Carbon nitride used as a reactive template to prepare mesoporous molybdenum sulfide and nitride

Carbon nitride C3N4 has been used as a sacrificial template to prepare inorganic materials with hierarchical pore structure. C3N4 impregnated with ammonium heptamolybdate was treated in reactive gas mixtures (H2S/H2 or NH3/H2). This approach allowed mesoporous molybdenum sulfide and molybdenum nitride materials to be obtained that replicate the morphology of the C3N4 template. Advantageous catalytic properties have been demonstrated in the thiophene hydrodesulfurization (HDS) and electrochemical hydrogen evolution reaction (HER). The highest rates in both reactions were observed for partially sulfidized Mo2N solid.


Introduction
Molybdenum is widely used in industrial catalytic processes such as hydrorening, 1-3 and currently studied as a promising alternative for highly expensive platinum group metals in processes for producing sustainable energy and reducing environmental pollution. [4][5][6] Many recent works describe application of molybdenum sulphides or selenides, nitrides, carbides and phosphides (both supported and unsupported) in the reactions of hydrogen evolution, 7-9 oxygen reduction, 10,11 or carbon dioxide reduction. 12 For both electrochemical and catalytic reactions porous materials with hierarchical systems of interconnected macropores and mesopores are preferable because such structure facilitates diffusion of the reactants. In order to synthesize mesoporous Mo suldes and nitrides versatile techniques have been proposed such as topotactic transformations of oxides 13,14 or of hybrid materials. 15 One of the most popular techniques to control the textural properties is template-based synthesis strategy. To prepare molybdenum sulde and nitride, templating with silica, 16,17 polymers and MOFs 18,19 or biotemplating 20,21 have been applied. In the majority of templating methods the removal of template must be carried out at the nal step, usually by etching. To avoid etching step the use of sacricial templates is preferable. Carbon nitride has been rst used as a sacricial and reactive template by Antonietti and coll. 22,23 Upon heating of mesoporous C 3 N 4 impregnated with metal oxide species chemical reaction occurs that converts oxides to the corresponding nitrides (Al-Ga-N) and (Ti-V-N). Carbon nitride was also applied as a template to obtain N-doped carbon layers with increased porosity by means of thermal treatment at 900 C. 24,25 When a process using C 3 N 4 as a template is carried out in the inert atmosphere, the temperatures as high as 800-900 C are required to fully transform metal oxide species to nitrides. While suitable to prepare nanoparticles of refractory nitrides, such temperatures are prohibitive for obtaining highly divided materials, because of advanced sintering.
In this work we demonstrate that in the reductive atmosphere carbon nitride can play a role of sacricial reactive template at much lower temperatures. Moreover, the presence of a transition metal could further decrease the decomposition temperature of C 3 N 4 . This allows obtaining highly divided mesoporous materials, as demonstrated for the case study of MoS 2 and Mo 2 N materials with hierarchical porosity and advantageous catalytic properties in HDS and HER reactions.

Materials preparation
To prepare carbon nitride (C 3 N 4 ), typically 30 g of urea was placed in a Pyrex tube, covered with a Pyrex cap and heated in static air for 2 h at 550 C, with heating rate 10 C min À1 . Ammonium heptamolybdate (AHM) was supported onto carbon nitride by incipient wetness impregnation from aqueous solution (10% wt. AHM/C 3 N 4 ). To carry out the treatments in reductive atmosphere, weighted amount of 10% wt. AHM/C 3 N 4 (ca. 0.8 g) was placed in a Pyrex reactor and treated in a reactive gas mixture (NH 3 /H 2 or H 2 S/H 2 ) for 2 h at 550 C with heating rate 5 C min À1 . The obtained samples are designated as Mo 2 N-CN and MoS 2 -CN, respectively. The solid obtained aer heating of AHM/C 3 N 4 in pure nitrogen ow was also characterized as a reference.

Characterizations
Textural properties of the C 3 N 4 template and Mo-containing samples were studied by N 2 adsorption-desorption volumetry at À196 C on a Micrometrics ASAP 2010 device. Pore distributions have been calculated using BJH equation. The samples were outgassed before the measurements at 400 C for 2 h. Phase composition was studied by X-ray diffraction (XRD) on a Bruker D8 Advance A25 diffractometer with CuKa emission. The phases were identied by comparison with JCPDS standards database. Phase composition was quantied using Rietveld renement as implemented in the Philips XPert soware. CHONS analysis was performed on a Thermo Fisher Flash 2000 device. Transmission Electron Microscopy (TEM) images were obtained on a JEOL 2010 instrument at 200 KV. TEM images were analyzed using Digital Micrograph Gatan program package. Temperature-programmed reduction (TPR) was carried out in a quartz reactor. The samples (ca. 0.01 g) were linearly heated under a hydrogen ow (50 ml min À1 ) from room temperature to 1050 C (heating rate 5 C min À1 ). The gases evolved upon reduction were detected by means of Thermo Fischer quadrupole mass-spectrometer. Thermogravimetric analysis (TGA) was carried out on a SETARAM device. A weighted amount of sample powder (5-10 mg) was placed in an alumina crucible and heated in nitrogen ow from room temperature to 800 C at a 10 C min À1 rate (NB: cyan and HCN released upon TPR and TGA experiments are toxic and should be neutralized at the reactor outlet).

Catalytic tests
The catalysts were tested in thiophene HDS right aer their preparation: 40 mg of the sample was placed in a quartz reactor in continuous ow of H 2 (50 ml min À1 ) passed through a bubbler with thiophene. HDS was studied at 320, 330, 340 C, respectively. The products were analyzed by gas chromatography on an Agilent 7820A device.
Electrochemical hydrogen evolution reaction (HER) was performed in a three electrode electrochemical cell in Arsaturated 0.5 M H 2 SO 4 electrolyte at room temperature. Glassy carbon rotating electrode was used as a working electrode, saturated calomel electrode as a reference electrode and graphite rod as a counter electrode. 10 mg of catalyst was suspended in 400 ml of Naon (0.5%) and 800 ml of EtOH and treated by ultrasound. 10 ml of the obtained catalytic ink was spread on the working electrode and dried. Activity of the catalysts in HER was measured in the potential range from À800 mV to 100 mV. LSV curves were obtained at 5 mW s À1 rate. Tafel slopes were calculated via Tafel equation. Electrochemically active surface area (ECSA) was calculated from the nonfaradaic parts of CV curves at sweep rates from 50 to 300 mV s À1 .

Results and discussion
Reactivity of C 3 N 4 vs. the nature of gas atmosphere and the presence of molybdenum If heated under nitrogen ow, bare carbon nitride starts to decompose approximately at 600 C and the decomposition completes at 780 C (Fig. S1 †). The mass loss is nearly 100% (the solid completely disappears from the crucible). These results are in good agreement with previous reports of the TGA of C 3 N 4 carried out in nitrogen. 26 As shown by mass spectrometry, 27 gaseous CN and C 2 N 2 species and N 2 are formed during the decomposition of C 3 N 4 . Addition of ammonium heptamolybdate (AHM) by impregnation results in the decrease of C 3 N 4 decomposition temperature by nearly 100 C (Fig. S1 †). This suggests a chemical reaction between AHM and C 3 N 4 and probably a catalytic effect of formed Mo species on the decomposition. The mass loss is 92%; a solid residual was observed in the crucibles. Changing the gas ow from N 2 to H 2 leads to a further decrease of the onset temperature of decomposition (Fig. 1). For bare C 3 N 4 it shis to 500 C and for the AHM/C 3 N 4 sample it becomes approximately 450 C. At the same time the nature of the released gases is modied: formation of NH 3 and HCN occurs in hydrogen ow instead of cyan release previously observed in the inert atmosphere.
Therefore, both addition of Mo species and applying hydrogen ow lead to a decrease of the C 3 N 4 elimination temperature. Moreover, simultaneous application of H 2 and addition of AHM provides a synergistic effect. Similar reaction onset temperature of 450 C is observed for AHM/C 3 N 4 sample in pure H 2 (Fig. 1) and in the NH 3 /H 2 mixture (Fig. S2 †).

Properties of the solid products
We further collected the solid products of AHM/C 3 N 4 reactions in different atmospheres (N 2 , NH 3 /H 2 and H 2 S/H 2 ) and analyzed their properties. The XRD patterns are shown in Fig. 2 and S3. † In the N 2 atmosphere highly divided g-Mo 2 N is formed at 650 C ( Fig. S3 †), but it is polluted with MoO 2 oxide (13% vol. according to Rietveld analysis). However, observing Mo 2 N as a major phase proves that C 3 N 4 plays the role of nitrogen source and therefore acts as a reactive template. Using NH 3 /H 2 and H 2 S/H 2 ows allows obtaining at 550 C almost pure phases of cubic g-Mo 2 N and hexagonal 2H-MoS 2 (Fig. 2), respectively (and small impurity of MoO 2 , 2-4% vol).
In agreement with XRD, chemical analysis shows the S and N content values close to the theory values for the Mo sulde and nitride, respectively (Table 1). Carbon is almost completely removed from the MoS 2 -CN sample, but is still present in the Mo 2 N-CN (probably as amorphous matter not detected by XRD). Small amount of oxygen in both samples is probably due to partial surface oxidation and due to a contribution from minor MoO 2 impurity. Therefore, the use of the reactive gas mixtures (NH 3 /H 2 or H 2 S/H 2 ) allowed us to obtain at 550 C the materials containing molybdenum nitride or sulde as major components. We further studied the morphology and the catalytic properties of these materials.
The C 3 N 4 template has the specic surface area (S BET ) 157 m 2 g À1 and pore volume 0.87 cm 3 g À1 ( Table 2). The isotherm shape corresponds to type II (IUPAC classication) which is characteristic of the macroporous materials (Fig. 3a). Pore size distribution is broad and has a polymodal shape (Fig. 3b), the pores with diameters between 30 and 50 nm being the most abundant ones. The amount of micropores is rather low; the narrow peak at pore radius ca. 2 nm is an artifact caused by cavitationinduced evaporation (Fig. 3b).
The isotherms of Mo 2 N-CN and MoS 2 -CN have similar shape to that of the parent C 3 N 4 material, but the pore volumes and the specic surface areas are smaller, obviously because of higher density of the corresponding molybdenum compounds ( Table 2). Comparison of isotherms and BJH pore size distributions suggests that Mo 2 N-CN and MoS 2 -CN replicate the main features of the pore structure of C 3 N 4 template and possess hierarchical mesoporosity. This is an important nding since mesoporous materials of this type are highly demanded and not readily available. Indeed, both Mo 2 N and MoS 2 obtained by conventional solid-gas reaction techniques are microporous. Bulk Mo 2 N prepared via widely used Volpe and    Boudart TPR method possesses high specic surface area (around 200 m 2 g À1 ), but pore size smaller than 30Å, fully lying in the microporosity range. 28,29 Similarly, MoS 2 synthesized by conventional techniques such as ammonium thiomolybdate (ATM) decomposition possesses low pore volume and considerable microporosity. 30 Transmission electron microscopy (TEM) study corroborates the results of other characterizations and provides additional insights into the morphology of the solids. Low-magnication TEM reveals lamellar morphology of the C 3 N 4 template ( Fig. 4a and S5 †). The layers of carbon nitride are randomly bent, having a rag-like aspect. The morphology of the reaction products Mo 2 N-CN and MoS 2 -CN bears a signicant similarity to the parent template. Lamellar morphology and open porosity with convex macropores were observed (Fig. 4b and c; see also video in the ESI †).
At higher magnications stacked layers of MoS 2 become visible in the MoS 2 -CN sample (Fig. 5b) and several nm-size Mo 2 N particles were observed in the Mo 2 N-CN solid (Fig. 5a).
Interplanar distances 0.205 nm and 0.242 nm correspond respectively to the (2 0 0) and (1 1 1) planes of g-Mo 2 N. In the MoS 2 -CN sample the measured interplanar distance is 0.67 nm, corresponding to the theory value for (0 0 2) plane of MoS 2 (0.62 nm) and to the measured XRD peak position (0.65 nm). Note that because of the slabs bending and stacking defects, the measured interplane distance in the nanoscopic MoS 2 samples is oen greater than in the bulk molybdenite. 31 Overall, the obtained mesoporous Mo 2 N and MoS 2 materials replicate the hierarchical porous structure and lamellar morphology of the template. Mo 2 N-CN and MoS 2 -CN were further tested in the model reactions of gas-phase thiophene HDS and liquid-phase HER.

Catalytic properties
Thiophene HDS rates are shown in Fig. 6 in comparison with the benchmark ATM-MoS 2 reference. The Mo 2 N-CN sample demonstrates one of the highest HDS activities reported for the non-promoted Mo catalysts, being at least ve times more active than the ATM-MoS 2 bulk reference and also much more active than bulk MoS 2 solids from our previous works that were tested in the HDS reaction under the same conditions. 32 (Fig. 6a).
The evolution of HDS activity versus time was signicantly different for sulde and nitride samples. At 320 C the nitride catalyst loses 50% of its activity during the rst hour on-stream, whereas MoS 2 -CN sulde was deactivated much less (Fig. 6b). The Mo 2 N-CN solid aer the HDS test preserved opened porous morphology (Fig. 6c). However, at high resolution short (1-3 nm) MoS 2 single slabs were observed. Therefore, partial suldation of molybdenum nitride occurred at the surface of pores. Steady state HDS activity can be attributed to these slabs, very short due to connement in the pores of Mo 2 N matrix. In agreement with our ndings, it was reported earlier that the initial thiophene HDS activity of g-Mo 2 N/Al 2 O 3 is high, but during the reaction rapid suldation of the surface occurs and the steady state activity is dened by the MoS 2 slabs formed on the surface. 33 Formation of MoS 2 occurs via suldation of the passivating surface oxide layer rather than a direct nitride-to-sulde transformation. 34 Exceptionally high apparent HDS rate during the initial period might be due to high intrinsic activity of Mo 2 N, but might be also related to the sulfur uptake by nitride.
The MoS 2 -CN sample demonstrates lower HDS activity than the Mo 2 N-CN one, but still higher than the ATM-MoS 2 reference (Fig. 6a). Indeed, on the TEM images of MoS 2 -CN, extended (>5 nm) and stacked layers of MoS 2 are mostly present (Fig. 5b). Lower dispersion of MoS 2 slabs and therefore lower amount of the active edge structures explains its lesser activity in comparison with Mo 2 N-CN. The selectivity distributions of HDS products are similar to those observed earlier for the nonpromoted Mo sulde catalysts (Fig. S6 †). 30,32 Beside C 4 HDS products, small amounts of methane were detected. Production of methane might be due to partial surface oxidation of Mo 2 N and MoS 2 with formation of Mo oxide species that possess acidic sites, able to perform cracking.
To access the potential of electrochemical applications, Mo 2 N-CN (fresh and aer HDS test) and MoS 2 -CN were tested in the electrochemical HER in 0.5 M H 2 SO 4 . The MoS 2 -CN material demonstrates poor HER activity, in accordance with its moderate HDS catalytic performance (Fig. 7a). Low activity of bulk 2H-MoS 2 was previously reported in the literature. 35 The Mo 2 N based samples demonstrate good activity, the sample taken aer HDS being more active than the initial nitride. The overpotential for Mo 2 N-CN aer HDS at 10 mA cm À2 is much lower (239 mV) than that of the initial nitride sample (390 mV) (Fig. 7a). The calculated Tafel slopes for three tested samples  are 86 mV dec À1 for Mo 2 N-CN aer HDS, 137 mV dec À1 for Mo 2 N-CN and only 284 mV dec À1 for MoS 2 -CN (Fig. 7b). The obtained values of Tafel slopes for Mo 2 N based catalysts indicate that HER reaction follows Volmer-Heyrovsky mechanism. For the sample Mo 2 N-CN collected aer HDS test, that shows the highest HER performance, the electrochemical surface area (ECSA) calculated from the non-faradaic CV at different scan rates is 136 m 2 g À1 (Fig. 7c and d). This value is close to the BET surface area and attests good availability of the surface for the electrochemical reaction.
The superior HER catalytic performance of the Mo 2 N-CN sample aer HDS text might be associated with high activity of short MoS 2 slabs on the surface of Mo 2 N, formed during the HDS reaction due to suldation of the catalyst surface by the released H 2 S. Oppositely to the metallic Mo 2 N, 2H-MoS 2 is a semiconductor, so it possesses relatively low electric conductivity 36 and, therefore, lower activity in HER, whereas the active sites for these two materials are similar (MoS 2 slabs edges). It appears that mesoporous Mo 2 N decorated with MoS 2 fringes provides optimal structure for an efficient HER catalyst. Thus in ref. 37 it was suggested that if MoS 2 edges are in a tight contact with the surface of Mo 2 N, then the electronic structure of Mo sites located at the interface could be tuned, boosting HER catalytic performance. Recently, in the same lines, for chemically similar W sulde it was shown that metallic W particles surrounded with WS 2 show high HER activity. 38 The interface between MoS 2 n-type semiconductor, and Mo 2 N would create ohmic or Schottky contact, depending on the difference between the work functions of two materials. For the similar Mo-MoS 2 contact, low Schottky barrier was observed that potentially provides easy charge injection from current collector to the catalyst sites. 39 For the Mo 2 N-MoS 2 interface, both experiment and DFT analysis show that an electric eld is created at the interface between Mo 2 N and MoS 2 that facilitates charge transfer. 40

Conclusions
Mesoporous MoS 2 and Mo 2 N materials replicating the hierarchical mesoporous morphology of C 3 N 4 were prepared using a simple topotactic solid-gas reaction. Due to the advantageous morphology these solids are well adopted for the applications in the heterogeneous catalysis and in the electrocatalysis. The most active catalyst in both HDS and HER reactions is mesoporous molybdenum nitride decorated at the surface by small MoS 2 fringes. The HDS performance of this sample is one of the highest ever reported for the non-promoted Mo systems.
Due to application of the reactive gases, our synthetic approach extends the potential use of C 3 N 4 as a template towards considerably lower temperatures and to the materials other than nitrides. The advantage of this method is its utter simplicity. Indeed, carbon nitride support is impregnated with ammonium heptamolybdate according to the standard procedure, commonly applied for the preparation of heterogeneous catalysts. Further activation with H 2 S/H 2 or NH 3 /H 2 is also carried out by following the standard procedures that are used to prepare sulde and nitride catalysts. Unlike in many other template-based techniques there is no template leaching steps. Obviously, mesoporous suldes and nitrides of other metals (such as V, W, Cr) and their combinations could be prepared using this approach under relatively so conditions. By the same token, preparation of mesoporous phosphides or borides could be considered, by means of the reductive treatment of phosphates or borates supported on C 3 N 4 . Beside catalysis, other applications could be considered in the elds where materials with open porosity are demanded, such as pseudo capacitors, electrodes or sensors.

Conflicts of interest
There are no conicts to declare.