Intramolecular London Dispersion Interactions in Single-Molecule Junctions

This work shows the first example of using intramolecular London dispersion interactions to control molecular geometry and quantum transport in single-molecule junctions. Flexible σ-bonded molecular junctions typically occupy straight-chain geometries due to steric effects. Here, we synthesize a series of thiomethyl-terminated oligo(dimethylsilmethylene)s that bear [CH2–Si(CH3)2]n repeat units, where all backbone dihedral states are sterically equivalent. Scanning tunneling microscopy break-junction (STM-BJ) measurements and theoretical calculations indicate that in the absence of a strong steric bias concerted intramolecular London dispersion interactions staple the carbosilane backbone into coiled conformations that remain intact even as the junction is stretched to its breakpoint. As these kinked conformations are highly resistive to electronic transport, we observe record-high conductance decay values on an experimental junction length basis (β = 1.86 ± 0.12 Å–1). These studies reveal the potential of using intramolecular London dispersion interactions to design single-molecule electronics.


■ INTRODUCTION
In isolation, a single London dispersion interaction is the weakest noncovalent interaction.However, when many concurrent London dispersion attractions occur simultaneously within a molecule, they can offer pronounced stabilization of sterically crowded molecular conformations that would otherwise be considered unfavorable. 1,2A groundswell of work exploiting this phenomenon has recently emerged across chemical disciplines.−14 This article describes the first example of using intramolecular London dispersion interactions to control charge transport in molecular electronics.
The motivation for our work lies in the broader interest of the single-molecule electronics community to design and use individual molecules as the active components (e.g., insulators, wires, rectifiers, switches) in electronic circuitry. 15A key strategy for controlling electronic transport in a singlemolecule junction is to control its conformation, since conformation dictates the strength of conjugation or coupling between electrodes through the molecular backbone. 16−25 Yang, and co-workers recently described London dispersion σinteractions to control the association between molecules in stacked supramolecular junctions; 26  This gap in the literature can be ascribed to the fact that most single-molecule wires with flexible σ-backbones have conformations that are overwhelmingly influenced by steric effects (Figure 1a).Though alkane wires are freely rotating, their backbones predominately exist in anti (ω = 180°) dihedral geometries due to repulsive steric crowding in their gauche (ω = 60°) conformations.The same is true in permethylated oligosilane and oligogermane wires, 27−29 where transoid (ω ∼ 165°) geometries are favored over gauche (ω ∼ 55°) and ortho (ω ∼ 90°) ones due to steric arguments (Figure 1a). 30These anti or transoid geometries become even more favored upon junction elongation, since the mechanical stretching axis is colinear with the straight-chain σ-bond axis.
But what happens if steric influences are removed from the backbone such that all rotational states around a single bond are energetically degenerate?This question motivated us to study carbosilanes [CSi] 2−4 (Figure 1b,2) with [CH 2 −SiMe 2 ] n repeat units in single-molecule junctions.−38 This article shows that, in the absence of a strong steric bias, intramolecular London dispersion σinteractions stabilize highly insulating gauche/ortho dihedral conformations in the backbone, even at full junction elongation.This leads us to observe the most insulating molecular junctions that have ever been measured on an experimental junction length basis. 39,40

RESULTS AND DISCUSSION
We end-functionalized carbosilanes [CSi] 2−4 (Figure 2) with gold-binding methylthiomethyl end groups (see Figure S1 and the Methods section for more details) 41−44 as these linkers form dative contacts with undercoordinated Au atoms that give well-defined conductance profiles. 28,45We synthesized alkane C n (n = 6,8,10) and oligosilane Si n (n = 2−6) (Figure 3d) as control compounds according to known procedures. 28,45We measured their single-molecule conductance properties using the scanning tunneling microscopy break-junction (STM-BJ) technique (see the Methods section). 46Ten thousand STM-BJ measurement traces for each molecule are compiled into a twodimensional (2D) conductance-displacement histogram (Figure 3a−c, Figure S2). 47,48We used a high voltage bias (1.0 V) to measure [CSi] 4 to clearly resolve the molecular peak from the noise floor of the instrument.In this specific case, increased voltage bias should not significantly change the conductance value we observe: the HOMO−LUMO gap of this molecule is large, so off-resonant transport should still occur at large applied biases.
We note that [CSi] 3 and particularly [CSi] 4 have a pronounced conductance-displacement slope, where conductance steadily decreases as displacement increases (Table S1, Figure S2).We thus focus on the average conductance value at full junction elongation (black circles, Figures 3a−c and S2) to compare conductance trends among all series.We define full junction elongation in this context as the point in the 2D histogram where there is a change in conductance-displacement slope near the breakpoint; an example of this analysis is given in Figure S3. 45The average conductance values at full elongation for the [CSi] n , Si n , and C n series are provided in Table 1 and plotted in Figure 3e against n, the number of backbone atoms between the distal sulfur atoms in each molecule.
This plot allows us to extract the length-dependent conductance decay parameter (β) value based on the equation, G ∼ e −β n .Τhe β value describes the strength of coupling between monomers in an oligomeric wire backbone, where low and high β values are ascribed to molecular conductors and insulators, respectively. 49The β values we obtain from this approach for the C n (β C = 0.96 ± 0.02 n −1 ) and Si n (β Si = 0.78 ± 0.03 n −1 ) wires are within error of previously reported β values for thiomethyl-terminated alkane (β = 0.94 n −1 ) and silane (β = 0.75 n −1 ) molecular wires.Surprisingly, we find that carbosilane backbones give a far steeper β value (β CSi = 1.50 ± 0.06 n −1 ) than purely carbon-based or silicon-based backbones.On a per atom basis, the [CSi] n β value is within range of the insulating oligo-(dimethylsiloxane) [SiO] n backbones (β SiO = 1.54 n −1 , Figure 2d,e) that give the largest β values reported in the singlemolecule literature. 40This steep conductance decay becomes more pronounced if we consider how conductance trends with experimentally determined junction length values.In Figure 3f, we plot conductance at full elongation against the 80th percentile step length. 53The [CSi] n series demonstrates a much steeper β value (1.86 Å −1 ) than all other σ-bonded wires, far exceeding that of any other material investigated in the literature including the previously reported [SiO] n oligo-(dimethylsiloxane)s (Figure 3f). 40This comparison highlights the potential of oligo(dimethylsilmethylene)s to serve as single-molecule insulators.
−56 While some degree of conductance attenuation is anticipated due to the polarity of the Si δ+ −C δ− backbones (EN Si = 1.8, EN C = 2.5), the β decay value that we observe experimentally is far greater than what we would expect simply from bond polarity arguments.
This point is captured clearly in our transmission calculations comparing the C n , Si n , and [CSi] n series.−59 with the PBE exchange-correlation functional 60 using the AITRANSS 30,31 postprocessor module within the FHI-aims 61 package (see the Methods section).While our transmission models of anti C n (β C, calc.= 0.96 n −1 ) and transoid Si n (β Si, calc.= 0.76 n −1 ) junctions give β values that match our experimental data (Figure S4), we find that all-transoid [CSi] n junctions give a significantly lower β value (β SiC, calc.= 0.99 n −1 ) than what we observe experimentally (β SiC, exp.= 1.50 n −1 ).This discrepancy is the first indicator that, despite the mechanical strain imposed on the carbosilanes upon junction stretching, the all-transoid geometries are not appreciably accessed at full junction elongation.
Indeed, one explanation for the high β value we observe experimentally is that gauche/ortho kinks remain in the backbone upon mechanical elongation (Figure 1b).−67 The introduction of these kinks also significantly lowers transmission in carbosilanes.Indeed, as we will show later in the manuscript, two ortho kinks in the backbone of [CSi] 3 lower transmission at the Fermi energy by 200-fold relative to all-transoid conformer.
The experimental step length trends in Table 1 further support the notion that [CSi] n oligomers maintain kinked backbone geometries at full junction elongation.Junction step length refers to the electrode displacement where a given molecular junction breaks.Though absolute step length values do not map directly with end-to-end molecular lengths due to electrode snapback 68−70 and linker group effects, 45 trends in step length provide valuable insights into the structural details of molecular junctions.The step lengths for all traces are plotted in the insets of Figure 3 and Figure S2, where a Gaussian fit gives the most probable step length value (z, Table 1).In Table 1, Δz denotes the average increase in z as C−C, Si−Si, or C−Si bonds are successively added to C 6 , Si 2 , and [CSi] 2 .We find that the Δz and bond length values are quite similar for the anti alkanes as well as transoid silanes, while the Δz value is much shorter than the C−Si bond length for the carbosilanes.Crucially, we find that carbosilane junctions (Δz = 0.13 nm) exhibit a much smaller Δz than alkane junctions (Δz = 0.16 nm), though C−Si bond lengths (0.19 nm) are much longer than the C−C bonds (0.14 nm) in alkanes.Previous studies on PDMSMs indicate that the CSiC and SiCSi bond angles are rigid and tetrahedral; 34,35,71 the only significant source of conformational flexibility is backbone torsion.As these kinked structures give shorter step lengths than fully transoid structures, the step length trends lead us to infer that gauche or ortho dihedral geometries persist in the backbone as the junction is stretched.
Three questions emerge from our central hypothesis that carbosilane junctions maintain coiled geometries upon mechanical stretching: (1) What is the energetic driving force that would cause kinked [CH 2 −SiMe 2 ] n backbones to be favored over transoid ones?(2) Why is this effect absent in the −CH 2 − (C n ) and −SiMe 2 − (Si n ) backbones?(3) If the σbonded backbone is colinear with the elongation axis, why does the mechanical strain of junction stretching not straighten the backbone to its longest transoid geometry?The DFT calculations below provide rationalization for these effects based on intramolecular London dispersion interactions.Despite extensive experimental and theoretical studies of PDMSM backbones in the polymer literature, a preference for kinked geometries has never been described.While most PDMSM conformational studies have been executed via molecular dynamics simulation, 33,34,38 Raptis and Melissas used DFT calculations (B3LYP/6-311G) to show that short PDMSM oligomer geometries are defined by transoid (t ± ) and ortho/gauche (o ± , g ± ) minima that are energetically degenerate. 35Crucially, the B3LYP hybrid functional used in this study did not account for intramolecular dispersion interactions.DFT corrections such as Grimme's D3 approach are necessary to account for dispersion in hybrid functionals. 72,73The impact of dispersion interactions is often quantified by evaluating relative energy differences between B3LYP and B3LYP-D3 calculations of the same set of structures. 1,2,74In Figure 4a,b, we apply this strategy to compare the energies of conformational minima in the linkerless [CSi] 3 -chain and Si 5 -chain analogues against their all-transoid (t + t + ) rotamers.As the only molecular differences distinguishing [CSi] 2 from [CSi] 4 are the additions of backbone C−Si bonds, we focus on the linkerless σ-chains to minimize complexity and simplify our analysis.Their specific dihedral geometries, end-to-end distances, and energies are also provided in Tables S2 and S3 (Supporting Information).
In line with Raptis and Melissas' finding, our DFT studies of [CSi] 3 -chain show that the kinked geometries are isoenergetic with the t + t + configuration when we similarly omit the D3 correction (gray bars, Figure 4a).However, calculations with the B3LYP-D3 functional indicate that the t + t + rotamer is the highest energy conformer (red bars, Figure 4a); all other geometries are lower in free energy, with the o + o + rotamer being the most stable (−1.5 kcal mol −1 relative to t + t + ).This difference indicates that dispersion interactions strongly favor coiled geometries in oligo(dimethylsilmethylene) chains.In contrast, applying the same treatment to Si 5 -chain gives the t + t + configuration as the lowest energy conformer regardless of whether dispersion interactions are accounted for (Figure 4b), since the strong steric influence of the SiMe 2 Si group biases oligosilane backbone geometries toward transoid conformations (Figure 1a). 75The alkane series resembles what is seen in the silanes.
We performed NCIPLOT calculations 76 on the B3LYP-D3optimized t + t + , t + o + , and o + o + conformers of [CSi] 3 -chain to visualize where these stabilizing London dispersion interactions occur in the backbone (Figure 4c).While there are CH•••H interactions between vicinal Si(CH 3 ) 2 groups in the t + t + conformer, these are comparatively weaker than the coiled The o + o + dispersion interactions are cooperative in the sense that they occur only when both dihedrals are ortho-disposed.In other words, breaking one strong contact by rotating to a transoid dihedral means that the other strong contact must also be broken.This creates an intramolecular stapling effect that enforces kinked geometries, leading to a 3-fold energetic stabilization for the o + o + geometry compared to the t + o + one.This difference in interdependency can be visualized in Figure 4c: the NCI interactions in the t + t + and t + o + conformers are separated into two independent strips of isosurface density but are joined together into one interconnected strip of NCI density for the o + o + conformer.
While it is a combination of entropic 77 and dispersion effects that lead to coiled carbosilane geometries being picked up in the junction initially (see Supporting Note 1), our findings suggest that intramolecular dispersion forces can keep coiled carbosilane geometries intact in spite of the mechanical strain applied along the backbone as the junction is stretched.To investigate this notion computationally, we modeled the stretching of coiled junctions between Au 10 electrodes to see whether the internal backbone kinks unravel to all-transoid states before a breaking force limit is reached (Figures 5, S6− S8).Conductive atomic force microscopy measurements have shown that thiomethyl-terminated alkane wires sustain average pulling forces up to 0.7 nN before the dative R 2 S−Au linkages break and junction rupture occurs. 78,79Our DFT calculations indicate that the donor−acceptor R 2 S−Au bond strengths are virtually identical between the previously reported 78 thiomethyl-terminated butane (14.7 kcal mol −1 ) and the [CSi] 3 geometries discussed herein (13.8−14.7 kcal mol −1 , Table S4).These similarities suggest that the Au−S dative bonds in the [CSi] n series can withstand average junction stretching forces up to 0.7 nN.
We note that there are many more low-lying conformers for [CSi] 3 junctions than the [CSi] 3 -chain models as shown in Figure 4 as there are now eight internal dihedrals, each with their own set of conformational states (see Supporting Note 2).Many of these conformers likely exist in the junction's overall conformational landscape and contribute to the observed conductance behavior and trends.While a comprehensive investigation of all possible junction geometries and their trajectories is outside the scope of this work, we carry out a focused analysis on representative geometries that help illustrate how dispersion can impact the evolution of molecular geometry and electronic transmission as carbosilanes are stretched to a 0.7 nN force limit (Figure 5).
We focus on the initial geometries of [CSi] 3 with central o + o + dihedral configurations as Figure 4 indicates that these internal geometries are the most stabilized by dispersion.Figures 5 and S6 plot the pulling trajectories of initial o + o + o + o + , t + o + o + t + , or o + o + o + t + junction geometries (see Supporting Note 2 for more geometry details).For each conformer, we performed DFT calculations at the B3LYP/6-31G(d,p)/def2-SVP(Au) level 47 with and without Grimme's D3 correction (filled circles and empty squares in Figure 5a−d, respectively) to model how junction geometry, energy, and applied force are impacted by dispersion interactions upon junction stretching.Meanwhile, Figure 5e models transmission at the zero-energy point or after a major change in internal dihedral geometry occurs (arrows in Figure 5a).We compare transmission calculations on structures with and without dispersion corrections invoked during geometry optimization as shown in Figure S9.In most modeled geometries, we find only marginal differences in transmission at the Fermi energy (E F ).We note that in some cases intramolecular dispersion interactions can alter molecular geometry in ways that lead to the onset of destructive quantum interference antiresonance features near E F (Figure S9d).The full details of these calculations are provided in the Methods section.
Broadly, we find that the dispersion correction increases both the relative energy and the force required to stretch molecular junctions as the Au−Au distance widens, likely due to the energetic penalty from disrupting the London dispersion interactions that require short contact.
Junction stretching of the o + o + o + o + conformer leads to a conformational transition to t + o + o + t + that is accompanied by an acute drop in energy and force (Figure 5a,c).This transition occurs regardless of whether dispersion corrections are included.Figure 5e shows that these outer ortho to transoid dihedral transitions give a modest decrease in transmission at the Fermi energy (E F ), which is consistent with the small dip in most frequent conductance that we observe experimentally starting around ∼0.3 nm as shown in Figure 3b.In the absence of Grimme's dispersion correction, we find that further junction stretching leads to a geometry change from the t + o + o + t + (blue square) to t + t + t + t + (orange square) configuration before the 0.7 nN limit is reached.If this change to the alltransoid geometry did occur, we would expect to find an increase in transmission by 2 orders of magnitude in the last ∼1 Å prior to its breakpoint (dotted orange line, Figure 5e).This is not consistent with what we observe experimentally in Figure 3a−c.When dispersion interactions are accounted for, we instead find that the t + o + o + t + junction geometry is maintained past the 0.7 nN force threshold as the fully elongated, dispersion-corrected geometry at the calculated breakpoint is 5.1 kcal mol −1 higher in relative energy than the equivalent point without the dispersion correction (Figure 5a).This suggests that the Au-linker contact breaks before the internal ortho staples are disrupted.The same general trends are observed with an initial t + o + o + t + junction geometry (Figure 5b,d), where the dispersion-corrected junction stretching trajectory ends with the poorly transmissive t + o + o + t + dihedral configuration rather than the all-transoid configuration.These findings suggest that London dispersion interactions such as the ones shown in Figure 4c may enable kinked geometries to be sustained even at full junction elongation, where mechanical force is highest.
We observe a different yet experimentally consistent outcome when we invoke an initial o + o + o + t + geometry (Figure S6).Regardless of dispersion, a transition to the o + t + o + t + junction configuration occurs.This geometry is sustained upon elongation without a switch to the all-transoid configuration.This transition is also accompanied by a mild decrease in transmission (Figure S6c) that is consistent with the decreasing conductance we observe upon junction elongation.Thus, in all three models studied here, we find that internal ortho dihedrals persist in the backbone at junction breakdown.

Journal of the American Chemical Society
Finally, we modeled elongation in Au 10 −Si 5 −Au 10 junctions with a starting o + o + o + o + configuration as a control (Figure S7).Regardless of whether dispersion interactions are accounted for, these junctions reach fully transoid geometries upon junction stretching prior to the 0.7 nN force limit as the dispersion interactions in permethylated oligosilanes are outweighed by steric influences.This concept reinforces that in the absence of strong steric biases intramolecular dispersion can play a dominant role in dictating molecular geometry in σbonded molecular junctions.

■ CONCLUSION
Intramolecular London dispersion σ-interactions have been characterized extensively in ensemble spectroscopic measurements. 80,81To the best of our knowledge, the present work is the first instance where these specific intramolecular interactions have been observed in a single-molecule context.These studies, along with related work by Venkataraman, Hybertsen, and co-workers investigating molecule-electrode van der Waals interactions, 82 point to mechanical breakjunction systems as useful tools to deepen our understanding of London dispersion forces where trends in conductance, step length, and mechanical force measurements provide physical readouts for the presence and strength of dispersion interactions in single-molecule circuits.
Finally, this work highlights the untapped potential of using intramolecular London dispersion interactions to engineer single-molecule electronics.Here, we show one specific use case, where London dispersion interactions structurally enforce kinked junction geometries that annul quantum transport and give single-molecule insulators with the highest experimental β values reported to date.However, it should be possible to apply London dispersion design concepts to deliberately install other features or functions into molecular junctions.For example, one can imagine using London dispersion interactions to instead enforce highly conductive yet sterically disfavored junction conformations, reversibly disrupt and restore these dispersion interactions with an external stimulus for switchable molecular electronic components, 11−14 or perform novel molecular electronic functions.

■ METHODS
Synthesis.General synthesis and characterization information and NMR spectra are provided in the Supporting Information.The synthesis for these compounds were adapted from previously reported procedures 29,83 and executed according to Figure 2.An oven-dried 250 mL three-necked flask equipped with a 100 mL addition funnel, reflux condenser, septa, and stir bar was connected to a Schlenk manifold under a nitrogen atmosphere.The flask was charged with magnesium turnings (0.69 g, 28.42 mmol, 1.00 equiv).The addition funnel was filled with a mixture of dichlorodimethylsilane (5.50 g, 42.62 mmol, 1.50 equiv) and chloro-(chloromethyl)dimethylsilane (4.07 g, 28.42 mmol, 1.00 equiv) in 80 mL of tetrahydrofuran (THF).This mixture was added dropwise at room temperature over the course of 5 h followed by overnight reflux.The next day, THF was removed in vacuo and 100 mL of anhydrous pentane was added to the flask.Salts were filtered over dry Celite using an air-free Schlenk filter into a 250 mL Schlenk flask.The solution was concentrated in vacuo to give 2.02 g of clear, colorless oil 1 (Figure 2) that was carried forward without further purification.
A separate oven-dried 50 mL Schlenk flask was equipped with a stir bar and rubber septa, then connected to a Schlenk manifold and charged with 2.5 M n-butyllithium solution in hexanes (4.60 mL, 62.52 mmol, 2.20 equiv).The Schlenk flask was cooled to 0 °C.N,N,N′,N′-Tetramethylethylenediamine (7.27 g, 62.52 mmol, 2.20 equiv) was added dropwise via a syringe, and the mixture was diluted with 3 mL of anhydrous pentane and stirred for 20 min.Dimethyl sulfide (3.89 g, 62.52 mmol, 2.20 equiv) was added dropwise over 5 min, then stirred for 4 h at room temperature.The flask containing 1 was dissolved in THF (110 mL) and cooled to 0 °C.The dimethyl sulfide solution was cannulated into the flask and stirred overnight.The next day, the reaction was quenched with methanol (10 mL) and the solvent was removed through rotary evaporation.The resulting yellow oil was redissolved in hexane and filtered through a silica plug using a 1:1 mixture of dichoromethane/hexane to yield 1.41 g of colorless oil.This was further purified via preparative gel permeation chromatography using a JAI Labo-Ace LC-5060 equipped with a JAIGEL-2.5HR and JAIGEL-2 HR gel permeation chromatography column in series using n-hexane (95%, Avantor) as the eluent.A characteristic chromatograph is shown in Figure S1.The following compounds were isolated: [CSi] 2 colorless oil (324 mg, 4.5% yield) 1 H NMR (600 MHz, CDCl 3 ): δ 2.15 (s, 6H), 1.80 (s, 4H), 0.13 (s, 12H), −0.06 (s, 2H). 13C NMR (151 MHz, CDCl 3 ): δ 23.32, 21.00, 1.44. 29 STM-BJ Measurements.Details of the STM-BJ technique have been reported elsewhere. 46In brief, a gold STM tip is brought in and out of contact with a gold substrate over a dilute solution of the target molecule (1 mM in 1,2,4-trichlorobenzene solvent) while recording the electrical current and displacement between the electrodes.The electrodes are first brought into contact by pushing the tip into the substrate until a conductance of 5 G 0 is achieved, then retracted at a rate of 20 nm/s for 5 nm.A single Au−Au point contact has a conductance (G) of 1 G 0 (=2e 2 /h), which serves as our fundamental unit of conductance.Once the point contact breaks, a proximal target molecule may bridge the electrodes to form a single-molecule junction.The junction is stretched by separating the electrodes until the Au-molecule-Au junction ruptures, thus ending a single measurement trace.2D conductance-distance histograms are generated using 100 bins/decade for conductance (y axis) and 1000 bins/nm along the displacement axis (x axis) and aligned at 0 nm displacement and 0.5 G 0 .
STM break-junction measurements were acquired on a home-built instrument with custom software and electronics.A PI 840.10 linear piezo (Physik Instrumente) controlled by a 24 bit NI-4461 DAQ module (National Instruments) is used for precise control of the tip electrode.Conductance measurements are conducted at a rate of 40 kHz using a FEMTO DLPCA-200 amplifier (FEMTO Messtechnik GmbH).A resistor in series with the amplifier and the molecular junction is used to set the sensitivity of the amplifier.Series resistor and voltage bias values are given in Figure S2 caption.Gold-on-steel substrates were prepared by polishing stainless 10 mm AFM/STM specimen discs (Ted Pella) using a hand rotary tool followed by ultrasonic cleaning.The discs were arranged in an Angstrom NextDep Thermal Evaporator and 200 nm of gold (Thermo Scientific Chemicals, 99.95% purity) were deposited onto the surface of the discs.Gold STM tips were formed by cutting gold wire (Alfa Aesar, 0.25 mm, 99.998% purity).Prior to measurement, both the gold tips and wires were cleaned via ozonolysis using an UV ozone generator (Novascan) for 30 min.Solutions of target molecules were made in 1,2,4-trichlorobenzene (>98%, TCI America).Measurements were done in both 1 mM and 0.1 mM 1,2,4-trichlorobenzene solutions, but we did not observe any significant variance in outcome.
Calculations.Thermochemical Calculations.Energies, ΔG 298 values, molecular dihedral angles, and molecular lengths of the molecules described in Figure 4 and Tables S1−S4 were obtained from gas-phase DFT calculations via Gaussian16. 84Molecular geometry optimizations were carried out at the B3LYP/6-311G(d,p) or B3LYP-D3/6-311G(d,p) level unless otherwise noted in the caption.In all frequency calculations, every stationary point was identified as a minimum with no imaginary frequencies observed.NCIPLOT calculations were carried out from B3LYP-D3/6-311G-(d,p)-optimized structures.Figure S5 plot was obtained from NCIPLOT; 76 the visual representations in Figure 4 were created in Jmol.
Junction Pulling Simulations.We first optimized the geometry of [CSi] 3 and Si 5 free molecules without constraint from initial dihedral angles of 165°(t + ) or 80°(o + ) for the listed internal backbone torsions and anti geometries for all other backbone dihedrals (Supporting Note 2).Rigid Au 10 pyramids were appended to each S atom in the optimized free molecule.We use fixed Au 10 pyramids as proxies for the electrodes at the B3LYP/6-31G(d,p)/def2-SVP(Au) level, 47 with and without Grimme's D3 correction. 72We note that using simpler Au 2 diatoms as electrodes reproduce the same qualitative effect (Figure S8).
The Au 10 -molecule-Au 10 system was relaxed without any other constraints to obtain our starting geometries for the junction stretching simulations.From these geometries, we either shortened or extended the distance between the Au apex atoms in 0.25 Å increments, allowing the molecule to relax at each stage between the fixed Au−Au distances to simulate junction compression or stretching.The energies of these constrained junction geometries are plotted relative to the energy of the optimized Au 10 -molecule-Au 10 without the distance constraint.In these calculations, we define the junction breakpoint as the point where the system crosses the 0.7 nN force limit or the molecule detaches from the Au 10 pyramid.We obtained stretching force by taking the derivative of energy with respect to Au−Au displacement. 45elect geometries (with and without D3 correction) from the pulling trajectory were extracted for subsequent transmission calculation.We replaced the Au 10 clusters with Au 20 clusters without changing any other details of the B3LYP-optimized molecular geometry.We calculated the transmission function with a DFTbased nonequilibrium Green's function method with the PBE exchange-correlation functional 60 and a light (double-ζ equivalent) basis set using the AITRANSS 30,31 postprocessor module within the FHI-aims 61 package to obtain transmission function plots.
Transmission β Value Plots.Single Au atoms are appended to the linker groups of molecular models with the following initial geometries: S−Au length of 2.49 Å, C−S−Au angle of 110°, Si− C−S−Au dihedral of 180°, and Si−C−S−Me dihedral of 90°.These models undergo an initial optimization with a maximum residual force per atom of 0.01 eV Å −1 .Au 20 pyramids are added to each Au atom to approximate the Au(111) structures.These junctions were used to calculate the molecule's energy-dependent transmission function according to the approach described above to give the plots as shown in Figure S4.
yet it remains an open question as to whether these interactions are strong enough to control backbone geometry in single-molecule junctions.Until this point, London dispersion forces have not been observed in single-molecule measurements.

d
Average increase in most probable step length across the C n , Si n , and [CSi] n series as C−C, Si−Si, or C−Si bonds are added to the backbone.e Average C−C, Si−Si, or C−Si bond lengths in the C n , Si n , and [CSi] n backbones based on B3LYP-D3/6-311G(d,p) calculations.

Figure 5 .
Figure 5. (a,b) Junction elongation plots track change in energy and backbone conformation as (a) oooo and (b) toot [CSi] 3 initial geometries are stretched between Au 10 pyramids until the junction breaks.Energies are plotted relative to the initial geometry optimized with either the B3LYP-D3 (filled circle) or the B3LYP (empty square) functional.Color changes indicate a major change in the backbone geometry.(c,d) Junction pulling force calculated from the change in energy with respect to electrode separation.Acute drops in force correspond to major configuration changes in the backbone dihedral geometries.(e) Transmission calculations and structures of Au 20 -molecule-Au 20 junctions with molecular geometries extracted from select points [marked by arrows in (a)] along the oooo pulling trajectories.The light gray vertical line is plotted at E F as a visual aid.General note: all ortho (o) and transoid (t) conformers refer to o + or t + dihedral geometries.

Table 1 .
Molecular Conductance and Junction Displacement Properties a Key.Number of atoms between the two S atoms.bGrefers to the most probable conductance at full junction elongation (see FiguresS2 and S3).c Most probable (50th percentile) step length values.The z values we obtain for the alkane and silane wires are in good agreement with previous measurements in STM-BJ setups (refs 28 and 50).