Effect of near-field optical angular momentum on molecular junctions

Understanding and controlling the molecular switching of a single molecule is fundamental for the development of molecular logic operations and for the further development of nanoscale computation. Recently, most research has focused on investigating molecular junctions based on electronic characterization^1–4. In addition to applied voltage, light can influence the molecular state and characterises molecular states complementarily^5–8. However, most classical optical spectroscopy methods are limited by diffraction, which makes it difficult to characterise nanoscale systems with ultrahigh spatial resolution. Near-field-enhanced techniques based on local surface plasmon (LSP) can break the diffraction limit and provide an excellent solution for ultramicroscopy⁹. Such near-field-enhanced techniques provide highly accurate methods for nano areas, such as coherent manipulation, processing, and measurement techniques for tunnelling currents in nano electronic devices^10–13.

In the past decade, many researchers have reported nanoscale optical resolution based on near-field-enhanced techniques with an apertureless probe tip^14–18. At such a nanoscale, the light field is intensely bounded by the LSP modes around the apex of the apertureless probe tip. The near field, especially its electric field component, is dramatically enhanced and, consequently, is used for efficient nondestructive detection of molecular states with local Raman spectroscopy^19–22. Actually, the Raman signal of the molecular junction can be strongly affected by the electrode geometry and the excited plasma^23,24. Simultaneously, asymmetric LSP modes can introduce a near-field angular momentum^25–27, which can have an effect on nearby molecules^28–30. Recently, the Raman response of oligo(phenylene ethynylene)-3 (OPE-3) molecular junctions with near-field angular momentum coupling was reported³¹. It was observed that such the coupling has an effect on the transport electrons in molecular junctions. However, the combined effects of the applied voltage and local electromagnetic field on the switching behaviour of the molecular junction must be investigated from electrical transport and electromagnetic viewpoints. Furthermore, the impact of the polarisation of light and the symmetry properties of the near-field system remain unexplored.

In this study, the combined effect of the near-field optical angular momentum and the bias voltage on the Raman response of a single molecular junction, creating a molecular switch, was investigated. Using a homemade molecular junction spectroscopy (MJS) platform³¹, optical and electrically driven conformational switching in the covalently linked metal–molecule–metal junctions of 2,2',2'',5'-tetramethyl-[1,1':4',1''-terphenyl]-4,4''-dithiol (TM-TPD) was characterised. The non-π-conjugated molecular ‘wire’ of TM-TPD electrically connects the gold-coated tip of a tunnelling microscope to a gold substrate (Fig. 1). Previously, a TM-TPD molecular junction was used¹⁷ to confirm a prior observation that the Raman activity of the molecule can be turned ON and OFF using the biased voltage because of the molecular conjugation during charge transport through molecular orbitals. In this study, it was further demonstrated that the molecular switch can be manipulated using an optical input. The effects of both the polarisation of light and the subsequent symmetry properties of the electromagnetic near field were investigated and used to control the switching behaviour. This change in the Raman response of the junction is associated with a modification of the conformation of the molecule. This is further supported by density functional theory (DFT) simulations. Consequently, both the polarisability tensor and the Raman response are tuned. By changing both the illumination side and the voltage applied to the junction, the Raman intensity can be turned ON and OFF with a difference of nearly five orders of magnitude between the two states (Fig. 2a).

Fig. 1  Optical and electrically driven Raman switching setup based on molecular junction. The schematic diagram shows conformational switching in a tip−molecule−substrate junction with Raman readout via optical coupling of the optical angular momentum to the molecule. Laser illuminations from the top (channel 1) or the side (channel 2) with a 45° incident angle are shown accordingly. The z-axis is defined as perpendicular to the gold substrate, and the triangular-pyramid tip is mirror-symmetric along the x-axis.

Fig. 2  Optical and electrically driven Raman switching in the molecular junction spectroscopy setup. a Raman spectra of TM-TPD corresponding to different voltage and illumination geometries. The red and blue lines are for voltages of 1.0 and 0.1 V, respectively, and the insets show the illumination geometries. b DFT calculations for the ON and OFF Raman modes. The Raman signature is mainly dominated by the molecular conformation of TM-TPD. The dihedral angle between the benzene ring at both ends and the central benzene ring is 50° for ON mode and 90° for OFF mode (see Supporting Information SI.1 for details). By changing both the illumination side and the voltage applied to the junction, the Raman intensity can be turned ON and OFF.

Upon preparation of a self-assembled monolayer of thiolated TM-TPD on an amorphous gold substrate, single molecules are contacted by an apertureless, gold-coated, triangular pyramid (TP)-shaped, near-field tip. The probe tip has three roles: serving as an electrical contact, light source input, and light collector output. Contacts between the tip and single molecules occur spontaneously and are stabilised by piezoelectric control in a typical timeframe of minutes under ultrahigh vacuum conditions. Unpolarised 632.8-nm laser inputs can be sent either from the top (channel 1) or from the side (channel 2) through the probe tip (Fig. 1). A near-field-enhanced Raman spectrum is collected from the top and used as the output. As several results have shown in recent years^17,31–33, the entire system is reliable for simultaneously measuring the electrical and optical properties of molecular junctions. With this setup, one can simultaneously measure the conductance and Raman spectra of molecular junctions. Using conductance measurements and statistical methods, the conductance of a single-molecule junction of TM-TPD was identified. This method is widely used to confirm the conductivity of single molecules. During the Raman measurements, the conductance of the molecular connections is maintained at a constant value to ensure that the Raman spectra are from the same molecular junction. In addition, p-terophenyl derivatives do not tend to form intermolecular π–π stacking because of the nonplanarity formed by twisting feet between adjacent benzene rings. Thus, the switching behaviour can be identified as a single-molecule junction and not as a supramolecular packing effect.

Fig. 2a shows the Raman spectra of the TM-TPD at different applied voltages and illumination geometries. For top illumination under a biased voltage of 1.0 V (first line from the top), a clear Raman signal was observed, which corresponds to the ON mode. However, when side illumination was adopted instead of top illumination (second line from the top), the Raman signal fell below the instrumental detection limits, turning off the Raman vibrational modes (OFF mode). For cases under a biased voltage of 0.1 V, no Raman signal was observed, regardless of the top illumination (third line from the top) or side illumination (fourth line from the top). The observed voltage-driven switching can be repeatedly turned on and off, which is limited solely by the junction stability.

According to the DFT calculation results (Fig. 2b), the ON and OFF signatures of the Raman spectra correspond to different molecular geometries¹⁷, with a dihedral angle of 50° corresponding to the charged TM-TPD, and 90° corresponding to the neutral one (see Supporting Information SI.1 for details). The Raman signature was strongly suppressed (OFF mode) with the highly sterically hindered nonplanar neutral TM-TPD (90°). During conduction, the charged TM-TPD molecule underwent conformational relaxation, leading to a dihedral angle change from 90° to 50°, and the TM-TPD molecules tended to be more conjugated (see Supporting Information SI.1 for details). The conjugated electrons of the phenylene rings tended to be shared by single bonds between the phenylene rings, thus enhancing the Raman activity of the molecule (ON mode).

In addition to the voltage and charge states, the Raman signature of TM-TPD was controlled by the illumination geometry. This result confirms previous observations with the OPE-3 molecule³¹, demonstrating the versatility of the molecule choice. To explain the switching behaviour and investigate the underlying mechanisms, it was necessary to investigate the near-field angular momentum around the probe tip and its impact on the molecular conformation and Raman response.

To fully characterise the electromagnetic near field and the related angular momentum around the probe tip, numerical simulations were performed using the finite-difference time domain (FDTD) method³⁴. Both the substrate and the tip shell were gold, while the tip core was silica. Unless otherwise stated, the tip had a TP shape with mirror symmetry along the x-axis (Fig. 1). There was a 2-nm gap between the tip and the substrate, and the coordinate origin was located at the gap centre. The incoming illumination was modelled using a Gaussian beam, incident either along the z-axis for top illumination or in the x–z-plane at 45° with respect to the z-axis for side illumination. The Gaussian beam was s-(p-)polarised with an electric (E) field parallel (perpendicular) to the y-axis.

The effect of the nanostructure is important in the near field. Using p-polarised incident light from the left side, the electric field intensity (|E|²) was computed first, as shown in Fig. 3. The main contributor to the intensity was the E_z component (see Supporting Information SI.2 for details). The combination of both the tip and the substrate is essential to create a cavity and then confine the light tightly when the local surface plasmon is excited, resulting in an enhancement of the near field (Fig. 3c, d).

Fig. 3  Electric field intensity distributions of near-field mode for side illumination with different nanostructures. The electric field intensity distributions are shown in the z = 0 plane for a tip only, b substrate only, and c tip with substrate. d Profile of the electric field intensity at y = 0 (white dash lines in a–c).

Such a numerical analysis also makes it possible to discriminate the contributions from the s- and p-polarised components of the unpolarised laser beam used in the experiments. Fig. 4 shows the electric-field intensity for s- and p-polarised illumination from both channels (side and top). The gap mode can be excited mainly by p-polarisation for side illumination (Fig. 4a–c). Side s-polarised illumination leads to a negligible intensity (Fig. 4b) because the gap mode mainly relies on the E_z component. Therefore, this side s-pol illumination, without the E_z component, hardly excites the gap mode. Next, top illumination was studied (see Supporting Information SI.3 for further details). The s-polarisation was the main contributor to the top illumination (Fig. 4d–f). The top p-polarised illumination showed poor excitation of the gap mode owing to the mirror antisymmetry of the electric field along the x-axis. Therefore, regarding polarisation, only the contributions of the side p-polarised and top s-polarised illumination must be further analysed.

Fig. 4  Comparison of gap modes for different illumination sides and polarisations. The electric field intensity distributions are shown in the z = 0 plane for a side p-pol, b side s-pol, d top p-pol, and e top s-pol illuminations. c, f Profiles of the electric field intensity at y = 0 (white dash lines in a, b, d, e).

After the electromagnetic near field and the polarisation dependency were identified, the impact of the gap mode on the near-field angular momentum was examined. Fig. 5 shows the simulation results of the gap-mode angular momentum excitations for different illuminations and tip structures. The illumination was either side p-polarised or top s-polarised. In addition to the TP tip, a rectangular pyramid (RP) tip was introduced to illustrate the influence of symmetry on the system. The angular momentum excitation of the gap mode is characterised by the angular momentum density (AMD) distribution, which can be directly obtained as

Fig. 5  Comparison of gap-mode angular momentum excitations for different illuminations and tip structures. The components l_i (subscript i is for x, y, z) of the angular momentum density distribution are shown in the z = 0 plane for a–c top s-pol illumination with a TP tip, d–f side p-pol illumination with a RP tip, and g–i side p-pol illumination with a TP tip. j Total angular momentum |L| and its components L_i. The different illuminations and tip structures result in different angular momentum responses limited by the symmetries of the angular momentum density distribution. Only side p-pol illumination with a TP tip can generate a z-component of angular momentum.

where the vectors of $ \vec r $, $ \vec p $, and $ \vec S = {1}/{2}{Re} \left( {\vec E \times {{\vec H}^*}} \right) $ correspond to the position, momentum density, and Poynting vector, respectively, while ε and μ are the electric permittivity and magnetic permeability, respectively. In the case of top s-pol illumination with a TP tip, $ \vec E $ and the tip structure are symmetric along the x-axis, but $ \vec H $ is antisymmetric. Therefore, the components of the AMD satisfy the following conditions (see Supporting Information SI.4 for details).

This indicates that the AMD is antisymmetric along the x-axis in the system, as confirmed numerically in Fig. 5a–c. For side p-pol illumination, an RP tip symmetric along the x- and y-axes can result in better symmetry in the near-field system. Similarly, one can deduce the components of the AMD for side p-pol illumination with an RP tip, which can be expressed as

Correspondingly, the AMD is antisymmetric along the y-axis, as confirmed by the results shown in Fig. 5d–f. The case for side p-pol illumination with a TP tip has results (Fig. 5g–i) similar to those with the RP tip; however, its AMD distribution is asymmetric along the y-axis, induced by the asymmetry of the TP tip along the y-axis. The near-field angular momentum can be deduced from the AMD as

(Fig. 5j), where V is the volume of the near-field space. Because of the antisymmetry of the AMD for top s-pol illumination with a TP tip, the y- and z-components of the angular momentum are equal to zero. Side p-pol illumination with an RP tip yields similar results for the x- and z-components. Therefore, these two cases cannot generate angular momentum in the z-direction, which is critical for coupling with the molecule. However, the TP tip breaks the structural symmetry along the y-axis. Accordingly, it results in an asymmetric AMD distribution for side illumination that generates near-field angular momentum in the z-direction (i.e., L_z is nonzero for side p-pol illumination with a TP tip in Fig. 5j).

According to the above analysis, the angular momentum excitation of the gap mode is closely related to the symmetry of the optical system, and some components of the near-field angular momentum can be cancelled by symmetry. Consequently, symmetry must be considered when designing optical systems for applications based on the manipulation of near-field angular momentum.

As mentioned previously, the Raman activity of TM-TPD is dominated mainly by the molecular conformation characterised by the dihedral angle between the benzene rings at both ends, and the central benzene ring (angle shown in Fig. 2b and Supporting Information SI.1). In Fig. 6a, the Raman spectra are presented for different dihedral angles from 50° to 90° for a TM-TPD molecule, clearly indicating a decrease in intensity when the angle is larger than 50°, with a minimum of 90°. These results correlate with the relationship between the dihedral angle and the Raman spectrum.

Fig. 6  Raman investigation of the molecular junction with both top and side illumination. a Raman spectra for TM-TPD molecule calculated at the B3LYP-D3(BJ)/6-31G(d,p) level with different dihedral angles. The four carbon atoms associated with the dihedral angle are marked in brown. The Raman spectra change simultaneously with the variation in the dihedral angle from 50° (ON mode) and almost disappear at 90° (OFF mode). b Change of Raman spectra of the 1.0-V biased molecular junction continuous under top illumination and side illumination with 1% intensity of top illumination. The Raman peak positions shift progressively, and the intensities of the Raman spectra decrease to be finally turned from ON mode to OFF mode completely after 40 s.

We attribute the observed change in the Raman activity with side illumination when a bias voltage is applied to a similar modification of the dihedral angle of the molecule in the junction. The angular momentum of light causes a torque on the molecule, modifies its conformation, and suppresses the Raman mode (Fig. 2a). More precisely, the change in conformation and the intense Raman signal observed when a bias voltage is applied is reversed when the molecular junction is illuminated from its side.

Specifically, the z-component of the angular momentum can produce a torque $ \vec \tau = {{d\vec L}}/{{dt}} $ along the z-direction. For top illumination, the z-component of the angular momentum is zero (Fig. 5j); thus, the molecular conformation is unaffected, and the Raman activity is influenced by the biased voltage only. However, Raman activity is significantly different under side illumination. Because of the nonzero L_z component for side illumination (Fig. 5j), torque is produced in the z-direction. This torque causes the molecule to follow a rotational motion, breaking down the electron density rearrangement caused by the biased voltage and rendering the conjugation vibration Raman inactive again. This change in the conformation of the molecule under side illumination is possible because of the low energy barrier of rotation of the charged molecule (see Figure S1 in the Supporting Information). Consequently, the Raman intensity of the molecular junction falls below the detection limit. It is, therefore, expected that the Raman signature of TM-TPD can be increased by temporal charging and then depressed if coupled with the angular momentum of the light where the π-conjugated character is dominant.

To confirm the mechanism behind the switching behaviour of the molecular junction induced by the near-field angular momentum transfer to the molecule, combined side/top illumination characterisation was performed. From the simulation presented above, asymmetry in the electromagnetic field is reintroduced in such an illumination geometry, causing a nonzero angular momentum in the z-direction and producing a z-axis torque acting on the molecule. Accordingly, it was found experimentally that the Raman peak disappeared again when side illumination was added to the well-established molecular junctions monitored with top-illumination Raman spectroscopy (bottom curve, Fig. 6b). Side illumination was introduced with only 1% of the energy of the top illumination, and the Raman spectra were monitored over time. The Raman spectrum of the current-driven molecular junction was initially monitored using only top-illumination geometry (ON mode at t = 0 s). When the side illumination channel was added, the Raman peak position shifted progressively, and the intensity of the Raman spectra decreased, finally turning off completely (OFF mode) after 40 s. The top illumination can turn the Raman mode ON because of the symmetry of the near-field system; however, the symmetry is broken when 1% intensity from side illumination is added to the system, and the conformation of the TM-TPD molecule is gradually changed to depress the Raman mode. For the turning-OFF process, the molecular junction exhibits a dynamic vibrational process. However, the theoretical DFT calculations were performed based on the static state of the stable conformations. The measured signals were not very strong and were easily affected by noise; thus, the experimental results did not perfectly match the calculated results. The turning-OFF process shows that the molecular junction is coupled progressively with the angular momentum of light and progressively changes the dihedral angle and, consequently, the Raman intensity. Furthermore, the Raman modes can be turned ON again in the single-molecule junction after switching OFF. The reproducibility of the process was confirmed experimentally.

It was shown that conformational molecular junctions can be controlled not only by the voltage applied to a single-molecule junction but also by the optical angular momentum in the near field, which is enhanced by the plasmon gap mode in the junction. A bias voltage exceeding 1.0 V results in an electron density rearrangement of the molecular junction, which activates the Raman mode. This Raman mode is enhanced by planarisation and increased π-conjugation of the TM-TPD molecule. In striking contrast to top illumination, side illumination of the junction leads to the absence of a detectable Raman signal for all biased voltages. The angular momentum of light causes a z-axis torque when the distribution of the electromagnetic field is asymmetric. This makes it possible to change the conformation of the TM-TPD, destroys the transport electron density rearrangement, and finally suppresses the Raman mode. These studies made it possible to identify experimentally that the optical angular momentum is a driving force for molecular switching. This conclusion was reached based on detailed numerical simulations of the near-field EM field. More precisely, it was found that the angular momentum excitation has a very close link with the symmetry of the near-field system. This study opens up new avenues for research on molecular logic and optical angular momentum in the near-field regime. The possibility of addressing physicochemical phenomena with a single-molecule resolution shown in this study serves as a departure point for a new generation of nanomechanical studies using MJS setups. Technologically, these investigations demonstrated single-molecule monitoring and manipulation with excellent control, opening new avenues for multilogic single-molecule computing.

The research described in this article was supported by the National Key Research and Development Program of China (No. 2021YFE0115700), Guangdong Basic and Applied Basic Research Foundation (Nos. 2020A1515110709, 2020A1515110712, and 2020B1515120068), and Jihua Youth Innovation Fund (No. X201021XF200). The authors acknowledge support from the WBI – MOST agreement. M.L. is a Research Associate of the Fonds de la Recherche Scientifique – FNRS.