Pressure-Induced Restructuring of a Monolayer Film Nanojunction

Different scaling regions were assigned to the change in the contact area, tunneling distance, number of conduction pathways, and structure of the fil...
0 downloads 0 Views 181KB Size
Download PDF
2288

Langmuir 2008, 24, 2288-2293

Pressure-Induced Restructuring of a Monolayer Film Nanojunction Produces Threshold and Power Law Conduction Alexei V. Tivanski,† James K. Li,‡ and Gilbert C. Walker*,‡ Department of Chemistry, UniVersity of Iowa, Iowa City, Iowa 52242, and Department of Chemistry, UniVersity of Toronto, Toronto, Ontario, Canada M5S 3H6 ReceiVed October 18, 2007. In Final Form: December 18, 2007 The electrical conduction of metal-molecule-metal junctions formed between Au-supported self-assembled monolayers of structurally different 1-hexanethiol, 1-decanethiol, and ferrocenyl-1-undecanethiol and a Pt-coated atomic force microscope (AFM) tip has been measured under different compression forces using conducting-probe AFM. The observed junction resistance had two distinct power law scaling changes with the compression force. Different scaling regions were assigned to the change in the contact area, tunneling distance, number of conduction pathways, and structure of the film under compression.

Introduction Recently, charge transport through metal-molecule-metal (m-M-m) junctions has been studied intensively because of their potential applications to molecule-based electronics. In these junctions, a molecular film is typically sandwiched between two metal or semiconductor electrodes. One of the fundamental goals of all electrical conduction measurements in nanojunctions is to understand how structural and electrical properties of nanocontacts and molecules forming junctions influence charge transport through such junctions. The scanning probe microscope (SPM), with the probe serving as one of the metal electrodes, has been commonly used to form and study the electrical and structural properties of nanojunctions directly, with examples including scanning tunneling microscopy (STM)1-6 and conducting probe atomic force microscopy (CP-AFM).6-20 In SPM, the electronic properties of a nanocontact are sensitive to the effects of * Corresponding author. E-mail: [email protected]. † University of Iowa. ‡ University of Toronto. (1) Bumm, L. A.; Arnold, J. J.; Dunbar, T. D.; Allara, D. L.; Weiss, P. S. J. Phys. Chem. B 1999, 103, 8122-8127. (2) Gorman, C. B.; Carroll, R. L.; Fuierer, R. R. Langmuir 2001, 17, 69232930. (3) Xu, B.; Tao, N. J. Science 2003, 301, 1221-1223. (4) Riposan, A.; Liu, G.-Y. J. Phys. Chem. B 2006, 110, 23926-23937. (5) Yokota, Y.; Fukui, K.; Enoki, T.; Hara, M. J. Phys. Chem. C 2007, 111, 7561-7564. (6) Davis, J. J.; Morgan, D. A.; Wrathmell, C. L.; Axford, D. N.; Zhao, J.; Wang, N. J. Mater. Chem. 2005, 15, 2160-2174. (7) Fan, F. F.; Yang, J.; Cai, L.; Price, D. W., Jr.; Dirk, S. M.; Kosynkin, D. V.; Yao, Y.; Rawlett, A. M.; Tour, J. M.; Bard, A. J. J. Am. Chem. Soc. 2002, 124, 5550-5560. (8) Cui, D.; Zarate, X.; Tomfohr, J.; Sankey, O. F.; Primak, A.; Moore, A. L.; Moore, T. A.; Gust, D.; Harris, G.; Lindsay, S. M. Nanotechnology 2002, 13, 5-14. (9) Gomar-Nadal, E.; Ramachandran, G. K.; Chen, F.; Burgin, T.; Rovira, C.; Amabilino, D. B.; Lindsay, S. M. J. Phys. Chem. B 2004, 108, 7213-7218. (10) Leatherman, G.; Durantini, E. N.; Gust, D.; Moore, T. A.; Moore, A. L.; Stone, S.; Zhou, Z.; Rez, P.; Liu, Y. Z.; Lindsay, S. M. J. Phys. Chem. B 1999, 103, 4006-4010. (11) Wold, D. J.; Frisbie, C. D. J. Am. Chem. Soc. 2001, 123, 5549-5556. (12) Lee, T.; Wang, W.; Klemic, J. F.; Zhang, J. J.; Su, J.; Reed, M. A. J. Phys. Chem. B 2004, 108, 8742-8750. (13) Tivanski, A. V.; Bemis, J. E.; Akhremitchev, B. B.; Liu, H.; Walker, G. C. Langmuir 2003, 19, 1929-1934. (14) Tivanski, A. V.; He, Y.; Borguet, E.; Liu, H.; Walker, G. C.; Waldeck, D. H. J. Phys. Chem. B 2005, 109, 5398-5402. (15) Ishida, T.; Mizatani, W.; Aya, Y.; Ogiso, H.; Sasaki, S.; Tokumoto, H. J. Phys. Chem. B 2002, 106, 5886-5892. (16) Suganuma, Y.; Trudeau, P.-E.; Dhirani, A.-A. Phys. ReV. B 2002, 66, 241405-241408. (17) Engelkes, V. B.; Frisbie, C. D. J. Phys. Chem. B 2006, 110, 1001110020.

deformation caused by the interaction force between the probe and the sample,9-15,17 thus it is important to measure force and current simultaneously in order to understand the mechanical and electrical properties of nanojunctions. This can be achieved directly using CP-AFM rather than STM where the contact force is not precisely known because it is not controlled independently. Recent studies9,11-15,17 employed CP-AFM to demonstrate the influence of the contact force on the electrical conduction through SAMs involving alkanethiols and conjugated molecules. These measurements showed two power law scaling changes in junction resistance with applied force. Although one of the power regions was related to the change in the contact area between the probe and compression sample, another was less understood and was assigned to apparent changes in the mechanical properties of the film. Here, we report detailed measurements and quantitative modeling of electrical conduction using CP-AFM for the nanojunction formed between a Pt-coated AFM tip and a Ausupported self-assembled monolayer under different compression forces with the objective being to correlate electrical conduction through m-M-m junctions with the mechanical and structural properties of the film. Specifically, we examine whether the measured dependence of the junction resistance on the compression force can be adequately described by the power law scaling relation originating from the exponential change in the elastic contact area, the tunneling gap between the probe and the sample, and through-space and through-bond hopping of electrons. Although similar studies performed earlier on C8SH-C12SH ordered alkanethiol SAMs8,11,17 showed that the observed scaling changes are nearly independent of the molecular length (apart from the obvious current magnitude change), we have chosen different molecular systems that are not expected to form closely packed and ordered films in the absence of applied force. An important aspect of these measurements is that they were undertaken under an insulating bicyclohexyl solvent, which significantly decreases the capillary forces due to water condensation and, as will be discussed below, prevents the formation of solvent layers at the AFM probe-sample interface. Earlier CP-AFM electrical conduction measurements11,17 performed in air were unable to examine the low stress (which is directly (18) Coffey, D. C.; Reid, O. G.; Rodovsky, D. B.; Bartholomew, G. P.; Ginger, G. S. Nano Lett. 2007, 7, 738-744. (19) McEvoy, T. M.; Long, J. W.; Smith, T. J.; Stevenson, K. J. Langmuir 2006, 22, 4462-4466. (20) Plane`s, J.; Houze´, F.; Chre´tien, P.; Schneegans, O. Appl. Phys. Lett. 2001, 79, 2993-2995.

10.1021/la7032498 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/16/2008

Letters

Langmuir, Vol. 24, No. 6, 2008 2289

proportional to the sum of the adhesion force and the loading force) region in detail because of the ∼10-15 nN adhesion force between the AFM probe and the monolayer due to capillary forces. The presence of a water layer at the interface between the probe and the monolayer could also lead to undesired and complicated conduction-force dependence under applied compression forces. In this work capillary effects are mitigated because adhesion forces are ∼0.5 nN or smaller, therefore allowing one to probe the low-stress region on the nanoscale directly. We explore a significantly lower force regime than has been previously reported for the conductive measurements in air.11,17 These improvements enable quantitative modeling that takes into account the mechanical and structural properties of the film in describing the change in electrical conduction with applied pressure. The molecular SAMs studied here were 1-hexanethiol (C6SH), 1-decanethiol (C10SH), and ferrocenyl-1-undecanethiol (FcC11SH). Molecular systems were chosen for the following reasons: whereas the charge-transport mechanism through these SAMs under relatively small applied biases (within (0.25 V) is expected to be nonresonant tunneling,21 the structure of C6SH assemblies is more disordered than for longer C10SH.22,23 Ferrocenyl-1-undecanethiol was chosen over similar-length dodecanethiol because of the presence of ferrocene end groups that are expected to decrease the contribution from chain-tochain coupling to electron tunneling through the film in the absence of applied force. It is important to mention that the FcC11SH molecules can undergo redox transitions under relatively large absolute biases (∼(1.7 V), and a negative differential resistance in the current-voltage curve can be observed.2,29 Here, significantly smaller applied biases are used, and the chargetransport mechanism through FcC11SH SAM is expected to be nonresonant tunneling. Having different structural properties of the films yet the same charge transport mechanism allowed us to probe how the film structure influences the measured nanojunction electrical conduction under compression. Experimental Details SAM Preparation. All self-assembled monolayers were formed by exposing the freshly prepared Au(111) facet of a single-crystalline bead to 5 mM ferrocenyl-1-undecanethiol (FcC11SH, Dojindo Molecular Technologies, Inc., Gaithersburg, MD) or 1 mM 1-hexanethiol and 1-decanethiol (C6SH and C10SH, Sigma-Aldrich Corp., St. Louis, MO) in purified (by distillation) tetrahydrofuran (THF) with soaking times between 4 and 24 h. After assembly, each sample was rinsed in purified THF and dried in a stream of nitrogen gas. All preparations were performed at room temperature, and all samples were used within 1 day of preparation. Conducting Probe AFM Measurements. The CP-AFM measurements were performed using a commercial contact mode AFM (Molecular Force Probe, Asylum Research, Santa Barbara, CA) modified for conducting probe experiments. To obtain detailed measurements of the current-contact force relationship, different fixed tip biases were applied, and currents through the film were measured (using a pico-ammeter, Chem-Clamp, Dagan Corp., (21) Simmons, J. H. Appl. Phys. 1963, 281, 1793-1803. (22) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am. Chem. Soc. 1990, 112, 4301-4306. (23) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559-3568. (24) Weihs, T. P.; Nawaz, Z.; Jarvis, S. P.; Pethica, J. B. Appl. Phys. Lett. 1991, 59, 3536-3538. (25) Burnham, N. A.; Colton, R. J. J. Vac. Sci. Technol., A 1989, 7, 29062913. (26) Joyce, A.; Thomas, R. C.; Houston, J. E.; Michalske, T. A.; Crooks, R. M. Phys. ReV. Lett. 1992, 68, 2790-2793. (27) Slowinski, K.; Chamberlain, R. V.; Miller, C. J.; Majda, M. J. Am. Chem. Soc. 1997, 119, 11910-11919. (28) Song, H.; Lee, H.; Lee, T. J. Am. Chem. Soc. 2007, 129, 3806-3807. (29) Tivanski, A. V.; Walker, G. C. J. Am. Chem. Soc. 2005, 127, 7647-7653.

Figure 1. Force (-) and absolute current (---) profiles measured simultaneously as a function of vertical piezo displacement for the Pt-coated tip over an FcC11SH SAM. Minneapolis, MN) as a function of the vertical piezo displacement, simultaneously with independent force detection between the tip and the sample. The sample was not scanned in horizontal directions; rather the AFM tip was allowed to drift thermally over the sample surface. Typically, three different tips were used for every SAM sample. Measured currents for different contact forces over different surface locations were averaged over the number of repeated measurements performed with different AFM tips to obtain averaged force-dependent current-voltage (I-V) characteristics of the nanojunction. Results obtained with different tips were similar to within experimental error, especially for contact forces smaller than ∼15 nN. The experimental error is dominated primarily by the uncertainty in the exact number of molecules forming the junction. Because the tip drifts over the surface, variations in the number of contacting molecules may cause fluctuations in the measured current. All experiments reported here were performed either in insulating bicyclohexyl or n-tetradecane solvents (99.0%, Fluka, Switzerland). Cleaned Pt-coated V-shape silicon cantilevers (MikroMash, Estonia) were used in this work where the force constants of each cantilever ranged from 0.25 to 0.6 N/m as determined by the thermal noise method. Cantilevers were cleaned in piranha solution (1:3 30% H2O2/ 98% H2SO4) for 5 min, rinsed in ultrapure water (>18 MΩ‚cm) for 1 min, soaked in hydrofluoric acid for 20 s, and finally rinsed again in distilled water for 1 min, followed by drying under vacuum. (Caution! Piranha solution is a Very strong oxidant and is extremely dangerous to work with. GloVes, goggles, and a face shield should be worn.)

Results and Discussion Typical force and current profiles as a function of vertical piezo displacement for the FcC11SH SAM in bicyclohexyl solvent under a fixed tip bias of -2.8 V are shown in Figure 1 by solid and dashed lines, respectively. Only approach data are shown. Positive piezo displacements imply that the probe was not in nominal contact with the surface, which occurred at 0 nm. Negative displacements reflect the continued motion of the cantilever toward the surface. It is important to mention that vertical piezo displacements do not correspond to the true motion of the tip in contact with surface, a combination of rotation and translation. The zero force was selected as the force when the tip was far away from the surface. Positive forces after the contact point are defined as a loading force with which the cantilever pushes the probe into the surface. Negative forces experienced by the cantilever before surface contact are primarily the attractive electrostatic forces due to the tip-sample capacitance.10,13 The adhesion force between the probe and the sample was defined as a pull-off force measured while the tip was retracting from the surface. The interaction or net force between the tip and underlying film at their contact is considered to be equal to the sum of the two forces: the adhesion force and the loading force. For electrical conduction measurements, bias magnitudes of less than 0.25 V were applied; therefore, the bias-dependent adhesion

2290 Langmuir, Vol. 24, No. 6, 2008

Letters

Figure 2. Force (black line) and current (gray line) profiles measured simultaneously as a function of vertical piezo displacement for the Pt-coated AFM tip over a crystalline gold substrate (no molecular film present) in (a) bicyclohexyl and (b) n-tetradecane solvents under a fixed tip bias of +0.7 V. Only approach data are shown. The maximum current was limited to 20 nA.

force contribution to the interaction force was not significant. However, under higher applied biases this factor would need to be considered. Because of the finite AFM tip-SAM contact resistance, a nonzero loading force was required to make good electrical contact between the conductive probe and sample and cause observable currents of ∼10 pA or larger. The results shown in Figure 1 clearly demonstrate an increase in current with applied loading force and confirm that this type of measurement can indeed be used to measure, with high force sensitivity, loaddependent nanojunction electrical conduction under compression. Control CP-AFM measurements over a single-crystalline gold surface (no molecular film present) in bicyclohexyl solvent showed that sufficient electrical contact was achieved for loading forces smaller than 1 nN (Figure 2a) and current increased steadily until a saturation value of 20 nA was reached. Current and force profiles shown in Figure 2 were obtained under a fixed tip bias of +0.7 V. However, similar experiments in n-tetradecane solvent showed that at least 15 nN of additional loading force was required to cause observable currents of ∼10 pA or larger and the currentforce dependence was significantly more complicated than that in bicyclohexyl solvent (Figure 2b). One of the possible explanations of the conduction dependence in n-tetradecane solvent is the formation of the solvent layers at the interface between the AFM probe and sample. The absence of such a layering effect for the bicyclohexyl solvent is consistent with its branched structure. Therefore, bicyclohexyl was chosen over linear alkane solvent to avoid this highly undesirable conductionforce dependence that can significantly complicate the interpretation of load-dependent nanojunction electrical conduction change under compression. Electrical conduction measurements were performed at different fixed applied tip biases with magnitudes smaller than 0.25 V so that the current through the junction could be measured as a function of the compression interaction force. Measured currents for different interaction forces were averaged over several hundred force plots to obtain average force-dependent current-voltage (I-V) characteristics of the junction. The observed I-V characteristics were linear at all interaction forces less than ∼30 nN within a (0.25 V bias range. The ohmic dependence observed

Figure 3. log-log plots of junction resistance vs interaction force for (a) FcC11SH, (b) C6SH, and (c) C10SH SAMs show the threshold and two power law scaling regimes. Symbols are averaged data, and solid lines are the corresponding linear fit.

in the I-V curves is consistent with the Simmons model in the low-bias region21 for nonresonant tunneling through an m-M-m interface, which was expected because of the large molecular gap between the highest occupied molecular orbital and the lowest unoccupied molecular orbital. Forces higher than 30 nN were not applied because of the noticeably quick decrease in the magnitude of observed currents, indicating the apparent damage of the conductive coating on the tip. The linear portions of the I-V curves were fit by straight lines, and the determined slopes were used to define junction resistances as 1/slope, which, as expected, were changing under different interaction forces. Figure 3 plots the log of the junction resistances in bicyclohexyl solvent for (a) FcC11SH, (b) C6SH, and (c) C10SH SAMs versus the log of the compression interaction force. Each data symbol shown in Figure 3 represents the obtained mean value of resistance for a series of repeated measurements typically with three different tips over the different surface locations. The standard deviation of the resistance found in different measurements is similar to, or smaller than, the size of the symbol. For these three molecular systems, two distinct power law scaling regimes were observed. Both scaling regions can be fit closely by straight lines (solid lines in Figure 3). In the low-load regime, the fitted resistance scales as (force)-0.7, (force)-9.8, and (force)-1.4 whereas at loading forces greater than ∼15 nN it scales as (force)-5.2, (force)-1.24, and (force)-0.7 for FcC11SH, C6SH, and C10SH, respectively. It is important to point out that although the threshold forces were somewhat similar (15 nN for FcC11SH and ∼13 nN for C6SH and C10SH) the slopes were significantly different for these SAMs.

Letters

Among possible factors that can change the observed junction resistances under applied interaction force are (a) changes in the contact area between the AFM tip and the sample, (b) change in the tunneling distance between two electrodes as a result of film compression, (c) intra- and interchain order of the molecular film, and (d) changes in conduction pathways that arise from a-c. First, a quantitative model that includes change in the elastic contact area and conduction pathways through ordered and closepacked SAMs will be introduced in an attempt to explain the different power law scaling regions observed in Figure 3. Changes in the junction resistance can be related to changes in the contact area and film indentation using a Hertzian elastic contact model with the inclusion of adhesion forces between the probe and sample.24 The contact area, a2, between a spherical tip of radius r penetrating into a uniform elastic film may be estimated as

(

I ∝ a2 exp(-βtbdm) +

Fr a ) K

( )

2

(1)

dm cosθ/dcc

ns!

N)1

(ns - N)!N!

∑

×

)

exp(-βtb(dm - Ndcc tan θ))exp(-βtsNdcc) (5) where ns equals the number of available hopping sites and is assumed to be equal to the number of carbon atoms along the molecular chain.27,28 By substituting eqs 1 and 2 into eq 5, the junction resistance R, which is inversely proportional to the tunneling current in the bias range applied here, depends on the interaction force as

R) dm cosθ/dcc

2/3

Langmuir, Vol. 24, No. 6, 2008 2291

∑ N)0

AF-2/3 ns! (ns - N)!N!

exp(-βtb(dm - Ndcc tan θ) - βtsNdcc) (6)

where F is the interaction force and K is an effective modulus equal to (4/3)[(1 - V2t)/Et + (1 - V2s)/Es]-1 (Es, Vs, Et, and Vt are the Young’s modulus and Poisson’s ratio of the sample and the Pt-coated AFM tip, respectively). The Poisson ratio for most materials is between 0.25 and 0.5, so assuming Vt ≈ Vs ≈ 0.33, an effective modulus can be approximated as K ) 1.5EtEs/(Et + Es). The indentation of the film is given by

δ)

a2 F2/3 ) 2/3 1/3 r (K r )

(2)

Although appropriate measured values for the elasticity modulus are not available, one can assume Et ) 170 GPa,25 Es ) 20 GPa26 for the close-packed SAMs and the tip radius of curvature to be 40 nm. The tunneling current can be related to the contact area by

I ∝ a2 exp(-(βd))

(3)

where d is the tunneling gap and β is the tunneling decay coefficient that represents both electron tunneling through-bond (βtb) and chain-to-chain (or through-space, βts) transport pathways. In the case of ordered and close-packed SAMs, most deformation due to film compression increases the tilting of molecules with respect to the substrate normal. Therefore, the change in the molecular tilt angle θ (relative to the substrate normal) with film indentation can be easily expressed as

θ ) cos-1

(

)

d cos θi - δ d

(4)

where θi is the tilt angle in the absence of applied force and the indentation of the film δ as a function of applied force is given by eq 2. To describe the tunneling current through such ordered and close-packed SAMs in terms of both through-bond and through-space tunneling transport, the multiple intermolecular hopping model for molecular tunneling pathways is employed.27,28 According to this model, for the maximum of N possible throughspace hops, the chain-to-chain tunneling distance along the molecular chain with the length dm tilted at an angle θ with the intermolecular distance of dcc decreases by Ndcc tan θ. The total current can then be described as the sum of two contributionss through-bond and through-space tunnelingsby the following equation28

where A is a scaling factor and the dependence of the tilt angle θ on the interaction force is given by eqs 2 and 4. The molecular lengths for molecules C6SH (1.23 nm), C10SH (1.73 nm), and FcC11SH (2.05 nm) were estimated using bond lengths and the van der Waals radii of each headgroup.13,27 The intermolecular distances for C6SH and C10SH SAMs (0.475 nm) were as used by Slowinski et al.,27 and for the FcC11SH SAM an approximate diameter of the ferrocene headgroups of 0.66 nm was used.22 The maximum number of hops N is restricted by the size of the system because the total through-space distance is limited by the molecular chain length. On the basis of the molecular lengths, the maximum number of hops is therefore 1 for the C6SH SAM and 2 for both the C10SH and FcC11SH SAMs. Through-space and through-bond tunneling decay coefficients of 1.31 and 0.91 Å-1 are assumed, respectively.8,12,27,29 The two regions of the resistance versus interaction force data for C6SH, C10SH, and FcC11SH SAMs were fit separately using eq 6 by adjusting three parameters: the scaling factor A, the elasticity modulus of the sample Es, and the initial tilt angle θi. The model follows the data closely only for the lower-force region for the FcC11SH SAM and the higher-force regions for C6SH and C10SH SAMs. The other regions cannot be fitted satisfactorily using this model. (These fits are not shown.) The successful fits are plotted in Figure 4 by solid lines with black lines corresponding to zero through-space hops and gray lines corresponding to a single through-space hop. The low-force region for the FcC11SH SAM and the high-force region for the C6SH SAM can be closely fitted with either 0 or 1 through-space hops, whereas the model with only 0 hops provides a close fit to the high-force region for the C10SH SAM. Within our experimental error, we were not able to distinguish between the 0 and 1 hop fits because both provide an accurate fit to the data for both FcC11SH and C6SH SAMs. From the analysis, the best-fit curves shown in Figure 4 have parameter values of Es ) 16 GPa and θi ) 20° for the FcC11SH SAM, Es ) 65 GPa and θi ) 35° for the C6SH SAM, and Es ) 17 GPa and θi ) 20° for the C10SH SAM. The initial tilt angles determined here agree reasonably well with the literature values.23 The elasticity moduli for the FcC11SH and C10SH SAMs are similar to the elasticity modulus for the close-packed SAMs (Es ) 20 GPa). However, the elasticity modulus for the C6SH deviates significantly from that value and is similar to the Au substrate with EAu ≈ 77 GPa. Whereas the origin of this deviation will be discussed in more detail below, it appears that the underlying

2292 Langmuir, Vol. 24, No. 6, 2008

Letters

Figure 5. Plot of ln(RF2/3) as a function of F2/3 for (a) FcC11SH (•), C6SH (2), and (b) C10SH (×) SAMs with the corresponding linear fit (-).

Figure 4. Plots of junction resistance vs interaction force for FcC11SH (×), C6SH (]), and C10SH (0) SAMs. Solid black and grey lines are fits to eq 6 for 0 and 1 through-space hops, respectively.

Au substrate defines the elastic response of the film to compression in the region of high applied forces. These results demonstrate that a single force region for each monolayer can be described by a model that includes zero hops and these regions can also be described by the multiple intermolecular hopping and increased tilting of molecules due to the monolayer compression for FcC11SH and C6SH SAMs. Because the change in the number of through-space hops cannot adequately describe other force regions, the possibility that the two observed power law scaling regions in the resistance versus interaction force plots shown in Figure 3 originate only from the different number of conduction pathways is excluded. The inability of this model to accurately describe other force regions indicates that the monolayer film compression does not simply tilt the molecules and another type of pressure-induced restructuring must take place. It is therefore necessary to further explore the possibility that force induces structural changes that alter the film stiffness under compression and thereby induce loaddependent conduction. Next, a quantitative model that includes a change in the elastic contact area and tunneling distance will be introduced, together with a consideration of the film structure to explain the different power law scaling regions observed in Figure 3. We assume that both through-space and through-bond tunneling pathways contribute to the conduction, and we will use an effective tunneling decay coefficient of (12 nm-1).27,29 Starting from eq 3, the junction resistance R, which is inversely proportional to the tunneling current in the bias range applied here, depends on

the interaction force as R(F) ≈ F-2/3exp(-SF2/3), with S ) β/(K2/3r1/3). Hence, a plot of ln(RF2/3) versus F2/3 should be linear with a slope of -S. For the tip radius of curvature of 40 nm, the expected slope for the electrical conduction change with the compression force over a closely packed and ordered SAM with an elasticity modulus of Es ) 20 GPa is -S ) -0.4 nN-2/3. It is worthwhile to consider that if the tunneling process through such a film is dominated by the through-bond tunneling then the slope in the plot of ln(RF2/3) versus F2/3 becomes zero and, therefore, the resistance is expected to be directly proportional to the interaction force to the -2/3 power. This conclusion stems from the fact that most of the deformation changes only the tilt angle of the molecules forming a closely packed, ordered SAM and does not change the molecular length. Because the molecular length is equal to the tunneling distance for the through-bond tunneling process, the latter remains unchanged in this case. The tunneling current then is defined by the change in the contact area between the AFM probe and the monolayer, which depends on the interaction force in accordance with eq 1. In terms of the previously described multiple intermolecular hopping model, it corresponds to 0 through-space hops. Figure 5a,b shows a plot of ln(RF2/3) versus F2/3 for FcC11SH (points), C6SH (triangles), and C10SH (crosses) SAMs where data in both scaling regions closely followed a linear dependence, as predicted by the above model. Data in different regions were fit to straight lines, as shown by the solid lines in Figure 5. As will be discussed later, the overall dependence for these films can be separated into three common regions: the first region with a slope of -0.01 nN-2/3 for FcC11SH, the second region with slopes of -0.93, -2.3, and -0.36 nN-2/3 for FcC11SH, C6SH, and C10SH, respectively, and the third region with slopes of -0.12 and ∼0 nN-2/3 for C6SH and C10SH, respectively. The first region in Figure 5 for FcC11SH shows a linear decrease with a slope of almost zero, thus the junction resistance is directly proportional to F-2/3, which represents the case when the contact area change is the dominant contribution to the junction resistance. The presence of Fc end groups in the FcC11SH SAM decreases the chain-to-chain coupling to the electron tunneling through the film (the coverage density of the FcC11SH SAM is roughly 2 times smaller than that for the similar length alkanethiol22), so under relatively small applied forces (less than 15 nN in our case), the through-bond charge transfer is the dominant tunneling mechanism. Again, within the small force regime, the change in the junction resistance appears to be limited by a change in the number of conductive contacts between the AFM tip and the

Letters

sample, which is proportional to the contact area. The slight deviation from zero slope is likely due to experimental error and possibly a small contribution coming from chain-to-chain tunneling. These results are consistent with the multiple intermolecular hopping model described above, where this region can be fit with both 0 and 1 hops corresponding to the throughbond and through-space tunneling mechanisms, respectively. The second region in Figure 5 showed a linear decrease with slopes of -0.93, -2.3, and -0.36 nN-2/3 for FcC11SH, C6SH, and C10SH, respectively. Whereas for the closely packed and ordered C10SH monolayer the observed slope is similar to the one predicted from the above model, slopes for the C6SH and FcC11SH films are ∼5.8 and ∼2.4 times larger than expected. The origin of this deviation can be found in the structural properties of the films. The structure of C6SH assemblies was shown to be significantly more disordered than that of longer chains such as C10SH.23 Such a structure would result from a combination of conformationally disordered and thermally disordered alkyl chains because of the presence of gauche kinks and weak interchain interactions, respectively. Under compression, the film structure becomes more ordered because of the increase in molecular packing, eventually forming closely packed C6SH assemblies at a compression force of around 13 nN, similar to that of the longer chains. Barrena et al.30 studied the change in the molecular packing in SAMs of similar molecular length alkanethiols on Au induced by external pressure. They observed, consistent with our work, changes in film thickness accompanied by a simultaneous change in friction versus applied load that were attributed to a collective transition to denser molecular configurations. In the case of the FcC11SH SAM, its structure is also expected to be more disordered as compared with that of C10SH because of the presence of Fc end groups that prevent the formation of a closely packed monolayer. In terms of the previously described model, a change in the molecular packing under compression forces can be represented as the tip indents through a more compressible film. Assuming the elasticity moduli of the C6SH and FcC11SH samples are Es ) 1.2 and 4.9 GPa, respectively, the observed slopes of -2.3 and -0.95 nN-2/3 can be obtained. In this range of forces, the assumption that film compression only increases the tilting of molecules is not consistent with changes in molecular packing under compression, and explains the inability of the multiple intermolecular hopping model to describe these regions accurately. The third region in Figure 5 showed a linear decrease with slopes of -0.12 and ∼0 nN-2/3 for C6SH and C10SH, respectively. A zero slope for C10SH similar to that in the first region for FcC11SH represents the case when the contact area change is the dominant contribution to the junction resistance and the charge transport is limited by the through-bond tunneling mechanism. This is also consistent with the multiple intermolecular hopping model that fits these two force regions for both molecular films assuming 0 hops. The two force regions for C10SH therefore correspond to the change from combined through-space and through-bond tunneling pathways to a single through-bond pathway. It is important to mention that the determined slope of ∼0 for C10SH is not exact because of somewhat higher variations in the measured resistance values in this region of applied forces as compared with the those in the lower force region. An increase in the number of gauche defects in the alkyl chains under such large compression forces (15-30 nN in this case) is likely the origin of these variations. Such drastic pressure-induced structural (30) Barrena, E.; Ocal, C.; Salmeron, M. J. Chem. Phys. 2000, 113, 24132418.

Langmuir, Vol. 24, No. 6, 2008 2293

changes in the alkyl chains decrease chain-to-chain coupling and lead to the predominantly through-bond charge transport mechanism. The observed slope of -0.12 nN-2/3 in the third region for C6SH is significantly smaller than the one predicted by the elastic compression model over the closely packed, compressed monolayer film presented above. The observed deviation in this range of applied forces (>∼13 nN) likely lies in the change in the mechanical properties of the compressed films. For these films, ∼13 nN corresponds to a film stress of ∼1-1.5 GPa (calculated from the force with the contact area estimated using eq 1. Others have observed significant changes in the elastic properties of alkanethiol SAMs at a stress of ∼1 GPa,8,30 which supports our hypothesis. It appears that in this force range the elasticity modulus of C6SH becomes similar or greater than that for the underlying Au substrate and the substrate dominates the elastic properties of the film. Assuming an elasticity modulus of the C6SH-Au nanostructure that is similar to that of the Au substrate with Es ≈ 77 GPa, the above model predicts a slope of -0.19 nN-2/3, which is reasonably close to the -0.12 nN-2/3 observed here. The significantly higher elasticity modulus obtained here is also in agreement with the multiple intermolecular hopping model, where the elasticity modulus of 65 GPa was determined. We note that unlike the C6SH and C10SH SAMs the FcC11SH sample shows a discrete transition between two scaling regions occurring at around 15 nN. We argue that the origin of the discrete change between the first and second regions for the FcC11SH SAM is due to significant structural reorganizations of ferrocene end groups. The absence of such a discrete change for the C6SH SAM is consistent with different structural properties of the films, as discussed previously. Additional experimental studies, beyond the scope of the present work, would be necessary to determine the origin of this discrete change. To underline the importance of the size of a nanocontact, we note a comparison with work from Joyce et al.,26 where a much larger tip radius of curvature (∼300 nm) was used than here (∼40 nm). This difference leads to a several orders of magnitude larger contact area between the tip and the monolayer and the number of molecules forming the junction. The corresponding stresses ranged from 0 to ∼100 MPa in this earlier work, which is significantly narrower than the range explored here. The fact that no distinct scaling changes were observed there suggests that a significantly larger contact area can hide nanoscopic properties of the film. In summary, electrical conduction through molecular m-M-m junctions under compression was investigated using CP-AFM. Observed junction resistances showed two distinct power law scaling changes with the compression interaction force for 1-hexanethiol, 1-decanethiol, and ferrocenyl-1-undecanethiol SAMs. These regions were quantitatively described using two analytical models for the change in contact area, tunneling gap between the AFM tip and the sample, number of conduction pathways, and different molecular structures of the films. Acknowledgment. We gratefully acknowledge financial support from NSF (CHE-0404579), ARO (W911NF-04-100191), ONR (N00014-05-1-0765), NSERC (312497), and CRC (202483). J.K.L. acknowledges an OGSST scholarship. LA7032498