Design, synthesis and characterization of the sulfoxide precursor
The photodeoxygenation of dibenzothiophene S-oxides, leading to the formal reduction of sulfur(0) to sulfur(-II), has been reported in solution on diversely substituted DBTO derivatives.15–18 In these systems, the dibenzothiophene core exhibits a fully planar geometry, but the pyramidal character of the sulfoxide moiety induces an out-of-plane location of the oxygen atom, thus leading to an overall bent geometry of the molecule. In anticipation of in-depth structural characterization and photoreactions on surface, the design of the DBTO precursor was adjusted to favor its face-on adsorption on weakly interacting insulating substrates.
Compared to metal surfaces on which the adsorption energy of planarly adsorbed aromatic molecules is relatively high due to the interaction between the molecule’s π-electrons and the electrons of the metal substrate, on insulating substrates, the fabrication of molecular layers in a planar adsorption geometry is rather challenging. Prior research on insulating surfaces, both theoretical and experimental, has emphasized the role of polar anchoring groups in the development of such highly ordered self-assemblies.19–22 Although DBTO molecules natively display a S = O sulfinyl group with significant polarity, it was envisioned to insert two additional electron-withdrawing bromine atoms on symmetrical ortho positions (with respect to the bisarylsulfoxide moiety), with the aim to optimize the adsorption energy on alkali halide substrates. 4,6-Dibromodibenzo[b,d]thiophene 5-oxide ((o-Br)2-DBTO, Fig. 1 left) was thus synthesized and fully characterized (see Methods section and Supplementray Information file). Its structure was unambiguously confirmed by single-crystal X-ray diffraction, revealing an angle of 117° between the plane of the polycyclic scaffold and the S = O bond (Fig. 1 and S18). In addition, thermogravimetric analysis (TGA) was carried out in anticipation of on-surface physical vapor deposition of this model compound, which highlighted its thermal stability up to ca. 200°C (Fig. S7).
In-solution photoreactivity of (o-Br)-DBTO
The absorption spectrum of (o-Br)2-DBTO in dichloromethane (DCM) displays a wide band peaking at λmax = 280 nm (Fig. S8a) and photoreactivity studies were thus performed under UVA-irradiation (280–365 nm). Noteworthy, wavelengths shorter than 280 nm were excluded as this leads to the cleavage of the bromine anchoring groups as a side reaction. On the first hand, the photoreactivity of (o-Br)2-DBTO was investigated in a dichloromethane solution at room temperature (RT). After 48h of UVA-irradiation, the expected deoxygenated and fully planar compound (o-Br)2-DBT was obtained as the major product in 53% isolated yield (Fig. 1). Besides, two byproducts incorporating one additional oxygen atom were isolated: the thiophene S,S-dioxide derivative (o-Br)2-DBTO2 (8%) and the oxathiine S-oxide derivative Br2-OTO (21%). The structure of the latter was unambiguously assigned thanks to an X-ray crystal structure, revealing the presence of the oxygen atom embedded in the central six-membered ring (Fig. S19).
From a mechanistic point of view, upon UVA-irradiation, the sulfur(0) precursor (o-Br)2-DBTO undergoes photodeoxygenation to give rise to the reduced sulfur(-II) product (o-Br)2-DBT and releases atomic triplet oxygen O(3P) in the reaction medium.15 The latter is then trapped by remaining (o-Br)2-DBTO molecules, leading to the formation of both oxidized sulfur(+ II) byproducts (o-Br)2-DBTO2 and Br2-OTO. Consequently, in solution, the efficiency of the photodeoxygenation reaction is hampered by intermolecular oxygen transfers, which inherently limit the selectivity of this process and the control of the resulting sulfur oxidation state.
Once the model compound designed, synthesized and its in-solution photoreactivity assessed, the next step was the investigation of (o-Br)2-DBTO sulfoxides on surface. Surprisingly, SPM studies involving sulfinyl derivatives on surface are scarce, mainly involving dimethylsulfoxide (DMSO) and tetramethylenesulfoxide on metallic substrates.24,25 A recent investigation also examined the impact of the molecular configurations of sulfonyl chloride compounds on desulfonylation reactions occurring on Au(111) and Ag(111) surfaces.26 To the best of our knowledge, sulfoxides have never been investigated by SPM on insulating substrates.
NaCl(100) thin films/ Au(111) substrate
Beyond the molecule's structure and the presence of polar anchoring groups, the adsorption of precursors on alkali halide substrates is critically influenced by the crystal's ionic structure.20,22,27–29 Indeed, matching the molecular network, i.e. the position of its electronegative anchoring groups, with the cationic lattice is crucial to control the self-assembly on ionic substrates. Therefore, the alkali halide surface NaCl(100) was chosen, since it features a cation distance (nearest neighbor: dNa−Na = 0.4 nm) matching quite well the Br − O distance (dBr−O = 0.36 nm) and, to a less extent, the Br − Br distance (dBr−Br = 0.66 nm) of the functional groups in the (o-Br)2-DBTO molecular tecton. Satisfyingly, upon deposition on bulk NaCl(100) at room temperature, (o-Br)2-DBTO molecules adsorbed in a planar geometry and formed stable monolayers (ML), with each polar functional group potentially located on a cationic site (Fig. 3c).
In this work, the main challenge lies in establishing the proof of a photocontrolled modulation of the oxidation state of sulfur-based molecules by high-resolution Scanning Probe Microscopy, using functionalized CO-tips to image the planarly adsorbed molecules.30 Since our aim is to characterize both structural (oxygen loss) and electronic modifications (change in sulfur oxidation state) upon irradiation, ultrathin insulating layers were deposited on a metal substrate to allow us to conduct STM measurements in combination with non-contact-Atomic Force Microscopy (nc-AFM) measurements (hereafter referred to as “dynamic STM”, see Methods section). Indeed, whereas STM cannot be performed on bulk insulating crystals, electron tunneling is transparent through ultrathin layers of wide-bandgap insulators (one to three monolayers), such as oxides or alkali halides, deposited on metal surfaces.31,32 Additionally, it has been demonstrated that in the case of ultrathin NaCl films, one double layer already exhibits the distinctive electronic structure of its bulk form,33,34 meaning that the NaCl layers on metals are thin enough to allow electrons to tunnel while preserving the optical properties of the adsorbed molecules.
NaCl islands have been successfully grown on a wide range of crystalline metal surfaces, including Ge(100),31 Al(111),32,35 Al(100),32,35 Cu(111),34 Au(111),36,37 Ag(100),38 Ag(001)39 and Ag(110).40 According to these experimental investigations and numerous theoretical studies, NaCl forms (100) oriented layers on these substrates. In the frame of this study, NaCl thin layers were deposited on gold surface Au(111) (see Methods section). As expected, STM images indicate that NaCl grows as (100) layers on the reconstructed (111) gold terraces. A high-resolution STM image of the NaCl unit cell (Fig. S2c) shows that its lattice constant of 5.65 Å is consistent with that of the NaCl bulk crystal. Larger-scale STM images (Fig. S2a) reveal a carpetlike growth of the initial NaCl(100) layer, with the formation of the second and third NaCl layers on the first one before the latter entirely covers the Au surface.
The NaCl layers on Au(111) form an ideal substrate for the present study, since they display an atomically clean surface on which i) the adsorption of the (o-Br)2-DBTO precursor molecules is expected to be in a planar geometry, ii) such molecules will be electronically decoupled from the metal substrate, thus preserving their intrinsic photophysical properties and allowing UV-induced deoxygenation, and iii) high-resolution SPM imaging with functionalized CO-tips is possible to observe the expected chemical modification of the molecules and characterize the change in the sulfur atom oxidation state.
Description of the supramolecular phase
A typical STM image following the deposition of 0.1 ML of (o-Br)2-DBTO onto a thin layer of NaCl(100) is displayed in Fig. 2a. It appears that the molecular islands, denoted by the white arrows, are substantially longer than they are wide, suggesting that in-line interactions are stronger than lateral ones. Additionally, the image shows that the molecules preferentially nucleate in the NaCl's polar directions [110] and [\(\:\stackrel{-}{1}\)10]. A close inspection on a zoomed STM image (Fig. 2b) revealed that the molecules self-assemble into parallel stripes, consisting of oppositely oriented molecular rows. The inter-rows distance is ∼12 Å, whereas the distance between two consecutive molecules within the same row is ∼8 Å. These values are coherent with a (6a × 2a) rectangular unit cell consisting of 2 molecules, where parameter a = 4 Å represents the distance between the nearest cations in a NaCl crystal. Figure 3b presents a structural model for this self-assembly, which illustrates a point-on-point adsorption, highlighting the match between dNa−Na and dBr−Br, as anticipated. Within a molecular row, (o-Br)2-DBTO molecules adsorb every second line of cations, while the electronegative atoms (both bromine atoms and the oxygen of the sulfinyl group, see below) match the cationic sites. We may therefore conclude that the growth of these lines is driven by the adsorption of the anchoring groups matching the periodicity of the substrate and stabilized by van der Waals (vdW) interactions. Residual electrostatic interactions are optimized by alternating the molecular orientation in neighboring lines while maintaining a rectangular unit cell. As evidenced by the work of Neff et al. for one dimensional lines of vertically adsorbed molecules,41 the parallel aligned dipole moments are sufficiently large to produce a remarkable long-range repulsion between molecular lines on an insulating substrate. In the present case, the molecules are planarly adsorbed with their dipole moment essentially aligned parallel to the surface (Fig. S5). We assume that the oppositely oriented lines of molecules are a result of maximized (attractive) electrostatic interactions between the intrinsic dipoles of the molecules.
Finally, the adsorption configuration of each single molecule within the supramolecular assembly was investigated, as the pyramidal shape of the sulfinyl group (Fig. 1, bottom left) leads to two scenarios for (o-Br)2-DBTO on the surface. Indeed, the molecule may adsorb with its polycyclic scaffold parallel to the surface and the out-of-plane oxygen atom pointing upwards, or towards the surface. To unambiguously determine the configuration of (o-Br)2-DBTO on NaCl thin layers, the Probe Particle Model (PPM), a numerical AFM model that simulates high-resolution non-contact AFM images with CO-functionalized tips,42,43 was used. Simple probe particle approaches combined with experimental measurements including dibenzothiophene have already been shown to be effective in providing rapid structural assignment.44
High-resolution nc-AFM images of (o-Br)2-DBTO issued from experimental observations with a CO-functionalized tip (Fig. S3b) were compared with those obtained from PPM simulations (Fig. S1b,c). The calculated images show that an oxygen-up configuration (Fig. S1c) would result in a very distinct contrast compared to a molecule with its oxygen facing the surface (Fig. S1b). Based on experimental images (Fig. S3b), which display a weak contrast in the central part of the molecules, it was thus inferred that (o-Br)2-DBTO is adsorbed with the oxygen pointing towards the substrate. This can occur through the favorable interaction between the oxygen and the cations of the NaCl thin layer (Fig. 3c).11 Importantly, in anticipation of photoreactivity studies, PPM-simulated images show very limited differences in the imaging contrast between a (o-Br)2-DBTO molecule with the oxygen pointing down (Fig. S1b) and a deoxygenated one, i.e. (o-Br)2-DBT (Fig. S1a). It is also worth noting that in these simulations, the carbon scaffold of all molecules was kept parallel to the substrate surface. In experimental conditions, this is expected for the planar C2v symmetric (o-Br)2-DBT molecule, whereas the adsorption of the bent (o-Br)2-DBTO sulfoxide derivative with the oxygen facing the surface is likely to be slightly tilted with respect to the surface. This induces a modification of the molecular contrast, allowing a clear distinction between the pristine (o-Br)2-DBTO and the deoxygenated (o-Br)2-DBT molecules (see below).
On-surface photoinduced deoxygenation
Once the self-assembly of (o-Br)2-DBTO molecules thoroughly characterized, their on-surface photoreactivity was explored by both, STM imaging and Local Contact Potential Difference (LCPD) measurements, as deduced from Bias spectroscopy (see below). The surface was thus exposed to a 280 nm wavelength irradiation, in line with in-solution photochemistry experiments. This was performed by using a light emitting diode (LED) and irradiating the sample inside the SPM head (see Methods section).
Following illumination, STM imaging in gentle conditions indicate that the molecules now exhibit two distinct contrasts, which is indicative of on-surface deoxygenation. Figure 3a displays a molecular island after 190 minutes of light exposure. The sulfur atoms of (o-Br)2-DBTO molecules are spot on the bright circle-shaped protrusions. These were less visible in Fig. 2b owing to smoother tip imaging conditions. Here, the molecules exhibit a “croissant-like” shape and a “butterfly-like” shape (see corresponding grey forms in the Fig. 3a.). To assign a particular contrast to the pristine sulfoxide, or to the deoxygenated compound, we performed high-resolution dynamic STM imaging of the molecular assembly using a CO-functionalized tip (Fig. S3a). In that figure, the deoxygenated (o-Br)2-DBT molecules are identified with white frames. The associated frequency shift map (Fig. S3b) shows only a minor change between (o-Br)2-DBTO and (o-Br)2-DBT molecules, as already expected from the series of PPM simulations (Fig. S1a,b). A slight modification of the contrast is noticeable: the tip-molecule interaction is no longer as repulsive above the sulfur atom. As explained earlier, this is rather an effect of a modified adsorption geometry of the now planar, deoxygenated (o-Br)2-DBT molecule compared to its tilted precursor, than a direct visualization of the missing oxygen atom. Therefore, these observations allow us to unambiguously assign in Fig. 3a the “croissant-like” shaped molecules to the pristine (o-Br)2-DBTO sulfoxides and the “butterfly-like” shaped ones to the (o-Br)2-DBT deoxygenated sulfoxides.
A statistical analysis of that sample presented in Fig. S4, involves 260 molecules and reveals that approximately 36% of them underwent the deoxygenation.
To complete this structural characterization and analyze the charge state of sulfur atoms before and after deoxygenation, individual molecules were selected among a statistical set of pristine/deoxygenated species. A series of Δf vs. bias voltage curves (∆f(V), so-called Bias spectroscopy curves) were performed on top of the sulfur atom of each of these molecules. Bias spectroscopy curves are known to exhibit a typical parabolic shape, whose maximum (Δf*, V*) is a measurement of the Local Contact Potential Difference (LCPD), a quantity that can be connected to the local work function difference between the tip and the surface. When the bias voltage matches V*, the local electrostatic forces occurring between the tip and the surface are canceled out. Any charge state manipulation of an atom on the surface (tip manipulation, bias pulses, light…) induces a change of its LCPD, which can therefore be tracked by bias spectroscopy. This was demonstrated in 2009 by L. Gross et al.45 using a qPlus operated at low temperature, which is like our experimental conditions. They have shown that a negatively charged single Au atom adsorbed on an ultrathin NaCl film on Cu(111) featured a larger LCPD than its neutral counterpart.
Before drawing any conclusions from our bias spectroscopy study, the charge distributions of the pristine and deoxygenated molecules must be compared. To this end, DFT calculations with Natural Bond Orbital (NBO) analysis were undertaken (see Methods section and Supplementary Information file), showing that the sulfur atom in the deoxygenated (o-Br)2-DBT molecule displays a higher density of negative charge than its pristine counterpart (Fig. S5). Therefore, as stated by Gross et al., we anticipate that, upon further measurement, the experimental spectra corresponding to the deoxygenated molecules will shift to more positive values when compared to those of the pristine (o-Br)2-DBTO molecules.
Figure 4a shows one of the molecular matrices on which the bias spectroscopy was performed. Three horizontal rows, each containing four molecules, are visible. Two bias spectroscopy curves were conducted on each molecule, with a total of eight measurements per line. The spectroscopy curves corresponding to the four molecules in the white frame are presented in Fig. 4b (red • and blue x correspond to the croissant-like pristine (o-Br)2-DBTO and butterfly-like deoxygenated (o-Br)2-DBT molecules, respectively). Experimentally, the tip was placed above a sulfur atom and a ∆f(V) curve was measured for each molecule. STM images were taken before and after each series of four curves to confirm that the molecules did not change position. Next, a parabolic fit was performed on the gathered data. In light of this, a global examination reveals that the bias spectra of the deoxygenated (o-Br)2-DBT molecules (blue x) are consistently shifted to the right with respect to those of the pristine (o-Br)2-DBTO molecules (red •). The average LCPD shift of S with respect to S = O is found to be (0.119 ± 0.045) V (see Fig. S6 for detailed explanations and a full bias spectroscopy study of 24 molecules). Bias spectroscopy experiments then confirm that the charge state of the sulfur atom is directly impacted by its chemical environment, and more particularly by the presence of the electronegative oxygen atom, with the sulfur displaying a higher density of negative charge in the deoxygenated (o-Br)2-DBT molecules. Most importantly, owing to the decoupling character of the NaCl thin layer, such difference in the charge state of sulfur atoms in parent dibenzothiophene derivatives such as (o-Br)2-DBT and (o-Br)2-DBTO can be directly probed on surface, which opens the way to in-situ charge state manipulation in purely organic compounds. In the present case, the photoinduced modulation of the charge state of a sulfur atom was successfully achieved on surface.
Finally, it is important to note that the photoreactivity of (o-Br)2-DBTO sulfoxide on surface features excellent chemoselectivity, since oxidized byproducts (o-Br)2-DBTO2 and Br2-OTO, resulting from intermolecular oxygen transfers, were never observed. This is a striking difference with the in-solution reaction, which furnishes large amounts of byproducts, and encourages the transposition of further photochemical reactions from solutions to surfaces.