Wei
Zhang
a,
Zhibin
Zhao
a,
Min
Tan
a,
Adila
Adijiang
a,
Shurong
Zhong
a,
Xiaona
Xu
a,
Tianran
Zhao
a,
Emusani
Ramya
a,
Lu
Sun
a,
Xueyan
Zhao
a,
Zhiqiang
Fan
*b and
Dong
Xiang
*ac
aInstitute of Modern Optics, Center of Single Molecule Sciences, Key Laboratory of Micro-scale Optical Information Science and Technology, Nankai University, Tianjin 300350, China
bSchool of Physics and Electronic Science, Changsha University of Science and Technology, Changsha 410114, China
cSchool of Materials Science and Engineering, Smart Sensing Interdisciplinary Science Center, Nankai University, Tianjin 300350, China. E-mail: zqfan@csust.edu.cn; xiangdongde@nankai.edu.cn
First published on 2nd October 2023
The molecular binding orientation with respect to the electrode plays a pivotal role in determining the performance of molecular devices. However, accomplishing in situ modulation of single-molecule binding orientation remains a great challenge due to the lack of suitable testing systems and characterization approaches. To this end, by employing a developed STM-BJ technique, we demonstrate that the conductance of pyridine-anchored molecular junctions decreases as the applied voltage increases, which is determined by the repeated formation of thousands of gold–molecule–gold dynamic break junctions. In contrast, the static fixed molecular junctions (the distance between two electrodes is fixed) with identical molecules exhibit a reverse tendency as the bias voltage increases. Supported by flicker noise measurements and theoretical calculations, we provide compelling evidence that the orientation of nitrogen–gold bonds (a universal coordinate bond) in the pyridine-anchored molecular junctions can be manipulated to align with the electric field by the synergistic action of the mechanical stretching force and the electric fields, whereas either stimulus alone cannot achieve the same effect. Our study provides a framework for characterizing and regulating the orientation of a single coordinate bond, offering an approach to control electron transport through single molecular junctions.
Meanwhile, for certain molecules with π-system coupling to electrodes, even a small change in the binding orientation can lead to a substantial alteration in the molecular junction conductance,27 making them an ideal system for investigating molecular binding orientation. Utilizing these specific molecules, we demonstrated that the orientation of a coordinate bond can be regulated by an external electrical field in a dynamic break junction, but cannot be regulated in a fixed junction. Employing the scanning tunneling microscopy break junction (STM-BJ) technique with a developed electrode suspension function, two types of contrasting molecular break junctions were constructed and the conductance evolution upon increased bias was monitored to quantify the change of bond orientation.28,29 The first type consists of pyridine-anchored molecular break junctions, where the π-binding can occur between the aromatic ring and electrode, and thus the conductance of the molecular junction is highly sensitive to the bonding orientation of the anchoring group. As the bias voltage increases, a decrease in the conductance of pyridine-anchored molecular junctions was observed by the repeated formation and breakage of thousands of single-molecule junctions. The second kind consists of amino-anchored molecular break junctions that are devoid of the π-binding between an aromatic ring and electrode. In line with this, we found that at larger bias voltages, the conductance of amino-anchored molecular break junctions essentially remained unaltered. Assisted by flicker noise measurement and density functional theory (DFT) calculation, it is revealed that the conductance change in the pyridine-anchored molecular junction originates from the alteration of coordinate bond orientation regulated by the synergistic action of electric field and mechanical stretching.
To reveal the role of mechanical stretching in regulating the bonding orientation, we established static fixed junctions by halting the movement of the top electrode once a molecule-bridged junction was formed. In contrast to the behavior demonstrated in the dynamic break junctions, the conductance derived from the I–V curves of both the pyridine-anchored and amino-anchored molecular junctions increases as the applied bias voltage increases. These differences indicate that the electric field applied in the experiment alone is insufficient to regulate the orientation of the nitrogen–gold bond. Instead, a synergistic action of mechanical forces and the electric field enables the regulation of coordinate bond orientation.
Fig. 1b shows typical conductance–displacement traces under different bias voltages. The gray traces show the conductance of a bare gold contact measured in the pure 1,2,4-trichlorobenzene (TCB) solvent. Integer multiples of G0 (G0 = 2e2/h ≈ 77.5 μS) can be observed with no additional conductance plateau below 1 G0. In contrast, the conductance plateaus below 1 G0 were clearly observed when the conductance measurement was performed in PY[4] containing solution. A close examination of the traces reveals that the conductance plateaus shift down as the bias voltage increases. Fig. 1c presents the one-dimensional (1D) conductance histograms of PY[4] molecular junctions at the bias voltages of 100 mV, 200 mV, 300 mV, and 400 mV, respectively. Each of them was constructed from approximately 3000 conductance–displacement traces without data selection. The peak position indicates the probable conductance of the PY[4] molecular junction. As the bias voltage increases from 100 mV to 400 mV, the conductance continuously shifts from 4.5 × 10−5G0 (≈3.51 nS) to 8.6 × 10−6G0 (≈0.66 nS). The corresponding two-dimensional (2D) conductance histogram is shown in ESI Fig. S3,† which confirms that the conductance of molecular junction decreases as the bias voltage increases. The conductance evolution under negative bias exhibits a similar trend, i.e., the conductance of the molecule junction decreases as the absolute value of the bias increases.
In contrast, for p,p′-diaminoquaterphenyl (labeled as NH2[4]) molecules which possess similar backbone structures to PY[4] but different anchoring groups (i.e., amino group) at both ends, the conductance remains almost unchanged when the applied bias voltage changed from 100 mV to 400 mV (Fig. 1d–f), and the corresponding 2D conductance histograms are shown in ESI Fig. S4.† Similarly, the conductance of the NH2[4] molecule remains constant when the applied negative bias changed from −100 mV to −400 mV. To verify the reliability of the observation that pyridyl-terminated molecules and amino-terminated molecules show discrepant behavior, we investigated a series of molecules with two different anchoring groups, i.e., 1,4-bis(pyrid-4-yl)benzene (labeled PY[3]), 4,4′-bipyridine (labeled as PY[2]), 4,4′′-diamino-p-terphenyl (labeled as NH2[3]), and p,p′-diaminoquaterphenyl (labeled as NH2[2]), respectively.
Fig. 2a and b show the 1D conductance histograms of PY[3] and PY[2] molecular junctions, respectively. Similar to PY[4], the conductance of PY[3] and PY[2] molecular junctions decreases when the applied bias voltage increases, as demonstrated by the conductance histograms. On close examination of Fig. 2b, additional minor peaks can be seen that are left-shifted as the bias voltage increases and situated at low conductance locations, as indicated by the star symbol. We assign the small peaks to the new type of molecular junctions (i.e., PY[2] dimer formed via π–π stacking)30–32 rather than the varied configuration of single PY[2] junctions upon mechanical stretching, since the low conductance plateau is parallel to the high conductance plateau (i.e., two conductance plateaus appeared simultaneously rather than in sequence), as shown in ESI Fig. S7.† This claim is further supported by the flicker noise measurement which will be explained in more detail later.
In contrast, as illustrated in Fig. 2d and e, the applied bias voltages do not affect the conductance of NH2[3] and NH2[2] molecular junctions. Further testing revealed that the conductance of the molecular junctions using thiol as an anchoring group did not exhibit bias dependence, see ESI Fig. S9.† These results confirm that the bias-dependent molecular conductance is strongly affected by the terminal anchoring groups of the molecules. Notably, minor peaks, as marked by the star symbol, can also be observed in the conductance histogram for NH2[2] molecular junctions, see Fig. 2e. We attribute the small peak to the NH2[2] dimer formation via an intermolecular hydrogen bond.9,33 The individual chemical structures of the molecules and the corresponding conductance under positive bias are summarized in Table 1. Accordingly, the probable conductances of different types of molecular junctions under negative bias are summarized in ESI Table S1.†
More intuitive analysis was carried out to investigate the bias voltage influence on the conductance of the molecules with different anchoring groups. Fig. 2c demonstrates a negative correlation between the conductance of pyridine-anchored molecules and the applied bias voltage, that is, the conductance of molecular junction drops as the absolute magnitude of the bias voltage increases. In contrast, the conductance of the amino-anchored molecules remains relatively constant, and is almost independent of the bias voltage, as shown in Fig. 2f. As an initial conjecture, the distinct behavior of these two types of molecules originates from either the energy structural changes as the molecule is exposed to an external electric field34,35 or the distinguished conformational changes (e.g., rotation of aromatic rings) regulated by the bias voltage.13 However, this speculation was ruled out by I–V measurements of the fixed molecular junctions, which demonstrates that the two types of molecular junctions exhibit the same behavior upon voltage sweeping, as shown below.
Fig. 3a and b illustrate the 2D logarithmic I–V and conductance–voltage (G–V) density plotting, which was constructed from approximately 3000 individual I–V curves of PY[3] molecular junctions. In contrast to earlier findings shown in dynamic break junctions (Fig. 2c), the G–V plot shows that the conductance increases with an increasing bias voltage (Fig. 3b), and Fig. 3c shows the schematic of the fixed PY[3] molecular junctions. For comparison, we also plotted the I–V and G–V density images of NH2[3] molecular junctions, as shown in Fig. 3d–f. It is evident that NH2[3] molecular junctions displayed an increased conductance with increasing bias voltage as in the case of PY[3]. Additional I–V measurement shows that the conductance of molecular fixed junctions for NH2[2] and NH2[4] also increases upon an increased bias voltage. This behavior is anticipated because the electrode's Fermi level typically approaches the molecular orbital (HOMO or LUMO) when a higher bias voltage is applied, leading to a lower energy barrier and higher conductance,34,36 see ESI Fig. S11.†
The interesting observation is that the conductance of the pyridine-terminated molecular break junction decreased with increasing applied bias voltage, but the conductance of static fixed junction shows the opposite trend. Notably, the molecule will be exposed to a similar electric field in both break junctions and fixed junctions as the same bias is applied, indicating that the distinct behaviors observed in the two types of junctions do not solely originate from the applied bias voltage. Also, the opposite behavior displayed by two types of junctions effectively eliminates the possibility that the rotation of aromatic ring modulated by the electric field results in conductance changes. Instead, we hypothesize that the distinct behavior of break junctions and fixed junctions originates from the properties of the anchoring group that are strongly influenced by the synergistic action of the electric field and mechanical forces.
For pyridine-anchored molecules, the pyridyl can couple with a gold electrode through both the σ bond of its N atom (nitrogen–gold, σ-binding) and the π-bond of its pyridine ring (π-binding),27,39–41 as schematically shown in Fig. 4c. For the π-binding configuration, the electron is mainly transported through space between the pyridine ring and electrode, which can be seen as a weak coupling compared to the one of σ bond.26,27 The electronic coupling between the molecular π*-electron and the gold d-electrons decreases as θ (the angle between the nitrogen–gold bond and the plane of the pyridine ring) decreases.42 The π-coupling is particularly strongest when the nitrogen–gold bond is perpendicular to the pyridine ring (i.e., the angle θ = 90°), and π-coupling is weakest when it is parallel to the pyridine plane (i.e., θ = 0°). The decrease of noise power (N) upon a higher bias voltage (Fig. 4a and b) indicates that the electron transport channel through space (π-coupling) is suppressed. In other words, the decrease of noise power indicates that the nitrogen–gold bond gradually aligns parallel to the direction of the pyridine plane (the same as that of the electric field) upon a higher bias in the dynamical break junctions.
In contrast, for the NH2[4] molecular junction, the donor–acceptor bond (σ-binding) is primarily formed between the electrode and the anchoring group,42,43 as schematically presented in Fig. 4f. Due to the lack of π-binding, the electron will primarily be transported through a bond, which results in N ≈ 1 and it remains unaltered upon a higher bias, as presented in Fig. 4d and e. The absence of π-coupling in NH2[4] molecular junctions can be attributed to two possible factors: (1) compared to PY[4], the aromatic ring in NH2[4] is far away from the electrodes by a C–N distance and thus the π-binding between the aromatic ring and the electrode is relatively weak;27 (2) the amine–gold link is naturally parallel to the plane of the pyridine ring,44 and this configuration makes the aromatic ring distant from the electrode and causes a weak π-coupling for the NH2[4] molecular junction.
Flicker noise spectrum of the PY[2] molecular junction at a low conductance plateau was also carried out, as we mentioned previously. The 2D histogram of the normalized noise power versus the average conductance of the molecular junction is displayed in ESI Fig. S13.† It can be found that N = 1.95, which indicates that the electron is transported across space rather than through a bond at this state. Combining the observation that the probable low conductance is one order of magnitude smaller than the probable high conductance (Fig. 2b) and the feature of the 2D conductance histogram (Fig. S7†), we believe that the minor peak with low conductance results from the PY [2] dimer formation via π–π stacking rather than a single PY[2] molecule with a perpendicular bonding configuration at the interface.
Moreover, we can find that the whole transmission coefficient decreases as the tilt angle decreases from 45° to 0°, which indicates that the conductance of the junction will decrease as the orientation of the nitrogen–gold bond changes from being out of the pyridine plane (θ = 45°) to being parallel to the pyridine plane (θ = 0°). This finding is in line with the calculated results of the electron transport pathway, which demonstrates that the volumes of electron transport through the nitrogen–gold bonds decrease when the tilt angle decreases, see ESI Fig. S14.† Transmission eigenstates at 0.3 eV for the PY[4] molecular junction at the tilt angle of 45° and 0° are shown in Fig. 5c and d. It is evident that the spatial distributions of transmission eigenstates at the interface become localized when the tilt angle changes from 45° to 0°, which will result in a decreased conductance. In contrast, for the NH2[4] molecular junction, both HOMO and LUMO are far from the Fermi level of the electrode and the wave function at the Fermi level is insensitive to the tilt angle of the molecule compared to PY[4], see ESI Fig. S15.†
We further calculated the electrostatic potential distribution of the molecules with different anchoring groups, see ESI Fig. S16.† It shows that the electrostatic potential around the amino group is positively charged, while the nitrogen atom of the pyridyl group is negatively charged. When the molecule comes in contact with the electrode, a coordination bond will be formed between the anchoring group and the electrode. Under a fixed electric field, the direction of the force undertaken by the amino group and the pyridyl group should be opposite. In that case the external field exerts a pushing force on the pyridyl-based bond (N–Au) and makes the bond tilt angle θ decrease (N–Au gradually parallel to the pyridyl plane) with the assistance of mechanical stretching; the amino-based bond will not experience such a pushing force and thus will not promote the decrease of θ. The change of bond tilt angle will lead to the pronounced variation of the molecular junction conductance, as demonstrated in Fig. 5 in the main text. Thus, the different behaviors of amino-terminated and pyridyl-terminated molecule junctions under the joint action of the electric field and mechanical stretching can also be explained by the distribution of electrostatic potential.
Therefore, the theoretical model calculations, electrostatic potential calculations as well as the measured noise spectra strongly support our hypothesis that the pronounced conductance change of the pyridine-anchored molecular junction highly likely results from the orientation change of the nitrogen–gold bond. With the assistance of mechanical stretching, the orientation of the nitrogen–gold bond gradually aligns with the direction of the increasing electric field. This is anticipated because the nitrogen–gold bond is a coordinate bond, where the electrons for the bond formation are solely supplied by the N atom (so-called semi-polar bond) and is highly sensitive to the direction of the external electric field.21 This bond orientation change reduces the coupling strength by suppressing the π coupling channel between the pyridine ring and electrode. The orientation change of the nitrogen–gold bonds with increasing bias not only makes the molecular coupling strength weaker in line with the decreased conductance, but also gives rise to the following features: (1) the length of the conductance plateau decreases, since a weak coupling facilitates the breakage of the junction and thus makes the lifetime of the junction short, which agrees well with the experimental observation as presented in ESI Fig. S2–S7;† and (2) the intensity of the peak in the conductance histogram decreases due to the decreased lifetime of the junction and decreased length of the conductance plateau, which is consistent with the feature seen in Fig. 1c and 2a, b.
We would like to emphasize that (1) the electric field activation alone does not result in a change in the orientation of the nitrogen–gold bond. This claim is supported by the I–V test, which shows that in a static fixed junction, the conductance decrease is not observed when the electric field is increased during the voltage sweeping process (Fig. 3); (2) the sole activation of mechanical stretching cannot trigger the change of nitrogen–gold orientation in a pyridyl-anchored molecular junction either. If mechanical stretching alone has the capability to alter the nitrogen–gold orientation, the conductance of the junction would no longer depend on the bias voltage; (3) only the synergistic action of dynamical mechanical stretching and electric field activation can regulate the orientation of the pyridine–gold link in the coordinate bond terminated molecular break junction.
As a discussion, although the conductance of the molecular junctions using thiol and amino as anchoring groups does not exhibit bias dependence, we cannot draw the conclusion that the orientation of the S–Au or amino–Au bond can't be modulated by the synergistic action of mechanical forces and electric field, because the π-coupling was absent in these molecular junctions and the orientational change of the S–Au/N–Au bond might not result in a notable conductance variation. Our additional calculations show that the binding energies of PY[4] and NH2[4] on the electrode surface are approximately 1.604 eV and 0.509 eV, respectively. The diverse binding energy may result in different behaviors of amino and pyridyl-anchored molecular junctions as well. For the amino-anchored molecular junction with weak bonding strength, the orientation of the N–Au bond may be modulated by the sole mechanical stretching already, resulting in the observation that the conductance does not depend on the bias voltage any more. To address whether it is feasible to regulate the orientation of amino–Au and S–Au bond by the same strategy, additional experiments are highly desired.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3sc03892k |
This journal is © The Royal Society of Chemistry 2023 |