Replacing –CH2CH2- with -CONH-Does Not Significantly Change Rates of Charge Transport Through AgTS-SAM//Ga2O3/EGaIn Junctions

: This paper describes physical-organic studies of charge transport by tunneling through self-assembled monolayers (SAMs), based on systematic variations of the structure of the molecules constituting the SAM. Replacing a -CH 2 CH 2 - group with a -CONH- group changes the dipole moment and polarizability of a portion of the molecule, and has, in principle, the potential to change the rate of charge transport through the SAM. In practice, this substitution produces no significant change in the rate of charge transport across junctions of the structure Ag TS -S(CH 2 ) m X(CH 2 ) n H//Ga 2 O 3 /EGaIn (TS = template stripped, X = -CH 2 CH 2 - or -CONH-, and EGaIn = eutectic alloy of gallium and indium). Incorporation of the amide group does, however, increase the yields of working (non-shorting) junctions (when compared to n- alkanethiolates of the same length). These results suggest that synthetic schemes that combine a thiol group on one end of a molecule with a group, R, to be tested, on the other (e.g. HS~CONH~R) using an amide-based coupling provide practical routes to molecules useful in studies of molecular electronics. with thiourea by hydrolytic cleavage of the resulting isothiuronium salt gave the target thiol. The reverse amide was synthesized in a similar manner. The thiols were obtained in 45 – 81 % yield over the four steps.


Introduction
Understanding charge transport through organic molecules and supramolecular structures is important in fields from biology [1][2][3][4][5] to materials science. [6][7][8][9][10][11][12][13][14][15][16][17][18] In biology, understanding the flow of electrons in redox biochemistry requires understanding the relation between molecular structure and rates of charge transport. In materials science, it is important in evaluating the potential of tunneling devices based on organic matter for use in electronics: the concept of "wave-function engineering" -that is, designing and shaping tunneling barriers by molecular design -has been an influential and theoretically attractive, but practically unproven, starting point for a number of concepts proposed for molecular electronics.
We, [19][20][21][22][23] and others,  are developing experimental systems for investigating charge transport by tunneling across self-assembled monolayers (SAMs) as a function of the structure of the molecules making up the SAMs. An ideal system would offer; i) convenience and reproducibility, ii) robustness (i.e. the ability to generate statistically significant numbers of data rapidly), iii) versatility i.e. the ability to modify, easily and rapidly, the structure of the organic part of the junction through synthesis. This paper describes the measurement of current density (J, amps/cm 2 ), as a function of applied bias (V), in molecular junctions comprising self-assembled monolayers (SAMs) formed from thiols having the structure 1, 2 or 3 adsorbed on so-called "ultra-flat", template-stripped silver (Ag TS ) substrates, and contacted by cone-shaped top-electrodes of the liquid HS(CH 2 ) m CONH(CH 2 ) n H HS(CH 2 ) m NHCO(CH 2 ) n H 1 m = 10, n = 0,1,3,4,6 3 m = 10, n = 4 2 m =11, n = 0-6 eutectic of gallium and indium with its surface film of native oxide (Ga 2 O 3 /EGaIn). 19,[21][22][23]48 The junctions described in this paper are similar to the Ag TS -S(CH 2 ) n H//Ga 2 O 3 /EGaIn junctions that we have described previously, 23 except for the substitution of -CONH-groups for -CH 2 CH 2 -groups.
This project had two objectives. i) We wished to make a controlled perturbation to the structure of the n-alkanethiolates (which have been the predominant subject of studies of processes involving SAMs), and to determine the influence of this perturbation on the rates of charge transport across these SAMs. The amide group (-CONH-) is one of the best understood in organic chemistry, 49,50 and while not isostructural with -CH 2 CH 2 -is similar in size, and is known, from prior work, to be compatible with the formation of SAMs. 51-55 ii) Perhaps more importantly, we wished to develop a system of SAMs to use in studies of charge transport that was more easily modified structurally than are derivatives of n-alkanethiolates. Preparing compounds of the structure HS(CH 2 ) n R, where n is 10 -20, and R is a group that we might wish to select, with as few restrictions as possible, from the full range of organic and organometallic groups, can be synthetically arduous. By comparison, compounds of the structure HS(CH 2 )~1 0 CONH(CH 2 ) n R are relatively simple to make, since amide-forming reactions are among the most versatile in organic synthesis in their ability to couple different groups. We knew-from other work-that amide groups are compatible with SAMs, and that the literature contains data suggesting that they are more stable (perhaps because of inter-chain hydrogen bonding) than are simple n-alkanethiolate-containing SAMs. [51][52][53][54][55] Examination of current as a function of voltage for these amide-containing SAMs yielded two important results: i) substituting an amide moiety, -CONH-, for an ethylene moiety, -CH 2 CH 2 -, resulted in no significant change in current density, and ii) introducing the amide group into the SAM raised the yield of non-shorting junctions from ~80-90 % 23 to ~100%. The former results indicate that even a large (from the vantage of organic chemistry) change in the electronic structure, dipole moments, and other properties of the SAM does not significantly influence the rate of charge transport by tunneling. It also provides a reality check on the idea that "wave-function engineering" may provide an easy method of designing new materials with currently unprecedented charge-transport properties. The latter result suggests that amides (and perhaps other functional groups capable of inter-chain hydrogen bonding) may provide the structural basis for a useful strategy to use in improving the robustness and practicality of SAMbased tunneling junctions and other devices. This project is complementary to a related study using the same type of junction. 56 In this other study, we used SAMs made up of a related structure (4), also containing an amide group. The objective of work involving 4 was different from that in this paper.

HS(CH 2 ) 4 CONHCH 2 CH 2 R 4
It was designed to examine the influence of the structure of the R group (chosen to include a number of different aliphatic and aromatic groups) on the tunneling current for SAMs of approximate constant thickness. It also compared the amide-containing compounds with homologous n-alkyl thiolates of the same length, and concluded that both compounds tunnel currents at almost similar rates (any difference was less than a factor of three). 6

Background
Charge Transport in Insulating Organic Molecules. The current consensus in the field of molecular electronics is that charge transport in SAMs of insulating organic molecules proceeds via non-resonant, through-bond tunneling. 20,23,34,42,[57][58][59][60][61][62][63][64] This behavior is typically modeled by a simple form of the Simmons equation (Equation 1). 65 In this equation, J (A/cm 2 ) is the current density flowing between the electrodes, d is the length of the molecule (in either Å or number of non-hydrogen atoms in the extended chain, n), J 0 is the current density in the hypothetical case of a junction with a SAM of zero thickness, but still including the contribution of all interfaces in the junction, and β (either Å -1 or per number of non-hydrogen atoms, n -1 ) is an attenuation factor related to the shape and height of the tunneling barrier posed by the SAM.
We 19,21-23 and others [66][67][68] have previously reported that J through SAMs of nalkanethiols is approximately log-normally distributed (albeit often with long, asymmetrical tails and significant outliers), rather than normally distributed, and have suggested that variations from junction to junction in thickness and in the number or type of defects in the SAM and electrodes would lead to a normal distribution in the effective thickness, d, of the SAM. 19,[21][22][23]66,69 Since J is exponentially dependent on a normally distributed parameter (eq. 1), J itself should be log-normally distributed.
We have reviewed the literature on charge transport through SAMs of alkanethiols, and found a consensus for the value of β = 0.8 -0.9 Å -1 (1.0 -1.15 n C -1 ) across many techniques; 70 we also found a much looser consensus for a value of J 0 (J 0 ~ 10 -10 3 A/cm 2 ) in junctions of the form metal-SAM//(protective layer)/liquid metal (We discuss the significance of the "protective layer" in another paper 48 ) We have demonstrated a statistically significant difference in J between alkanethiols with odd and even numbers of carbon atoms (the so-called "odd-even" effect). 23 Specifically, J for odd-numbered alkanethiols is roughly one order of magnitude smaller than what one would predict for the same thickness from an interpolation of J for even-numbered alkanethiols. In this work, we infer that this "oddeven" effect persists in SAMs containing amide moieties.

Why Secondary (-CONHR-) Amides? This work focuses on secondary amides
because; i) They have the potential to form intermolecular hydrogen bonds when incorporated into a SAM. ii) The absence of a second N-alkyl group leads to less interference with self-assembly than would more hindered structures such as -CON(R')R, and therefore the charge transport characteristics of SAMs incorporating -CONH-groups can be compared to those of n-alkanethiolate SAMs (which are often seen as a baseline/standard system). iii) Although a -CONH-group has only small structural/steric differences from a -CH 2 CH 2 -group, amides have a large (μ~ 4 D) 80-84 group dipole moment. Tertiary amides, are sterically larger than -CH 2 CH 2 -groups, and cannot be directly compared to n-alkanethiols. Primary amides place a polar -CONH 2 group at the interface with the Ga 2 O 3 film, and as such can perhaps not be directly compared with n-

alkanethiolates.
Position of the Amide: We synthesized thiols, HS(CH 2 ) m CONH(CH 2 ) n H, with the amide moiety separated from the thiol group by 10-11 methylene units (m =10 or m =11). We chose a C 10 or C 11 spacer to allow for a well-ordered region between the amide moiety and the thiol. 73,85 Use of n-Alkanethiolate SAMs as Bracketing Standards: Data were collected over long intervals of time (days to months apart), and random and/or systematic errors (environmental and seasonal variations, differences among operators, changes in equipment) were probably unavoidable. To make sure that data collected at different times by different operators were comparable, we used two n-alkanethiols

(octadecanethiol and dodecanethiol) that bracket the values of J(V)of interest in this work
to calibrate values obtained with new compounds or from multiple users; we call these thiols (C 12 and C 18 ) "bracketing standards". We took measurements of these two SAMs periodically during collection of data on amide-containing SAMs, and compared these data to those previously reported 21,23 using the EGaIn/Ga 2 O 3 top electrode. Measurements using the standards were collected randomly throughout the study, with the first threefive tunneling junctions measured in any set of experiments being from the two standards, before starting to measure the amides. Whenever the measured standards deviated from the literature by more than an order of magnitude, the experiment was stopped, a new tip formed, and data re-collected. By applying a randomized system of measuring the standards, we minimized uncontrolled variations in the measurements.

Management of Measurement Variability:
To minimize and manage experimental errors, we have developed a standard operating procedure and well-defined statistical tools for this work. 69 We collected data generated by multiple users, to avoid a singleuser bias in the results. We pooled all non-shorting (working junctions only) data from different users, and summarized the pooled data as histograms. To these histograms, we fitted Gaussian curves using a non-weighting algorithm to avoid biasing the data due to outliers. 69 Statistical Analysis of log(|J|) Rather than J: As noted previously by us 22,23 and others, 43,66,67,86 J is log-normally distributed, rather than normally distributed -that is, log(|J|) is (approximately) normally distributed. We chose, therefore, to plot and fit histograms of log(|J|), rather than J. When necessary, we used two-sample t-tests 69 to determine whether the distributions of log(|J|) for two compounds had distinguishable or indistinguishable means, at the 99% confidence level. Since the t-test assumes normality, and since the distributions we observe sometimes deviate from normality, the statistical inferences from t-tests are suggestive but not conclusive.

Results and Discussion:
Synthesis of Amides. Scheme 1 summarizes the general procedure -two consecutive two-step reactions -we used to synthesize all molecules (see supporting information for details). Synthesis followed by chromatographic purification gave the target thiols in 45-81% yields over four steps, and in high purity (as determined by 1 H NMR). As previously discussed, 23 we purified the thiols carefully, by column chromatography using 15% ethyl acetate in n-hexanes as eluant, before forming SAMs.
The purified thiols were stored under N 2 and refrigerated when necessary. When, or if, the molecules degraded, they were re-purified by column chromatography, and purity was confirmed by 1 H NMR. For convenience, we abbreviated the names of the compounds using the assignments in Table 1  (sometimes non-consecutive) days, although data collected over short periods of time had fewer variations and gave more tightly clustered data. Before collecting data for a particular amide-containing SAM, we measured current densities across each of the two bracketing standards from three -five junctions, and then randomly repeated throughout the analysis after every 10 junctions.
The  Table 1). By contrast, the yields of working junctions incorporating SAMs of n-alkanethiols (S(CH 2 ) n-1 CH 3 , n = 9 -18) averaged ~ 80-90 %. 23 We attribute the high yields of junctions derived from amide- Estimates of β and J 0 for Alkyl Amides: For these amide-containing SAMs, we do not have data over a sufficiently broad range of lengths to generate confident estimates of β and J 0 by fitting <log|J|> vs n to the Simmons equation. 91 We exclude from our analysis compounds 10,0 and 11,0, since they terminate in a -CONH 2 group rather than a -CH 3 group. We draw three qualitative conclusions from the data. i) Within the uncertainty of the measurements, it is not possible to distinguish between n-alkanethiolates and amides with the same length ( Figure 1). This statement is not the same as an assertion that there is no difference, only that within the uncertainties of these measurements (≤ ±1 unit in log|J|), we cannot distinguish them.
ii) It appears that the difference between chains with odd and even numbers of atoms in the backbone of the chain, observed previously, 23 is still preserved in the amides (Figure 2). A crude estimate of the difference between  lines are inserted to highlight how the molecules segregate into two groups of "odd" and "even" lengths. Substituting an amide for an ethylene unit has no discernible effect on σ log . We also compared the log-standard deviation, σ log , of data derived from amides to those derived from n-alkanethiols (Table 2). We observed no consistent difference in σ log between the two classes of compounds. This result tentatively suggests that the main advantages of having an internal amide -high yields of working junctions and ease and flexibility in synthesis -comes without a trade-off in the spread of log(|J|).
Reversing the orientation of the amide, -CONH-vs -NHCO-, does not make a significant difference to either <log|J|> or σ. We synthesized and measured one compound with the amide moiety reversed [10,4*, S(CH 2 ) 10 NHCO(CH 2 ) 3 CH 3 )]. Figure   3 compares the histograms of 10,4*, and 10,4. Reversing the orientation of the amide moiety lowered the mean value of <log|J|> by 0.6 log units (ΔJ=10 0.6 A/cm 2 ), but this   We have shown a simpler way of achieving rectification than the Aviram-Ratner approach: placing a single accessible molecular orbital asymmetrically between two electrodes (i.e. at the terminus of an alkanethiol chain). 21, 22 Others have proposed that even an accessible molecular orbital is not required -that a dipole with a component along the axis of charge transport is sufficient to cause rectification by "tilting" the tunneling barrier and breaking the symmetry between the wave-functions of electrons approaching the barrier from the left and right. 101 An amide has a large dipole moment (μ ~ 4 D). 81,83,84 Placing an amide within the SAM does not produce an easily quantifiable change in that component of the dipole moment that might be relevant to the shape of the tunneling barrier for a number of reasons (the orientation of the dipole, partial cancellation of dipoles on adjacent molecules, and uncertainties concerning the path of the electron during tunneling, among others). Nonetheless, it seemed worthwhile to test our set of amide-containing SAMs for rectification. We calculated the raw rectification ratio, r, as the ratio of the current density at opposing values of applied bias;  Table 1

Fig 4
For all molecules synthesized, we calculated the value of r, at V = ± 0.5 V, for each trace, plotted values of log|r| in histograms (r, like J, is log-normally distributed), and fitted Gaussians to the histograms to determine the mean and standard deviation of log|r|. All of the amides gave values of log|r| that were close to zero, and similar to those of n-alkanethiols in magnitude and polarity (Table 3). The widths of histograms for the measurement of rectification, r, are narrower than those for histograms for the measurements of current density J. A comparison of width of histograms of current density, J, to those of the rectification ratio, r, demonstrate again 23 that the former had much larger distributions than the latter. Figure 5 shows examples, and histograms, for three types of molecules; two amides and an alkanethiol.
The difference in the width of the distributions in the histograms of J and r reflects, we presume, the fact that the measurements of rectification ratios are self-referencing, 22 i.e. current density on reverse bias is compared to current density in the same junction at forward bias. The junction at one bias thus acts as a reference for the same junction at the opposite bias.  group, which is closer to the silver electrode in 4 than in 1 -3. 56 We draw three conclusions. i) The amide-based compounds are easier to synthesize than those with an all-carbon backbone, and allow easy synthetic access to a wide range of compounds with which to test hypotheses relating structure to tunneling current.
ii) The presence of the dipole moment embedded in the SAM by the amide has no apparent influence on either the tunneling currents, or, perhaps more significantly for the theory of these systems, on their rectification ratios (r): the values of r for amide-containing and all-hydrocarbon compounds are indistinguishable. iii) The orbital structure of the -CONH-group is thus, apparently, not sufficiently different from that of a -CH 2 CH 2 -group to influence the rate of charge transport by tunneling across these junctions significantly.
The quality of the data we report in this paper is somewhat more broadly distributed than that in previous papers focused on n-alkanelthiolates, and also on amides of structure 4. 56 We do not know the reason for this difference yet.
The most important problem in the design of compounds 1-3 is the long -(CH 2 ) [10][11] chain connecting the thiol and the amide group. This length limits the size of the groups, R, that can be placed on the other side of the amide, since for large R groups, values of J(V) become too small to measure reliably with our electrometer. We are thus constrained to use relatively small R groups, where the molecular order of these groups in the SAM is not established, but is almost certainly less than in n-alkanethiolates. [51][52][53][54]77 Whatever the reason, the data in Figures 1 and 2  Wave-function Engineering. One of the hopes at the beginning of the study of organic tunneling junctions based on SAMs was that variations in the HOMO and LUMO energies of the organic groups-in principal easily achieved through synthesis-would allow the design and generation of tunneling barriers with designed energetic topographies, and the discovery of new types of tunneling behaviors. Variations in the structure of the functional groups included in the SAMs that include common groups (e.g., simple aromatics, amides, saturated hydrocarbons) seem to have little effect on rates of tunneling. 56 The theory of tunneling through junctions containing SAMs is not sufficiently developed at present to give any guidance to these studies, and it is not clear whether this low sensitivity is expected or not. It does, however, empirically constrain the range of organic groups that seem worthwhile to study, when looking for interesting influences on tunneling currents to those that have much larger changes in orbital energies than simple organic functional groups. The large rectification observed with terminal ferrocene groups may point toward a useful direction. 56 Rectification. The largest values of r so far reproducibly observed have been with ferrocene (Fc). The most plausible mechanism underlying these values is based on a difference in mechanism of charge transport at opposite bias in these systems (from pure tunneling to a combination of hopping and tunneling). 21, 22 Other mechanisms 22