Understanding Keesom Interactions in Monolayer-Based Large-Area

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE

Surfaces, Interfaces, and Catalysis; Physical Properties of Nanomaterials and Materials

Understanding Keesom Interactions in MonolayerBased Large-Area Tunneling Junctions Jiahao Chen, Miso Kim, Symon Gathiaka, Soo Jin Cho, Souvik Kundu, Hyo Jae Yoon, and Martin M. Thuo J. Phys. Chem. Lett., Just Accepted Manuscript • DOI: 10.1021/acs.jpclett.8b01731 • Publication Date (Web): 20 Aug 2018 Downloaded from http://pubs.acs.org on August 20, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Understanding Keesom Interactions in Monolayer-Based Large-Area Tunneling Junctions Jiahao Chen,1† Miso Kim,2† Symon Gathiaka,3 Soo Jin Cho,2 Souvik Kundu,4 Hyo Jae Yoon,2* and Martin M. Thuo1* 1

Department of Materials Science and Engineering, Iowa State University, Ames, Iowa, 50010

(USA) 2

Department of Chemistry, Korea University, Seongbuk-gu, Seoul 02841 (South Korea)

3

School of Pharmacy and Pharmaceutical Science, University of California, La Jolla, CA 92093-

0657 (USA) 4

Department of Electrical and Computer Engineering, Iowa State University, Ames, Iowa, 50010

(USA)

Abstract Charge transport across self-assembled monolayers (SAMs) has been widely studied. Discrepancies of charge tunneling data that arise from various studies, however, call for efforts to develop new statistical analytical approaches to understand charge tunneling across SAMs. Structure-property studies on charge tunneling across SAM-based junction have largely been through comparison of average tunneling rates and associated variance. These early moments (especially the average), are dominated by barrier width—a static property of the junction. In this work, we show that analysis of higher statistical moments (skewness and kurtosis) reveals the dynamic nature of the tunnel junction. Intra-molecular Keesom (dipole-dipole) interactions dynamically fluctuate with bias as dictated by stereo-electronic limitations. Analyzing variance in

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the distribution of tunneling data instead of the first statistical moment (average), for a series of nalkanethiols containing internal amide and aromatic terminal groups, we observe that direction of dipole moments affects molecule-electrode coupling. An applied bias induces changes in the tunneling probability, affecting distribution of tunneling paths in large-area molecular junctions. TOC graphic

Understanding charge tunneling across molecular system often relies on structure-property relation studies.1-4 Large-area junctions, based on self-assembled monolayers (SAMs), have emerged as a versatile tool to delineate tunneling characteristics via a statistically validated ensemble average.5-6 Dependence of tunneling characteristics on molecular and supramolecular structure properties within SAM such as odd-even parity,7-9 redox-active moieties,10 terminal or head groups,8-9, 11-12 SAM isomorphism,7 conformational order,13 and dipole moments,7 has been reported. Among these properties, understanding the effect of dipole moments on tunneling behavior has been limited, in part due to complexity of SAMs14 and potential for field-driven effects15-16—the latter remains unresolved. Molecular dipoles can, however, induce large built-in fields that significantly affect band alignment in tunneling devices.17 Using isomorphic carboranethiol SAMs, it was shown that molecular dipole orientation plays a major role in tunneling currents.7 The angular orientation of the dipole affects molecule-electrode coupling due, in part, to band broadening (Fermi-level pinning) when the dipole moment is not oriented


Page 2 of 24

Page 3 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


orthogonal to the surface normal (cosθ ≠ 0).18-19 On the other hand, changing polarity of terminal groups (R) in HS(CH2)4CONH(CH2)2-R does not significantly affect the tunneling rate.3, 20 The seemingly opposing/incoherent theoretical17 and experimental2-3, 7 findings, coupled with other reported data,21-27 point to a need for further insights into how embedded molecular dipoles (magnitude, position and resulting interfaces) in SAMs influence rates of charge tunneling across SAM-based large-area junctions. This dilemma can be at least due to the following two reasons: i) either the empirical data is different—an unlikely premise, or ii) there is need to revisit the data from a different point of view. We infer the latter since in a tunnel junction dipole moments are not the primary defining variable (at least from Simmons’ tunneling equation28 in which the size of the barrier width plays a larger role). Dipoles and other secondary/dynamic effects may not be readily captured in the early analytical approach. This paper explores the influence of field-driven intra-molecular dipole interactions in SAMbased tunneling junctions on the distribution of tunneling current density (J, A/cm2) data. This effect is revealed through statistical mapping of changes in the Gaussian distribution fits to logJ histograms as a function of applied voltage, V. We focus on large-area EGaIn-based junctions having the structure AgTS-SAM\\Ga2O3-EGaIn, (where AgTS is template-stripped silver29, ‘-’ and ‘\\’ indicate chemisorbed and physisorbed interfaces respectively, and EGaIn is eutectic galliumindium covered by a passivating Ga2O3 layer).30 Target SAMs are formed with molecules of the general form HS(CH2)4CONH(CH2)2R (R=Ph, or 2-, 3-, or 4-pyridyl; Figure 1a). Similar structures have been previously studied10, 31-33 except for R=3-pyridyl. In these SAMs, two dipoles (internal amide and terminal R group) are embedded in an n-alkyl backbone, and are electronically separated by a two saturated (sp3) carbon spacer.



Figure 1. a) Structure and dipole moments of HS(CH2)4CONH(CH2)2R molecules (segment of the active moieties of the molecular structure are drawn next to each ball-and-stick structure for clarity) with different terminal R groups such as methyl benzene (MB), 2-methyl pyridine (2MP), 3-methyl pyridine (3MP) and 4-methyl pyridine (4MP). The arrows point towards regions of low electron density. b) Average current densities (J, A/cm2) across molecular junctions of SAMs derived from the molecules are statistically indistinguishable. c) The electrostatic potential (ESP) maps of the molecules show significant differences in electronic structure of terminal groups. d-e) Bias-dependent changes in Gaussian fits for the distributions of the measured current density for MB and 4MP. Dipole moments in freely rotating molecules align to an applied electric field to a lower thermodynamic potential state.34 Similarly, in highly polarizable isotropic molecules, a field induced asymmetry in electron density distribution leads to an induced directional dipole. In SAMs, however, such alignment is dependent on sterics, chemisorption, and inter-chain interactions. In solid-like SAMs (≥C10 n-alkanethiols), molecules are well packed hence rotation


Page 4 of 24

Page 5 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


is energetically prohibitive as it is analogous to solid state molecular motion.35 It can, therefore, be anticipated that only through space and/or field driven polarization will dominate Keesom-type interactions in solid state SAMs. Such dipole (vector) interactions, however, are proportional to the cosine of the angle between the interacting electrostatic tensors (cosθ).36 To delineate the effect of dipoles, it is, therefore, critical to consider the direction of the dipole relative to the tunneling pathway as previously shown—albeit with isomorphic SAMs.7 An objective of this work is to test the hypotheses above using structurally simple thiol derivatives (Figure 1) and statistically significant sets of tunneling data, obtained in well-defined molecular junctions. Electrode oxidation/deformation and junction inhomogeneity have been shown to be minor variables in EGaIn junction characteristics4 (discussed in detail in SI). To this end, we focused on the shape of Gaussian fitting curves of logJ in the context of skewness and kurtosis (third and fourth statistical moments). Skewness indicates a shift in the mode and median values of the population of logJ relative to the mean. Even when the average values (Gaussian mean) of logJ are similar for different molecules, positive and negative skewness indicate a shift to more resistive or conductive modes respectively (schematically shown in supporting information Figure S6). Kurtosis, on the other hand, is an indicator of peakedness (convergence) of the distribution and can thus be associated with increased degeneracy in the number of tunneling pathways/states (Maxwell-Boltzmann statistics) or coherence in resistivity of charge tunneling channels. A larger kurtotic value (leptokurtic) corresponds to a narrower distribution of logJ, and vice-versa for lower values (platykurtic) as schematically demonstrated (supporting information, Figure S7). We first calculated the dipole moments for the two non-anchored moieties in HS(CH2)4CONH(CH2)2R molecules (Figure 1a). For brevity and clarity, we independently



Page 6 of 24

considered structure of the amide and the R groups: methyl benzene (MB), and 2-, 3-, 4-methyl pyridines (abbreviated 2MP, 3MP and 4MP, respectively, Figure 1c). The effective dipoles before surface adsorption were estimated using the Gaussian software package and compared to literature.37, 32 Figure 1a shows the calculated magnitude and direction of each group’s dipole moment. The amide group within the alkyl chain backbone has a dipole of ca. 3.9 D while the pyridyl units have dipole moments ranging from 1.9 D to 2.8 D. Considering the tilt angle of the molecules adsorbed on Ag surface (~10° for n-alkanethiols), we expect a smaller tilt for these molecules, in part due to possible stronger intermolecular (hydrogen) bonding relative to an nalkyl chain.38-39 The amide dipole, therefore, is likely almost orthogonal to the surface normal as observed with cage compounds with analogously oriented dipoles40-42. This deduction is further supported by the decrease in cant angle upon hydrogen bonding and other secondary interactions.39 The energy of interaction between permanent dipoles (Keesom interactions) depends on the inverse sixth power of distance (~1/d6) between them.43 Therefore, the group dipoles in our molecules are likely electrostatically isolated at zero bias. We hypothesized that perturbation of this equilibrium with an external field, however, is likely to induce interactions, hence, coupling responses between the two moments. The angular dependence of this coupling implies that changing the position of a heteroatom in the terminal group will lead to different coupling points and response (effective alignment and work needed to align them)—hence changes in the distribution in the Gaussian distribution of logJ. Synthesis of tested molecules (except 3MP) have been published elsewhere.9,

31

The

Supporting Information contains detailed information of synthesis and characterization of 3MP as an example. Electrical characterization of the corresponding junctions was conducted following literature procedures.7, 44 The resulting EGaIn junctions gave high yields of working junctions


Page 7 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(>~80% for n-alkanethiolates) hence a large amount of J-V traces, sufficient for rigorous statistical analysis.9 When the R group changed, no significant difference in the average value of J was observed (Figure 1b), even at 95% confidence interval (highlighted bands, Figure 1b), which is consistent with previous result.31, 45 This result is surprising since the R//Ga2O3 top-interface is significantly different across the series—based on terminal group polarity. For junction measurements, we applied biases ranging from 0 V to ±0.5 V because other work on EGaIn based junctions have widely used this bias window, and the use of identical bias range allows for direct data comparison.1-4, 7, 27, 46-50 Furthermore, low bias window increase yield of working device, and avoid the resonance-induced non-linear variation in tunneling behavior or possible electrode geometry variation.51 and/or SAM breakdown.52 Electrostatic potential (ESP) maps of the terminal moieties (Figure 1c) showed that the distribution of electron density for the aromatic terminal group differed significantly from the others even though the average current densities was very similar to the others. We infer that the average and standard deviation (first and second statistical moments) are limited in delineating some obvious differences in tunneling junctions. Tunneling efficiency, however, is based on statistically averaged probability (the SAM being an ensemble of tunneling pathways), as such subtle structural perturbation can affect distribution of J. Figure 1d and 1e indeed capture the bias-dependence of the shape of Gaussian fits to logJ histograms (raw data) for MB and 4MP molecules. Tunneling characteristics for these two molecules were significantly different in the context of the peakedness of Gaussian fits as a function of applied voltage. In conjunction with the fact that MB and 4MP exhibit largely different dipole moment for the R group, this observation motivated us to examine in detail how the applied bias affects distribution of logJ relative to molecular dipoles.



Figure 2. Summary of charge tunneling characteristics of studied junctions. a) Heat maps showing distribution of current density at different bias for junctions derived from the four molecules. The calculated b) skewness, and c) kurtosis in the distribution of the data at each voltage where a consistent trend is observed. d) Skewness and e) kurtosis for junctions derived from 3MP where fluctuations at low bias is observed. *Solid symbol: at reverse bias (top-electrode negatively biased); open symbol: at forward bias. Figure 2a shows the heat-maps of all J-V traces as analyzed using a previously developed method, Large-Area Junction Analysis (LAJA).6 Although these heat-maps indicate ostensible differences in data distribution (especially on the bias-dependent variance), they are not quantitative. To this end, we determined kurtosis and skewness for individual fitting curves over


Page 8 of 24

Page 9 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


all applied biases. Trends in skewness (Figure 2b and 2d) and kurtosis (Figure 2c and2e) against applied bias show significant differences across the series. For skewness we observe that; i) nDodecanethiolates (C12 as reference) and 2MP show no statistically significant bias-dependence, irrespective of direction of charge flow. ii) In contrast, decrease in skewness (ΔS= negative) − towards more conductive tails with increasing bias was observed for MB and 4MP (|∆𝑆𝑆𝑀𝑀𝑀𝑀 |=0.67

− and |∆𝑆𝑆4𝑀𝑀𝑀𝑀 |=0.75). iii) For 4MP and MB, the overall trends in skewness for reverse and forward

biases were same. iv) However, the rate of decrease with increasing voltage was different for 4MP + + − − and MB (|∆𝑆𝑆4𝑀𝑀𝑀𝑀 − ∆𝑆𝑆4𝑀𝑀𝑀𝑀 |= 0.12 and |∆𝑆𝑆𝑀𝑀𝑀𝑀 − ∆𝑆𝑆𝑀𝑀𝑀𝑀 |= 0.2; in Figure 2b). v) For 3MP, an increase

− in skewness was observed (∆𝑆𝑆4𝑀𝑀𝑀𝑀 =1.81), with yet unexplained fluctuations at low bias (Figure

2d). As with MB and 4MP, 3MP showed no statistically significant differences between skewness trends in forward and reverse biases. The fluctuations in skewness for 3MP stopped at ca. -0.3V

and 0.2V. Interpreting this observation was difficult because it might be due to high noise-to-signal ratio at low bias regime or a yet to be determined field-dependent effect that is dependent on the angular orientation of the dipole moment.17 Nonetheless, there is a general increases in skewness with increasing voltage indicating that the distribution is tailing towards higher resistance. Uniqueness in each of the terminal group, therefore, manifest in the bias-dependent tails (skewness) in their distribution, a phenomena unobservable by looking at the mean and variance. For kurtosis (Figure 2c and 2e, and Figure S2 in the Supporting Information), we observed that; i) C12 and 2MP did not show significant change in peakedness over the range of biases − evaluated, ii) MB showed an increase in peakedness (∆𝐾𝐾𝑀𝑀𝑀𝑀 =1.91 at forward bias) with increasing

the voltage indicating a trend towards convergence of resistance, iii) The leptokurtotic behavior

(see supporting information Figure S7) of MB slightly differs with change in direction of bias. Rate of increase in kurtosis was higher under reverse bias (that is; when EGaIn is negatively biased,



+ − |∆𝐾𝐾𝑀𝑀𝑀𝑀 − ∆𝐾𝐾𝑀𝑀𝑀𝑀 |= 1.05) suggesting that the behavior could be dependent on effective field on the

top electrode. iv) 4MP, which has a permanent dipole moment oriented along methylpyridyl’s C2symmetry axis, showed a decrease in kurtosis. This trend is inverse to that observed with MB, even though both molecules showed similar trends in skewness. The applied bias rendered the 4MP distributions less peaked (platykurtic) with no significant dependence on the direction of bias + − (|∆𝐾𝐾4𝑀𝑀𝑀𝑀 − ∆𝐾𝐾4𝑀𝑀𝑀𝑀 |= 0.3). v) In comparison, 3MP having dipole moment oriented away from the

surface normal and at an angle relative to the head group C2 symmetry line, showed a gradual − decrease in kurtosis (∆𝐾𝐾3𝑀𝑀𝑀𝑀 =4.4) as the voltage increased albeit with oscillatory fluctuations at

low biases (Figure 2e). vi) The fluctuations in kurtosis trends for 3MP resemble those in skewness but inverted, i.e. gradually decreasing. In conjunction with the information about molecular dipoles shown in Figure 1a, the results of skewness and kurtosis could be summarized as follows: molecules bearing a readily polarizable moiety, MB, exhibited an applied field-induced negatively skewed distribution of logJ with concomitant narrowing of the distribution (higher kurtosis) while moieties having permanent fixed dipole moments did not show similar correlation between skewness and kurtosis. These findings point to Keesom interactions that readily adopt to Coulombic perturbations (electric field).53 At zero bias, the terminal moieties’ dipole (abbreviated as D1) is electronically isolated from that of amide moiety (abbreviated as D2). As voltage increases, Keesom interaction between D1 and D2 becomes significant due to likely polarization of the spacer.54 Since the C-N bond (between amide and methylene) can rotate, the dipoles relax to the lowest Keesom potential, hence the most stable D1-D2 angular arrangement dominates. In this regard, it is expected that for 2MP the dipole moments will orient anti-parallel, which would support the lack of field-induced changes in current density distributions. In MB, field-induced asymmetry in electron density in the phenyl ring will


Page 10 of 24

Page 11 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


align to the applied field rendering it more conductive (hence negative skewness) with a concomitant increase in coherence of the tunneling pathway: that is, the distribution become peaked (higher kurtosis) due to convergence around the mean. Inductive effects due to para relation between the pyridyl N and the methylene in 4MP induces significant polarization, as such, any perturbation by the applied field either increases the polarization, hence increased coupling between D1 and D2. Since D1 and D2 are related by an obtuse angle (

Understanding Keesom Interactions in Monolayer-Based Large-Area

Recommend Documents