Molecular Functionalization of 2H-Phase MoS2 Nanosheets via an Electrolytic Route for Enhanced Catalytic Performance

The molecular functionalization of two-dimensional MoS2 is of practical relevance with a view to, for example, facilitating its liquid-phase processing or enhancing its performance in target applications. While derivatization of metallic 1T-phase MoS2 nanosheets has been relatively well studied, progress involving their thermodynamically stable, 2H-phase counterpart has been more limited due to the lower chemical reactivity of the latter. Here, we report a simple electrolytic strategy to functionalize 2H-phase MoS2 nanosheets with molecular groups derived from organoiodides. Upon cathodic treatment of a pre-expanded MoS2 crystal in an electrolyte containing the organoiodide, water-dispersible nanosheets derivatized with acetic acid or aniline moieties (∼0.10 molecular groups inserted per surface sulfur atom) were obtained. Analysis of the functionalization process indicated it to be enabled by the external supply of electrons from the cathodic potential, although they could also be sourced from a proper reducing agent, as well as by the presence of intrinsic defects in the 2H-phase MoS2 lattice (e.g., sulfur vacancies), where the molecular groups can bind. The acetic acid-functionalized nanosheets were tested as a non-noble metal-based catalyst for nitroarene and organic dye reduction, which is of practical utility in environmental remediation and chemical synthesis, and exhibited a markedly enhanced activity, surpassing that of other (1T- or 2H-phase) MoS2 materials and most non-noble metal catalysts previously reported for this application. The reduction kinetics (reaction order) was seen to correlate with the net electric charge of the nitroarene/dye molecules, which was ascribed to the distinct abilities of the latter to diffuse to the catalyst surface. The functionalized MoS2 catalyst also worked efficiently at realistic (i.e., high) reactant concentrations, as well as with binary and ternary mixtures of the reactants, and could be immobilized on a polymeric scaffold to expedite its manipulation and reuse.


Post-processing of the functionalized products: nanosheet extraction and dispersion in water
To obtain individualized MoS2 nanosheets (NSs) from the electrochemically treated material, the latter was first cut into smaller pieces with the aid of a scalpel and then transferred to water at nominal concentration of 2 mg mL -1 , where it was bath-sonicated (J.P. Selecta Ultrasons system, 40 kHz) for 3 h. After standing undisturbed for 24 h to allow sedimentation of the poorly exfoliated fraction of the material, the upper ~75% of the supernatant volume was collected from the sonicated dispersion and kept for subsequent use.

Characterization techniques
The material was characterized by UV-vis absorption spectroscopy, zeta-potential measurements, atomic force microscopy (AFM), field-emission scanning electron microscopy (FE-SEM), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy, electron paramagnetic resonance (EPR) spectroscopy as well as by measurement of its water contact angle. UV-vis absorption (extinction) spectra were obtained for colloidal dispersions of the exfoliated MoS2 material in a double-beam Heios  spectrophotometer (Thermo Spectronic) using quartz cuvettes with an optical path length of 1 cm. The average lateral size and thickness of the dispersed NSs along with their concentration in the supernatant suspension were estimated by means of UV-vis absorption (extinction) spectroscopy using the metrics developed by Backes et al for this 2D material [1]. Zeta-potential measurements were carried out at a temperature of 25 ºC with a Malvern Zetasizer Nano ZSP apparatus from Malvern Instruments Ltd.
using ~1 mL of MoS2 dispersion at a concentration of 0.02 mg mL -1 . AFM measurements were carried out with a Nanoscope IIIa Multimode apparatus (Veeco Instruments) in the tapping mode of operation, using silicon cantilevers with nominal spring constant and resonance frequency of ~40 N m -1 and 250-300 kHz, respectively.
Samples for AFM were prepared by drop-casting a small volume (~10 L) of a lowconcentration MoS2 NS dispersion in isopropanol (~0.03 mg mL -1 ) onto highly oriented pyrolytic graphite and allowing it to dry under ambient conditions first and then, under vacuum at room temperature. FE-SEM images were recorded on a Quanta FEG S-4 apparatus (FEI Company) operated at 25 kV. Raman spectra were recorded on a Horiba Jobin-Yvon LabRam instrument at a laser excitation wavelength of 532 nm (green line).
A low incident laser power (~0.2 mW) was employed to avoid damage to the NSs. XPS was accomplished on a SPECS system, working at a pressure below 10 -7 Pa with a monochromatic Al K X-ray source (1486.7 eV) operated at 14.00 kV and 150 W. For Raman spectroscopy, XPS and contact angle measurement, specimens were prepared in the form of thin films by drop-casting MoS2 NS dispersions (in isopropanol, in the case of the nonfunctionalized NSs and in water, in the case of the functionalized ones) onto stainless steel discs, which were allowed to dry at room temperature. ATR-FTIR spectra were registered on a Nicolet 8700 spectrometer (Thermo Scientific) with a diamond ATR crystal. Specimens for ATR-FTIR spectroscopy were prepared in the form of freestanding, paper-like films by vacuum-filtering aqueous dispersions of the functionalized MoS2 material through silver membrane filters. EPR spectra were recorded with a X-Band (9.4 GHz) Bruker ELEXSYS E500 spectrometer using a magnetic field modulation amplitude of 1 G, a modulation frequency of 100 kHz and a microwave power of ~20 mW. Contact angles were measured by dropping 2 L of water on the films with a pipette and immediately taking images of the droplet with a standard digital camera with an attached macro lens. Analysis of the recorded images with ImageJ software was carried out to determine the value of the contact angle.

Catalytic tests
For the catalytic tests, aqueous aliquots (2.5 mL) containing MoS2 catalyst at concentration of 7 g mL -1 and the substrate molecule (a nitroarene or an organic dye at a concentration of 0.12 or 0.06 mM, respectively) as well as a large excess of NaBH4 (72-110 mM) were prepared in quartz cuvettes with an optical path length of 1 cm.
Immediately after preparation, the cuvettes were transferred to an UV-vis absorption spectrophotometer and the reaction progress was monitored by recording the corresponding kinetic profiles. The latter were obtained by measuring the temporal evolution of absorbance at the wavelength of a characteristic peak of the substrate  To increase the amount of loaded NSs, the soaking/drying process was carried out several times (typically three), after which the original white color of the polyurethane foam turned into an olive green tone. For the catalytic tests, the MoS2-loaded foam cylinders were immersed into the substrate/NaBH4 reaction mixture. After reaction completion, the cylinders were removed from the solution, rinsed with water and allowed to dry before being used in the next catalytic cycle.

Additional XPS spectra of MoS2 nanosheets functionalized with iodoacetic acid
The main bands are labeled for clarity. The fact that no signal was detected in the 635-615 eV binding energy range, which is characteristic of the main XPS core level of iodine (I 3d) confirmed that this element was absent from the material.
where θ is the angle of emission of the electrons from the surface normal and L is the attenuation length, which gathers both elastic and inelastic scattering effects. By considering electrons that emerge at θ=90 º to the sample surface, which was the configuration used for the XPS measurements reported here, the equation simplifies to: Incidentally, the approximate probe depth or information depth (ID) is defined as the sample thickness from which a specified percentage (e.g., 95%) of the detected signal originates. From eq. (2), some 95% of the signal will emanate from a depth lower than 3L and thus: In earlier approaches to estimate L, the inelastic mean free path () was calculated first and then, in a simplistic way, L was considered to be 10 per cent less than  [2]: More recently, independent universal curves for  [3] and L [4] have been derived, which display the lowest deviations from the experimental data reported to this day amongst the different available equations [including the well-known TPM-2M equation by Tanuma, Powell and Penn [5], which is offered as an option for the calculation of for inorganic compounds in the NIST electron inelastic mean free path database from predictive formulae [6]]. Specifically, the equation of the universal curve for L is: Apart from the energy of the photo-emitted electrons E, the only parameters involved in the calculation of L are the average atomic number Z and the average atomic size a for the corresponding material. In addition, for strongly bonded materials, such as oxides and alkali halides, a small extra term is included for the heat of formation H (where W= 0.06H). In the case of 2H-MoS2, H=276.14 KJ mol -1 .
The simplest approach to estimate Z for an inorganic compound with chemical formula GgHh is to calculate an average Z from the molecular formula as follows: In the case of MoS2, Z ~24.67 Da.
The atomic size a is deduced from the relation: where  is the density in g cm -3 , NA is Avogadro's number, and g and h are the stoichiometric coefficients in the molecular formula GgHh of a molecule with molecular mass M. In the case of 2H-MoS2, using M=160.07 Da and =5.03 g cm -3 yields a~0.26 nm.
The kinetic energy of the electrons photo-emitted from the S 2p core level using an Al K X-ray source is 1326.6 eV. Substituting E by the latter value and the aforementioned S-9 values of H, Z and a for 2H-MoS2 in the expression for L (eq. 5) yields a value of ~2. S-10 Figure S3. Background-subtracted, high resolution XPS C 1s core level spectrum of bulk MoS2 crystal functionalized through cathodic treatment with iodoacetic acid.

Presence of bismuth in some of the MoS2 materials
It has been recently reported that geological MoS2 crystals, such as the starting material in this work, contain bismuth impurities in concentrations high enough to be detected by XPS [8]. Specifically, the bismuth impurity has been found to appear mainly as the most thermodynamically stable form of bismuth oxide, Bi2O3. In the present work, the presence of micron-size clusters of bismuth in the MoS2 starting material was indeed confirmed by EDX. Correspondingly, bismuth was detected in some of the samples analyzed by XPS. Bi 3+ 4f5/2 spin-orbit component appears in a similar range to that where the S 2p component for S-C bond is expected to appear (~163-164 eV) [9,10].
Thus, special care was taken not to mistake Bi 3+ 4f5/2 for the S-C component in S 2p spectrum, by widening the acquired range for the S 2p spectrum downwards to include Bi 3+ 4f7/2 as well, thus confirming the presence of bismuth. eV, respectively, which corresponded to Bi(III).

Difference C 1s spectrum of functionalized and nonfunctionalized MoS2
S-13 Figure S7. Difference C 1s XPS spectrum of the functionalized and the nonfunctionalized MoS2 nanosheets (green and black traces in Figure 2g of the main text, respectively).

ATR-FTIR spectrum of functionalized MoS2
The ATR-FTIR spectrum of functionalized MoS2 exhibited some additional indication of functionalization. Specifically, a weak band located at ∼630 cm −1 can be ascribed to S−C stretching vibrations [9,11] arising from acetic acid groups grafted onto the NSs via sulfur atoms.
This band was not very intense even in the case of functionalized 1T-phase MoS2 monolayers, where the degree of derivatization with iodoacetic acid was ~20 percent [10]. Thus, it is expected to be weak for the current materials, where, as explained in the main text, the multi-layered nature of the MoS2 NSs implies that the percentage of (surface) sulfur atoms that are covalently bonded to carbon must be very small.
The sharp absorption peak at ~470 cm -1 is due to the A 1 2u phonon of 2H-MoS2, which involves asymmetric translation of both Mo and S atoms parallel to the c axis [12]. The bands associated to the acetic group were also observed (range not shown) but they S-14 would be compatible with just adsorption (noncovalent fuctionalization) of iodoacetic acid. Figure S9. Representative first-derivative X-band EPR spectrum of the functionalized MoS2 NSs. The unsaturated molybdenum atoms in edges and sulfur vacancies give rise to an EPR signal with g~2 [13,14]. (not shown) [15], which is followed by -HN-NH-bond dissociation to yield N,Ndimethyl-benzene-1,4-diamine and 4-amino-benzenesulfonate (both shown in violet trace) [16]; (g) MB (cyan trace), MB in the basic medium provided by NaBH4 (blue trace) and its reduction product, leucomethylene blue (purple trace) [17,18]. The wavelength chosen to follow the kinetics of the corresponding reactions, which was in all cases the location of an absorption maximum of the reactant, is indicated for clarity.     nanostructures, and non-noble metal-based catalysts reported in the literature.

Reaction mechanism
The mechanism of nitroarene and dye reduction catalyzed by the present functionalized MoS2 is expected to be the same that we previously proposed for vacancy-decorated 2H-MoS2 [19], i. e., a mechanism involving the catalytically active sites in the NSs, such as sulfur vacancies and edges, which possess local metallic character. The fact that the functionalization takes place near an active site of the MoS2 NSs does not promote the deactivation or blocking of the latter. As the acetic moiety is bonded to the NSs via a sulfur atom near a sulfur vacancy, the sulfur vacancy remains "vacant", and it is thus S-22 preserved and active. In contrast, a usual passivation method would involve filling of the vacancy with a sulfur atom provided by an external source, e.g., through chemical reaction with alkanethiols [75][76][77]. Indeed, the EPR signal of the functionalized MoS2 NSs (Fig. S9) is indicative of the presence of Mo-S dangling bonds. The presence of the negatively charged acetic moiety near the active site does affect the process in some way, as discussed below and in the main text, but the reaction mechanism remains basically the same.
When nanoparticles (NPs) of certain noble metals such as gold are used as catalysts for the catalytic reduction of nitroarenes with NaBH4, the BH4 − anion readily oxidizes, injecting electrons to the NPs, which thereby become a reservoir of excess electrons [78]. Upon adsorption of the substrate molecules onto the NPs, the excess electrons trigger their reduction. In such case, the catalytic reaction is insensitive to the pH of the aqueous medium [79]. In contrast, when metallic 1T-MoS2 NSs are used as catalyst for the same type of reaction, the reduction does not occur at pH values above ∼11 [80], which is the pH range where the spontaneous hydrolysis of BH4 − anion becomes arrested [81,82]. This implies that, in the latter case, some product of the hydrolysis of the BH4 − anion, rather than BH4 − anion itself, is involved in the reduction reaction. As the active sites in 2H-MoS2 (sulfur vacancies and edges) possess a locally metallic character, the same mechanism proposed for 1T-MoS2 applies for 2H-MoS2, as has been indeed reported before for vacancy-decorated 2H-MoS2 [19].
In basic solution below pH~11 the BH4 − anion hydrolyzes spontaneously to give molecular hydrogen and BH3OH − species as reaction products.
Although the resulting adsorbed H atoms could in principle react with a nitroarene/dye molecule to trigger its reduction, such a reduction reaction is inhibited by the fast recombination of the adsorbed H atoms into H2 molecules, which is a very favorable process at the catalytic sites of MoS2 (i.e., the Tafel step of hydrogen evolution reaction, HER) [85,86]. Indeed, previously reported control experiments in which a H2 gas flow several orders of magnitude larger than the hydrolysis rate of BH4was passed through a nitroarene solution in the presence of MoS2 NSs showed that no reduction of the substrate took place [20].

S-23
In addition to H2, the spontaneous hydrolysis of NaBH4 yields the BH3OH − anion as a rather stable intermediate. In the basic medium of the reaction, the BH3OH − species is oxidized at a sulfur vacancy site on the MoS2 surface with the assistance of hydroxide anions, to release the metaborate anion (BO2 − ), water, hydrogen and electrons [reaction (1) in Fig. S12 ] [87]: As previously reported control experiments where NaBH4 was substituted by NaBO2 in the reaction medium have shown, the generated BO2 − is inactive as a reductant and thus will not participate in the catalytic reduction [20]. Molecular hydrogen is activated by the released electrons to give a negatively charged hydride (H − ) [88], which is very unstable, and thus would not form on the surface of defect-free MoS2, but becomes stabilized when it adsorbs at sulfur vacancies by transferring excess electron charge to their neighboring unsaturated molybdenum atoms [89] [reaction (2)]. The vacancyanchored, negatively charged hydride should readily trigger the hydrogenation of the nitro group in the nitroarene molecules through the well-known three-step sequence [36,90]; i.e., the nitro group is first converted to the nitroso group [reaction (3)], which is then reduced to a hydroxylamine species [reaction (4)] and finally to the amino group [reaction (5)]. In the case of the dyes, the generation of highly active hydride species at the sulfur vacancy sites by way of reactions (1) and (2)

S-25
As aforementioned, according to the proposed mechanism, the catalytic activity of a MoS2 NS relies on the availability of stable adsorption sites for the H − species, that is, unsaturated molybdenum atoms in MoS2 edges and sulfur vacancies. Thus, the colloidal stability of the MoS2 catalyst will have a major influence on the catalytic activity, as the adsorption sites for the H − species in the MoS2 layers will only be available as long as their surface remains exposed to the aqueous medium where catalysis takes place. If the layers aggregate, catalytic activity will consequently decrease.
The fact that the current materials perform better as catalysts in aqueous medium than previously reported surfactant-stabilized MoS2 catalysts [19] could be partly related to the different way in which colloidal stabilization affects the catalytic sites in one and the other. In the present case, the covalently attached acetic group which imparts colloidal stability is grafted to a sulfur atom near the sulfur vacancy/edge, and thus the unsaturated molybdenum atoms remain available for adsorption of Hat any time.
However, in the surfactant-assisted stabilization, guanosine monophosphate (GMP) molecules adsorbed on the surface of MoS2 by acid-base interaction between their nucleobase and some of the unsaturated molybdenum sites [22], thus blocking them.
Hence, in the latter case, colloidal stability is attained at the expense of catalytic activity although not all catalytic centers will be blocked with GMP and, even those with GMP on, will be temporally available for catalysis as GMP is not irreversibly adsorbed. This was confirmed by the fact that the surfactant-stabilized MoS2 NSs aggregated if many washing cycles were done, as washing led to their progressive removal from the NS surface and thus to a loss of colloidal stability. Thus, GMP will be desorbing and adsorbing from the surface of surfactant-stabilized MoS2 NSs and unsaturated molybdenum sites will be available for catalysis when GMP eventually desorbs. We note that, in the case of the current functionalized MoS2 NSs, the presence of the acetic group near the sulfur vacancy could introduce some steric impediment for the approaching substrate molecules. However, this effect will be minor in the case of diatomic vacancies, which are easily formed and stabilized [91].