Odd-Even Effects in Charge Transport across n-Alkanethiolate-Based SAMs

Mostafa Baghbanzadeh,a Felice C. Simeone,a Carleen M. Bowers,a Kung-Ching Liao,a Martin M. Thuo,a Mahdi Baghbanzadeh,b Michael Miller,c Tricia Breen Carmichael,c and George M. Whitesidesa,d*

a

Department of Chemistry and Chemical Biology, Harvard University,

12 Oxford Street, Cambridge, MA 02138, United States
b c

Department of Statistics, Shiraz University, Shiraz, Iran

Department of Chemistry and Biochemistry, University of Windsor, 401 Sunset Avenue,

Windsor, Ontario, Canada, N9B 3P4
d

Wyss Institute of Biologically Inspired Engineering, 60 Oxford Street, Cambridge, MA 02138,

United States

* To whom correspondence may be addressed: gwhitesides@gmwgroup.harvard.edu

1

ABSTRACT This paper compares rates of charge transport across self-assembled monolayers (SAMs) of nalkanethiolates having odd and even numbers of carbon atoms (nodd and neven) using junctions with the structure of MTS/SAM//Ga2O3/EGaIn (M = Au or Ag). Measurements of current density, J(V), across SAMs of n-alkanethiolates on AuTS and AgTS demonstrated a statistically significant odd-even effect on AuTS, but not on AgTS (that could be detected using this technique). Statistical analysis showed the values of tunneling current density across SAMs of n-alkanethiolates on AuTS with nodd and neven belong to two separate data sets, and while there is a significant difference between the values of injection current density, J0, for these two series (log|J0Au,even| = 4.0±0.3 and log|J0Au,odd| = 4.5±0.3), the values of tunneling decay constant, β, for nodd and neven alkyl chains are indistinguishable (βAu,even = 0.73±0.02 Å-1, and βAu,odd= 0.74±0.02 Å-1). A comparison of electrical characteristics across junctions of n-alkanethiolate SAMs on gold and silver electrodes yields indistinguishable values of β and J0 and indicates that a change that substantially alters the tilt angle of the alkyl chain (and, therefore, the thickness of the SAM) has no influence on the injection current density across SAMs of n-alkanethiolates.

Keywords: Odd-even effect; tunneling; interface; self-assembled monolayer; EGaIn

2

INTRODUCTION Understanding charge tunneling through thin molecular films is a current focus of nanoscience.1-7 Using self-assembled monolayer (SAM)-based large-area junctions with the structure AgTS/S(CH2)nT//Ga2O3/EGaIn (one type of so-called “EGaIn junction”),8,9 we were able to study the influence of the changes in the structure of the SAM on the tunneling current density.10-14 (Here AgTS is a template-stripped silver substrate; S(CH2)n is the n-alkanethiolate SAM covalently bonded to the AgTS substrate, T is the terminal functional group; and EGaIn is the liquid alloy of gallium and indium, covered with a thin, electrically conductive surface oxide—mostly Ga2O3—that forms spontaneously upon exposure to air.15,16) Those studies showed that alterations in the structure of the insulating organic layer, S(CH2)nT,‒‒for example, changing the anchoring group of the SAMs10 or introducing polar organic groups either into the backbone of a polyethylene chain11 or at the van der Waals T//Ga2O3 interface12-14‒‒do not influence the rates of charge transport at a level that is statistically significant (less than a factor of three). In surprising contrast, we and others observed that the addition of one CH2 group to the alkyl chain‒a change that, in essence, simply interchanges an exposed terminal methyl group for an ethyl group‒appears to be sufficient to make a small but statistically significant difference in the electronic properties of the junction.17-20 This phenomenon is one manifestation of the socalled odd-even effect,21 and its origin is a continuing puzzle in understanding SAM-based tunneling junctions. In this work, we focused on these specific questions about the odd-even effect: i) Is there indeed‒‒when examined using current technology‒‒a systemic difference in the rate of charge transport across SAMs of n-alkanethiolates (SCn) with an odd number of carbons (nodd) and SAMs with an even number of carbons (neven)? (Much of the historical evidence for odd-even

3

effects in charge transport has come from earlier and less accurate studies,17 and from studies of phenomena such as wetting, which are of unknown relevance to quantum tunneling.21) ii) If there is an odd-even effect, then, mechanistically, how can such a small change in the structure of a SAM influence tunneling probabilities? iii) How does changing the substrate for the SAM from Au to Ag—a change that also changes the tilt angle of n-alkanethiolates as well as properties, such as wetting and work function of electrode—influence tunneling current density across SAM-based large-area junctions? There are three plausible explanations for the observed odd-even effect in the rates of charge transport: i) Changes in the nature of the van der Waals interface between the SAM and the top electrode—changes that are presumed to be responsible for the odd-even effect in wetting21—might also influence the ability of the SAM and the Ga2O3 to “wet” and conform to one another, or to condense a water film, or to adsorb adventitious contaminants at the interface. ii) Differences in rates of charge transport might reflect the differences in the orientation of the terminal group at the interface of the SAM with the Ga2O3 (e.g., CH2CH3 vs. CH3, Figure S2), which might influence the strength of electronic coupling between the HOMOs of the n-alkyl groups and the Ga2O3. iii) There might be a systematic difference in the thickness of the SAM between odd and even n-alkanethiolates. (The tunneling path might have a through-space component and thus be sensitive to the thickness, independently of the length of the n-alkyl chain.) A fourth (in principle) possibility—difference in the electric dipole moment along the surface normal, which would be induced by the odd-even differences in the orientation of the terminal ethyl (‒CH2CH3) group—seems unlikely to be important since the difference is undoubtedly small, and introducing even large dipoles into the interface has little or no effect.12,13

4

In our previous studies, although we were able to identify an odd-even effect in tunneling current density, J (A/cm2) at ±0.5 V, (i.e., the results from SAMs with neven and nodd belonged to separate data sets and demanded separate fits to eq. 1) we were not able—because the logstandard deviation (σlog) in those studies was large (σlog = 0.6-1.4) relative to those we now generate (σlog = 0.3-0.6)—to determine the mechanism of the odd-even effect, or to establish whether there was a statistically significant difference in the two parameters, β, and, J0, of the simplified Simmons equation (eq. 1).17 In Eq. 1, J(V) is current density (current divided by the geometrical contact area; A/cm2) at an applied voltage V.22,23 J(V) = J0(V)e-βd = J010-βd/2.303 In this study, as in prior work, we have focused on trends in current densities as a function of the number of methylene (-CH2-) units in SAMs of structure M(Au or Ag)/SCn//Ga2O3/EGaIn, rather than on absolute values of these measurements. We have described details of the procedure we used previously.15 Here we used so-called “flattened” EGaIn tips, V = ±0.5V and template-stripped (TS) Ag and Au substrates (Figure 1).8,9,15 (Our experiments, however, showed that the J value is independent of the type of the tip—selected14 or flattened15—but using flattened tips yields in data with slightly narrower distribution). These modifications in procedure generate data with smaller standard deviations, and over larger range of data, than those data generated using unflattened/unselected tips. Surface of Au and Ag have similar work functions (ФAu = 4.9 eV and ФAg = 4.4 eV).24 The tilt angle of the backbone from the surface normal for n-alkanethiolate monolayers on Au(111) (~30°) is larger than that on Ag(111) (~10°) (Figure S2).25 Accordingly, comparing the (1)

5

Figure 1. Sch F hematic struc cture of a AuTS/SAM//G 2O3/EGaIn junction. H u Ga n Here, d repres sents th length of the tunneling barrier established by the alkyl chains in the S he SAM. Assum ming a through-bon mechanis d coincid with the length of th alkyl chain J0 is defin as nd sm, des he n. ned th current de he ensity across a hypotheti junction where d = 0 s ical n 0.

6

rate of charge transport across the n-alkanethiolate SAMs on Au and Ag substrates makes it possible to distinguish several characteristics of a SAM, such as the tilt angle (and thus the relative thickness of the SAM along the surface normal) and the SAM packing density (21.7 Å2/molecule on Au and 19.1 Å2/molecule on Ag).26 In particular, we hoped to answer two questions in this comparative study: i) Does the charge move along the shortest distance between the two electrodes (a through-space path)? or ii) does the charge follow the molecular orbitals along the backbone regardless of the tilt of the n-alkyl group (a through-bond path)?27,28 Density of chains, tilt angle, and the properties of the interface (T//Ga2O3, T = —CH3, —CH2CH3) are obviously convoluted, and no simple comparison of SAMs on Au and Ag will distinguish them all unambiguously; our results do, nonetheless, both allow some possibilities to be excluded, and help to frame questions for future study. We observed a statistically significant odd-even effect in the rates of tunneling across SAMs of n-alkanethiolate on AuTS, where n-alkanethiolates with nodd gave higher current densities (J(V)) than did SAMs of n-alkanethiolate with neven. We show that the tunneling current densities across SAMs of n-alkanethiolates with nodd and neven on AuTS substrate fit in separate data sets, when analyzed using the simplified Simmons model (eq. 1); that is, analyzing the two series separately led to more consistent interpretations of the results than analyzing them together. In contrast, we notice that charge tunneling across SAMs of n-alkanethiolate on AgTS does not show an odd-even effect that we can detect in either  or J0. Background Odd/Even effects in tunneling across SAMs. The odd-even effect in SAMs has been observed (and reviewed21) in a number of different physical properties, such as wetting by liquids,29-32 molecular packing,33 and electronic,19 frictional ,34 and electrochemical properties.35

7

Odd-even effects have been studied in the surface treatment of dielectric materials for pentacenebased field-effect transistors,36 and in the rate of charge transport across carbon-chain-based molecular devices.20 Odd-even alternation in rectification of tunneling current density has also been reported.18 Using EGaIn-based top electrodes in molecular junctions, we have studied oddeven effects in charge transport across n-alkanethiolate SAMs on AgTS substrates.17,11 Using statistical tools, we showed that the data for current densities for SAMs with neven and nodd belong to separate data sets.37 Theory of charge tunneling across SAMs. SAMs of alkanes are considered to be good electrical insulators for current transport at low bias (we use ±0.5 V), and charge transport takes place by hole tunneling.38,39 When such molecules are placed between two electrodes, the charge transport has usually been interpreted ( in the absence of a more detailed model) in terms of a rectangular barrier model using the Eq. 1.22 The injection current density at an applied voltage V, J0(V), gives, ideally, the current flowing across a hypothetical junction in which the SAM has zero (d = 0) thickness, but the characteristic of the interfaces between the SAM and the electrodes remain the same as those when there is a SAM. The tunneling decay constant β (Å-1) contains, according to the Simmons model, information about the shape of the tunneling barrier, which, in principle, is determined by the molecular/electronic structure of the backbone in the SAM. The width of the barrier across which charges traverse is d (Å), and is usually assumed to be determined by the molecular structure, not the shortest path between the two electrodes, but this assumption is not based either on solid empirical or theoretical grounds.

8

RESULTS AND DISCUSSION We collected J(V) using “flattened” conical tips. The procedures for formation and use of these flattened tips are described in detail elsewhere.15 (See the Supporting Information for details of experimental design.) Charge transport across n-alkanethiolates on AgTS and AuTS substrates. Values of J are approximately log-normally distributed. For each SAM, we could fit, by nonlinear leastsquares fitting,37 Gaussian curves to the distributions of log|J|. Figure S3 (in Supporting Information) shows the histograms for the values of log|J| (J, Acm-2) for SAMs of nalkanethiolates on AgTS. Figure S4 shows analogous data for SAMs formed on AuTS substrates. Table S1 summarizes the values of log|J| at V = +0.5 V for SAMs of n-alkanethiolates (SCn) on AgTS and AuTS with both odd and/or even number of carbons (nodd and neven). Table 1 summarizes the values for injection current density, log|J0|, and tunneling decay coefficient, β, for the n-alkanethiolates with neven and nodd on both Ag and Au substrates. Figure 2 shows plots of linear regression of log|J| versus molecular length (number of carbons) for SAMs of nalkanethiolate on Ag (Figure 2a) and Au (Figure 2b) with neven and nodd. Figure S5 shows plots of log|J|, at V = +0.5V, versus d(Å) for different lengths of n-alkanethiolates on both AgTS and AuTS substrates. We calculated the molecular length from the sulfur atom to the distal hydrogen atom of the final methyl group in contact with Ga2O3 (Figure 1). For SAMs on AuTS, junctions comprising SAMs of n-alkanethiolates with nodd consistently showed higher values of current densities than their analogs with neven (Figures 2b and 3). Based on a careful statistical analysis, we concluded, with 95% probability, that there is a significant difference between two regression lines (Figure 2b). We determined that the

9

Table 1. Values for injection current density, log|J0|, and tunneling decay coefficient, β, for the n-alkanethiolates with neven and nodd on both AgTS and AuTS substrate.
a

Length (Å)b log|J0| AgTS/even AgTS/odd AuTS/even AuTS/odd 3.7±0.3 3.6±0.3 4.0±0.3 4.5±0.3 β (Å-1) 0.73±0.02 0.71±0.02 0.73±0.02 0.74±0.02

Length (nC)c log|J0| 3.4±0.3 3.3±0.3 3.6±0.3 4.1±0.3 β (nC-1) 0.93±0.02 0.91±0.02 0.93±0.02 0.94±0.02

M /odd or even

a

The metal substrate. bMolecular length from the sulfur atom to the distal hydrogen atom of the

final methyl group in contact with Ga2O3 (Figure 1). cLength based on the number of carbons of SCn.

10

Figure 2. log|J| at V = +0.5V ver 2 rsus number of carbons f SAMs of n-alkanethi r for f iolates (SCn) on a) AgTS and b) AuTS. Solid line (f n-alkanethiolate SAM with neven) and dashe line (for n a for Ms ed nn alkanethi iolate SAMs with nodd) represent the linear regre s r e ession analys which gi ses ives β/2.303 (slopes) and log|J0| (i a intercepts at d = 0). Erro bars repres or sent the stan ndard deviati for the lo ion og|J| values (0 0.1≥σlog≥ 0.6 6).

11

difference in the values of the intercepts (log|J0|) of the two lines is statistically significant, but the slopes (β) are indistinguishable (for details of statistical analysis; see Supporting Information). We did not, however, observe any significant difference in the trends of log|J| (i.e., an odd-even alternation) for SAMs of n-alkanethiolates with neven or nodd on AgTS (Figure 2a). Statistical analysis, similar to what we have done for SAMs on AuTS, for SAMs of nalkanethiolates with neven or nodd on AgTS showed that the two lines are indistinguishable and values of the injection current density, log|J0|, and of tunneling decay coefficient, β, for the nalkanethiolates with neven were indistinguishable from log|J0|, and β for n-alkanethiolates with nodd. For SAMs on AuTS, values of log|J| of most of the n-alkanethiolate SAMs with neven (SCn) were within 0.1 of the values of log|J| of their next homolog with nodd (SCn+1) (Figure 3a). Figure 3b shows that the Δlog|J| (Δlog|J| = log|Jobserved| - log|Jfitted|) for SAMs with nodd are constantly ≤ 0 while Δlog|J| for SAMs with neven are constantly ≥ 0. To examine the statistical difference between the tunneling current density values, log|J|, of SAMs of n-alkanethiolates on AuTS we calculated the 95% confidence intervals (CI) for the whole set of log|J| values (for details see SI). Error bars in Figure 3 show the 95% CI for the log|J| values on AuTS substrate (i.e., the smaller the overlap of bars, or the larger the gap between bars, the stronger the evidence for a true difference).40 Figure 3b shows an interesting phenomenon; that is, although the odd-even effect alternation can be observed for the whole series of SCn (n = 5-18), it is only statistically significant for 10≤n≤14. We rationalize this difference as a reflection of the conclusion—drawn largely from spectroscopic studies41—that SAMs with longer SAMs (n ≥ 10) are crystalline and well ordered, while those with 9 ≥ n are disordered, probably because of weak interactions between the shorter alkyl chains. The SAMs

12

with long chain—n ≥ 15—may be less ordered and have chains with gauche configurations (similar observations have been reported for some other types of odd-even effect18,42). Our data are consistent with suggestions that the magnitude of odd-even effect depends on the degree of order of the SAM. One of the factors that, potentially, can influence the quality of the SAM is the incubation time (i.e., the time that metal substrate stayed in a solution of the thiols (1 mM in toluene)). To examine the effect of incubation time, we prepared SAMs of SC10 and SC11 on both AgTS and AuTS substrates with incubations of 3 h, 1 day, 2 day, and 7 day. The rate of charge tunneling for SAMs formed over 3 h is not distinguishable from the 7 day sample (Table S2). Comparison between Au and Ag. To examine the influence of geometry of SAMs on tunneling transport, we compared J(V) of n-alkanethiolates on AuTS and AgTS. Figure 4 shows plots of log|J| at V= +0.5 versus molecular length (in number of carbons) for n-alkanethiolate SAMs with neven and nodd on both AuTS and AgTS. Statistical analysis, similar to that which we have carried out for the odd-even effect, showed that SAMs with neven on both AuTS and AgTS result in indistinguishable values of log|J0|and β (Table 1, Figure 4a). When, however, we compared the tunneling current density between SAMs with nodd on AgTS and AuTS, we observed a significant difference in values of log|J0|, while the values of β were indistingushable (Table 1, Figure 4b). These comprisons lead to two important conclusions: i) They indicate that replacing Au/SR with Ag/SR (for SAMs with neven)—a change that substantially alters the tilt angle of the alkyl chain (and, therefore, the thickness of the SAM)29— had no influence (to the precision that we can measure) on the injection current density across these SAMs (Figure 4a).

13

TS Figure 3. Odd-eve effect in charge transp across n en c port n-alkanethio olates on AuT . a) Plot of f

log|J| at V = +0.5V versus num V mber of carb bons for SAM of n-alka Ms anethiolates o AuTS ( SCn, on C n=5-18). The solid line represents the line regression analysis to all the data Error bars d ear n o a. sent J| D (Δlog|J|) betw ween the obs served curre ent repres 95% CI for the log|J values. b) Difference ( density (log|Jobserved|) and the calculated cu c urrent densit (log|Jfitted|) based on th regression line ty ) he n (Δlog|J| = log|Jobse | - log|Jfittted|). Error bars represen standard d nt deviation (da ashed) and 95 5% erved CI (so olid) for the log|J| values. l

14

ii) The odd-even effect in charge tunneling across the SAMs of n-alkanethiolates is related to the orientation of the terminal methyl group. This study showed that SAMs of nalkanethiolates with neven on AuTS behave, in EGaIn junctions, similarly to the SAMs with neven and nodd on AgTS, and that only SAMs with nodd on AuTS conduct differently (Figure 4 and Table 1). The significant structural difference between SAMs with nodd on AuTS with the other SAMs studied in this comparison is the orientation of terminal ethyl group: for SAMs with neven on AuTS, and SAMs with neven and nodd on AgTS, the surface consists predominantly of methyl groups. The terminal CH2-CH3 moiety of the SAMs with nodd on AuTS is tilted away from the surface normal, producing a surface that is composed of a mixture of methyl and methylene groups (that is, an “ethyl” rather than “methyl” surface) (Figure S2). The same behavior was reported in wetting properties of SAMs on Ag and Au; SAMs of n-alkanethiolate on Ag showed no29 or slight42 odd-even effect; the effect was more distinguishable on Au.43 CONCLUSION The odd-even effect in charge transport by tunneling is real. Our statistical analysis for SAMs of n-alkanethiolates on AuTS showed i) the values of tunneling current density across SAMs of n-alkanethiolates with nodd and neven belong to two separate data sets, and ii) while there is a significant difference between the values of J0 for these two series, the values of β for nodd and neven alkyl chains were indistinguishable. We conclude that there is an odd-even effect in tunneling transport across SAMs on AuTS; an odd-even effect, however, was not observable (at the precision that we could measure) for the series on AgTS.

15

Figure 4. log|J| at V = +0.5V ver 4 rsus number of carbons f SAMs of n-alkanethiolates a) wi an r for f ith
S even num mber of carbo on AuTS and AgTS an b) with an odd numbe of carbon on AuTS an ons nd n er ns nd

AgTS. Solid line (for n-alkanethio olate SAMs on AgTS) an dashed lin (for n-alka nd ne anethiolate nt a ich .303 (slopes) and log|J0| SAMs AuTS) represen the linear regression analyses whi give β/2. (intercepts at d = 0). Error bars re epresent the standard dev viation for th log|J| values. he

16

The geometry of the n-alkyl chain determines the odd-even effect. The contrast of junction measurements between SAMs of n-alkanethiolates on gold (where we observed an oddeven effect) and on silver (where we did not observe a significant trend) indicates that the oddeven effect in tunneling transport corresponds to the composition (“methyl” vs. “ethyl”) of the surface of the SAM (Figure S2). The tunneling path is “along single molecule” rather than “across multiple molecules”. Our results demonstrate that although a SAM on gold is thinner than one on silver, the values of J0 and β for SAMs of n-alkanethiolates with neven on AuTS and AgTS are statistically indistinguishable (Figure 4a). We suggest—based on this observation—that charge carriers travel through the backbone of the alkyl chain in the SAM, rather than across the shortest (and thus across multiple molecules of n-alkanethiolate) distance between the two electrodes. The SAM//Ga2O3 interface is responsible for the odd-even effect on AuTS. Significant differences in values of injection current densities, J0, and indistinguishable values of tunneling decay, β, for SAMs of n-alkanethiolates with nodd and neven imply that the odd-even effect is influenced by some property of the interfaces (e.g., Au/S or SAM//Ga2O3). We suggest that a determining contribution of the SAM//Ga2O3 interface to odd-even effects is more plausible than one for the Au/S interface for two reasons: i) Our previous studies,10,44 and current study (Figure 4a), showed that changing the bottom interface does not significantly influence the tunneling current densities. ii) Results of this study indicate that changing the terminal group from CH3 to CH2CH3 correlates with a change in the injection current density (Figures 2b and 4b). The origin of the odd-even effect is not completely defined by currently available data. While we were able to identify the odd-even effect with respect to J, β, and J0, between nalkanethiols with nodd and neven, we cannot pinpoint the origin of the observed odd-even effect on

17

AuTS. We can, however, identify and assess four possibilities (although—based on our current and previous studies—some possibilities are less plausible than others). (i) Differences in the Orientation of the Terminal (T) Group at the Ga2O3 Interface

(e.g., CH2CH3 vs. CH3). A larger exposure of terminal CH2CH3 group at the surface of the nalkanethiolate SAMs with nodd results in larger van der Waals interactions (by changing the polarizability of surface and increasing the area of contact) between the monolayer and the EGaIn tip, and may consequently give rise to a difference in J(V) (similar to wetting properties of SAMs on Au21). If the orientation of terminal CH2−CH3 bond is responsible for both the oddeven effect in charge transport, and the odd-even effect in wetting, then increasing the van der Waals interactions as a result of different surface polarization is a possible origin of the former. This hypothesis—that the odd-even effect in charge tunneling across SAMs terminating in ethyl and methyl groups reflects polarizability and van der Waals interactions—is not immediately easily reconciled with the fact that introducing groups such as halogens or aromatics into this interface does not produce large differences in J0 in our experiments.12,13 (Cahen and others have reported larger differences with specific terminal halogen substituents when using a mercury top electrode.45) (ii) Systematic Alternations in Energy of the Frontier Orbital Energies (i.e.

HOMO/LUMO levels) of Odd and Even n-Alkanethiolates. Difference in rates of charge transport might reflect the difference in the orientation of the terminal group at the SAM interface between the Ga2O3 (CH2CH3 vs. CH3, Figure S2) which might influence the strength of electronic coupling between the HOMOs of the n-alkyl groups and the Ga2O3. The HOMO in these SAMs is, however, concentrated at the Au-S interface, with little amplitude at the SAM/Ga2O3 interface.46

18

(iii)

Systematic Differences in the Thickness of the SAM between Odd and Even n-

Alkanethiolates. Naively, it seemed that SAMs on gold with nodd might be thinner than SAMs with neven. Our calculation of thicknesses of odd and even SAMs suggests that there is no significant (<0.5 Å) systematic alternation in the thickness between nodd and neven. Figure S6b (in the SI) shows an essentially monotonic increase in the thickness for n-alkanethiolate monolayers with a tilt angle of 30° on AuTS. These calculations indicate that there is no significant alternation in the thicknesses of the SAMs and thus that thickness does not correlate with the odd-even alternation in J(V) (Figure S7). In addition, by comparing the rate of tunneling across SAMs of n-alkanethiolates on Au and Ag, we have shown empirically that charge transport is insensitive to the thickness of the SAMs (as measured along the perpendicular to the mean plane of the surface of the metal.). We conclude that there is no evidence to support a contribution of the thickness of the SAM to the presence of odd-even effect. (iv) Difference in the Dipole along the Surface Normal, which might Correlate with

Odd-Even Differences in the Orientation of the Terminal CH3 Group (Figure S2). Different orientations of the terminal methyl group in nodd and neven could cause an odd-even alternation in the dipole along the surface normal,47 which might contribute to an odd-even effect through an influence on the work function of the surface.19,47 Although we have no direct evidence to prove or exclude this hypothesis, our previous studies12,13 showed that introducing even a with a large dipole moment at the SAM//Ga2O3 interface does not change the tunneling barrier enough to influence the rate of charge transport; thus, we think this very small difference in dipole along surface normal seems unlikely to be important. To be proved or excluded, this hypothesis, however, needs more investigation (e.g., using a series of odd and even SAMs having terminal

19

polar groups such as CF3 or OCH3 that introduce a dipole at the interface but do not change the topography of the SAM). If there is an observable odd/even effect for n-alkanes, why was there no detectable difference in much larger changes in terminal groups for SAMs? This paper confirms that there is an odd/even effect for n-alkanethiolates, but that it is small (a factor of ~3 when appropriately compared, or ΔlogJ0 ~ 0.5), and only reliably detectable given the availability of a substantial number of structurally related compounds, with measurements taken carefully, and in parallel. Previous papers have shown that a range of groups with quite different polarities, degrees of saturation, composition, and polarizability have a surprisingly small influence on tunneling current densities.13 These two sets of observations seem at odds. The problem in comparing these data is that they are not, strictly speaking, comparable. Most of our studies of terminal groups did show variability, but the variation (which might have been a factor of 10 across of series of groups) was not designed to detect differences as small as those characterizing the odd-even effect, and would require comparing groups known to have different sizes and shapes. In those studies, we compared the current densities of compounds having polar terminal groups (R) with those of hypothetical n-alkanethiolates of the same length. If those comparisons demonstrate that the difference in log|J| (Δlog|J| = log|Jpolar| − log| JCH3|) was ≤ 0.5, or less than a factor of three in |J|, we considered the compound with polar terminal group and n-alkanethiol to be indistinguishable. In the current study, however, we did not compare two single data points, but studied a series of compounds (S(CH2)nH, n = 5-18) and analyzed the results based on eq. 1. Although the Δlog|J| for each adjacent pair of nalkanethiolate was ≤ 0.4 (Figure 3b), analyzing the two series separately led to more consistent

20

interpretations of the results than did analyzing them together (neven and nodd produce two statistically distinguishable sets of data). The surprise in these studies was thus that substituting, for example, a terminal methyl group for an amide, or a terminal cyclohexyl group with a phenyl, produced a change in current density (J(V)) of approximately a factor of 4.0 (∆log|J(0.5V)| = 0.6),12 and that was smaller than we had—perhaps naively— expected, while substituting a terminal methyl for a terminal ethyl produced a change of a factor of 2.5 (∆log|J(0.5V)| = 0.4), which was larger than we had expected. The precision of the former studies was, however, substantially less than the latter, and the studies of polar and aromatic groups, as they were designed, would not have been able to detect the subtle differences characterizing the odd/even effect. Differences in size, composition, and other factors restricted our ability to make meaningful comparisons of current densities with ∆log|J(0.5V)| ≤ 0.6. So, what is the origin of the odd/even effect? Although this paper cannot answer the question unambiguously, it has eliminated a number of possibilities (differences in the thickness of the SAM, certain differences in the HOMOs for the SAMs), and leaves a focus on the contribution of the non-covalent interface between the SAM and the Ga2O3 film on the EGaIn electrode. This interface has been characterized empirically by studies such as those described in this paper, but have not yet been included in relevant theoretical calculations. In particular, the height of the barrier to tunneling is probably high in the region where there is only physical contact (relative to the rest of the SAM), but the width of this region is small. Understanding the mechanism of charge transport across this interface, and its contribution to overall tunneling currents, remains an unresolved problem with this junction (and, in fact, with all junctions involving SAMs as insulating layers between two conducting electrodes).

21

ASSOCIATED CONTENT Supporting Information. The experimental procedures, synthesis characterization of the alkanethiols, histograms of J(V) data, and statistical analysis details. This material is available free of charge via the Internet at http://pubs.acs.org. AUTHOR INFORMAIONS Corresponding Author *E-mail: gwhitesides@gmwgroup.harvard.edu Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS The work at Harvard University was supported by a subcontract from Northwestern University from the United States Department of Energy (DOE, DE-SC0000989). The DOE grant from Northwestern also supported the salary for M.B., C.M.B, and K.-C.L. Aspects of this work (SAM preparation and characterization) was supported by the US Department of Energy, Division of Materials Sciences, under Award No. DE-FG02-00ER45852. REFERENCES (1) (2) (3) (4) (5) 2010, 22, 140. (6) McCreery, R. L. Chem. Mater. 2004, 16, 4477. 22 Bergfield, J. P.; Ratner, M. A. Phys. Status Solidi B 2013, 250, 2249. McCreery, R. L.; Bergren, A. J. Adv. Mater. 2009, 21, 4303. Salomon, A.; Cahen, D.; Lindsay, S.; Tomfohr, J.; Engelkes, V. B.; Frisbie, C. D. Wang, W. Y.; Lee, T.; Reed, M. A. Rep. Prog. Phys. 2005, 68, 523. Vilan, A.; Yaffe, O.; Biller, A.; Salomon, A.; Kahn, A.; Cahen, D. Adv. Mater.

Adv. Mater. 2003, 15, 1881.

(7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24)

Coskun, A.; Spruell, J. M.; Barin, G.; Dichtel, W. R.; Flood, A. H.; Botros, Y. Y.; Chiechi, R. C.; Weiss, E. A.; Dickey, M. D.; Whitesides, G. M. Angew. Chem. Dickey, M. D.; Chiechi, R. C.; Larsen, R. J.; Weiss, E. A.; Weitz, D. A.; Liao, K.-C.; Yoon, H. J.; Bowers, C. M.; Simeone, F. C.; Whitesides, G. M. Thuo, M. M.; Reus, W. F.; Simeone, F. C.; Kim, C.; Schulz, M. D.; Yoon, H. J.; Yoon, H. J.; Shapiro, N. D.; Park, K. M.; Thuo, M. M.; Soh, S.; Whitesides, G. Yoon, H. J.; Bowers, C. M.; Baghbanzadeh, M.; Whitesides, G. M. J. Am. Chem. Bowers, C. M.; Liao, K.-C.; Yoon, H. J.; Rappoport, D.; Baghbanzadeh, M.; Simeone, F. C.; Yoon, H. J.; Thuo, M. M.; Barber, J. R.; Smith, B. J.; Whitesides, Barber, J. R.; Yoon, H. J.; Bowers, C. M.; Thuo, M. M.; Breiten, B.; Gooding, D. Thuo, M. M.; Reus, W. F.; Nijhuis, C. A.; Barber, J. R.; Kim, C.; Schulz, M. D.; Nerngchamnong, N.; Yuan, L.; Qi, D. C.; Li, J.; Thompson, D.; Nijhuis, C. A. Tao, Y. T.; Wu, K. Y.; Huang, K. H.; Perng, T. P. Org. Electron. 2011, 12, 602. Qiu, M.; Liew, K. M. J. Phys. Chem. C 2012, 116, 11709. Tao, F.; Bernasek, S. L. Chem. Rev. 2007, 107, 1408. Simmons, J. G. J. Appl. Phys. 1963, 34, 1793. Simmons, J. G. J. Appl. Phys. 1963, 34, 2581. de Boer, B.; Hadipour, A.; Mandoc, M. M.; van Woudenbergh, T.; Blom, P. W.

Stoddart, J. F. Chem. Soc. Rev. 2012, 41, 4827. Int. Ed. 2008, 47, 142. Whitesides, G. M. Adv. Funct. Mater. 2008, 18, 1097. Angew. Chem. Int. Ed. 2014, 53, 3889. Whitesides, G. M. J. Am. Chem. Soc. 2012, 134, 10876. M. Angew. Chem. Int. Ed. 2012, 51, 4658. Soc. 2014, 136, 16. Simeone, F. C.; Whitesides, G. M. Nano Lett. 2014, 14, 3521. G. M. J. Am. Chem. Soc. 2013, 135, 18131. M.; Whitesides, G. M. Chem. Mater. 2014, 26, 3938. Whitesides, G. M. J. Am. Chem. Soc. 2011, 133, 2962. Nat. Nanotechnol. 2013, 8, 113.

M. Adv. Mater. 2005, 17, 621. 23

(25) (26) (27) (28) 117, 14272. (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43)

Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. Dubois, L. H.; Nuzzo, R. G. Annu. Rev. Phys. Chem. 1992, 43, 437. Slowinski, K.; Chamberlain, R. V.; Miller, C. J.; Majda, M. J. Am. Chem. Soc. Foti, G.; Sanchez-Portal, D.; Arnau, A.; Frederiksen, T. J. Phys. Chem. C 2013, Laibinis, P. E.; Whitesides, G. M.; Allara, D. L.; Tao, Y. T.; Parikh, A. N.; Srivastava, P.; Chapman, W. G.; Laibinis, P. E. Langmuir 2005, 21, 12171. Tao, Y. T. J. Am. Chem. Soc. 1993, 115, 4350. Colorado, R.; Lee, T. R. J. Phys. Org. Chem. 2000, 13, 796. Cyganik, P.; Szelagowska-Kunstman, K.; Terfort, A.; Zharnikov, M. J. Phys. Ramin, L.; Jabbarzadeh, A. Langmuir 2012, 28, 4102. Aguilar-Sanchez, R.; Su, G. J.; Homberger, M.; Simon, U.; Wandlowski, T. H. J. Stoliar, P.; Kshirsagar, R.; Massi, M.; Annibale, P.; Albonetti, C.; de Leeuw, D. Reus, W. F.; Nijhuis, C. A.; Barber, J. R.; Thuo, M. M.; Tricard, S.; Whitesides, Paddonrow, M. N.; Shephard, M. J.; Jordan, K. D. J. Phys. Chem. 1993, 97, 1743. Lin, L. L.; Jiang, J.; Luo, Y. Physica E 2013, 47, 167. Cumming, G.; Fidler, F.; Vaux, D. L. J. Cell Biol. 2007, 177, 7. Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. Walczak, M. M.; Chung, C. K.; Stole, S. M.; Widrig, C. A.; Porter, M. D. J. Am. Lee, S.; Puck, A.; Graupe, M.; Colorado, R.; Shon, Y.-S.; Lee, T. R.; Perry, S. S.

Rev. 2005, 105, 1103.

1997, 119, 11910.

Nuzzo, R. G. J. Am. Chem. Soc. 1991, 113, 7152.

Chem. C 2008, 112, 15466.

Phys. Chem. C 2007, 111, 17409. M.; Biscarini, F. J. Am. Chem. Soc. 2007, 129, 6477. G. M. J. Phys. Chem. C 2012, 116, 6714.

1987, 109, 3559. Chem. Soc. 1991, 113, 2370. Langmuir 2001, 17, 7364.

24

(44) (45) (46) (47) 107, 11690.

Bowers, C. M.; Liao, K.-C.; Żaba, T.; Breiten, B.; Baghbanzadeh, M.; Rappoport, Haj-Yahia, A.-E.; Yaffe, O.; Bendikov, T.; Cohen, H.; Feldman, Y.; Vilan, A.; Mirjani, F.; Thijssen, J. M.; Whitesides, G. M.; Ratner, M.A.; Submitted. Alloway, D. M.; Hofmann, M.; Smith, D. L.; Gruhn, N. E.; Graham, A. L.;

D.; Krzykawska, A.; Aspuru-Guzik, A.; Cyganik, P.; Whitesides, G. M. Submitted. Cahen, D. Adv. Mater. 2013, 25, 702.

Colorado, R.; Wysocki, V. H.; Lee, T. R.; Lee, P. A.; Armstrong, N. R. J. Phys. Chem. B 2003,

25

Table of content graphic. f a

26