Functionalization of MoS2 with 1,2-dithiolanes: toward donor-acceptor nanohybrids for energy conversion

The covalent functionalization of exfoliated semiconducting MoS2 by 1,2-dithiolanes bearing an ethylene glycol alkyl chain terminated to a butoxycarbonyl-protected amine and a photoactive pyrene moiety is accomplished. The MoS2-based nanohybrids were fully characterized by complementary spectroscopic, thermal, and microscopy techniques. Markedly, density functional theoretical studies combined with X-ray photoelectron spectroscopy analysis demonstrate preferential edge functionalization, primarily via sulfur addition along partially sulfur saturated zig-zag MoS2 molybdenum-edges, preserving intact the 2D basal structure of functionalized MoS2-based nanohybrids as confirmed by high-resolution transmission electron microscopy and electron energy loss spectroscopy. Furthermore, in the MoS2-pyrene hybrid, appreciable electronic interactions at the excited state between the photoactive pyrene and the semiconducting MoS2 were revealed as inferred by steady-state and time-resolved photoluminescence spectroscopy, implying its high potentiality to function in energy conversion schemes. MoS2 can be chemically functionalized with organic functional units, to develop innovative hybrid materials suitable for energy conversion. Collaborative efforts between experimental and modeling teams at the Theoretical and Physical Chemistry Institute—National Hellenic Research Foundation, Greece, Institut des Materiaux Jean Rouxel—CNRS Nantes, France, Instituto de Nanociencia, Universidad de Zaragoza, Spain, and Chimie des Interactions Plasma-Surface, University of Mons, Belgium, reveal that chemical functionalization occurs preferentially on MoS2 edge sites. When photoactive components are attached to MoS2 edges, visible light illumination promotes efficient charge transfer within the hybrid material. Research along those lines may result in the construction of innovative solar and photoelectrochemical cells with high efficiencies.


INTRODUCTION
Transition metal dichalcogenides (TMDs) are layered nanomaterials consisting of an atomic plane of a transition metal sandwiched between two atomic planes of chalcogens. Bulk TMDs are semiconductors with an intrinsic band-gap and exhibit noteworthy optical and electronic properties desirable for energy conversion and storage, 1-7 sensing, [8][9][10] and photodynamic/ photothermal therapy. [11][12][13] Analogously to graphite, numerous van der Waals forces keep together the layered sheets of TMDs forming stacks. However, despite their close similarity to graphene, TMDs offer significant advantages related to their band-gap dependent electronic properties, 14,15 which can be tuned by altering the transition metal and/or the chalcogen composition. In addition, although functionalization of graphene is greatly developed, allowing the preparation of a plethora of modified graphene-based hybrid materials, the corresponding field for TMDs is still at its infancy. Particularly, a central aspect of layered TMDs chemistry, which has yet to be fully exploited, concerns their functionality in energy conversion applications. In this context, it is imperative the functionalization of TMDs, permitting their better manipulation and processing in wet media by improving solubility in common organic solvents. At the same time, of particular importance is the incorporation of organic electron donor units by robust and stable chemical bonds with the network of TMDs in order to alter their properties and entirely harness their capabilities.
Diverse methodologies for exfoliating TMDs from the bulk exist, giving access predominantly to the metallic polytype lattice structure with octahedral coordination geometry. [16][17][18][19][20][21][22] On the other hand, we recently reported a facile approach for the exfoliation of MoS 2 and WS 2 polytypes with trigonal prismatic geometry and semiconducting properties with a direct band-gap, by treating the corresponding bulk materials with chlorosulfonic acid. 23 The TMD layers are protonated by the superacid and kept apart due to developed electrostatic repulsive interactions, however, without being oxidized. Nevertheless, despite the progress achieved so far targeting the exfoliation of MoS 2 , its covalent functionalization is still hampered. In fact, the limited functionalization procedures reported to date mainly utilize exfoliated MoS 2 by n-butyl lithium (n-BuLi), thus of the metallic polytype 1T-MoS 2 in which the negative charges on MoS 2 due to charge-transfer from n-BuLi, are quenched by alkyl halides, 24 and diazonium salts, 22 while thiols are conjugated at sulfur vacancies 21,[25][26][27][28] and metal carboxylate salts coordinate to surface sulfur atoms of 2H-MoS 2 . 29 Therefore, it is absolutely desirable and timely to functionalize semiconducting MoS 2 species with organic electron donor units. 30 Additionally, it is necessary to understand where and how functionalization occurs, as to date the characterization performed is rather vague.
Herein, we report for the first time on the covalent functionalization of exfoliated semiconducting MoS 2 by employing 1,2dithiolane units as parts of organic derivatives bearing either an ethylene glycol alkyl chain terminated to a butoxycarbonyl (BOC)protected amine 1a or a pyrene 1b (Fig. 1) as an electron donor. Moreover, we disclose the location ("where") and the manner ("how") this functionalization occurs based on density functional theoretical studies supported by X-ray photoelectron spectroscopy (XPS) and confirmed by high-resolution transmission electron microscopy (HRTEM) imaging and electron energy loss spectroscopy (EELS). Notably, the 1,2-dithiolane component possesses high binding affinity for Mo atoms, particularly those located at the edges of the corresponding TMDs, where sulfur vacancy sites were naturally introduced after chemical exfoliation from the bulk. 31 Such a facile approach for the chemical functionalization of MoS 2 offers the advantage of keeping intact the basal plane of MoS 2 and preserving to great extent their novel surface properties, as well as their semiconducting character, while also allows the incorporation of a plethora of suitably modified organic derivatives targeting diverse applications. Importantly, the presence of defects at the edges of MoS 2 , as generated by either chemical exfoliation 21,[25][26][27][28] or ion irradiation, 32 guarantees the universal utility and versatility of the functionalization route. The newly prepared MoS 2 -based nanohybrids were fully characterized, while examination of the optical properties revealed the occurrence of significant electronic interactions in the MoS 2 -pyrene 2b hybrid.

RESULTS
The exfoliated semiconducting MoS 2 was prepared from the bulk upon chlorosulfonic acid treatment by following our recently published procedure. 23 The general scheme for the functionalization process is illustrated in Fig. 1 (structural characterization data for 1,2-dithiolanes 1a and 1b are presented at the Supplementary Information, Figs. S1-S4). The newly functionalized MoS 2 -based nanohybrids 2 found to be soluble in N,N-dimethylformamide (DMF), ca. 1.2 mg/mL, while their solubility in dichloromethane, acetone, and THF was limited. The so-derived MoS 2 -based nanohybrids can be stored in solution for months, without observing precipitation, hence justifying their stability in wet media. Treatment of 2a with gaseous HCl cleaves the BOCprotecting group, yielding the cationic ammonium derivatized MoS 2 -based material 2c, soluble in protic solvents and aqueous media. The amino-loading of 2c was calculated by performing the Kaiser test and found to be 103 μmol g −1 . The latter result allows manipulating and processing such modified MoS 2 -based materials in biotechnological applications, where handling in physiological media is a prerequisite. The MoS 2 -based nanohybrids 2a-c were fully characterized by complementary spectroscopic and thermal techniques. Markedly, the functionalization appears to primarily occur via sulfur addition along partially sulfur saturated zig-zag MoS 2 molybdenum-edges. Density functional studies combined with XPS analysis demonstrated the preferential edge functionalization, preserving the 2D basal structure of the functionalized MoS 2 -based nanohybrids as confirmed by HRTEM imaging and EELS. This is an extremely important finding, shedding ample light on the location and the manner the functionalization of exfoliated MoS 2 with 1,2-dithiolanes proceeds, hence allowing the design and development of a plethora of interesting hybrid materials for diverse applications. A natural consequence of this finding is that chemical functionalization behavior of exfoliated MoS 2 should occur similarly to that for bulk material, where edge functionalization is likely to be predominant. Thus, we would expect the possibility of direct transfer of functionalization strategies from bulk to exfoliated MoS 2 , for example transition metal doping of edge sites to improve catalytic behavior. 33 Particularly in the MoS 2 -pyrene 2b hybrid material, appreciable electronic interactions at the excited state between the photoactive pyrene and the semiconducting MoS 2 nanosheets were revealed as inferred by steady-state and time-resolved photoluminescence spectroscopy studies. An electron donor-acceptor combination, such as the one found in 2b hybrid material, mimics the simple photosynthetic system and imitates charge-transfer processes that ensue in nature. Moreover, such MoS 2 -based hybrid materials, incorporating photo-active and/or electro-active species, feature novel light harvesting, charge-separation and/or charge-transport properties, hence considered ideal for photovoltaics and optoelectronic devices. Thereby, assays on artificial photosynthetic models capable of undergoing charge-separation phenomena are of primary importance in constructing economic and light-toelectricity converting devices. Currently, we are extending the applicability of the current functionalization methodology to other layered transition metal dichalogenides incorporating diverse photoactive organic electron donors en route the preparation of advanced hybrid materials suitable for energy conversion applications. For example, solar cells convert the energy captured from sunlight into electricity by exploiting the generation of electrons and holes, namely the generation of charge-separation, that move to the corresponding electrodes of the photovoltaic cell. Needless to mention that the development of a long-lived charge-separated state is desirable for achieving high energy conversion efficiency, since it can effectively stream electrons and  holes to the respective electrodes. Efforts along those lines are underway in our laboratories.

DISCUSSION
Our functionalization largely deviates from recent studies based on thiol reactions, which were shown to oxidize to the corresponding disulfides which eventually only adsorb onto MoS 2 and are not chemically anchored. 34 In stark contrast, in our experimental protocol a mixture of exfoliated MoS 2 and 1,2dithiolane derivatives 1a or 1b in DMF was stirred at 70°C for 36 h. Then, the reaction mixture was filtered over a polytetrafluoroethylene (PTFE) membrane (0.1 μm pore size) and the solid residue collected onto the filter was extensively washed with dichloromethane to remove organic impurities, yielding MoS 2based nanohybrids 2. The strong binding of the organic addends on the MoS 2 was proved by examining their electronic absorption profile after extensive washing cycles to remove any loosely bound (physisorbed) organic compound. Evidently, after four consecutive washing cycles, the UV-Vis spectrum of functionalized MoS 2 -based materials 2 remain unaffected, namely the characteristic absorptions due to pyrene, as well as due to the MoS 2 retained their intensity, while also the filtrate fail to show characteristic absorptions owed to the pyrene species. Attenuated-total-reflectance IR assays gave spectroscopic proof for the success of the functionalization. Briefly, strong C-H stretching and bending modes are discernible in the 2840-2970 cm −1 range, while the ethylene glycol unit gives rise to vibrational modes in the region 1130-1190 cm −1 for 2. Particularly for 2a, two carbonyl moieties are identified at 1650 and 1710 cm −1 owed to the amide and BOC protecting units, respectively-the latter is absent in the IR spectrum of 2c, in which free amine functionalities are present-while for 2b, only the ester carbonyl is identified at 1730 cm −1 (Fig. 2a, b).
Next, Raman spectroscopy was employed as a valuable tool to obtain meaningful insights on the functionalization of MoS 2 . Upon in-resonance excitation at 633 nm, more rich spectra as compared with measurements conducted under off-resonance conditions at 514 nm were acquired. Upon comparison of the Raman spectra for functionalized MoS 2 with those owed to exfoliated MoS 2 flakes the following points are deduced: (a) the intensity of the 2LA(M) mode at 447 cm −1 is decreased for the functionalized MoS 2 nanohybrids, for spectra normalized at the A 1g mode located at 404 cm −1 , in accordance with Raman studies performed on differently functionalized MoS 2 species, 22 and (b) phonon modes directly related to the metallic polytype of MoS 2 at 150, 225, and 325 cm −1 , 35, 36 are totally absent, hence proving the semiconducting behavior of MoS 2 in the functionalized nanohybrids 2. A representative Raman spectrum for 2b is shown in Fig. 2c (for 2a, see Supplementary Information, Figure S5). Additional proof for the success of the covalent functionalization of MoS 2 with 1,2-dithiolane derivatives 1a and 1b was delivered by thermogravimetric analysis (TGA). Exfoliated MoS 2 is thermally stable up to 250°C under nitrogen atmosphere. However, in MoS 2based nanohybrid 2b 8.1% weight loss in the temperature range 250-550°C is observed (Fig. 2d). The latter is attributed to the thermal decomposition of the organic functionalities covalently bonded to Mo at the edges of MoS 2 , i.e., at sulfur vacancies. Actually, the loading of pyrene onto MoS 2 is calculated to be one per every   Information, Figure S6), the maximum loading of one organic addend is calculated to be per every 24 MoS 2 units. Markedly, the relatively low mass loss observed in the TGA performed for nanohybrids 2 clearly proves that the organic functionalities were covalently anchored onto MoS 2 , contrasting the case where physisorption occurs, in which the coverage of MoS 2 is very high reaching a loading of one organic unit per every two Mo atoms. 34 HRTEM images show that the flakes owed to functionalized MoS 2 -pyrene 2b vary from single-layer to oligo-layer (the exfoliated MoS 2 was composed of 6 layers on average according to calculations performed on their extinction spectra), 37 with largely damage-free basal plane (Fig. 3a, b). Although the edge structure is difficult to be determined conclusively, there is evidently strong preference for zig-zag edge orientation, consistent with theoretical predictions 38,39 and experimentally observed structures, [39][40][41][42][43] suggesting that 50% S and 100% S saturated zigzag molybdenum-edges should be thermodynamically preferred. [43][44][45] The HRTEM images also show the presence of material at the edges, which may correspond to carbonaceous species. In order to investigate this point, we carried out spatially-resolved EELS studies, as the most appropriate technique for such kind of analyses. 46,47 In more detail, Fig. 3c displays the EEL spectrum recorded in a region close to the edge of a flake belonging to functionalized MoS 2 -pyrene 2b nanohybrid-see also highangular annular dark field, scanning TEM image displayed as inset of Fig. 3c. Apart from the expected Mo and S contributions, the C-K edge is visible. The fine structure of this C-K signal (shown in Fig. 3d) includes a significant π* peak and other contributions at 288 eV and in the σ* region (290-310 eV), consistent with attribution to the organic addend incorporated in 2b hybrid material and different to the one of amorphous carbon, further proving the success of functionalization of semiconducting MoS 2 with pyrene species. [46][47][48][49] We next turn to density functional theory (DFT) modeling to better understand the functionalization mechanism. In order to gain insight into the covalent binding behavior, we first model the interaction of 3-methyl-1,2-dithiolane with the MoS 2 edge and basal plane. The former molecule represents the covalent linking segment of 1a-c with the carbon chain attached to neighboring S atoms of MoS 2 either in the basal plane or at zig-zag 100% S molybdenumedge (Fig. 4a-d and Supplementary Information, Figure S7). Preferential binding for the edge compared to the basal plane is found (2.54 eV per molecule for the most stable configuration).
At the edge, 3-methyl-1,2,-dithiolane is most stable with its carbon chain approximately perpendicular to the edge (Fig. 4b) breaking apart a single S 2 pair. The other less stable edge configurations result in two broken S 2 pairs, which then reconstruct horizontally (Fig. 4c, d). Neighboring bonds in MoS 2 around the covalent anchoring site are also distorted, notably in Fig. 4c and d, the distance between Mo atoms connected to the S atoms bound to the carbon chain of 3-methyl-1,2,-dithiolane is reduced from 3.13 Å for the bulk to around 2.96 Å.
We next perform calculations on covalently attaching the full 1,2-dithiolane-based derivative 1b to both basal plane and edge sites of MoS 2 (Fig. 4e and Supplementary Information Figure S8). The edge functionalization configuration is constructed based on the 3-methyl-1,2,-dithiolane configuration in Fig. 4c, since this maintains the same conformation of the aliphatic chain and pyrene moiety as in the basal case. For MoS 2 basal plane functionalization we assume direct substitution of S into neighboring S vacancies on the basal plane, resulting in neighboring carbon atoms bound directly to S atoms in the basal plane. One of these carbon atoms sits directly above the lattice S, while the second is displaced, pulling its S neighbor 0.32 Å out of the plane and distorting neighboring Mo-S bonds.
In all cases, the pyrene and aliphatic tail preferentially stack parallel to the MoS 2 basal plane. The pyrene moiety and chain stack at approximately their van der Waals radii distance above the MoS 2 surface (3.16 Å above the S plane, 4.73 Å above the Mo plane; Fig. 4e and Supplementary Information Figure S8). In the basal plane case this adds 2.0 eV binding enthalpy per molecule compared to alternative non-interacting configurations, of which 1.1 eV comes from the pyrene moiety (Supplementary Information, Figure S9).
Once again edge binding is preferred, with the 1b molecule 1.96 eV more stable at the edge than on the basal plane. This value is very close to that for 3-methyl-1,2-dithiolane in the equivalent binding configuration (2.03 eV). In addition, the distances between pyrene and the S atoms from the top plane of MoS 2 is the same as for the basal functionalization model (3.16 Å). This suggests that the preference for the edge in this model is mainly due to the changes produced by the covalent linkage between the molecule and MoS 2 , while the non-covalent interaction with the basal plane, due to pyrene and the aliphatic chain, is similar in both basal and edge functionalization models. Moreover, structures 1a, 1b, and 1c all connect to MoS 2 through the same functional group, and the 3-methyl-1,2-dithiolane molecular fragment is a valid model for all three. While the noncovalent binding of 1a and 1c may not behave similarly to that of 1b, even in the unlikely case that the non-covalent interactions show a preference for basal plane binding, such a difference would have to be larger than 2.54 eV to give net favorable basal plane binding. Thus, the calculations suggest the edge binding preference of 1b will be shared by 1a and 1c.
Complete energy cycles are given in Supplementary Information, including either vacancy formation followed by molecular insertion into the vacancy sites, or S 2 removal from the molecule followed by direct addition to either the basal plane (which is endothermic in this case) or the edge. The relative energy between basal and edge functionalization is independent of the energy cycle chosen, i.e., while the sulfur reference state energy will depend on the sulfur chemical potential, this is a 'constant' in both cases. Actual energy cycles could involve other processes such as creation of monovacancies and vacancy migration, and in practice the experimental cycle followed is likely to be considerably more complex (e.g., dealing with oxygen impurities). We note also that even if the basal plane is successfully functionalized, there is then a 2.54 eV driving force for the functional group to migrate to the sheet edge.
Hence, we suggest that edge binding is strong energetically favored, and functionalization will likely progress via physisorption to the basal plane followed by migration and chemisorption to the edge. This leads us to propose a new model for covalent functionalization of semiconducting MoS 2 with the 1,2-dithiolanes. There is evidence that edges play an important role in chemical processes like catalysis, in contrast to the less reactive basal plane. 50,51 Literature studies of edge configurations give the 50% S and 100% S saturated zig-zag molybdenum-edges as the most thermodynamically favored (Fig. 4f), with the 50% S edge more stable over a wide range of S chemical potential. [40][41][42][43][44] These edges have been both observed experimentally and it has been shown that the edge configuration depends on the synthesis conditions. [40][41][42][43] Although further studies need to be done in order to fully unveil the processes occurring at these edges, some theoretical studies have already shown the possibility of their modification through adsorption and desorption processes. [52][53][54] Starting with the most stable 50% S zig-zag edges, 1,2dithiolanes can directly bind to the exposed Mo at this edge, saturating and resulting in 100% S coverage with covalently attached species, preferentially in a binding configuration perpendicular to the edge (cf. Fig. 4b). This converts the sulfur bonding at the extensive edge sites to a configuration much closer to the bulk, and can also be viewed as filling 'vacant sulfur' edge sites. The model is supported by XPS studies of the semiconducting MoS 2 samples before and after functionalization with 1b (i.e., MoS 2 -pyrene 2b hybrid material) as depicted in Fig. 5. Before functionalization, the S 2p signal shows a superposition of the conventional semiconducting MoS 2 signal, the split peak pair at 163.4 and 162.2 eV, with a second pair of peaks at 163.1 and 161.9 eV, typically associated with "vacancy" or damaged material, 55 which we associate here with the unsaturated sulfur edges. After functionalization, these additional peaks are removed as the edges all become fully sulfur saturated. At the same time, there is very little change in the Mo 3d signal, the slight peak shifts consistent with small charge transfer from π-stacked pyrene found in our calculations. We note that the XPS signal shows no secondary peaks corresponding to oxidized S and Mo proving the high purity of functionalized MoS 2 -pyrene 2b hybrid material. The same was also confirmed by the EELS measurements (cf. Fig. 3c). The UV-Vis spectrum for 2a in DMF exhibits the characteristic features due to the semiconducting polytype of MoS 2 .
Discernible are the excitonic transitions at 400, 500, 630, and 690 nm ( Supplementary Information, Figure S10), justifying the semiconducting nature of the functionalized MoS 2 -based species. 56 For the pyrene modified MoS 2 -based nanohybrid 2b, additional bands directly derived from the optical absorption of the pyrene are identified at 315, 330, and 345 nm (Fig. 6a), hence, directly implying the success of the functionalization reaction. Overall, the UV-Vis spectrum of MoS 2 -pyrene nanohybrid 2b is a simple superimposition of the spectra owed to the semiconducting polytype of MoS 2 and to the pyrene derivative 1b, implying the absence of appreciable electronic interactions between the two species in the ground state.
Next, focusing at the excited states, strong electronic communication between MoS 2 and pyrene within 2b is identified. More precisely, based on steady-state photoluminescence assays, the characteristic emission bands of pyrene in 1b, located at 376, 396, and 418 nm upon photoexcitation at 340 nm, were quenched in 2b (Fig. 6b), for samples possessing equal absorptions at the excitation wavelength. Notably, a blank sample prepared by mixing exfoliated MoS 2 with pyrene derivative 1b failed to show photoluminescence quenching of the pyrene emission (Supplementary Information, Figure S11). Hence, intrahybrid transduction of electron and/or energy from the pyrene singlet excited state to MoS 2 for 2b occurs, implying that MoS 2 acts as electron/energy acceptor. Moving further, based on the time-correlated-singlephoton-counting method, the fluorescence lifetime profiles for 1b were obtained and analyzed. The evaluation of the time profile of the fluorescence decay at 396 nm for the singlet-excited state of free pyrene in 1b was monoexponentially fitted with a lifetime of 3.79 ns. Conversely, two decay components in MoS 2 -based nanohybrid 2b were identified. Biexponential fitting for the fluorescence decay gives rise to a slower component with 5.01 ns, attributed to non-interacting pyrene and a faster one with 280 ps, corresponding to the fluorescence quenching of the emission intensity of the singlet excited state of pyrene in MoS 2 -pyrene 2b. Then, based on Eq. (1) the quenching rate constant k S q for the singlet excited state of pyrene in MoS 2 -pyrene 2b was evaluated as 3.3 × 10 9 s −1 . Furthermore, by employing the Eq. (2), the  quantum yield Φ S q for 2b was determined to be 0.85.
where T f refers to the lifetime of the fast-decaying component in 2b, and T o refers to the lifetime of free pyrene in 1b.
In conclusion, we demonstrate here that covalent chemical functionalization of exfoliated MoS 2 with 1,2-dithiolane derivatives occurs preferentially at the sheet edges, with functionalization likely progressing via physisorption to the basal plane followed by migration and chemisorption to the edge. The presence of the 1,2dithiolane end-group in our attach-ing species facilitates incorporation at the available edge sites. This is mediated through easy variation in sulfur saturation (50-100%) along the edge sites. Preferential edge reactivity is highly reminiscent of the chemistry of non-exfoliated MoS 2 , known for its strong catalytic capability, and suggests that other bulk MoS 2 functionalization and treatment strategies may be transferable to the 2D system. Moreover, covalent incorporation of pyrene modified 1,2-dithiolane 1b on exfoliated MoS 2 furnished donor-acceptor 2b hybrid material featuring charge-transfer interactions potentially suitable for managing energy conversion schemes. Similar edgefunctionalization behavior is likely to also hold true for other 2D TMDs such as WS 2 , while we are currently extending our investigations with the incorporation of diverse photoactive molecules to yield novel functional materials.

METHODS
General procedure for the synthesis of 1,2-dithiolanes 1a and 1b In a round bottom flask, αlipoic acid (1.6 mmol), tert-butyl (2-(2-(2aminoethoxy)ethoxy)ethyl)carbamate (2.00 mmol, 1.25 equiv.) for 1a or 1pyrenebutanol (1.6 mmol, 1 equiv.) for 1b, EDCI (4.8 mmol, 3 equiv.), and DMAP (4.8 mmol, 3 equiv.) were added in dry dicloromethane (100 mL). The reaction mixture was stirred under nitrogen at room temperature for 18 h. Then, the organic phase was extracted with H 2 O (5 × 100 mL), dried over MgSO 4 and purified by column chromatography (petroleum ether/ ethyl acetate 50%). The experimental procedure was replicated five times in our laboratories. Spectroscopic data for 1a: 1  General procedure for the preparation of functionalized MoS 2based nanohybrids In a round bottom flask, exfoliated MoS 2 (20 mg) and 1,2-dithiolane derivative (1a or 1b; 10 mg) in DMF (10 mL) were stirred at 70°C for 36 h. After that period, the reaction mixture was filtered through a PTFE membrane (0.1 mm pore size), the solid residue was extensively washed with DMF and dichloromethane and then collected as a dispersion in dichloromethane. The experimental procedure was replicated ten times in our laboratories.
Computational methodology DFT calculations were performed using the local density approximation (LDA) as implemented in the density functional code AIMPRO. [57][58][59] The spin-averaged charge density is fitted to plane waves with an energy cut- for C, 40 for O (l ≤ 2), and 12 for H (l = 0), giving a highly versatile and transferable basis description. A finite electron temperature of 0.04 eV was used to aid self-consistent convergence. Absolute energies were converged in the self-consistency cycle to better than 10 −5 Ha. Atomic positions and lattice parameters were geometrically optimized until the maximum atomic position change in a given iteration dropped below 10 −4 a 0 (a 0 : Bohr radius) and/or energy change less than 10 −5 Ha.
While the LDA does not include explicit van der Waals interaction terms, its known over-binding 'mimics' these interactions. Additionally the focus of these calculations is on the relative covalent bonding behavior of the species, which is largely independent of the physisorbed pyrene moiety where van der Waals effects are most important. This can be seen in the similar relative energies between the basal and edge sites for the complete 1b molecule (1.96 eV) and the truncated molecule (3-methyl-1,2-dithiolane) at 2.03 eV, the same species as 1b without the pyrene moiety present.
The continuous monolayer MoS 2 models use a hexagonal Mo 64 S 128 repeating unit cell ( Large cells were used to avoid self-interaction, with a fixed interlayer distance of 31.1 Å (giving a vacuum spacing between MoS 2 monolayers of 27.9 Å), and an in-plane vacuum spacing between edges of 34.1 Å. This ensures that the stable functionalized structures are all more than 23 Å from atoms in neighboring slabs.

Code availability
Further details of the code AIMPRO are available from the website, http:// aimpro.ncl.ac.uk/. Calculations presented here used version 3.21. The code can be accessed by arrangement with the code authors Patrick Briddon and Mark Rayson at Newcastle University, UK, who should be contacted directly (contact details available through the website).