Photo-Programmable Processes in Bithiophene–Azobenzene Monolayers on Gold Probed by Simulations

1. Introduction

Molecular switches (MS) represent the molecules that respond to stimuli by changing between two or more conformational or configurational states. In the former case, the switch involves a physical transformation in geometry of the molecule, while configurational switches are related to bond-breaking and bond-forming reactions [1]. An example of a configurational molecular switch is one with an azobenzene (Azo) group, which undergoes reversible photoisomerization depending on the wavelength of the light trigger. In the dark or upon irradiation with visible (usually blue) light, azobenzene adopts a trans-state, which represents a global minimum on the potential energy surface. Under UV light, a cis-state is formed as a product of isomerization. The process is fully reversible, and under visible light or upon heating, the metastable cis-isomers change their configuration back to the trans-state. Responsive monolayers of Azo-MS [2] are the key building blocks for advanced applications ranging from smart surfaces with switchable wettability for textile industry [3] and biotechnology, where the binding and release of ions [4] and biomacromolecules [5] can be controlled, to modern elements of organic electronics, such as molecular junctions (MJ), where MS are bound to metal or semiconductive electrodes  [6,7], and sensors [8].

Embedding MS into semiconductive material is a straightforward method for implementing the versatility of optoelectronic devices. The organic semiconductor (OSC)/electrode interfaces can also act as places where MS can function [9]. According to recently published review of Chung et al. [9], MJ engineering can be divided into two categories: photo-programmable tunneling injection control and work function tuning of the electrode.

Literature surveys have demonstrated that azobenzene-based molecules organized as molecular junctions on electrode surfaces can work as photoswitches of (i) conductance [10,11,12,13,14,15], (ii) electrode work function [13,16,17,19], and (iii) magnetization/magnetic transitions [20,21]. Changes in conductance are mainly related to the tunneling barrier length between the top probe/electrode (such as Hg droplet, C-AFM or STM tip) and the bottom electrode, usually a gold surface, Au nanoparticle (Au–NP) [22,23] or cobalt [15]. In other words, they are related to the configurational changes that directly affect the molecule–probe intervals. The height of MS layer between two electrodes for trans-isomers is larger, and in the majority of experimental papers this isomer is attributed as the OFF–state with higher resistance, whereas the shorter cis-form is the ON–state. The ON/OFF conductance ratio sometimes reaches dramatic values [11,12,14,23,24,25]. However, the identification of the cis-state as the conducting one is not unambiguous. For example, Kumar and colleagues [26] have shown that the cis-azobenzenes incorporated in 1-decanethiolate SAM (self-assembled monolayer) films, i.e. in two-component SAM, have properties of OFF-state. Here, the azobenzene-containing molecules work as approximately 2.1±0.3Å and ca. 0.7±0.2Å apparent protrusions from the 1-decanethiolate matrix for trans- and cis-isomer, respectively, with a longer protrusion providing a contact with the top electrode. Another example by Del Valle et al. [27] illustrates by first-principles calculations that the conductance of azobenzenes is possibly electrode-sensitive, with the cis-isomer being the OFF-state for carbon nanotube electrodes and the cis-isomer being the ON-state for silicon contacts.

The photo-programmable electrode work function tuning is attributed to differences originating from significant metal-to-molecule electron charge transfer, or to the role of the dipole moment of the MS. The latter effect is especially intriguing in light of the fact that azobenzene isomers vary significantly in the magnitude of dipole moment and its direction [28,29,30], and can lead to more effective and controlled tuning of the electrical characteristics in biosensing device applications [19].

In general, the photo-programmable MS layer in MJ is situated on the upper surface of the electrode, and ideally, a thin and uniform semiconductor 2D structure is formed on the surface of MS layer to maximize the switching effect of the device [9]. However, this upper layer of OSC on MS is difficult to achieve for large scale solution processing; additionally, its thickness and uniformity are the issues of incident light quenching. The new interesting and at the same time more complex systems to circumvent the aforementioned challenges are the MJ layers consisting of azobenzene molecules comprising the Azo group conjugated with aromatic units or semiconducting oligomers. For example, additional benzene [14,15,16,31,32,33,34,35,36,37,38,39] or fluorinated benzene rings [17,18,40,41], the second covalently attached azobenzene [42], biphenyl [43], azothiophene [44], bithiophene [10,11,15,23,45,46] were included into the chemical structures of one-component Azo-MS. Newly introduced blocks have:

• influenced the extent of conjugation;
• affected dipole moment;
• changed backbone rigidity,
• brought new functionalities such as strong stacking between the aromatic units,
• altered the system’s response to light stimulus (i.e. accelerated collective switching behavior or caused significant broadening of the optical absorption spectra);
• emphasized the role of spacer length [11]: a shorter spacer should facilitate the charge transfer rate through the junction and increase conductance, while a longer spacer is prone to decoupling of the Azo from the electrode, therefore
• allowing larger dynamics of the switching event, and consequently
• a larger ON/OFF conductance ratio.

The findings listed above have important implications for understanding charge transport mechanisms in MS devices. Developing correlations between structure and function is necessary for fully understanding the properties of molecular junctions platforms. Therefore, it is important to account for molecular–level variations that can occur in the bound layers. In this context, optical and transport experimental and theoretical characterizations can provide insights into the structural aspects of MS layers.

Here, we report the structural, optical and transport properties of a molecular switch in which the azobenzene moiety is linked to a bithiophene spacer and a short alkanethiol (Figure 1). This design (Figure 1b) tethered to a gold electrode or Au–NP has been previously developed by Lenfant and others, and has been studied in a series of works [10,11,23,45,46,47]. The incorporation of bithiophene segment, which is redox active at moderately positive potentials, allowed for preliminary characterization of the immobilization process using simple electrochemical experiments [10]. The bithiophene unit also helps maintain a high level of current in the BT-Azo molecular junction, as compared to a fully saturated n-alkyl spacer with the same length [11]. In addition, the bithiophene segment promotes possible stacking interactions in the molecular junction layer, as well as endows the molecule with more conformational rigidity [48], as suggested by Lenfant and colleagues.

In addition to the synthesis and experimental characterization of the BT-Azo molecular junctions, Lenfant et al. [23] have simulated the isolated trans- and cis-isomers of this switch applying density functional theory (DFT) with B3LYP/6-311G options and calculated the optical absorption spectra for the optimized isomers in vacuum using time-dependent DFT. The same BT-Azo states in a contact with the top electrode distanced at 0.2nm above the molecule and modeled as Au${}_{11}$ cluster were described employing DFT with B3LYP functional and 6-31G and LANL2DZ basis sets for organic and inorganic parts of the system, respectively [11]. The main theoretical output was that the high ON/OFF cis/trans conductance ratio is primarily due to the changes in both the molecule–electrode coupling and the electronic structure of the metal–molecules–metal junction, rather than due to the variation of SAM thickness. Specifically, a top mechanical electrode contacts the “insulating” methyl group in trans-SAM (here, –CH${}_3$ is acting as a tunnel barrier) and the Azo bond or aromatic benzene ring in cis-SAM under UV light (these groups are strongly coupled electronically to the Au${}_{11}$ cluster). Besides, the electronic density delocalization and the charge transfer from the switch to the top electrode was found to be more pronounced for the cis-isomer.

Five years later, Lenfant, Cleri et al. have conducted classical molecular dynamics (MD) simulation of realistic SAMs utilizing MM3 empirical force field in order to emulate the dynamical structure of the interfaces. In focus was the evaluation of the intermolecular interactions and mutual orientations of two contacting flat or curved layers of interdigitating densely packed BT-Azo switches (Figure 1b) from adjacent spherical or icosahedral Au–NPs forming the so-called nanoparticles self-assembled networks (NPSANs) [23]. The main conclusions were that switching between configurations changes the overlap (here, the physical touch of the molecules, their interdigitation and the stacking are meant) between the molecules which causes variations in conductance. Interestingly, the overlap between the cis-switches at the NPs interface in NPSANs is drastically reduced already at 100 K and vanishes at 300 K, whereas the trans-switches keep the overlap at all temperatures tested, and electrons have a larger tunnelling probability to jump between NPs. Nevertheless, the cis-isomer, as measured in experiments, still plays a role of ON–state, and the mentioned effect coming from more efficient molecular overlap does not prevail over the tunneling barrier length, i.e. the SAM thickness, which is larger for trans-switch (OFF–state). As a result, the smaller ON/OFF conductance ratio up to 620 has been detected, as compared to previously observed value 7×10${}^3$ for the planar SAMs in a contact with mechanical top electrode [10,11]. In the latest published work dedicated to BT-Azo MJ [15], the MS effects on cobalt electrode work function have been characterized. The DFT model included both electrodes (Co and the gold top electrode) and single BT-Azo molecule in different configurational and conformational states between them. We should note, that in contrast to the previously studied systems [11,23], where the molecules were standing almost upright on the surface, the MS molecules on Co are tilted with respect to the substrate with tilting angle of 73±4${}^o$. In this system, the trans-isomer has been identified as ON–state, while the cis-isomer is responsible for lower conductance.

To the best of our knowledge, only these three publications [11,15,23] have described theoretically the azobenzene switches with bithiophene segments coupled in one architecture. The focus, as we mentioned, has laid on the explanation of the conductance changes happening under the light stimulus and the high or moderate ON/OFF cis/trans conductance ratio for studied samples. The motivation is clear, the authors have pursued the goal of elucidating their own experimental findings. Meanwhile, in the conclusions to their published papers [10,11,23,45,46,47], they have indicated several interesting directions for the future development of research including purely theoretical approaches, which could uncover the phenomena and issues that have not been gauged in the mentioned works. For instance, the overlap of the molecules found in molecular dynamics (MD) simulations, depends not so much upon the interdigitation and the mutual molecular orientations, as seen from the analysis of the orientation dihedral angle between the conjugated planes of trans-azobenzenes [11,23]. The reason for this is unknown. Another factor that may come into play is the intermolecular stacking distance, d, which has not been altered in experiments. Besides the variations of d, which have a strong impact on the electron transport properties, the electronic coupling between the neighbouring molecules and the charge transfer within the MJ layer are also interesting tasks for computer modelling. These and other questions related to the rational design of azobenzene–bithiophene molecular switches (Figure 1a) we would like to address in the present paper.

2. Models and Methods

Optimized isolated molecules have been taken from our previous study [30]. Further calculations with B3LYP/6-311G(d,p) functional and basis set concern the isomers geometries in implicit dichloromethane (the solvent, that used in experiments on BT-Azo [10]) modeled as polarizable continuum using the integral equation formalism [49]. This functional has been widely used to optimize geometries of various azobenzene-containing molecules assembled in layers [11,23,39,50,51]. Later, the isomers optimized in dichloromethane are subjected to time-dependent DFT calculations of absorption spectra with CAM-B3LYP/6-311G(d,p). Here, the vertical transition energies to the first hundred excited states are evaluated. The electronic spectra are simulated employing Gaussian functions with the half-width at half height of 0.333 eV to build a continuous spectrum from a collection of transitions corresponding to the time-dependent DFT transition energies and oscillator strengths.

Another property defined for isolated MS is the reorganization energy $\lambda$. This one of the parameters for calculation of the charge transfer rate $k_{CT}$ following the Marcus theory:

$$k_{CT}=\frac{2\pi }{\hslash }{\left|V\right|}^2\frac{1}{\sqrt{4\pi \lambda k_{B}T}}exp\left(-\frac{{(\lambda +\Delta G^0)}^2}{4\lambda k_{B}T}\right),$$

(1)

where V (the charge transfer integral) is the second variable of $k_{CT}$. The value of $\Delta G^0$=0 for the self-exchange reactions, i.e. where the electron or hole is transferred between chemically identical molecules, and therefore [52]

$$k_{CT}=\frac{2\pi }{\hslash }{\left|V\right|}^2\frac{1}{\sqrt{4\pi \lambda k_{B}T}}exp\left(-\frac{\lambda }{4k_{B}T}\right).$$

(2)

The reorganization energy is calculated following the approach described by Nelsen, Blackstock and Kim [53], and the V values for interacting neighboring molecules are evaluated applying the so-called energy splitting in dimers [54] for various intermolecular stacking distance d. These approaches have been utilized in some recent publications [55,56,57]. The values of charge transfer rate $k_{CT}$ later are used for the calculation of electron/hole mobility in one-dimensional molecular stacks via the Einstein–Smoluchowski equation for diffusion of charged particles:

$$\mu =\frac{eD}{k_{B}T}=\frac{e{d}^2{k}_{CT}}{2k_{B}T},$$

(3)

where D is diffusion coefficient, T is temperature (is set at 298K), d is either the intermolecular stacking distance for dimers or the separation between centers of masses of the molecules in interacting dimer (the details are provided below). All DFT and time-dependent DFT calculations are conducted in Gaussian 09, Revision A.01 [58].

The collective behavior chemisorbed layers is modeled in MD runs using Materials Studio [59]. The simulations are carried out in the slab models, which are in the size of 53×57×103 Å with 81 Azo-BT or BT-Azo pure cis- or pure trans-layers anchored to the Au(111) with surface concentration of ca. 2.7 molecules per nm${}^2$, which is close to the experimentally evaluated ligand density of 3.2 molecules per nm${}^2$ for gold nanoparticles [23]. We should note, that this relatively high concentration leads to quite dense layers of azobenzenes anchored to the surface. The starting configurations of the layers are constructed by an algorithm developed by us, which is stored at bitbucket, t.ly/81cWk. In short, the optimized geometries (rigid bodies) of isomers have been placed vertically as a grid 9×9 without additional rotations with a distance between any neighboring molecules (atoms) in the rows of 6.8–12 Å. Next, the molecules are moved in x, y or in a diagonal direction with a certain step. The algorithm stops, when the minimal chosen distance between the molecules is reached. As a result, denser SAMs of vertically oriented molecules are prepared. Built layers are placed in the simulation box with their thiol groups next to the frozen gold surface modeled as two layers. Further, the thiol –SH groups are constrained. This is needed to mimic chemisorbed layers of MS. We omitted a step of real chemical attachment of the molecules to the gold with dissociation of the thiol proton, since it needs then a parameterization of newly formed S–Au bonds, we should not check their initial lengths (since molecules are not on the grid on the surface, the distances between S and the neighboring Au vary), and at the same time, chosen simplification is not limiting the model of MJ – we are mostly interested in the phenomena happening at the rim of the layer, distant from the surface. The intra- and intermolecular parameters, as well as the interactions with gold are parameterized using the polymer consistent force field (PCFF), previously applied for the simulations of the assembled Azo-layers [32,60].

Each simulation box containing pure trans- or cis-Azo-BT or BT-Azo is subjected to preliminary equilibration stage using Smart algorithm, followed by initial NVT-MD run with the following parameters: T=298K (Nose thermostat, Q ratio is 0.01), 500ps with a time step of 1fs, PCFF partial charges (Ewald summation method with accuracy of 0.001kcal mol${}^{-1}$ and a buffer width of 0.5Å), atom-based summation method for the calculation of the van der Waals interactions (9-6 Lennard-Jones potential with a cutoff distance of 12.5Å). The NVT-MD production runs are performed under the same settings with trajectories of 10ns. Later, these snap-shots are used for the analysis of the structural properties of the layers, like the inclination angles of the molecules, the thickness, etc. Besides, from MD data the interacting dimers are taken for calculation of the averaged values of transfer integrals, as described above.

As an auxiliary property characterizing changes in the wettability of the monolayer upon light stimulus, the solvation free energies ($\Delta G$) are evaluated for the molecules dissolved in dichloromethane or water. Here, trans- or cis-isomer is placed in a cubic simulation box with 1500 CH${}_2$Cl${}_2$ or 1700 H${}_2$O molecules, equilibrated in NPT-MD (P=1atm, Berendsen barostat) for 500ps until the solution density reaches the value close to the experimental one: 1.071g cm${}^{-3}$ (dichloromethane) and 1.018g cm${}^{-3}$ (water); pure solvents at 25${}^o$C have densities 1.316g cm${}^{-3}$ and 0.997g cm${}^{-3}$, respectively [61]. The prepared solutions are calculated later using thermodynamic integration technique (NPT-MD). Other settings for this calculation are used as described elsewhere [28,29,62]. The averaging is done for three independent calculations for each modeled system.

The obtained data are compared with experimentally derived layer thicknesses [23,30] and inclination angles [15], measured [10,23] and calculated [23] optical absorption spectra, wetting behavior of the layers [11].

3. Results

3.1. Properties of isolated molecules

As mentioned above, we study intermolecular interactions of two azobenzene switches combining in their structure bithiophene with a short alkanethiol tail (Figure 1). The latter one is the part of the molecule, which is chemisorbed via thiol group to the bottom electrode in experiments with BT-Azo [10,11,23,45,46,47]. Structural properties of the isomers of Azo-BT and BT-Azo have been described in our previous study [30]. In that publication, we have considered trans-, cis-Azo-BT and BT-Azo states as isolated molecules and being in a contact with top golden tip modeled as Au${}_{10}$ cluster. The most relevant data from Ref. [30] for the present paper are:

• the values of dipole moments (1.34 and 3.22 D for trans- and cis-BT-Azo, 1.67 and 3.52D for trans- and cis-Azo-BT);
• the differences in the length of trans- and cis-isomers $\Delta$h is 6.05Å and 14.69Å for BT-Azo and Azo-BT, respectively; the experimental value of $\Delta$h=5.1Å for the layer of BT-Azo [23];
• even though the molecular volume reduces upon trans-cis isomerization for both MS, the intermolecular sterical clashes may arise for cis-populated layers, especially for Azo-BT MS.

3.1.1. Optical properties

First, let’s consider the optical properties predicted for the isolated molecules immersed in implicit dichloromethane. The time-dependent DFT calculated spectra are compared in Figure 2. In the experiment from Ref. [10], it was reported that upon irradiation of the BT-Azo solution at 360 nm, the band at 342 nm ($\pi -{\pi }^{*}$) progressively decreases and broadens, while the absorbance of the n–${\pi }^{*}$ transition in the 400–450 nm region increases, suggesting the trans-cis isomerization of the molecular switch (MS). The reduction of absorbance close to 340–350 nm during the transition from the trans to the cis state was also observed for the chemisorbed BT-Azo on Au nanoparticles in dimethylformamide [23]. The calculated spectrum reported by Viero et al. [23] shows a major active transition at $\lambda$=353 nm for the trans-isomer, while the cis-state is characterized by three absorption bands located at 300, 337, and 520 nm.

Our results predict the position of ${\lambda }_{max}$=342 nm for the trans-isomers of both MS (Figure 2a,b, top panels), which quantitatively coincides with the experimentally found value. The curves of the spectra are not sensitive to the positions of Azo and BT blocks in the molecule. In contrast, the data for the cis-isomers are slightly different for Azo-BT and BT-Azo: the band is blue-shifted for Azo-BT at 310 nm (325 nm for BT-Azo) and is characterized by a smaller value of the oscillator strength f. At the same time, the absorbance of the n–${\pi }^{*}$ transition is seen at 455 nm for both molecules. The latter one for BT-Azo is closer to the experimental value mentioned above, compared to the predictions of Viero et al. [23]. The bands around 310-325 nm include excitations coming from the bithiophene fragment linked to the azobenzene moiety via the sulfur bridge, and they are classified as $\pi -{\pi }^{*}$ transitions, which is in agreement with experimental findings [10]. Since the bithiophene fragment in Azo-BT and BT-Azo has different neighboring connections (one –S– and two –S– bridges, respectively), and participation in the conjugation due to lone electron pairs, it explains the observed shifts in the absorption spectra.

3.1.2. Reorganization energies

Table 1 collects the values of the reorganization energies in eV for the hopping of holes (${\lambda }^{+}$) and electrons (${\lambda }^{-}$) calculated for the isolated molecules in implicit CH${}_2$Cl${}_2$ as described in the previous section.

Larger values of the reorganization energy suggest more substantial conformational changes of the molecule during the hopping process. In the case of the experimentally known BT-Azo molecule, the reorganization energy required for hole transport is up to three times higher compared to electron hopping. For both isomers of the designed Azo-BT molecule, ${\lambda }^{+}$ is two times smaller than that of BT-Azo. However, there is no clear dependence on the isomeric state for this molecule. Notably, the restructuring of cis-Azo-BT during electron uptake is less energy favorable. With the exception of cis-Azo-BT, electron transport is preferred in the remaining cases. In the following sections, we will explore how the reorganization energy, together with the hopping distance, determines the charge transfer rate and charge carrier mobility.

3.1.3. Gibbs free energy of solvation and its changes upon light stimulus

The next property of the studied molecules is the free energy of solvation, modeled for two solvents - dichloromethane (CH${}_2$Cl${}_2$), which is thermodynamically good and in which BT-Azo has been synthesized [10], and water, which is thermodynamically poor and was used in the wetting experiments [10,11]. Table 2 presents the values of $\Delta G$ with standard deviations, and the mean values of the ideal, van der Waals, and electrostatic terms.

In the experiments, it was observed that under UV stimulus for 90 minutes (with an intensity of 70–250 $\mu$W cm${}^{-2}$) [10,11], the water contact angle decreases from 98±2${}^o$ (in the trans state) to 93±2${}^o$ (in the cis state). The process of partially improved wetting is fully reversible, as shown in those experiments, and the contact angle dynamically changes upon blue (trans-rich layer) to UV (cis-rich layer) illumination. In the experiments conducted by Lenfant and co-authors [10,11,47], this behavior was explained by the difference in the static dipole moment of the trans and cis-isomers of BT-Azo. In our recent publication, we confirmed that the cis isomer is a more polar molecule, characterized by a dipole moment of 3.22 D (while the trans-state value is 1.34 D) [30].

The values presented in Table 2 complement the aforementioned explanation of the wetting experiments and, at the same time, appear more reliable as they are statistically averaged. These values were obtained through classical simulations at room temperature in systems with explicit solvent molecules, ensuring meaningful intermolecular interactions and preferred molecular solvation. It is worth noting that the partial charges for the isomers were determined from electrostatic potentials, as described elsewhere [28,62].

The solvation calculations in dichloromethane serve as a validation of the model, as all the values are negative, indicating the dissolution of the molecules in this solvent. The solvation energies are independent of the isomerization state and the position of BT and Azo in the molecular structure.

In contrast, the hydration process shows positive $\Delta G$ for both isomers of the BT-Azo molecule, but the value is two times smaller for the cis-isomer compared to the trans-counterpart. This observation rationalizes the experimental findings [10,11,47]. For the Azo-BT molecule, more significant changes in hydration are predicted. The wetting of the layer under UV light may become complete, as $\Delta G<$0. Therefore, dewetting/wetting switching could be observed under blue/UV illumination. Several factors contribute to this phenomenon, such as the larger dipole moment of the molecule, improved accessibility of the molecular groups for water, lesser conformational rigidity of the backbone, and, consequently, coordination of water near hydrophilic sites of the molecule.

3.2. Properties of chemisorbed monolayers

3.2.1. Structural properties

Now, let’s switch from considering the isolated molecules in implicit or explicit solvents to the properties of the chemisorbed layers. The latter ones are visualized in Figure 3. For the chemisorbed layers, we chose a surface concentration of 2.7 molecules per nm${}^2$, which is less dense compared to the anchoring on the facets of the gold nanoparticles in experiments [23]. We reduced the surface concentration in our model as we are considering flat surfaces, not 3D surfaces, where more molecules can be chemisorbed due to geometrical considerations.

The snapshots of the systems depict that the trans-layers are rather ordered and densely packed. We note here that the all-trans layer serves as the initial state of the simulations and experiments before any light stimulus is applied [11]. However, based on the isomerization kinetics revealed in experiments [11], the maximum possible population of the cis isomers under the conditions mentioned by the authors is slightly above 60%. This indicates that the isomerization reaction is quite effective for such densely packed layers, although it does not lead to a complete trans-to-cis conversion. Therefore, the modeled all-cis-layers are merely hypothetical systems. Nonetheless, this approximation does not diminish the value of theoretical investigation; it represents the first logical step towards studying mixed layers, such as 40% trans–60% cis, in future research. Moreover, one effect that may occur on larger scales remains unclear - the so-called cooperativity of isomerization, which could lead to cis-rich or trans-rich domains and their cooperative switching under light.

Figure 3 illustrates that the structure of the layers is preserved for all four modeled systems, thanks to the chemisorption of the molecules on gold. The trans-layers appear to be more ordered due to their stretched and planar geometry. Among the cis-layers, the BT-Azo one also exhibits a higher degree of structure due to the positioning of the kink in the molecule away from the anchoring surface. On the contrary, the cis-Azo-BT is a more disorganized system, as the cis-azo bond of each molecule is close to the gold surface, and therefore, no order is expected from this sample. The concentration profiles shown in Figure 4 are consistent with the ellipsometry results, which indicate a shrinkage of the layer height of BT-Azo by approximately 5±1 Å, in agreement with the DFT calculations [11,30]. Such a clear height alteration is explained by the intrinsic ordering of the molecules in the trans- and cis-states. In the literature, there are also examples showing that for relatively dense anchored systems, the all-cis-layers also demonstrate some organization [34].

On the other hand, the designed layers of Azo-BT do not exhibit noticeable switching in thickness, as deduced from the comparison of concentration profiles. Firstly, this is caused by the tilting of the molecules in the trans-layer (Figure 5a), which is relatively narrow and centered at an angle of 45${}^o$ to the surface. Secondly, the cis-Azo-BT layer is the least ordered system, with molecules having the potential to adopt almost any orientation relative to the surface. In fact, the range of tilting angles spans from 7 to 75 degrees.

The experimentally known molecule BT-Azo also demonstrates a greater extent of tilting to the surface in the trans-state compared to the cis-isomer. The most populated angle is 30${}^o$ for the trans-isomer and approximately 45${}^o$ for the cis-isomer. However, due to the differences in the length of the cis-isomer, the resulting layer height is smaller for the layer under UV stimulus.

3.2.2. Photo-programmable charge transfer

Table 3 collects the values of the transfer integrals, the defined com-com distances for dimers, the calculated charge transfer rates and charge carrier mobilities.

The transfer integral for the hopping of electrons or holes has similar values for all the studied systems, regardless of their isomerization state or the position of the azobenzene fragment in the backbone. The separations between the center of masses in the considered dimers are also very close to each other, differing by at most 1 angstrom. In this regard, the decisive factor affecting the charge transfer rate is the reorganization energy. Molecules with smaller magnitudes of reorganization energy will facilitate faster hopping of charge carriers. For example, electron transfer is favorable for the BT-Azo molecule, whereas the hopping of holes is minimal. Upon UV stimulus, the electron transport decreases by a factor of 4. Interestingly, electron hopping for Azo-BT decreases by 52 times under UV light and becomes almost negligible, whereas hole transfer is promoted by the trans-to-cis isomerization by a factor of 2. In absolute terms, the electron mobility in the layers of BT-Azo is rather high. In our opinion, this presents an interesting addition to the properties of the molecular switch BT-Azo, already known from the experiment, which is located between two electrodes.

4. Discussion and conclusions

In the published papers [10,11,23,45,46,47], the authors have suggested several intriguing avenues for future research, including purely theoretical approaches that may uncover phenomena and issues not yet explored in their previous works. For instance, exploring the electronic coupling between neighboring molecules and charge transfer within the molecular junction layer are intriguing tasks for computer modeling. In this paper, we addressed these questions pertinent to the rational design of azobenzene-bithiophene molecular switch.

Our primary idea was to investigate how changing the positions of Azo and BT blocks in the molecule impacts the properties of the isolated molecules, as well as the molecular layers attached to the gold surface. Systematically exploring different configurations, we aimed to gain insights into the structural and electronic characteristics of these molecular switches. Additionally, studying the influence of these changes on the charge transfer and solvation energies would allow us to assess the suitability of various configurations for molecular electronics and optoelectronic devices.

Regardless of the block positions, the studied molecules exhibit similar absorption spectra for the trans-isomers, while the cis-state optical spectra show slight shifts. Interestingly, the calculated hydration free energies provide an explanation for the observed dewetting/partial wetting transition in BT-Azo layers under UV stimulus, as observed in experiments. In the case of the designed Azo-BT layers, the $\Delta G$ values change sign, indicating more drastic changes in wettability, shifting from dewetting to wetting upon exposure to UV light. These findings suggest that the molecular design and arrangement play crucial roles in dictating the wetting behavior of the layers, making them potential candidates for future applications in surface engineering.

The electron coupling V within the layers is found to be insensitive to the isomerization state or the position of the azobenzene and bithiophene blocks. As a result, the transfer rates and charge carrier mobilities are primarily determined by the reorganization energies and, to a lesser extent, the hopping distance. In our approach, the hopping distance is approximated by the mean distance between the centers of masses in the interacting dimers. Our predictions indicate that the BT-Azo trans layer exhibits high electron mobility, which can be reduced by 4 times under UV illumination. On the other hand, if the layer consists of Azo-BT molecular switches, the electron mobility can be significantly reduced or even completely turned off. These results highlight the tunability and sensitivity of the electron mobility in these molecular layers, depending on their isomeric states and the specific arrangement of the azobenzene and bithiophene blocks. Such findings are of great interest for potential applications in molecular electronics and provide valuable insights into the design and optimization of molecular switches for various electronic devices.