The azobenzene derivatives, tetra-acidic azobenzene (NN4A) and di-acidic azobenzene (NN2A), as seen in Scheme 1, were synthesized by the reported process (Wang et al., 2014). Heptatonic acid was used as the solvent bought from J&K Chemical Ltd. (Beijing, China), and the solvent was used without any further purification in this work.

All the studied samples were dissolved in heptatonic acid at less than 1.0 × 10⁻⁴ M concentrations, then a droplet of NN4A or NN2A solution (0.5 µL) was deposited onto the freshly cleaved highly oriented pyrolytic graphite (HOPG, grade ZYB, Advanced Ceramics Inc., Cleveland, United States), forming the self-assembly samples. The samples were prepared under ambient conditions.

Characterization of self-assembled surface morphology was performed with a Nanoscope IIIA STM (Bruker, United States) under atmospheric conditions, and the STM images were obtained in constant current mode using a mechanically formed Pt/Ir (80/20) tip at the liquid-solid interface. The figure captions have included the specific tunneling conditions (i.e., tunneling current and bias). The drift for all the images was calibrated using an atomic-resolution HOPG lattice as a reference.

In addition, we have also done a series of other characterizations of the photosensitivity of the self-assembled molecular layer and the carboxylic azobenzene, using instruments including the X-ray photoelectron spectroscopy/ATR-FTIR/Contact angle goniometry/spectroscopic ellipsometry/light scattering measurements. Supplementary Material for detailed results.

The theoretical calculation was performed using density functional theory (DFT) provided by the DMol3 code (Delley et al., 2000). Herein, we used the periodic boundary conditions (PBC) to describe the 2D periodic structure on the graphite. A self-consistent field procedure was performed with a convergence criterion of 10^–5 on the energy and electron density. Combined with the experimental data, we have optimized the unit cell parameters and the geometry of the adsorbates in the unit cell. To evaluate the interaction between the adsorbates and HOPG, we designed the model system. In our work, we performed our calculations on infinite graphene monolayers using PBC. In the super lattice, graphene layers were separated by 40 Å in the normal direction. As for the adsorbates on graphene, we used graphene supercells and sampled the Brillouin zone by using a 1 × 1 × 1 k-point mesh (Shi et al., 2018).

The microscopic lateral forces (Friction) of self-assembly samples were measured by the MFP-3D AFM (Asylum Research, United States) with the silicon CSG10 probe (nominal normal spring constant 0.1 N/m) at room temperature. The measurement was performed at the gas-solid sample interface after the solvent evaporated. Before each measurement, the genuine normal and lateral factors need to be calibrated. The sensitivity of the normal photodetector was calculated from the slope of the force curve acquired on a stiff substrate. The power spectral density of thermal noise fluctuations under environmental conditions can be used to estimate the normal spring constant. The voltage result was converted into a force value, and the lateral factor was calculated based on the improved wedge calibration method by scanning the commercial TGF11 silicon grating (MikroMasch). In the contact mode, the microscopic friction was measured by Scanning perpendicular to the direction of the cantilever beam. The scanning area was 100 nm × 100 nm, and the scanning frequency was 1 Hz. In addition, the adhesion force was recorded the retrace signals with 15 nN load.

The photo-irradiation experiments were carried out in order to study the isomerization reversion, with UV light at 365 nm for 40 min, and the re-examined samples were illuminated by visible light at 560 nm for 20 min using the Xenon lamp (PLS-SXE 300, 50 W); the 365 nm glass filter and 560 nm filter were bought form Peking Perfect Company (CHN).

The self-assembled nanostructure of NN4A is shown in Figure 1. The self-assembled molecular layers of NN4A substances have been characterized in previous studies (Li et al., 2008), as seen in Figure 1A. Combined with previous research results, the assembly structure of NN4A on the HOPG surface forms a Kagome´ image in which two adjacent carboxylic groups form a pair of hydrogen bonds and make up the network structure, which included two cavities formed by six symmetry benzene rings and three NN4A molecules, respectively. As one typical photosensitive substance, NN4A will isomerize under light irradiation. The NN4A self-assembly layer on HOPG could isomerize from trans to cis isomers upon irradiation with UV light at 365 nm and reversibly isomerize from cis to trans when irradiation with visible light at 560 nm. As revealed in Figure 1B, after UV light irradiation, the Kagome networks of NN4A are broken, and liner characteristics appear in the assembled structures of NN4A-uv. Unlike the case of the NN4A, one NN4A-uv molecule is linked by two other NN4A-uv molecules, and the two adjacent NN4A-uv molecules arrange in an opposite direction via H-bonds, thus every isomerized NN4A molecule forms two pairs of hydrogen bonds, as shown in Figure 1B, the corresponding molecular model calculated by DFT method based on STM observations. The measured unit cell is also superimposed on the molecular model with a = 0.8 ± 0.1 nm, b = 1.1 ± 0.1 nm, α = 87 ± 1°, as shown in Table 1. In addition, it can be seen in Figures 1Aiii,Biii that the 3D structure of the molecular layer changed from the tile type to the bulge shape at the hydrogen bond connection position after being irradiated by light.

Another sample substance NN2A was studied. However, unlike NN4A, NN2A has only a symmetric pair of carboxylic acids groups at both ends. As presented in Figure 2A, the self-assembled molecular layer of NN2A is detected and shows an order structure: every single molecule in the region is lined up. The length and width of the unit cell are measured according to the length size of the central NN2A backbone. However, the width of the bright spot stripe of NN2A can be known to be less than that of the NN2A molecule, so every molecule is sloping on the HOPG. The unit cell parameters are calculated with a = 0.76 ± 0.1 nm, b = 1.48 ± 0.1 nm, α = 33 ± 1°, as shown in Table 1. In Figure 2A, the proposed molecular model of the STM image can be seen. After UV light irradiation at 365 nm, a different well-ordered network can be obtained, as shown in Figure 2B. Due to structural changes in the STM molecule image patterns, it shows good agreement with the isomer transformation from trans to cis. The distance between two molecules in the length direction becomes smaller after UV light illumination. The parameters of a unit cell in the Figure 2B model are also presented: a = 0.56 ± 0.1 nm, b = 1.17 ± 0.1 nm, α = 65 ± 1° (as seen in Table 1). Similarly, Figures 2Aiii,Biii show that the 3D structures of the molecular layer of NN2A also have obvious changes. The longitudinal inclined benzene ring formed through the tightly bound of the hydrogen bond between the molecules of NN2A, while the head-to-head linkage between the molecules of NN2A-uv result in the projection of the carbon and hydrogen on the benzene ring into a labile wrinkle-like structure.

All the well-ordered formed patterns are stable by virtue of the van der Waals interactions and hydrogen bonding (Zhu et al., 2019). The van der Waals interactions mentioned here mainly come from the interactions between the molecule and HOPG. Meanwhile, the hydrogen bonding is expressed as O-H...O, which is formed from an O atom in the side chain in the carboxyl group and an H atom in the carboxyl group of the adjacent molecule. The hydrogen bonding and van der Waals interactions together control the self-assembled structures, when the UV light triggers the self-assembled molecules.

Generally, the self-assembled molecule structures result from the balance of a variety of interactions, which include the molecule-molecule and molecule-substrate interactions (Zhu et al., 2019). In order to further understand the self-assembled mechanisms based on observed phenomena, we performed DFT calculations to investigate the arrangement of self-assembled nanostructures based on the related interactions. The DFT results exhibit a difference in the energies before and after light irradiation (as listed in Table 2). The total energies of NN4A self-assembled structure (−101.506 kcal/mol) are higher than those of the NN4A-uv structure (−71.704 kcal/mol), which is possibly caused by isomerization. Simultaneously, the total energies of the NN2A self-assembled structure (−41.839 kcal/mol) are smaller than those of the NN2A-uv structure (−44.597 kcal/mol). For comparison, the energies of the NN4A molecules are higher than those of NN2A, which is attributed to the stronger interactions between the NN4A molecules. Other than the intermolecular interactions, the interactions between molecules play an important role in the surface supramolecular assembly process. As shown in the first column in Table 2, the interaction energies between molecules of NN2A (−14.873 kcal/mol) are lower than those of all the others. It can be concluded that the van der Waals interaction between molecules and HOPG is strongest due to the much tighter alignment of molecules, which can evaluate the interaction strength and thermodynamic stability of the assembly systems with different unit cells.

There have been some studies on the tribological properties of the self-assembled organic molecule layers (Nalluri et al., 2011; Shi et al., 2018), while the studies on the tribological properties of the photo-regulated molecular layers have been reported rarely (Pace et al., 2007; Tang et al., 2020). For the self-assembly layers of the NN4A and NN2A molecules, we had investigated the tribological lateral forces using AFM at room temperature. Under the light stimulation, the contrast between the illuminated molecular layers and the original molecular layers produced a significant friction contrast. The experimental results of AFM friction performance are summarized in Figure 3, the lateral forces (Friction) of the NN4A-uv/NN4A/NN2A/NN2A-uv/HOPG are listed as a function of the applied load. It can be seen in Figure 3A that the lateral force of the system has a good linear fitting relationship with the load. By contrast, the lateral force of NN4A is generally higher than that of NN2A, which corresponds to the total energy in Table 1. Meanwhile, we defined the slopes of the linear fitting curves of the lateral force-load as the corresponding friction coefficient, labeling this in Figure 3A. For NN4A, the friction coefficient of the NN4A-uv structure self-assembled layer after light illumination is smaller than that of the original material NN4A, with about a 56% reduction. However, in comparison, the friction coefficient of the NN2A-uv structure after light illumination has the opposite trend, which is higher than that of the original material NN2A, with approximately more than twice. Therefore, light stimulation has a better regulation effect on the self-assembly system of NN2A. Especially, comparing the three other self-assembled systems, the NN2A has the lowest friction coefficient below 0.001, indicating it has excellent microscale superlubricity properties.

Moreover, we also investigated the adsorption energy by the AFM force measurement mode to further research tribological properties of the self-assembled system (as seen in Figure 3B). As for NN4A, combining with the results of simulation analysis, the structure of NN4A-uv after exposure to light has smaller total energy values (consisting of interactions between absorbates as well between absorbates and substrate). It is mentioned here that the unit cell area of the NN4A-uv substance after light illumination is larger than that of the original substance NN4A, the obtained per unit area energy is smaller. However, by observing the 3D molecular layer structure before and after light illumination (shown in Figure 4A), it can be found that due to the increase of the fluctuation of molecular layer structure after illumination, leading to the adhesion force (F_adhesion) between the probe and the molecular layer increases. As a result, the NN4A-uv material molecular layer after illumination has a larger initial lateral force as shown in Figure 3A.

Similar to NN2A self-assembled systems, the 3D structure of the NN2A self-assembled molecular layer transformed from the original approximately tile-like state to the state of molecular erection closely arranged after light illumination, and the energy between adsorbates increases. Therefore, more energy barriers need to be overcome when the self-assembled molecular layer comes into contact with the AFM probe, which is reflected in the increase of the friction coefficient of the system. The more important is that the hydrocarbon structures on the benzene ring in the NN2A molecular layer are exposed after light illumination (as 3D structures shown in Figure 4B), which will interact with the hydroxyl on the silicon tip to increase the lateral friction force. However, the conjugation between the benzene ring and the HOPG decreases due to the molecule isomerization, so that the adsorption energy between the adsorbates and the HOPG decreases, and the adhesion force thus decreases after light irradiation.

In conclusion, the experimental results indicate that light-stimulation has a good effect on the regulation of photo-isomerizable molecular self-assembly systems, and the regulation of dicarboxylic azobenzene (NN2A) has a more obvious result than that of tetra-carboxylic azobenzene (NN4A) molecule. As shown in Figure 5, the friction properties of the two self-assembled molecular layers under photo regulation have displayed excellent fatigue resistance. This has great implications for our choice to optimize the friction performance of self-assembly coatings on micro-surface interfaces.