosing the sensors with the analytes at 5000 ppm in purified and dried air. Figure 4 shows the response isotherms for the films AuDT and AuSF4 for analyte concentrations from 100 to 5000 ppm. The AuDT and AuSF4 resistor devices responded within a few seconds to the organic analytes with almost ideal rectangular

Figure 3. Response traces of AuDT, AuSF4, and AuTPT toward 5000 ppm at 0% relative humidity of the indicated analyte. (top) Chemiresistor response traces; (bottom) concentration in the film determined by QCM measurements.

transients. In the case of the flexibly interlinked AuDT film, the response to all analytes is characterized by an increase in resistance, as we reported earlier.4 In striking contrast, the AuSF4 film containing the rigid SF4 linker responded with a decrease in resistance to all analytes. To our knowledge this is the first example of an Au-nanoparticle-based chemiresistor that gives negative responses (resistance decrease) even to analytes with relatively low permittivity, such as toluene. Remarkably, the AuTPT chemiresistor with the rigid and more conductive TPT linker was almost insensitive, though the QCM transducer revealed a significant uptake of analyte. As shown in Figures 3 and 4, the response amplitudes of the AuDT-coated chemiresistor decreased in the order toluene, 4-methyl-2-pentanone, and 1-propanol. The corresponding QCM measurements revealed that this decrease in sensitivity correlates qualitatively with a decrease of analyte partitioning in the film. In general, the QCM results of all films suggest that partioning of the analytes in the films was mainly determined by the match of solubilities between the analytes and linker compounds. According to the charge transport model mentioned in the Introduction, the observed resistance increase of the AuDT film indicates that the flexible DT interlinkage allows for swelling during vapor sorption. Consequently, increasing interparticle distances seem to be the dominating component of the sensing mechanism. We note that AuDT chemiresistors from different preparations gave qualitatively always the same positive responses (resistance increase) to the four analytes, as shown in Figure 3. However, we occasionally observed some variations in the magnitude of response amplitudes when comparing

12858 J. Phys. Chem. C, Vol. 111, No. 34, 2007

Joseph et al. The insensitivity of the AuTPT film to vapor sorption suggests that the use of more conductive and rigid linker molecules results in a quite robust charge transport process. Conclusions

Figure 4. Response isotherms of AuDT and AuSF4 showing concentration-dependent responses from 100 to 5000 ppm for toluene, 4-methyl-2-pentanone, and 1-propanol.

different sensor batches (Supporting Information). We tentatively assign these variations to slight differences in the size distribution and composition of the nanoparticles, as well as to possible slight differences in the degree of interlinkage. In the case of the AuSF4 material comprising the rigid, nonconjugated interlinkages, the ability of the film material to swell must be significantly reduced. In agreement with this assumption, the negative responses (resistance decrease) indicate that swelling is not the dominating component of the sensing mechanism. Instead, the sensor response seems to be determined by an increase in permittivity of the nanoparticle environment. Because swelling should be efficiently suppressed, we assume that the increase in permittivity involves mainly filling of empty voids by analyte and not the displacement of organic material. In this context it should be recalled that the thickness-normalized UV/vis spectra along with the XP spectra led to the conclusion of a more open structure of the AuSF4 material compared to the AuDT film. Comparing the chemiresistor and QCM responses of the AuSF4 films (Figures 3 and 4), it is noted that comparable concentrations of sorbed toluene and 4-methyl-2-pentanone resulted in very different response amplitudes of the chemiresistor. The toluene response is about only half as strong as that to 4-methyl-2-pentanone. This finding further supports the assumption that the response of the rigid chemiresistor coating was mainly determined by the dielectric constant of the analyte (see above). Along the same line, it is seen that the response amplitudes of the chemiresistor to toluene and 1-propanol are comparable, although the concentration of sorbed 1-propanol was only about half that of toluene.

In conclusion, our results confirm that swelling and permittivity changes are of major importance for the sensing mechanism of chemiresistors from organically networked gold nanoparticles. In particular, we summarize the following: 1. The selectivity of the QCM sensors indicates that the relative partioning of analytes in the sensor coatings is mainly controlled by the solubility match between the analyte and the interlinking molecules. 2. The response characteristics of AuDT films suggest that flexibly interlinked films typically respond to analyte sorption with an increase in resistance because swelling can be the dominant component of the sensing mechanism. 3. The response characteristics of AuSF4 film suggests that rigidly interlinked coatings respond with a decrease in resistance to analyte sorption. The sensitivity of this response, related to the concentration of sorbed analyte, increases with increasing permittivity of the analyte. This result supports the assumption that for such coatings void filling accompanied by an increase in the permittivity of the nanoparticles’ environment is the overriding component of the sensing mechanism. 4. The low sensitivity of TPT-interlinked chemiresisors suggests that the sensitivity of rigidly interlinked coatings significantly decreases with increasing conductivity of the linkage. These results reveal that further rational development of such chemiresistor coatings is possible by carefully balancing the chemical and physical properties of the linker molecules, including their solubility, flexibility/rigidity, and conductivity. Acknowledgment. We thank B. Guse for support in sensor film preparation, Dr. H.-G. Nothofer and M. Rosenberger for support in organic synthesis of the interlinking molecules, and K. Wiegers for support in QCM measurements and data evaluation. Supporting Information Available: Activated tunneling model, synthesis of 4,4′-terphenyldithiol, thickness-normalized UV/vis spectra, composition of AuDT, AuTPT, and AuSF4, and chemiresistor responses of a second set of sensors. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Franke, M. E.; Koplin, T. J.; Simon, U. Small 2006, 2, 36-50. (2) Joseph, Y.; Krasteva, N.; Besnard, I.; Guse, B.; Rosenberger, M.; Wild, U.; Knop-Gericke, A.; Schloegl, R.; Krustev, R.; Yasuda, A.; Vossmeyer, T. Faraday Discuss. 2004, 125, 77-97. (3) Grate, J. W.; Nelson, D. A.; Skaggs, R. Anal. Chem. 2003, 75, 1868-1879. (4) Joseph, Y.; Besnard, I.; Rosenberger, M.; Guse, B.; Nothofer, H. G.; Wessels, J.; Wild, U.; Knop-Gericke, A.; Su, D.; Yasuda, A.; Vossmeyer, T. J. Phys. Chem. B 2003, 107, 7406-7413. (5) Joseph, Y.; Guse, B.; Yasuda, A.; Vossmeyer, T. Sens. Actuators, B 2004, 98, 188-195. (6) Krasteva, N.; Besnard, I.; Guse, B.; Bauer, R. E.; Mu¨llen, K.; Yasuda, A.; Vossmeyer, T. Nano Lett. 2002, 2, 551-555. (7) Leopold, M. C.; Donkers, R. L.; Georganopoulou, D.; Fisher, M.; Zamborini, F. P.; Murray, R. W. Faraday Discuss. 2004, 125, 63-76. (8) Han, L.; Daniel, D. R.; Maye, M. M.; Zhong, C. J. Anal. Chem. 2001, 73, 4441-4449. (9) Evans, S. D.; Johnson, S. R.; Cheng, Y. L.; Shen, T. J. Mater. Chem. 2000, 10, 183-188. (10) Zhang, H. L.; Evans, S. D.; Henderson, J. I.; Miles, R. E.; Shen, T. Nanotechnology 2002, 13, 439-444.

Networked Gold Nanoparticle Chemiresistors (11) Foos, E. E.; Snow, A. W.; Twigg, M. E.; Ancona, M. G. Chem. Mater. 2002, 14, 2401-2408. (12) Pang, P.; Guo, Z.; Cai, Q. Talanta 2005, 66, 1343-1348. (13) Lu, C.-J.; Tian, W.-C.; Steinecker, W. H.; Guyon, A.; Agah, M.; Oborny, M. C.; Sacks, R. D.; Wise, K. D.; Pang, S. W.; Zellers, E. T. “Functionally integrated MEMS micro gas chromatograph subsystem,” Proceedings of the SeVenth International Conference on Miniaturized Chemical and Biochemical Analysis Systems-µTAS ’03, Squaw Valley, CA, October 5-9, 2003; Transducers Research Foundation, Inc., Cleveland Heights, OH, pp 411-415. (14) Steinecker, W. H.; Rowe, M.; Matzger, A.; Zellers, E. T. IEEE Transducers 2003, 3E44P, 1343-1346. (15) Krasteva, N.; Guse, B.; Besnard, I.; Yasuda, A.; Vossmeyer, T. Sens. Actuators, B 2003, 92, 137-143. (16) Krasteva, N.; Krustev, R.; Yasuda, A.; Vossmeyer, T. Langmuir 2003, 19, 7754-7760. (17) Kaszynski, P.; Friedli, A. C.; Michl, J. J. Am. Chem. Soc. 1992, 114, 601-620. (18) Krasteva, N.; Fogel, Y.; Bauer, R. E.; Mu¨llen, K.; Joseph, Y.; Matsuzawa, N.; Yasuda, A.; Vossmeyer, T. AdV. Funct. Mater. 2007, 17, 881-888.

J. Phys. Chem. C, Vol. 111, No. 34, 2007 12859 (19) Wessels, J. M.; Nothofer, H. G.; Ford, W. E.; von Wrochem, F.; Scholz, F.; Vossmeyer, T.; Schroedter, A.; Weller, H.; Yasuda, A. J. Am. Chem. Soc. 2004, 126, 3349-3356. (20) Bourg, M. C.; Badia, A.; Lennox, R. B. J. Phys. Chem. B 2000, 104, 6562-6567. (21) Maye, M. M.; Luo, J.; Lin, Y.; Engelhard, M. H.; Hepel, M.; Zhong, C. J. Langmuir 2003, 19, 125-131. (22) Castner, D. G. Langmuir 1996, 12, 5083-5086. (23) Cavalleri, O.; Oliveri, O.; Dacca, A.; Parodi, R.; Rolandi, R. Appl. Surf. Sci. 2001, 175-176, 357-362. (24) Freeman, T. L.; Evans, S. D.; Ulman, A. Langmuir 1995, 11, 44114417. (25) Luedtke, W. D.; Landman, U. J. Phys. Chem. 1996, 100, 1332313329. (26) Handbook of Chemistry and PhysicssA Ready-Reference Book of Chemical and Physical Data; CRC Press: Boca Raton, FL, 2000.