Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

A single atom change turns insulating saturated wires into molecular conductors

Abstract

We present an efficient strategy to modulate tunnelling in molecular junctions by changing the tunnelling decay coefficient, β, by terminal-atom substitution which avoids altering the molecular backbone. By varying X = H, F, Cl, Br, I in junctions with S(CH2)(10-18)X, current densities (J) increase >4 orders of magnitude, creating molecular conductors via reduction of β from 0.75 to 0.25 Å−1. Impedance measurements show tripled dielectric constants (εr) with X = I, reduced HOMO-LUMO gaps and tunnelling-barrier heights, and 5-times reduced contact resistance. These effects alone cannot explain the large change in β. Density-functional theory shows highly localized, X-dependent potential drops at the S(CH2)nX//electrode interface that modifies the tunnelling barrier shape. Commonly-used tunnelling models neglect localized potential drops and changes in εr. Here, we demonstrate experimentally that \(\beta \propto 1/\sqrt{{\varepsilon }_{r}}\), suggesting highly-polarizable terminal-atoms act as charge traps and highlighting the need for new charge transport models that account for dielectric effects in molecular tunnelling junctions.

Introduction

Significant effort has been dedicated to study and manipulate tunnelling rates across molecular wires, which serve as model systems to improve our understanding of the mechanisms of charge transport across molecules which, in turn, play a central role in, e.g., biological processes, catalysis, and energy conversion1,2,3,4. It is well-known that the tunnelling current density (J in A/cm2) decreases exponentially with the length of the molecular wire (d in Å) given by the general tunnelling equation

$$J={J}_{0}(V){\mathrm{e}}^{-\beta d}={J}_{0}(V){10}^{-\beta d/2.303},$$
(1)

where J0 is a pre-exponential factor and the tunnelling decay coefficient (β in Å−1) determines how quickly the measured current decays with d2,5,6,7. In this context, unsaturated molecules with conjugated π-bonds are usually thought of as “molecular conductors” with low values of β (0.1–0.4 Å−1)2,5,8,9 and saturated molecular wires with localized σ-bonds provide “molecular insulators” with large values of β (0.8–1.2 Å−1)2,5,7,10,11. This rule of thumb stands in sharp contrast with the high tunnelling rates established for various biomolecules12,13, molecular wires of oligo-peptides14,15, and oligosilanes16. These all have saturated molecular backbones yet they exhibit low values of β (0.1–0.5 Å−1), and support long-range tunnelling over remarkably large distances of up to tens of nanometres13,17.

So far, it has been challenging to engineer β in experiments, and this difficulty is also reflected in various established mechanisms of charge transport across molecular wires. Often, coherent tunnelling is assumed (Eq. 1), where β can be related to the tunnelling barrier height \(\delta {E}_{\mathrm{{ME}}}\) (defined by the offset in energy between the energy of the Fermi level, EF, of the electrode and the energy of the molecular frontier orbital relevant for charge transport), as \(\beta \propto \sqrt{\delta {E}_{\mathrm{{ME}}}}\)2,5,18. This explains why conjugated molecules, which often have frontier orbitals aligned close to EF, have lower values of β than saturated molecules, which have frontier orbitals further from EF. Conversely, in the McConnell superexchange model, charge carriers tunnel via virtual states defined by the repeat units of the molecular wire; here, the tunnelling rate depends on the interaction strength between the repeat units of the molecular bridge2,5,19,20,21. This model has been used to explain low β values (0.2–0.5 Å−1) measured across tunnel junctions with self-assembled monolayers (SAMs) that have σ-bond backbones of oligoglycines14, oligoprolines22, and oligoglycols15. Furthermore, for very long molecules (e.g., proteins), a flickering resonance model has been proposed to explain long-range tunnelling and low β values13,23,24. Finally, different types of hopping models have been proposed to explain low β values of, for instance, bacterial nanowires23, DNA25, proteins26, and long conjugated molecular wires16,27,28; here the value of β also depends on the coupling strength between the repeat units, but these models predict a thermally activated component27,28. To summarize, all previous models suggest the necessity of tuning the chemical nature of the molecular wire to change the value of β.

We note that the value of β also depends, besides the chemical nature of the molecular backbone2,5,13,14,15, on the coupling strength between the molecules and electrodes (Γ) that is naturally related to \(\delta {E}_{{\mathrm{{ME}}}}\)28,29,30. For molecular wires, where \(\delta {E}_{\mathrm{{ME}}}\) decreases with the number of repeat units due to an increase in conjugation with increasing molecular length, extremely low (<0.1 Å−1)27,28,31,32 and even negative β values have been reported31,33,34,35. Such low β values are also a signature of incoherent hopping and these junctions, in particular those containing redox centres, may operate in this hopping regime (also called incoherent tunnelling regime)27,28,33. Lambert and co-workers36 were able to tune the β value between 0.06 and 0.39 Å1 in Au–S(CH2)nFG(CH2)nS–Au junctions with a functional group FG = α-terthiophene, phenyl, or viologen. They found that changing the anchoring group from dithiol to dithiolmethyl for FG = phenyl resulted in an increase of the β value from 0.14 to 0.50 Å1 from which they concluded that localized states on the Au–S bond are involved in tunnelling along the FG units. In contrast, Frisbie and co-workers37 found that β values are similar for Au–S(CH2)nCH3//Au and Au–S(CH2)nS–Au junctions, implying that localized states on the Au–S bond are not important for tuning β (but note that they still significantly affect the contact resistance). Frisbie and co-workers38,39 suggested that Stark effects are important to consider as they can cancel the potential effects of localized anchoring group-electrode states. Indeed, strong Au–S interaction results in severe broadening of the molecular states and therefore the Au–S states only occur as weak features in valence band spectra of aliphatic SAMs38 (as also observed in the present study), highlighting the need to optimize the Γ such that the molecular states remain localized in the molecule. Recently, Chen and co-workers40 reported a method using bimetallic electrodes to enhance the conductance of HO2C(CH2)nCO2H single-molecule junctions via the surface d-band. They improved the interfacial interactions between molecules and transition metal electrodes, promoting interfacial electron transport. Here, we use junctions of the form Ag–S(CH2)nX//EGaIn (n = 10, 12, 14, 16, or 18, and X = H, F, Cl, Br, or I) where the weak interaction between the top electrode and the SAM allows us to investigate in detail how the terminal group X affects the tunnelling rates across the junctions.

So far, the influence of electrostatic effects in molecular tunnelling junctions on β has been largely ignored. One way of quantifying trends in the electrostatics of various systems is by studying the static dielectric constant (εr) of molecular junctions, a macroscopic observable that can be measured via impedance spectroscopy41. Previous work focused mainly on π-conjugated systems and established that in densely packed SAMs, εr hardly changes when the polarizability of the molecules of a SAM, α, is tuned due to depolarization effects (e.g., induced dipoles in neighbouring molecules)42,43,44. However, it is not known how α affects the tunnelling behaviour of junctions in which depolarization effects are reduced to a minimum. In addition, the molecular ionization potential directly relates to α and, consequently, changing α affects molecular frontier orbital energies45 and the energy level alignment of molecule–electrode interfaces45,46, but it is disputed whether an increase in α changes the conductance of the junction45,46,47. Also not currently understood is how α affects the relationship between εr and β. In principle, polarizable groups screen applied electric fields48 or result in an induced dipole and, therefore, also affect the potential drop profile inside junctions45,46. Thus polarizable atoms or moieties are expected to have a large effect on the measured tunnelling rates, but so far experimental examples are rare and conflicting45,46,47. For instance, Whitesides and co-workers reported that the charge transport rates in metal-S(CH2)nFG//EGaIn junctions with aliphatic SAMs are independent of FG with FG being terminal aromatic groups49, polar groups50, ionic and/or hydrogen bonding groups51, or halogen atoms47, and concluded that changes in terminal group does not affect the charge transport rates. In these studies they used large junction areas of >1000 µm2, but we have shown that such large junctions are prone to defects masking molecular effects and that, for EGaIn-based methods, stable junctions that are dominated by molecular effects should have an area of 300–500 µm2 (ref. 52). Indeed, the Whitesides’ group could reproduce our results and also found a factor of 600 in the charge transport rates when X = H was replaced with X = Br when small junctions were used47.

Here, we show that the value of β of molecular wires with an alkyl chain backbone can be reduced from 0.75 to 0.25 Å−1, in effect turning them from insulators into conductors without changing the chemical structure of the backbone of the molecular wire, by introducing one distal polarizable atom at one end of the molecular wire of the form HS(CH2)nX. Changing X from H to I in the long S(CH2)18X molecular wire gives a factor of 10≈5 increase in J. For S(CH2)10X, the currents change by a factor of 10≈2. As we will discuss below, these observations cannot be explained by changes in the molecule–electrode interfaces, or contact resistances, alone. While we have shown before that the halide group affects the current and εr in Ag–S(CH2)11X//GaOx/EGaIn junctions45, here we demonstrate that the value of β can be controlled by changing X without the need to modify the chemical structure of the molecular backbone. This change in β explains why the largest change in current is found for the longest molecules studied in this work. On the basis of experimental and theoretical data, we discuss how introducing this polarizable atom changes the electrostatic potential profile of the tunnelling barrier, the εr of the junction, and the contact resistance (or Γ), which are important to consider when modifying tunnelling efficiency across molecular wires.

Results

The junctions

Figure 1 shows a schematic illustration of the Ag–S(CH2)nX//GaOx/EGaIn junctions, and indicates how the coupling and energy level alignment (i.e., Γ and δEME) change with X as discussed in detail below. The schematic also includes the equivalent circuit consisting of the contact resistance (RC, in mΩ cm2) in series with a parallel combination of the SAM resistance (RSAM, in Ω cm2) and the capacitance of the SAM (CSAM, in µF/cm2) in the junction. The equivalent circuit and the associated physical meaning of each circuit component has been explicitly discussed in our previous work41 (and is summarized in Supplementary Section 7). Briefly, the RC includes the resistances of the contacts of the SAM with the top and bottom electrodes, and the resistance of the electrodes and wires connecting the junction with the electrometers. The SAM itself behaves as a capacitor (CSAM) with associated resistance (RSAM) as expressed in Eqs. (2) and (3). It highlights that the junctions are essentially parallel plate capacitors in which dielectric behaviour depends on the chemical structure of the junctions, which, as we show below, is also important to explain tunnelling rates. All SAM precursors were synthesized following previously reported methods and characterized with 1H NMR, 13C NMR, and mass spectroscopy (Supplementary Sections 1 and 2). The SAMs were formed on template-stripped Ag electrodes using well-established methods and the junctions were completed with cone-shaped GaOx/EGaIn top contacts53 (Supplementary Sections 3 and 6). Previously, we have reported that for EGaIn junctions with S(CH2)nX SAMs (there only n = 11 was studied) the measured current increased by three orders of magnitude and the value of εr increased by a factor of 4, when X was changed along the halogen series from F to I45. However, the evolution of J and ɛr with increasing molecular length and the corresponding β values for different X have so far not been studied. Here we address whether this increase in current is caused by changes in RC or by changes from coherent tunnelling to incoherent process (i.e., β). Changing the value of n for each X allows us to investigate in detail how and why β changes as a function of X while keeping the nature of the molecule–electrode interfaces and the molecular backbone the same.

Fig. 1: The junctions, equivalent circuit, and energy level diagram.
figure 1

a Schematic illustration of the Ag–S(CH2)nX//GaOx/EGaIn junction (shown for n = 14, EGaIn is for eutectic alloy of Gallium and Indium, “-” represents covalent bond, “//” represents non-covalent contact, “/” means the interface between GaOx and EGaIn) together with the equivalent circuit diagram. In this work we investigated junctions with n = 10, 12, 14, 16, or 18, and X = H, F, Cl, Br, or I. b Energy level diagram of the junction showing how the coupling strength between molecules and electrodes (Γ) and tunnelling barrier height (δEME) change with X.

Characterization of the SAMs

We characterized the SAMs on Ag with n = 14 for X = H, F, Cl, Br, or I, and n = 10, 14, or 18 for X = Br with angle-resolved X-ray photoelectron spectroscopy (ARXPS) and molecular dynamics (MD) simulations (for all combinations of n and X) and all results are summarized in Table 1 (see Supplementary Sections 4 and 5 for details). Figure 2a shows a representative snapshot from the MD simulations of Ag–S(CH2)14I SAM with computed molecule heights in excellent agreement with film thicknesses dSAM measured by XPS (Fig. 2b, c and Table 1) indicating that the S(CH2)14I precursor readily forms dense layers with all molecules in a fully-upright position. We determined the relative values of surface coverage (ΨSAM) with XPS, which confirms that all SAMs have indistinguishable packing densities (Fig. 2b, c) within experimental error. The value of dSAM,MD increases by about 1.4 Å overall on increasing van der Waals radius of X from H to I (Supplementary Table 1), but this small increase falls within the experimental error of dSAM,XPS (Fig. 2b). Figure 2c shows that for X = Br, dSAM,XPS increases linearly with n with a slope of 1.5 ± 0.1 Å per carbon (solid blue line, error represents standard error from linear fit), which is in close agreement with the MD value of 1.3 ± 0.1 Å per carbon (dashed blue line; see Supplementary Fig. 11 for dSAM,MD values of all the SAMs). Figure 2d shows the packing energies per molecule (Emol,MD, in eV) and per methylene CH2 unit (Emeth,MD, in meV) extracted from the MD calculations. The values of Emol,MD and Emeth,MD improve slightly as X shifts from H (−1.8 ± 0.1 eV per molecule) to Br (−2.4 ± 0.2 eV per molecule), which is due to the increasing intermolecular van der Waals interaction. For SAMs with X = I, the packing energies weaken slightly due to small competing effects caused by mild steric repulsion between the large I headgroups. These observations confirm that the halogen functionality does not significantly disrupt the supramolecular structure of the SAM. Finally, we determined the energy level alignment of the SAMs on Ag using ultra-violet photoemission spectroscopy in Supplementary Section 4 and Supplementary Fig. 10, which we used to validate our density-functional theory (DFT) calculations as discussed in more detail below (Table 1).

Table 1 Summary of properties of the Ag–S(CH2)nX SAMs.
Fig. 2: Characterization of the self-assembled monolayers (SAMs).
figure 2

a Representative slice-through of a large-area Ag–S(CH2)14I SAM structure calculated by molecular dynamics (MD) computer simulations. b Surface coverage (ΨSAM) of Ag–S(CH2)14X SAMs as a function of X determined with angle-resolved X-ray photoelectron spectroscopy (ARXPS, filled circles) and thickness of SAM (dSAM) determined with ARXPS (filled triangles) and MD (empty triangles). c ΨSAM of Ag–S(CH2)nBr SAMs as a function of n determined with ARXPS (filled circles) and dSAM determined with ARXPS (filled triangles) and MD (open triangles). The solid and dashed blue lines are linear fits to the experimental and MD data with R2 of 0.94 and 0.99, respectively. The horizontal dashed line in panels b and c indicates the ΨSAM used in the MD calculations. d Computed MD packing energy per molecule Emol,MD and per methylene –CH2– unit Emeth,MD of Ag–S(CH2)14X SAMs as a function of X. Dashed lines are guides to the eye. The errors on the XPS data represent instrumental and fitting errors of 10% in total (see Section S4). The error bars in the MD data represent the standard deviations in the time- and molecule-averages calculated across 500 snapshots taken during the final 50 ns of 100 ns of room temperature MD of 128-molecule Ag–S(CH2)14X SAMs with the average experimental coverage of 1 nmol/cm2 on Ag(111).

Electrical characterization of the junctions

To study how the halogen functionality affects the tunnelling rates across the SAMs, we measured the electrical characteristics of the junctions as a function of X and n using J(V) measurements and impedance spectroscopy. The SAMs were contacted with cone-shaped GaOx/EGaIn electrodes following a previously reported method53. To minimize leakage currents and to ensure that molecular effects dominate the junction characteristics, we used junctions with a small contact area of ~350 μm2, as large junctions suffer from leakage currents across defective sites52. We recorded statistically large numbers of J(V) curves to determine the Gaussian log-average J(V) curves, <log10|J| > G, and associated Gaussian log-standard deviations (σlog,G) which are plotted in Fig. 3a for junctions with X = F, and in Fig. 3b for junctions with X = I, for n = 10–18 (all Gaussian log-average J(V) curves and histograms of the log10 |J| at ±0.5 V are given in Supplementary Section 6). Clearly, the tunnelling rates are more attenuated for X = F than for X = I. Figure 3c shows the decay of <log10|J| > G at –0.5 V as a function of dSAM,MD (Supplementary Section 5 and Supplementary Table 1) for all X. The solid lines are fits to Eq. (1) from which we determined the values of β which are listed in Supplementary Table 7. Supplementary Figure 18 shows the plot of β vs. X. Interestingly, the value of β steadily decreases from 0.75 ± 0.01 Å−1 for X = H—a typical value for tunnelling along alkyl chains—to 0.25 ± 0.01 Å-1 for X = I which is a typical value for tunnelling along π-conjugated molecules (the error in β represents the standard error of the fit to Eq. 1). We measured the J(V) characteristics as a function of temperature, T in K, of Ag–S(CH2)14X//GaOx/EGaIn junctions for all X using top electrode of EGaIn confined in a microfluidic network in polydimethylsiloxane following a previous reported method41 (see Supplementary Section 6 for details). Figure 3d shows that the tunnelling rates are independent of T in the range of T from 250 to 340 K, which is consistent with coherent off-resonant tunnelling54.

Fig. 3: Electrical characterization of junctions.
figure 3

Gaussian log-average values of the current densities <log10|J| > G vs. applied bias V obtained from Ag–S(CH2)nX//GaOx/EGaIn junctions with X = F (a) or I (b) and n = 10 (solid black line), 12 (solid red line), 14 (solid blue line), 16 (solid pink line), and 18 (solid green line). The dashed-line error bars represent the Gaussian log-standard deviation, σlog,G. c Decay plots of <log10|J| > G at –0.5 V against dSAM,MD with X = H (black square), F (red circle), Cl (blue triangle), Br (pink inverted triangle), or I (green diamond). The solid lines are fits to Eq. (1). The dashed lines represent the 95% confidence bands. d Plots of log10|J| vs. V as a function of T (T = 250–340 K) recorded from Ag–S(CH2)14X//GaOx/EGaIn junctions.

Dielectric constant of the junctions

To characterize the dielectric response of the junctions, we conducted impedance spectroscopy using a sinusoidal voltage perturbation with an amplitude of 30 mV around 0 V in the frequency range of 100 Hz to 1.00 MHz and the data were fitted to the equivalent circuit shown in Fig. 1a following a previously reported method41 (see Supplementary Section 7 for details). Supplementary Figure 23 shows the Bode, Nyquist, and the corresponding phase angle (ϛ) vs. frequency (f) plots along with the fits to the equivalent circuit (Supplementary Tables 810 list all fitting results). Figure 4a shows that RC decreases by a factor of 5 when X is changed from H or F to I while RC is independent of n (Fig. 4b). This change in RC indicates that Γ substantially increases as a function of X. This increase in Γ can be rationalized by the increase in polarizability α and associated induced dipoles as a function of X resulting in an increase in the van der Waals interaction strength between the SAM and the top contact45. Supplementary Table 9 shows the decrease of RSAM with X which is mainly caused by lowering of \(\delta {E}_{\mathrm{{ME}}}\) and increase of \(\varGamma\) (see the “DFT Calculations” section). Frisbie and co-workers9,38 have shown that a decrease in RC by increasing the work function of the bare metal electrodes increases the conductivity of molecular junctions (with H or S terminal atoms), which was mainly driven by a large increase in Γ, with changes in \(\delta {E}_{\mathrm{{ME}}}\) and β playing only a minor role.

Fig. 4: Characterization of junctions with impedance spectroscopy.
figure 4

a Contact resistance RC vs. X for Ag–S(CH2)14X//GaOx/EGaIn junctions at DC of 0 V and sinusoidal perturbation of 30 mV. Log-resistance of SAM, log10RSAM, (b) and RC (c) vs. n for Ag–S(CH2)nBr//GaOx/EGaIn junctions. The solid black line represents a fit to Eq. (2). d Corresponding dielectric constant εr vs. X for Ag–S(CH2)14X//GaOx/EGaIn junctions. The error bars are the standard deviations of three independent measurements. Dashed lines are visual guides.

To confirm the consistency between the J(V) and impedance measurements, we determined the value of β from the impedance measurements for junctions with X = Br. The value of RSAM increases exponentially with n (Eq. 2)

$${R}_{{\rm{SAM}}}={R}_{{\rm{SAM}},0}(V){\mathrm{e}}^{\beta {d}_{{\rm{SAM}},{\rm{MD}}}}={R}_{{\rm{SAM}},0}(V){10}^{\beta {d}_{{\rm{SAM}},{\rm{MD}}}/2.303}$$
(2)

where RSAM,0 is a pre-exponential factor. Figure 4c shows the plot of log10RSAM vs. n along with a fit to Eq. (2) from which we extracted the values of β = 0.41 ± 0.03 Å−1 and log10RSAM,0 = −2.0 ± 0.2 Ω/cm2 (or RSAM,0 = 1.0 × 102 Ω/cm2). The value of RSAM,0 is essentially equivalent to J0 (defined in Eq. 1) derived from a current decay plot at 30 mV (since the sinusoidal perturbation used in the impedance measurements was 30 mV). The value of J0 at 30 mV is 5.1 ± 2.0 A/cm2 and the β = 0.46 ± 0.03 Å−1 (Supplementary Fig. 17). RSAM,0 ≈ V/J0 = 0.59 × 10−2 Ω/cm2, which is within a factor of 2 of the value measured with impedance spectroscopy (RSAM,0 = 1.0 × 10−2 Ω/cm2). The contribution of RC is minor since RC is a parallel circuit element, but it is included in J0. The β and RSAM,0 values are, within error, the same as the values determined with the J(V) measurements.

To gain further insight into the dielectric properties of the junctions, we used the parallel plate capacitor equation (Eq. 3) to determine εr as a function of X and n

$${C}_{{\rm{SAM}}}={\varepsilon }_{0}{\varepsilon }_{r}\frac{{A}_{{\rm{geo}}}}{{d}_{{\rm{SAM}},{\rm{MD}}}}$$
(3)

wherein ε0 is the vacuum permittivity and Ageo is the geometrical area of the junction. Figure 4d shows that εr increases by a factor of 3 when changing X from H or F to I, yet εr is independent of n (Supplementary Tables 910). Although this factor 3 increase in εr is expected for bulk systems which can be described via the Clausius–Mosotti relation55, this observation cannot be explained as an intrinsic electrostatic property of the molecular wires, as we will show and discuss below. Moreover, even though we can quantify the contributions of different circuit components from impedance spectroscopy, how these components are influenced by each other are not directly revealed. Therefore, we referred to DFT and Landauer modelling for further explanations.

DFT calculations

To provide further microscopic insight into the electrostatic properties and electronic structure of the molecular wires, we performed first-principles calculations based on DFT using the VASP code56 and a 3 × 2√3 Ag surface unit cell containing four molecules arranged in a herringbone pattern (see Supplementary Section 8 for full details).

Figure 5a shows that the shape of the potential energy towards the tail of the alkyl chain strongly depends on the functionalization at the X-site. Specifically, the vacuum level changes with X functionalization owing to the polarity of the C–X bond, which translates into a change of the Ag work function, Φ, as expected for SAMs with different tail groups45,46,57,58. Comparing the DFT-calculated Φ to the experimental Φ, it can be seen that the agreement is good for all terminations except X = F (Fig. 5b). We tentatively ascribe the quantitative deviations to the often observed overestimation of polar effects in periodic DFT calculations of metal–SAM interfaces due to the assumption of perfect molecular order and periodicity59, while practical systems have defects (e.g., step edged, grain boundaries, or phase domains) and are dynamic in nature52,60.

Fig. 5: Density-functional theory (DFT) calculations.
figure 5

a DFT-calculated plane-averaged electrostatic potential of Ag(111)–S(CH2)14X where X = H, F, Cl, Br, or I, along the surface-normal coordinate. b Work function (Φ) of the SAMs, calculated from DFT (red squares) and measured experimentally (black dots). Density of states (DOS) projected onto the molecular backbone (c) and onto the X-site (d) of Ag(111)–S(CH2)14X. Note that X contributes very little to the band edges for X = H, F.

Figure 5c reports the density of states (DOS) projected onto the molecular part of Ag–S(CH2)14X. All systems show a feature at ~1.5 eV (marked by * in Fig. 5c) that is due to Ag–S hybridization (Supplementary Section 8). Interestingly, we find that lower lying occupied states as well as the lowest unoccupied state strongly shift in energy (on the order of 1–2 eV) with varying X (see arrows in Fig. 5c). These energy shifts clearly correlate with X functionalization and increase in magnitude along the halogen series, so that the X = I SAM shows pronounced new features close to the band edges when compared to the X = H or X = F SAM. Figure 5d shows the DOS projected onto just the X-site in Ag–S(CH2)14X, which confirms that these new occupied and unoccupied states are due to the halogen functionalization. For both the occupied and unoccupied parts of the DOS, these halogen-derived states do partially overlap in energy with other features but are localized primarily at the tail of the SAM (Supplementary Section 8). The X groups are not redox-active, and even for X = I the HOMO is still ~1.7 eV below EF (Fig. 5c, d). Hence, the HOMO cannot enter the applied bias window of ±0.5 V (note the molecules with n = 10 tend to break down at higher voltages)61.

Finally, we determined εr as a function of X for the free-standing and hydrogen-terminated HS(CH2)14X SAMs using previously reported protocols44,62 (see Table 1 and Supplementary Section 8). Hereby, εr is calculated from the change in the dipole moment induced by an applied static electric field, with the atomic positions fixed at their equilibrium positions. Thus, the εr obtained in such a manner represents the instantaneous response of the electronic charge density to a static electric field. In contrast to the above-discussed experimental results, we find that εr hardly changes with X functionalization in our DFT calculations. This result is expected from purely electrostatic reasoning and fully in line with previous work by various groups42,43,44. Briefly, in these studies it has been shown from electrostatic and DFT calculations that varying the molecular polarizability of the SAM-forming molecules does not result in significant changes of εr in the densely packed conjugated SAMs due to depolarization effects arising from the neighbouring molecular dipoles in the SAM42,43,44. Therefore, the calculations show that tuning the molecular polarizability by changing X does not strongly impact the calculated εr of the HS(CH2)14X SAMs.

As pointed out by Natan et al.43, a competition between suppression of in-plane polarization and enhancement of out-of-plane polarization occurs in SAMs. The suppression dominates for densely-packed SAMs and, thus, the substituent X should not affect the calculated εr of the SAM in sharp contrast to our herin reported experimental findings. Our calculations as well as previous theoretical studies, however, only probe the intrinsic dielectric properties of the isolated highly organized SAM without contacts. The interaction between the SAM and the top electrode that is naturally present in the experimental determination of εr could affect the dielectric behaviour of the junction considerably, which would be consistent with the experimentally recorded trends for the RC shown in Fig. 4a. Given the high electric fields on the order of GV/m and the polarizable nature of X, the substituents may be partially charged during charge transport (especially iodines are well-known to readily accommodate electrons)63,64. We note that previously reported DFT calculations of εr of the HS(CH2)11X SAMs45 were incorrect due to simulation artefacts of uncompensated dipoles in the unit cell, which created a spurious correlation with experimentally measured εr values.

Single-level Landauer model

In the following, we discuss our results in the context of commonly used models to interpret charge transport through the S(CH2)nX molecular junctions. The single-level Landauer model is frequently used to model the current flowing across molecular tunnel junctions65. Here we modelled the current using the following expression:

$$I=\frac{Nq}{h}{\int }^{}{\int }_{\!\!\!\!-\infty }^{\infty }{\mathrm{d}}E{\mathrm{d}}{E}^{{\prime}}{D}_{{E}^{\prime}}(E){G}_{\delta {E}_{\mathrm{{ME}}}}({E}^{{\prime}})\frac{{\gamma }_{\mathrm{L}}{\gamma }_{\mathrm{R}}}{{\gamma }_{\mathrm{L}}+{\gamma }_{\mathrm{R}}}[{f}_{\mathrm{L}}(E)-{f}_{\mathrm{R}}(E)]$$
(4)

where \({\gamma }_{\mathrm{L}}\) and \({\gamma }_{\mathrm{R}}\) are the tunnelling rates between the molecule and the left and right electrodes (respectively), \({D}_{E^{\prime}}(E)\) is the electronic density of states of the molecular level having the shape of Lorentzian and is given by

$${D}_{{E}^{{\prime}}}(E)=\frac{\frac{\gamma}{2\pi }}{{\left(\right.E-\left({E}^{{\prime}}+\left(\eta -\frac{1}{2}\right) \times V\right)}^{2}+{\left(\frac{y}{2}\right)}^{2}}$$
(5)

centred at energy \({E}^{\prime}+\left(\eta -\frac{1}{2}\right)V\), where \(\eta ={V}_{\mathrm{R}}/({V}_{\mathrm{L}}+{V}_{\mathrm{R}})\) is the voltage division parameter accounting for the capacitive coupling with the left and right electrodes, and with a level width \(\gamma ={\gamma }_{\mathrm{L}}+{\gamma }_{\mathrm{R}}\). The \({f}_{\mathrm{L}}(E)\) and \({f}_{\mathrm{R}}(E)\) are the Fermi functions representing the electronic occupation of the left and right electrodes, respectively, which are given by65

$${f}_{{\mathrm{{L,R}}}}(E)=\frac{1}{1+\exp \left[\frac{E\pm \frac{V}{2}}{{K}_{\mathrm{B}}T}\right]}$$
(6)

Equations 46 provide the model to which we fitted the experimental data. In addition, we attached a Gaussian to the model with the inherent dispersion (\(\sigma\)) of the molecular level energy (\(\delta {E}_{\mathrm{{ME}}}\)) in an ensemble of molecules (rather than a single-molecule junction), as given by the following expression:

$$\begin{array}{c}{G}_{\delta {E}_{\mathrm{{ME}}}}({E}^{{\prime}})=A\,\exp\left(\frac{{\left({E}^{{\prime}}-\,\delta {E}_{{\rm{ME}}}\right)}^{2}}{2{\sigma }^{2}}\right)\end{array}$$
(7)

We accounted for the behaviour of a group of molecules by setting the number of such molecules fixed at N = 150. As obviously seen in the above model, the current is directly dependent on \({\gamma }_{\mathrm{L}}\times {\gamma }_{\mathrm{R}}\). There is a trade-off between the Gaussian and the density of states which is in the shape of a Lorentzian centred at the energy level \(\delta {E}_{{\mathrm{{ME}}}}\). All the molecules appeared to be symmetric and what accounts for the difference in conductance is the terminal atom on the molecular unit. In this case, the ligands were X = H, F, Cl, Br, or I. Therefore five different set of fittings were done for each X, fixing \(\delta {E}_{{\mathrm{{ME}}}}\) to the values extracted from DFT and leaving \({\gamma }_{\mathrm{L}}\), \({\gamma }_{\mathrm{R}}\), \(\eta\) and \(\sigma\) as fitting parameters to obtain best fits to the data of junctions of Ag–S(CH2)14X//GaOx/EGaIn (Supplementary Table 12). Figure 6a shows the fits of the theoretical model (orange lines) to the experimental data (symbols) for each S(CH2)14X molecule. Figure 6b, c shows the two parameters that vary across molecules: the energy \(\delta {E}_{\mathrm{{ME}}}\) of the frontier orbital (extracted from DFT), which decreases from 4.3 to 1.7 eV on moving through the sequence H–F–Cl–Br–I (Fig. 6b), and the overall tunnelling rate through the junction (i.e., the Γ), defined as \(\varGamma =\frac{{\gamma }_{\mathrm{L}}{\gamma }_{\mathrm{R}}}{{\gamma }_{\mathrm{L}}+{\gamma }_{\mathrm{R}}}\), which increases exponentially along the halogen sequence (Fig. 6c) and accounts for the observed exponential increase of the current through the junctions.

Fig. 6: Single-level Landauer model analysis.
figure 6

a The modelled current through the Ag–S(CH2)14X//GaOx/EGaIn junctions using Landauer theory (orange solid lines are Landauer fits, symbols represent experimental data). The values of tunnelling barrier height \((\delta {E}_{{\rm{ME}}})\) (b) and the coupling strength \((\varGamma )\) (c) used for modelling the current through the junctions. d Tunnelling decay coefficient β vs. \(\sqrt{\delta {E}_{{\mathrm{{ME}}}}}\) with a linear fit (red line), the error bars represent the standard deviations of the β values from linear fits to Eq. (1). e Double-log plot of RC vs. 1/Γ2 (RC represents contact resistance) where the red line is a power-law fit with a slope of 0.25 and R2 = 0.99, error bars of RC represent the standard deviations of three independent measurements. f Double-log plot of β vs. \({\varepsilon }_{r}\) where the red line is a fit with a slope of −0.82 and R2 = 0.99. The error bars of β represents the same as panel b, and of \({\varepsilon }_{\mathrm{r}}\) represent the standard deviations of three independent measurements.

Figure 6d shows a linear relationship between calculated \(\sqrt{\delta {E}_{\mathrm{{ME}}}}\) and measured β, which agrees with commonly used coherent tunnelling models2,5,18 including the Simmons model which also accounts for εr. However, the Simmons model also predicts a decrease of the tunnelling rates with increasing εr due to a reduction of the image charge effects in the electrodes due to screening within the SAM18,66. This reduction of image charge in effect increases δEME and, consequently, β, but we observe the opposite trend. Using the same model, a reduction of the effective electron mass could also account for an increase in tunnelling rates, but it is not clear how the effective electron mass would change as a function of X with the essentially localized features derived from the HOMO and LUMO. Furthermore, Vilan67 argued that changes in the electron mass are equivalent to changes in the \(\delta {E}_{\mathrm{{ME}}}\) within the Simmons model, which further complicates the interpretation of our findings within this framework.

The experimentally determined values of \({R}_{{\rm{C}}}\,\) have been related to \(\varGamma\) as \({R}_{{\rm{C}}}\propto {\varGamma }^{-2}\) (ref. 30), i.e., the coupling of the molecules with the electrode we have determined above (Fig. 6e). To test whether this holds for the SAMs studied here, Fig. 6e shows a double-log plot of RC vs. 1/\(\,\varGamma\)2, indicating that our results can be explained, at least qualitatively, using this picture: changes in both \(\delta {E}_{\mathrm{{ME}}}\) and \(\varGamma\) can lower β, in accordance with findings by others27,28,29,30,31,32,33,34,35. This approach, however, does not capture the observed changes in the dielectric response of the junctions directly, and, of course, it does not explicitly account for the local changes in the electrostatic potential profile induced by X observed in the DFT calculations; these effects are essentially compensated by the large change in Γ of 29 times.

An interesting finding was reported by Berlin and Ratner68 based on an alternative model to describe tunnelling across barriers with charge traps, with the finding that \(\beta\propto 1/\sqrt{{\varepsilon }_{r}}\). In this framework, the distance dependence of the conductance is related to a thickness-dependent barrier akin to the one inherent to the Simmons model67,69. In their model, however, the barrier arises from the presence of localized charge traps along the path of charge migration leading to a non-linear potential drop between the macroscopic leads. Figure 6f shows the linear relation of the double-log plot of β vs. \(\,{\varepsilon }_{\mathrm{r}}\) with a slope of −0.82 which is lower than the expected −0.5 from the model by Berlin et al.68, but note that a change in the contact resistance or further changes in the barrier shape are not taken into consideration in this model. In our experiments, however, the contact resistance changes and our DFT calculations show that the barrier shape at the SAM//top electrode interface is affected by X.

Discussion

This work shows that substitution of a single highly polarizable atom can have a pronounced effect on the energy level alignment, charge transport rate, and dielectric response of molecular junctions. We were able to tune β over a wide range from 0.25 to 0.75 Å1 across saturated alkyl chains by changing one atom per molecule inside large-area (~350 μm2)52 junctions. The largest effects of X on J are found in the longest molecules of S(CH2)18X where the J increases by a factor of 10≈5 when X changes from H to I. In contrast, for the shortest molecule S(CH2)10X, J increases by a factor of 10≈2, indicating that the observed changes in J, and the corresponding values of β, are driven by more than just changes in the interfaces, which has been discussed previously by Frisbie and co-workers9,38. Combining experiment with DFT and Landauer charge transport models, we established three factors that contribute to the dramatic change in β of these aliphatic halogenated junctions with varying X and associated increase in α and εr: (1) The HOMO-LUMO gap5,45 and associated δEME is reduced which lowers β, (2) the shape of the tunnelling barrier is modified at the SAM-top electrode interface, resulting in larger potential drops at this interface, and (3) the electronic coupling Γ of the molecular orbitals with the electrodes increases (potentially because of an increase in the van der Waals interactions along the halogen series).

In a broad context of widely used charge tunnelling mechanisms, our findings point out their limitations highlighting the need for improved models that take dielectric (or collective) effects of the junctions into consideration. Specifically, the popular Simmons model predicts that image charge effects in the electrodes are reduced with increasing εr, resulting in lowering of the tunnelling rates67, which is in sharp contrast to what we find. Superexchange models19,20,21 also fail to explain our observations, since the coupling between the molecular repeat unit (i.e., the CH2 units) was not changed here. Conversely, the Landauer model65 could explain our results at least qualitatively, but not quantitatively. This is because it does not treat electrostatic effects in the junctions explicitly and self-consistently which resulted in our case in a large increase in the values of Γ (29 times) even though the \({R}_{{\rm{C}}}\,\) only changed by a factor of five in our experiments. Interestingly, a mechanism proposed by Berlin and Ratner68 that is based on charge traps provides a hypothesis for how the value of β could decrease with increasing εr. Although the physical interpretation differs as the tunnelling behaviour is explained in terms of charge traps rather than the electrostatic response of the SAM inside the junction, this line of thought stimulates further theoretical and experimental testing of the presence of “impurities”—here in the form of polarizable atoms—as charge carriers that move across the energy band profiles. Although our findings suggest a correlation between β and εr, the increase of εr as a function of X could not be reproduced in our DFT calculations, which may be because the calculations do not take the SAM–metal interface into consideration, and perhaps other factors are important such as (partial) charging of highly polarizable molecules during charge transport inside the junctions. To summarize, our work proposes an effective way of tuning the tunnelling rates across molecular junctions without chemically altering the backbone of the molecules and highlights the importance of understanding dielectric effects in these junctions. We hope that our findings will stimulate further experimental and theoretical investigations towards establishing improved transport mechanisms for junctions in their in situ physicochemical environment and electronic states inside working devices.

Data availability

The data that support the findings of this study are available within the article and Supplementary Information file, or at https://doi.org/10.7910/DVN/WZAIGUSource data are provided with this paper.

Code availability

The code for the analysis of J(V) results in this study is available at https://doi.org/10.7910/DVN/WZAIGU.

References

  1. Shah, A. et al. Electron transfer in peptides. Chem. Soc. Rev. 44, 1015–1027 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  2. Vilan, A., Aswal, D. & Cahen, D. Large-area, ensemble molecular electronics: motivation and challenges. Chem. Rev. 117, 4248–4286 (2017).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  3. Jortner, J., Nitzan, A. & Ratner, M. A. in Introducing Molecular Electronics (eds. Cuniberti, G. et al.) 13–54 (Springer, 2005).

  4. Jia, C. & Guo, X. Molecule–electrode interfaces in molecular electronic devices. Chem. Soc. Rev. 42, 5642–5660 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  5. Metzger, R. M. Unimolecular electronics. Chem. Rev. 115, 5056–5115 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  6. Jeong, H., Kim, D., Xiang, D. & Lee, T. High-yield functional molecular electronic devices. ACS Nano 11, 6511–6548 (2017).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  7. Hylke, B. A. & de Bert, B. Electrical conduction through single molecules and self-assembled monolayers. J. Phys. Condens. Matter 20, 013001 (2008).

    Article  CAS  Google Scholar 

  8. Nitzan, A. Electron transmission through molecules and molecular interfaces. Annu. Rev. Phys. Chem. 52, 681–750 (2001).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  9. Xie, Z., Bâldea, I. & Frisbie, C. D. Determination of energy-level alignment in molecular tunnel junctions by transport and spectroscopy: self-consistency for the case of oligophenylene thiols and dithiols on Ag, Au, and Pt electrodes. J. Am. Chem. Soc. 141, 3670–3681 (2019).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  10. Kumar, K. S., Neoh, E. H. L., Vilan, A. & Nijhuis, C. A. Protective layers based on carbon paint to yield high-quality large-area molecular junctions with low contact resistance. J. Am. Chem. Soc. 142, 3513–3524 (2020).

    Article  CAS  Google Scholar 

  11. Simeone, F. C. et al. Defining the value of injection current and effective electrical contact area for EGaIn-based molecular tunneling junctions. J. Am. Chem. Soc. 135, 18131–18144 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  12. Karuppannan, S. K., Pasula, R. R., Lim, S. & Nijhuis, C. A. Long-range tunneling processes across ferritin-based junctions. Adv. Mater. 28, 1824–1830 (2016).

    Article  CAS  Google Scholar 

  13. Christopher, D. B. et al. Protein bioelectronics: a review of what we do and do not know. Rep. Prog. Phys. 81, 026601 (2018).

    Article  CAS  Google Scholar 

  14. Baghbanzadeh, M. et al. Charge tunneling along short oligoglycine chains. Angew. Chem. Int. Ed. 54, 14743–14747 (2015).

    CAS  Article  Google Scholar 

  15. Baghbanzadeh, M. et al. Anomalously rapid tunneling: charge transport across self-assembled monolayers of oligo(ethylene glycol). J. Am. Chem. Soc. 139, 7624–7631 (2017).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  16. Su, T. A., Li, H., Steigerwald, M. L., Venkataraman, L. & Nuckolls, C. Stereoelectronic switching in single-molecule junctions. Nat. Chem. 7, 215–220 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  17. Yan, H. et al. Activationless charge transport across 4.5 to 22 nm in molecular electronic junctions. Proc. Natl Acad. Sci. USA 110, 5326–5330 (2013).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  18. Simmons, J. G. Generalized formula for the electric tunnel effect between similar electrodes separated by a thin insulating film. J. Appl. Phys. 34, 1793–1803 (1963).

    ADS  Article  Google Scholar 

  19. Naleway, C. A., Curtiss, L. A. & Miller, J. R. Superexchange-pathway model for long-distance electronic couplings. J. Phys. Chem. 95, 8434–8437 (1991).

    CAS  Article  Google Scholar 

  20. McConnell, H. M. Intramolecular charge transfer in aromatic free radicals. J. Chem. Phys. 35, 508–515 (1961).

    ADS  CAS  Article  Google Scholar 

  21. Todd, M. D., Nitzan, A. & Ratner, M. A. Electron transfer via superexchange: a time-dependent approach. J. Phys. Chem. 97, 29–33 (1993).

    CAS  Article  Google Scholar 

  22. Juhaniewicz, J. & Sek, S. Peptide molecular junctions: distance dependent electron transmission through oligoprolines. Bioelectrochemistry 87, 21–27 (2012).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  23. Beratan, D. N. et al. Charge transfer in dynamical biosystems, or the treachery of (static) images. Acc. Chem. Res. 48, 474–481 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  24. Zhang, Y., Liu, C., Balaeff, A., Skourtis, S. S. & Beratan, D. N. Biological charge transfer via flickering resonance. Proc. Natl Acad. Sci. USA 111, 10049–10054 (2014).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  25. Berlin, Y. A., Kurnikov, I. V., Beratan, D., Ratner, M. A. & Burin, A. L. in Long-Range Charge Transfer in DNA II (ed. Schuster, G. B.) 1–36 (Springer, 2004).

  26. Amdursky, N. et al. Electronic transport via proteins. Adv. Mater. 26, 7142–7161 (2014).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  27. Sangeeth, C. S. S. et al. Comparison of DC and AC transport in 1.5–7.5 nm oligophenylene imine molecular wires across two junction platforms: eutectic Ga–In versus conducting probe atomic force microscope junctions. J. Am. Chem. Soc. 138, 7305–7314 (2016).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  28. Nguyen, Q. V. et al. Highly efficient long-range electron transport in a viologen-based molecular junction. J. Am. Chem. Soc. 140, 10131–10134 (2018).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  29. Sepunaru, L. et al. Electronic transport via homopeptides: the role of side chains and secondary structure. J. Am. Chem. Soc. 137, 9617–9626 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  30. Xie, Z., Bâldea, I., Smith, C. E., Wu, Y. & Frisbie, C. D. Experimental and theoretical analysis of nanotransport in oligophenylene dithiol junctions as a function of molecular length and contact work function. ACS Nano 9, 8022–8036 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  31. Gunasekaran, S. et al. Near length-independent conductance in polymethine molecular wires. Nano Lett. 18, 6387–6391 (2018).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  32. Sedghi, G. et al. Long-range electron tunnelling in oligo-porphyrin molecular wires. Nat. Nanotechnol. 6, 517–523 (2011).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  33. Leary, E. et al. Bias-driven conductance increase with length in porphyrin tapes. J. Am. Chem. Soc. 140, 12877–12883 (2018).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  34. Zhang, Y. et al. Tunneling probability increases with distance in junctions comprising self-assembled monolayers of oligothiophenes. J. Am. Chem. Soc. 140, 15048–15055 (2018).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  35. Algethami, N., Sadeghi, H., Sangtarash, S. & Lambert, C. J. The conductance of porphyrin-based molecular nanowires increases with length. Nano Lett. 18, 4482–4486 (2018).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  36. Sangtarash, S. et al. Gateway state-mediated, long-range tunnelling in molecular wires. Nanoscale 10, 3060–3067 (2018).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  37. Engelkes, V. B., Beebe, J. M. & Frisbie, C. D. Length-dependent transport in molecular junctions based on SAMs of alkanethiols and alkanedithiols: effect of metal work function and applied bias on tunneling efficiency and contact resistance. J. Am. Chem. Soc. 126, 14287–14296 (2004).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  38. Xie, Z., Bâldea, I. & Frisbie, C. D. Energy level alignment in molecular tunnel junctions by transport and spectroscopy: self-consistency for the case of alkyl thiols and dithiols on Ag, Au, and Pt electrodes. J. Am. Chem. Soc. 141, 18182–18192 (2019).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  39. Xie, Z., Bâldea, I. & Frisbie, C. D. Why one can expect large rectification in molecular junctions based on alkane monothiols and why rectification is so modest. Chem. Sci. 9, 4456–4467 (2018).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  40. Gu, M. W., Peng, H. H., Chen, I. P. & Chen, C. H. Tuning surface d bands with bimetallic electrodes to facilitate electron transport across molecular junctions. Nat. Mater. 20, 658–664 (2021).

  41. Sangeeth, C. S. S., Wan, A. & Nijhuis, C. A. Equivalent circuits of a self-assembled monolayer-based tunnel junction determined by impedance spectroscopy. J. Am. Chem. Soc. 136, 11134–11144 (2014).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  42. Van Dyck, C., Marks, T. J. & Ratner, M. A. Chain length dependence of the dielectric constant and polarizability in conjugated organic thin films. ACS Nano 11, 5970–5981 (2017).

    PubMed  Article  CAS  PubMed Central  Google Scholar 

  43. Natan, A., Kuritz, N. & Kronik, L. Polarizability, susceptibility, and dielectric constant of nanometer-scale molecular films: a microscopic view. Adv. Funct. Mater. 20, 2077–2084 (2010).

    CAS  Article  Google Scholar 

  44. Romaner, L., Heimel, G., Ambrosch-Draxl, C. & Zojer, E. The dielectric constant of self-assembled monolayers. Adv. Funct. Mater. 18, 3999–4006 (2008).

    CAS  Article  Google Scholar 

  45. Wang, D. et al. Tuning the tunneling rate and dielectric response of SAM-based junctions via a single polarizable atom. Adv. Mater. 27, 6689–6695 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  46. Haj-Yahia, A.-E. et al. Substituent variation drives metal/monolayer/semiconductor junctions from strongly rectifying to ohmic behavior. Adv. Mater. 25, 702–706 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  47. Baghbanzadeh, M., Pieters, P. F., Yuan, L., Collison, D. & Whitesides, G. M. The rate of charge tunneling in EGaIn junctions is not sensitive to halogen substituents at the self-assembled monolayer//Ga2O3 interface. ACS Nano 12, 10221–10230 (2018).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  48. Stone, A. J. The Theory of Intermolecular Forces (Oxford University Press, 2013).

  49. Yoon, H. J. et al. The rate of charge tunneling through self-assembled monolayers is insensitive to many functional group substitutions. Angew. Chem. Int. Ed. 51, 4658–4661 (2012).

    CAS  Article  Google Scholar 

  50. Yoon, H. J., Bowers, C. M., Baghbanzadeh, M. & Whitesides, G. M. The rate of charge tunneling is insensitive to polar terminal groups in self-assembled monolayers in AgTSS(CH2)nM(CH2)mT//Ga2O3/EGaIn junctions. J. Am. Chem. Soc. 136, 16–19 (2014).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  51. Bowers, C. M. et al. Introducing ionic and/or hydrogen bonds into the SAM//Ga2O3 top-interface of AgTS/S(CH2)nT//Ga2O3/EGaIn junctions. Nano Lett. 14, 3521–3526 (2014).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  52. Chen, X., Hu, H., Trasobares, J. & Nijhuis, C. A. Rectification ratio and tunneling decay coefficient depend on the contact geometry revealed by in situ imaging of the formation of EGaIn junctions. ACS Appl. Mater. Inter. 11, 21018–21029 (2019).

    CAS  Article  Google Scholar 

  53. Nerngchamnong, N. et al. The role of van der Waals forces in the performance of molecular diodes. Nat. Nanotechnol. 8, 113–118 (2013).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  54. Wang, W., Lee, T. & Reed, M. A. Mechanism of electron conduction in self-assembled alkanethiol monolayer devices. Phys. Rev. B 68, 035416 (2003).

    ADS  Article  CAS  Google Scholar 

  55. Ashcroft, N. W. & Mermin, N. D. Solid State Physics (Holt, Rinehart and Winston, 1976).

  56. Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).

    ADS  CAS  Article  Google Scholar 

  57. Lee, H. J., Jamison, A. C. & Lee, T. R. Surface dipoles: a growing body of evidence supports their impact and importance. Acc. Chem. Res. 48, 3007–3015 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  58. Chen, X. et al. Interplay of collective electrostatic effects and level alignment dictates the tunneling rates across halogenated aromatic monolayer junctions. J. Phys. Chem. Lett. 10, 4142–4147 (2019).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  59. Abu-Husein, T. et al. The effects of embedded dipoles in aromatic self-assembled monolayers. Adv. Funct. Mater. 25, 3943–3957 (2015).

    CAS  Article  Google Scholar 

  60. Jiang, L., Sangeeth, C. S. S., Yuan, L., Thompson, D. & Nijhuis, C. A. One-nanometer thin monolayers remove the deleterious effect of substrate defects in molecular tunnel junctions. Nano Lett. 15, 6643–6649 (2015).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  61. Yuan, L., Jiang, L. & Nijhuis, C. A. The drive force of electrical breakdown of large-area molecular tunnel junctions. Adv. Funct. Mater. 28, 1801710 (2018).

    Article  CAS  Google Scholar 

  62. Heitzer, H. M., Marks, T. J. & Ratner, M. A. First-principles calculation of dielectric response in molecule-based materials. J. Am. Chem. Soc. 135, 9753–9759 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  63. Rozhkov, I. N., Raevsky, A. M., Gushchin, S. I., Ignatova, J. L. & Semin, G. K. Evaluation of charge changes on chlorine, bromine and iodine atoms in perfluoroalkyl halogenides based on nuclear quadrupole resonance data. J. Fluor. Chem. 58, 234–234 (1992).

    Article  Google Scholar 

  64. Probst, A., Raab, K., Ulm, K. & von Werner, K. Synthesis and chemistry of perfluoro-2-iodo-2-methyl-alkanes. J. Fluor. Chem. 37, 223–245 (1987).

    CAS  Article  Google Scholar 

  65. Garrigues, A. R. et al. A single-level tunnel model to account for electrical transport through single molecule- and self-assembled monolayer-based junctions. Sci. Rep. 6, 26517 (2016).

    ADS  CAS  PubMed  PubMed Central  Article  Google Scholar 

  66. Simmons, J. G. Electric tunnel effect between dissimilar electrodes separated by a thin insulating film. J. Appl. Phys. 34, 2581–2590 (1963).

    ADS  MATH  Article  Google Scholar 

  67. Vilan, A. Analyzing molecular current-voltage characteristics with the simmons tunneling model: scaling and linearization. J. Phys. Chem. C 111, 4431–4444 (2007).

    CAS  Article  Google Scholar 

  68. Berlin, Y. A. & Ratner, M. A. Conduction of metal–thin organic film–metal junctions at low bias. J. Phys. Chem. C 122, 7557–7563 (2018).

    CAS  Article  Google Scholar 

  69. Karuppannan, S. K., Hu, H., Troadec, C., Vilan, A. & Nijhuis, C. A. Ultrasmooth and photoresist-free micropore-based EGaIn molecular junctions: fabrication and how roughness determines voltage response. Adv. Funct. Mater. 29, 1904452 (2019).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We acknowledge fruitful discussions with Ayelet Vilan (Weizmann Institute of Science) and Gemma Solomon (University of Copenhagen). Fundings by the Ministry of Education (MOE) for supporting this research under award No. MOE2019-T2-1-137 and R-143-000-B30-112 are acknowledged. Prime Minister’s Office, Singapore under its Medium sized centre programme is also acknowledged for supporting this research. The authors would furthermore like to acknowledge the Singapore Synchrotron Light Source (SSLS) for providing the facilities at the Surface, Interface and Nanostructure Science (SINS) beam line under NUS core support C-380-003-003-001. The Laboratory is a National Research Infrastructure under the National Research Foundation Singapore. We moreover acknowledge funding from the Alexander von Humboldt Foundation within the framework of the Sofja Kovalevskaja Award, endowed by the German Federal Ministry of and Research, and the Technical University of Munich—Institute for Advanced Study, funded by the German Excellence Initiative and the European Union Seventh Framework Programme under Grant Agreement No. 291763. D.T. thanks Science Foundation Ireland (SFI) for support (awards Grant Numbers 15/CDA/3491 and 12/RC/2275_P2) and for computing resources at the SFI/Higher Education Authority Irish Centre for High-End Computing (ICHEC). Finally, we also acknowledge support from the U.S. National Science Foundation (Grant No. ECCS#1916874). The authors gratefully acknowledge the Gauss Centre for Supercomputing e.V. (www.gauss-centre.eu) for funding this project by providing computing time through the John von Neumann Institute for Computing (NIC) on the GCS Supercomputer JUWELS at Jülich Supercomputing Centre (JSC).

Author information

Authors and Affiliations

Authors

Contributions

X.C., D.A.E., and C.A.N. conceived and designed the project. X.C. synthesized the compounds, performed electrical characterizations, and associated data analysis; B.K. and D.A.E. performed the DFT calculations; F.A., C.N., and E.d.B. performed the Landauer model analysis; X.C. and X.Y. conducted the XPS and UPS measurements; D.T. conducted the molecular dynamics; all the authors discussed the results and prepared the manuscript.

Corresponding authors

Correspondence to David A. Egger or Christian A. Nijhuis.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Source data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Chen, X., Kretz, B., Adoah, F. et al. A single atom change turns insulating saturated wires into molecular conductors. Nat Commun 12, 3432 (2021). https://doi.org/10.1038/s41467-021-23528-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s41467-021-23528-8

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing