Gating of Quantum Interference in Molecular Junctions by Heteroatom Substitution

Abstract To guide the choice of future synthetic targets for single‐molecule electronics, qualitative design rules are needed, which describe the effect of modifying chemical structure. Here the effect of heteroatom substitution on destructive quantum interference (QI) in single‐molecule junctions is, for the first time experimentally addressed by investigating the conductance change when a “parent” meta‐phenylene ethylene‐type oligomer (m‐OPE) is modified to yield a “daughter” by inserting one nitrogen atom into the m‐OPE core. We find that if the substituted nitrogen is in a meta position relative to both acetylene linkers, the daughter conductance remains as low as the parent. However, if the substituted nitrogen is in an ortho position relative to one acetylene linker and a para position relative to the other, destructive QI is alleviated and the daughter conductance is high. This behavior contrasts with that of a para‐connected parent, whose conductance is unaffected by heteroatom substitution. These experimental findings are rationalized by transport calculations and also agree with recent “magic ratio rules”, which capture the role of connectivity in determining the electrical conductance of such parents and daughters.


Recentexperimentsdemonstratingroomtemperaturequan-
tum interference (QI) in single-molecule junctions have stimulated intense interest in the development of molecular switches, [1] transistors [2] and thermoelectric devices [3] with improved performance.M uch effort has been devoted to controlling QI effects by modifying electronic structures or molecular topologies. [2b, 4] However,adirect correlation between QI and chemical structure remains ambiguous.F or example,d estructive QI has been observed in linearly conjugated molecules,w hile constructive QI occurs in crossconjugated molecules of the close-loop type. [5] Amongst all molecules investigated to date,n onalternant and alternant hydrocarbon systems have been intensively investigated. [6] Although heterocyclic aromatics are particularly interesting due to the presence of spatially separated conductance pathways,w hich offer opportunities for varying QI, only af ew papers about the effect of heteroatoms on singlemolecule junction conductances have appeared in the literature. [7] Recently,w ed eveloped an ew magic ratio rule (MRR) showing that the conductance ratio of two molecules through the same aromatic core with different connectivties is equal to the square of the ratio of their magic integers. [8] The accuracy of the MRR for polycyclic aromatic hydrocarbons (PAHs) is demonstrated in Ref.
[8b],i nw hich MRR predictions for the ratios of conductances for arange of PA Hs are compared with experiments.Aremarkable feature of the MRR is that the associated most-probable conductance ratios are independent of the binding geometry,e longation etc., even though the conductances themselves do depend on such features.A sd iscussed in Ref.
[7c] and in the Supporting Information (SI), this insensitivity can be expected, provided the statistical distribution of contact geometries is independent of connectivity.Ageneralization of this theory [7c] that predicts three simple rules governing the effect of heteroatom substitution in alternate hydrocarbon parents (see also SI), has not yet been verified experimentally.A saconsequence, we set ourselves the task of investigating how destructive QI in the "parent" meta-phenylene ethylene-type molecule (m-OPE)c hanges when one nitrogen atom is inserted into the central benzene ring at different positions,t oy ield the "daughter" molecules M1, M2 and M3 in Scheme 1. To assess the heteroatom effect on QI and to gain insight into the correlation between connectivity and electrical conductance, the parent para-OPE (p-OPE)a nd its daughter molecule P (Scheme 1) were also studied.
Thet arget compounds m-OPE, p-OPE, P and M1-M3 were synthesized via Sonogashira cross-coupling reactions of S-4-iodophenyl ethanethioate with the corresponding diethynylbenzene/pyridine.All of them were purified using column chromatography and characterized by NMR and high-resolution mass spectrometry (see SI). Charge transport measurements of single-molecule junctions with molecules shown in Scheme 1w ere performed using am echanically controllable break junction (MCBJ). [9] Figure 1A displays typical conductance (G)versus distance (Dz)stretching traces,asplotted in asemi-logarithmic scale.The single-molecule conductance plateaus of 6molecules vary from 10 À3.5 to 10 À6 G 0 besides the plateau presenting the gold-gold atomic contact located at G 0 ,w hile the blank experiment using THF/TMB solvent without molecules shows no single-molecule plateau. Figure 1B,C show the 1D conductance histograms of molecules,constructed from 1000 experimental conductancedistance traces for each compound without data selection. After the rupture of gold-gold atomic contacts at G 0 ,asingle conductance peak, corresponding to the most probable conductance for each molecule was detected (also summarized in Table 1). Briefly,t he conductance values follow the trend M2 > M3 > M1 = m-OPE,s uggesting that heteroatom substitution leads to an increase in conductance only when the substituted nitrogen is ortho to both contacts or para to one connection and ortho to the other. On the other hand, there is an egligible change in electrical conductance when the nitrogen is meta to both contacts,i na greement with Ref.
[7c].A sd epicted in Figure 1C,t he presence of the nitrogen atom shows no distinguishable effect on the constructive QI in the junctions through the para-coupled molecules p-OPE and P,w hich is in good agreement with Ref.
[7c] and the results obtained by Gonzalez et al. [7b] Thet wo-dimensional (2D) conductance versus distance histogram of molecule P demonstrates ac lear molecular density cloud at around 10 À4.0 G 0 ( Figure 1D), which is quite similar to that of p-OPE. [6c] Interestingly,t here is no significant conductance feature of the molecular junctions with one of Au electrodes connected directly to the central pyridine ring via the AuÀNb onds,a so bserved in the analogous system containing ap yrimidine ring. [7b] This observation can be attributed to the significant binding force difference between the Au À Sc ovalent bond (1.2 nN) [10] and the Au À Nb ond (0.8 nN) [11] that is close to the AuÀS(methylsulfide) bond (0.7 nN), [12] leading to apreference for the covalent AuÀSover the AuÀNbinding motif in the present case.Furthermore,the conductance plateau is ca. 0.8 nm (after adding the snap back distance of 0.5 nm it will be 1.3 nm), suggesting that the most pronouncing feature stems from the molecules trapped only via Au À Sb onds.H owever, the 2D histograms of the meta-coupled molecules ( Figure S9, SI) show that the conductance cloud becomes less clear as previously reported in the literature. [4d, 13] Apparently,t he bent molecular geometry strongly affects the evolution of molecular junctions,r esulting in as lanted conductance region. Thus,i ti si mpossible to plot the plateau-length distribution and further calculate the molecular length in the meta-coupled systems.
To further verify our hypothesis of an end-to-end configuration via the AuÀSbinding motif and to completely rule out the possibility of the conductance plateaus being due to the molecules bonded to one of the electrodes through the N atom at the central ring, we have performed control experiments of the reference compounds R1-R3 containing only one SAc group (Figure 2).
All of them provide clear evidence for creation of molecular junctions with distinctly higher conductance values compared to those of M1-M3.T hey are arranged in the following order of their decreasing conductance,the paracoupled molecule R2 (10 À3.2AE0.1 G 0 ) > the ortho-coupled molecule R3 (10 À3.6AE0.1 G 0 ) > the meta-coupled molecule R1 (10 À3.9AE0.1 G 0 ), which is ac lear signature of the QI effect. As illustrated in Figure 2B-D,aclear intensity cloud is observed for each case and the plateau length for R2 is longer than for R1 and R3.T he most probable stretched distance Dz*i s determined from the plateau distribution histograms to be 0.8 nm (R2), 0.7 nm (R1)a nd 0.6 nm (R3), respectively.A s aresults,the most probable absolute distance z*(z* = Dz* + 0.5 nm) [9c] between two gold tips is 1.3 nm for R2,1.2 nm for   M3 (red). B,C) Alldata-point1 Dconductancehistogramsc onstructed from 1000 MCBJ traces of each molecule. D) All-data-point 2D conductanceversus relative distance (Dz)ofthe molecule P.The stretching distance distribution is determined from the range of 0.7 G 0 to 10 À5 G 0 . R1 and 1.1 nm for R3,w hich agrees well with theoretically determined junction lengths (1.16 nm for R2 1.09 nm for R1 and 0.96 nm for R3,T able S3). It can therefore be deduced that the distinct variations in conductance values of M1-M3 and R1-R3 are attributed to the different electronic transport pathways whereby the former molecules are trapped between two Au leads via the Au À Sb onds while the latter ones are connected on the one side via the Au À Sb onds,o nt he other side via the AuÀNb onds.A ll these results provide direct evidence for an unambiguous preference for the covalent AuÀSover the AuÀNbinding motif in the junctions through M1-M3,w hich verifies the proposed junction configurations.
Thea bove heteroatom effect can be understood by applying the theory of Ref.
[7c],w hich identifies the ratio of two molecular conductances with the ratio of two "energydependent core transmission functions", evaluated at the Fermi energy (see SI). As discussed in section 4o ft he SI, these energy-dependent core transmission functions are proportional to the Greensf unction of the core of the molecule and are independent of the nature of the coupling to the electrodes.
Fort he three-ring molecules of Scheme 1, the energy dependences of the core electron transmission coefficients are shown in Figure 3A.F or m-OPE,t he parental M-function and the corresponding core transmission function vanish at E = 0, due to destructive QI in the middle of the HOMO-LUMO gap.However, by adding aheteroatom to the central benzene ring, core transmission coefficients of the molecules M2 and M3 become non-zero while the core transmission coefficient of the molecule M1 remains zero at E = 0. On the other hand, the transmission functions of para-coupled systems p-OPE and P are non-zero at E = 0a nd overlap with each other, indicating that the heteroatom effect is negligible in the constructive QI systems.Illustrated in Figure 3B are the full electron transmission coefficients obtained from DFT mean field Hamiltonian. [14,15] Table 1s hows the experimental values for the most probable conductances of the molecules m-OPE, p-OPE, P and M1-M3,w hich are divided by conductance value of molecule M3.C omparison with the corresponding coretransmission ratios of Figure 3A also reveals that the best agreement with experimental data is obtained with adaughter Fermi energy of E F d = 0.211, which differs from the mid-gap of the daughter (E HL d = À0.0795) and parent (E HL p = 0), indicating the variation of the alignment of intrinsic molecular orbital (MO) levels due to heteroatom substitution. Although the lone-pair electrons of the Natom are held in an sp 2 hybrid orbital perpendicular to the orbitals in the p-system, its presence has an unambiguous effect on the energy levels and nodal properties of the MOs ( Figure S6, SI). Both experiment and DFT-based theory reveal that the presence of the nitrogen atom shows no discernible effect on the constructive quantum interference in the junctions through the paracoupled molecules as ar esult of the fact that the core transmissions are only slowly varying functions of energy.I n the absence of aFermi energy shift, the ratio of conductances for M1 and M3 should vanish. Taking as hift of E F d = 0.211 into consideration, M-theory yields ac onductance ratio of Figure 2. A) 1D conductanceh istograms of R1 (black), R2 (blue) and R3 (red), constructed from 1000 MCBJ traces of each molecule. B-D) 2D conductance histograms of R1-R3,r espectively, and stretching distance Dz distributions (inset). The stretching distance distribution is determined from the range of 0.7 G 0 to 10 À5 G 0 . Figure 3. A) Core transmission coefficients t ij (E)ofthe molecules m-OPE, p-OPE, P, M1-M3 against E/d,where d is half of the HOMO-LUMO gap of the parental core;i .e. d = 1, using DFT transport approach implemented in Gollum. [15] B) The calculated transmission coefficients T ij (E)ofthe molecules m-OPE, p-OPE, P, M1-M3,c onnected to gold electrodes using the mean-field Hamiltonian from Siesta. [14] Dashed lines correspond to "parents" and solid lines to "daughters". 0.37 that compares well with the experimental ratio of 0.32. Thesame holds true for the conductance ratio of M2/M3.All of these results are captured by the minimal theory, [7c] which identifies the conductance ratios with the square of the ratio of two "magic numbers" (Table S2, SI).
In conclusion, it has been demonstrated that the destructive QI can be alleviated by the heteroatom substitution whereas constructive QI is almost unaffected. Theincrease in conductance for meta-coupled OPEs originates from two distinct effects.F irst, the change in the Hamiltonian modifies the interference pattern within the core,t he energy dependence of the associated M-functions and core transmission coefficients.S econdly,h eteroatom substitution shifts the Fermi energy relative to the parent, which is also analyzed in the MO viewpoint. Our results provide aqualitative guide for tuning QI effect in single-molecule junctions that is of prime importance in designing molecular devices with desirable functions.