Partial oxidation of methane to methanol on boron nitride at near critical acetonitrile

Direct catalytic conversion of methane to methanol with O2 has been a fundamental challenge in unlocking abundant natural gas supplies. Metal-free methane conversion with 17% methanol yield based on the limiting reagent O2 at 275 °C was achieved with near supercritical acetonitrile in the presence of boron nitride. Reaction temperature, catalyst loading, dwell time, methane–oxygen molar ratio, and solvent-oxygen molar ratios were identified as critical factors controlling methane activation and the methanol yield. Extension of the study to ethane (C2) showed moderate yields of methanol (3.6%) and ethanol (4.5%).

Indirect, high temperature (600-1100 °C) steam reforming coupled with high pressure (400-800 psi) syngas conversion using Cu/ZnO/Al 2 O 3 catalysts afford methanol 1 . Direct partial oxidation at mild temperatures (< 450 °C) are afforded as two main strategies used in conversion of methane to methanol [2][3][4][5] . Oxidation of methane to methanol using O 2 and H 2 is known to produce methanol in water [6][7][8][9][10] . Even though the direct partial oxidation of methane is thermodynamically feasible, the overoxidation of subsequent products such as methanol, formaldehyde and formic acid to CO 2 have less activation barrier than activation of methane 11-13 . In the direct route, a trade-off between methane bond activation (E a 175 kJ/mol on Cu(111)) and product (methanol and other oxygenates) protection (methanol has ~ 50 kJ/mol lower bond C-H dissociation energy than methane) against overoxidation govern the overall yield of methanol 3,11,14 . Based on the above concept, Nørskov et al. recently established a mathematical model 11 to explain the reason for low methanol yields (< 1%) despite years of research. This model recognizes solvation free energy (SFE) modification of methanol as one approach to improve product yield by decreasing the activation free energy difference between methane and methanol (product protection). Similarly, minimization of mass transfer limitations (MTL) in conventional homogenous catalytic systems which arise due to limited solubility of oxygen and methane, may lead to improved methanol yields (methane activation). These low solubility and mass transfer limitations can be avoided by going into a supercritical or near critical phase. Unusual selectivities were previously observed by Debendetti et al. 15 for toluene disproportionation over ZSM-5 that was ascribed to near critical clustering. The authors hypothesized the near critical clustering of toluene resulted in more surface reactions as the diffusion inside the zeolite pores was reduced. For two phase reaction systems, the near critical clustering phenomenon is not hitherto explored. We have studied a number of two phase systems comprising of solvents such as CO 2 16 and acetonitrile. We observed A3240HC6EB (Parr Instruments, Moline, IL) utilized with reactor controller 4848. The reactor controller was operated in the PID controlling mode and the rotor was at the highest rotation speed of 60 rpm. The system pressure and temperature were continuously digitally monitored with the associated software. All reactant gases including ultra-purity nitrogen, helium, oxygen, methane, ethane, carbon dioxide, and propane were purchased from Airgas, Inc, North Franklin, CT. Graphite, h-BN, acetonitrile, fluoroacetonitrile, trichloroacetonitrile, deuterated acetonitrile, and copper perchlorate hexahydrate were purchased from Sigma Aldrich. Liquid reactants were introduced with a 1000 μL micropipette (one-time use tips) and sampling was carried out with the help of single-use sterilized syringes and PTFE microfilters.
Method. In a typical reaction, solvent and catalyst were loaded to the reactor. The reactor was cooled down to − 30 °C with liquid nitrogen and pressurized with the calculated amounts of O 2 , CH 4 , and inert gas (N 2 or He) while the temperature was stable at − 30 °C. Then the reactor was heated to the precalculated (to avoid explosion range) temperature with a ramp rate of 2.5 °C/min (PID) and maintained there for the desired dwelling time.
The reactor was then cooled down to ambient temperature by natural convection. Products were analyzed by GC-MS and NMR to determine the methanol concentration. Product mixtures were extracted with a single-use sterilized syringe and filtered with 0.22 μm PTFE microfilters. The complete reactor flow diagram can be found in Fig. 1.  www.nature.com/scientificreports/ Methanol and CO 2 yield, and selectivity were calculated using following formulas (assuming CO 2 is the only by product formed),

Results
Optimization of the reaction parameters. Optimization of methane:oxygen molar ratio. The methane to oxygen molar ratio was systematically increased by increasing methane and decreasing oxygen amounts to validate the product overoxidation hypothesis as shown in Fig. 2. An increase in the oxygen-based methanol yield from 0.1 to 5.2% was noticed as the methane/oxygen molar ratio was increased from 0.5 to 16.8. Even at lower oxygen (3 mmol) amounts, higher methanol yield (4.7%) was noticed. A significant increase (27%) in the methanol yield was noticed when the methane amount was maintained at a constant (118 mmol) value and the oxygen amount was decreased from 13.5 to 7 mmol, see Fig. 2  Optimization of the temperature. The effect of temperature on methanol yield was studied by maintaining an optimum methane/oxygen ratio of 16.8 and without any dwell time as shown in Table 1. An increase in the oxygen-based methanol yield from 0.3 to 6.6% was noticed when the reaction temperature was increased from 250 to 300 °C. The lowest carbon dioxide selectivity was observed when the reaction was performed at 275 °C. A 25 °C higher or lower reaction temperature than 275 °C resulted in higher carbon dioxide selectivity. The optimal reaction temperature was selected as 275 °C for the rest of the study.
Effects of different supercritical solvents and optimization of acetonitrile:oxygen molar ratio. Effects of different solvents on methane activation were studied as shown in Table S1. The molar ratio and solvent volume have been adjusted to avoid the flammability range of methane. Even though a direct comparison cannot be attained due to changes in conditions, a general idea about the reactivity trend can be obtained from these experiments. The methanol yield was halved (2.7%) when deuterated acetonitrile was used instead of non-deuterated acetonitrile (5.3%). Low methanol yield (1%) compared to deuterated or non-deuterated acetonitrile was noticed when a trichloroacetonitrile rich acetonitrile mixture was used in the reaction. Low methanol yields, 5% and 0.4%, were noticed when reactions were performed with conventional supercritical solvents such as carbon dioxide and water. The lowest methanol yield (0.04%) was observed in apolar solvent benzene as compared to other solvents.
The methane/oxygen molar ratio (same amount of methane and oxygen) was held constant at the optimum value (~ 17) while changing the acetonitrile content in the system at 275 °C as shown in Fig. 3. A linear increase in the methanol yield (O 2 based) from 0.2 to 4.6% was noticed as the solvent to oxygen molar ratio was increased from 1.3 to 8.

Effects of different nitrides, stirring and optimization of h-BN:oxygen molar ratio.
After optimizing the acetonitrile only conditions, h-BN was introduced into the system to investigate synergetic effects. A 70% (from 4.6 yield) increase in the methanol yield was observed when 1.15: 1 h-BN: oxygen molar ratio of hexagonal boron nitride was used as a catalyst (Table 1). Relatively low methanol yields were observed in the presence of other nitrides (Supplementary Table S2) including C 3 N 4 (0.4%), InN (1.7%), and TiN (0.8%). A relationship between the system stirring and the methanol yield was observed as shown in Supplementary Table S3. The methanol yield was increased by 50% upon stirring the system at 60 rpm compared to the non-stirring system. A methanol yield of 3%, lower than the acetonitrile only yield (5.2%), was noticed even in the absence of acetonitrile and in the presence of h-BN. A 118% increase in the methanol yield was identified upon increasing the boron nitride loading to 2.30: 1 h-BN: oxygen molar ratio. A steady decrease in the methanol yield was observed when the boron nitride loading was increased from 2.30: 1 to 4.60: 1 (h-BN: oxygen molar ratio) as shown in Fig. 4. Both oxygen and methane-based methanol yields are depicted for comparison to the literature. A maximum oxygenbased methanol yield of 17.3% and methane-based yield of 2.0% were observed at 2.30: 1 h-BN to oxygen molar loading. Use of an oxygen-based yield can be justified by considering the limited oxygen (1: 17; oxygen: methane) amount in this study.
Optimization of acetonitrile:oxygen molar ratio with h-BN. The acetonitrile to oxygen ratio was changed in the presence of h-BN initiator as shown in Fig. 5. A linear increase in the oxygen-based methanol yield with (up to Table 1. The effect of temperature on direct methane oxidation to methanol in near supercritical acetonitrile. *With 400 mg h-BN (2.30: 1 h-BN: oxygen molar ratio), all other runs were conducted with 3 mL acetonitrile, 7 mmol of O 2 , 118 mmol of CH 4 , and 163 mmol He. All gases were cold fed to a total of ~ 1570 psi at − 30 °C. The reactor was heated at a 2.5 °C/min ramp rate to 275 °C. The heater cut off was set at 275 °C (no dwell time). The stirring speed was set at 60 rpm. The reactor was cooled to ambient at the natural convection rate. Standard deviations (S. D.) were based on deviations in loading pressure. Methanol yield was calculated based on a GC peak area calibration plot and reconfirmed with NMR. The standard deviation of the yield was calculated by replicating one experiment and comparing those data to NMR data. Small amounts of acetamide and acetic acid were formed as hydration products of acetonitrile and for simplicity purposes those products were excluded from the calculations. † Higher tendency towards formation of acetamide and acetic acid relative to other reaction conditions was observed. As based on the GC traces these products were formed in negligible amounts for entries 1, 2 and 4.  Table S4. There were no significant changes in the selectivity of the products with the change of the support.

Cut off T (°C)
Extension to C2 activation. The substrate scope of the reaction was investigated with higher hydrocarbons (C2) as shown in Table 2. Both methanol and ethanol were observed in the presence of ethane and the collective yield (8.1%) was more than the methanol yield (6.5%) in the methane reaction.
Isotopic label experiment. In order to eliminate the assumption of methanol formation from solvent acetonitrile, 13    Pre-and post-catalytic studies. According to the Auger electron spectroscopic (AES) data a slight decrease in the intensity of the oxygen KLL signal in the post-reaction h-BN catalyst was seen in comparison to the pre-reaction h-BN (Fig. S3). 11 B solid state NMR (SS-NMR) data also indicated the decrease in the B(OH) x O 3-x bond intensity at 15.5 ppm (Fig. 6a, Table 3) upon comparison between pre-and post-reaction h-BN 33 . Furthermore, post-reaction h-BN 15 N SS-NMR showed a downshift in comparison to the pre-reaction h-BN (Fig. 6b). The X-ray photoelectron spectroscopy (XPS) analysis showed an increase in O-H%, decrease in O-B%, increase in N-B%, increase in N-H%, increase in B-N%, and decrease in B-O% in post-reaction h-BN compared to pre-reaction h-BN [37][38][39][40] . The binding energies of both B and N of post-reaction h-BN shifted to lower binding energies compared to pre-reaction h-BN (Table S5). The crystallinity of the h-BN was increased after the reaction when compared to pre-reaction h-BN (Fig. 7).

Discussion
Control of the available amount of oxygen is critical since oxygen is responsible for both methane conversion and product overoxidation 41 . Sato et al. used a lower O 2 /CH 4 molar ratio (0.03) for methane conversion in supercritical water 42 . The methane to oxygen ratio study showed that relatively higher oxygen amounts (22 mmol) lead to lower methanol yields (Fig. 1) probably due to product overoxidation. As the oxygen amount decreased (22 mmol to 7 mmol) the available oxygen molecules that participate in both overoxidation and conversion reactions may have been diminished which would result in a tradeoff in yield. Increase in the yield even upon decreasing the oxygen content (constant methane) suggests that at lower oxygen partial pressures the methane activation reaction surpasses the overoxidation 3 . Another explanation can be introduced in terms of the supercritical solvent clustering effect explained elsewhere 21 . Methane can be converted to methanol inside the acetonitrile cluster by activated oxygen, and methanol can be subsequently released from the cluster into an oxygen-rich bulk solution where the oxidation takes place. As the oxygen content decreased the spectating oxygen in the bulk can be decreased which results in improved yield. At even lower oxygen content (3 mmol) oxygen could be acting as a limiting reactant inside the cluster resulting in slightly lower yield (4.7%) compared to 7 mmol reactions (5.2%).  www.nature.com/scientificreports/ Even though evidence to isolate the exact mechanism has not been established, both these explanations stem from the overoxidation hypothesis. Experiments conducted at three temperatures; below (250 °C), just above (275 °C), and above (300 °C) critical temperatures further corroborate the overoxidation phenomena. When the solvent temperature stays below the critical temperature, the clustering effect can be limited which would result in lower to no methanol yield (0.3%) due to overoxidation. Higher carbon dioxide selectivity at 250 °C (50%) compared to 275 °C eliminates the possibility of lack of enough energy for activation. Just above supercritical conditions, 275 °C, CO 2 selectivity  www.nature.com/scientificreports/ decreased to 15% suggesting product protection by cluster formation at supercritical conditions. Even though more yield through cluster formation is possible at 300 °C, higher temperatures may have contributed to the total combustion of methane to carbon dioxide (80% CO 2 selectivity). The static dielectric constant and density of acetonitrile depend on the temperature and pressure (reduced density) of the system 43,44 . As temperature increases, density and dielectric constant decrease whereas pressure has a completely opposite effect. At the critical point (272.35 °C, 703.43 psi) the density of acetonitrile was reported as 0.225 g/mL whereas the density of methane, methanol, benzene, carbon dioxide, and nitrogen are 0.162, 0.270, 0.210, 0.225, 0.313 g/mL respectively 44 . At critical pressure, the dielectric constant of acetonitrile is 40.3 (25 °C) and at critical temperature extrapolated values go down to 0.66 (at 1 bar) 45 . Due to the above reasons, acetonitrile is expected to behave as an apolar gaseous solvent with high apolar compound solubility (10 3 to 10 8 ) which can facilitate methane solubility 46 . Organic solvents in high-pressure or supercritical pressure can arrange into high density localized microreactor pockets due to high compressibility 47,48 . The size of these clusters can be altered based on the pressure of the system 48 . Oxygen and methane can be activated inside the apolar clusters via strong solvent-gas interactions and product molecules can be protected against overoxidation. We have probed high pressure differential scanning calorimetry (DSC) of acetonitrile and O 2 under the same reaction conditions as for methane partial oxidation and have seen no oxidation of acetonitrile at short residence times. We have also probed the potential of clustering phenomena by using diffusivity measurements via high pressure NMR. These clustering phenomena seem to be dynamic and shorter than the NMR timescale. NMR chemical shift measurements show that the acetonitrile remains unchanged under the reaction conditions in the presence of O 2 .
Strong interactions between oxygen/methane and acetonitrile clusters are believed to activate the gaseous reactant molecules. As the solvent to oxygen ratio increased, the number of available solvent molecules to make strong interactions with solutes (methane, oxygen), hence clustering, would have been increased. Low product yield (0.2%) at low acetonitrile (10 mmol) concentrations could be attributed to insufficient cluster formation. The results infer that in order to form an adequate number of clusters, the solvent to oxygen ratio should exceed a critical value (in this case 5.2). The molar fraction of solvent (acetonitrile) in a mixture determines the critical temperature and pressure. As the acetonitrile molar fraction decreases, the critical temperature also decreases while the critical pressure can show a positively skewed curve 21 . If similar behavior is assumed, the critical condition (T, P) should be decreased compared to pure acetonitrile within the operating acetonitrile molar fraction regime of this study (0.03 to 0.17).
Use of acetonitrile as the subcritical solvent was justified by studying conventional supercritical systems such as water and carbon dioxide. The results showed that supercritical acetonitrile derivatives deliver higher methanol yields compared to conventional supercritical systems. All the reactions were performed at 275 °C (despite the critical point of water at 375 °C) due to significant methanol overoxidation at temperatures above 275 °C. The hydrogen-bonded cluster network of water decreases at supercritical temperature transforming into a tetrahedral packing structure 49 . The energetically most favorable five-membered cluster can be formed by hydrogen bonding between H and N of different acetonitrile molecules 50 . The hydrogen bonding strength, a common factor in both systems, can be disrupted by another hydrogen bonding additive/product. Electronegative -fluoro and -chloro derivatives of acetonitrile that have very similar boiling points (e.g., CH 3 CN 82 °C, FCH 2 CN 79-80 °C, Cl 3 CCN 83-84 °C) potentially impact the cluster properties and O 2 solvation in a way that leads to lower methanol yields. Apolar carbon dioxide and benzene, on the other hand, can only participate in London dispersion forces and resulted in lower methanol yield (0.5% and 0.04% respectively). The critical points of benzene (289 °C) and acetonitrile (272 °C) are similar.
The 70% and 260% (from 4.6% yield, Fig. 3 (Table S2), a 2D structure similar to h-BN (Fig. S3), suggests a structure independent methane   (Fig. 3) suggests mass transfer limitations in the system. The optimal solvent: oxygen molar ratio of 8.2 with or without h-BN further suggest a synergetic effect of supercritical acetonitrile and h-BN systems. The doubt of methanol formation from the solvent acetonitrile rather than methane was cleared using 13 C isotope labeled methane for the reaction while keeping other parameters unchanged. Upon analyzing the products using GC-MS, mass fragments for methanol peak m/z = 32 and 33 were shown as the highest abundant species respectively for 13 CH 3 O − and 13 CH 3 OH fragments (Fig. S2).
The decrease in the intensity of the oxygen KLL peak of the post-reaction catalyst in comparison to the prereaction catalyst (Fig. S3) in the AES spectra could be due to oxygen terminated B-O bond dissociation during the methane oxidation reaction. Also, the decrease in the 15.7 ppm peak representing B-O in MAS 11 B SS-NMR (Fig. 6a) upon comparison between pre-and post-reaction h-BN further confirm the observations shown from AES data. Furthermore, post-h-BN 15 N-SSNMR showed a downshift in comparison to the pre-h-BN (Fig. 6b). This could be mainly due to the proton abstraction by N atoms in h-BN during the reaction. The observation of the decrease in the B-O (break in B-O-O-N bridge) after the reaction (Fig. 8) in XPS data agrees quite well with the AES and SS-NMR data. These data corresponded to a previously reported oxidative dehydrogenation (ODH) reaction mechanism (active site; an oxygen terminated armchair edge of BN bridge, B-O-O-N) of C2-C4 alkanes, to some extent 30 . However, methane conversion could not be explained with the ODH pathway. Methane to ethene oxidative coupling reaction mechanism on h-BN (active site; B-OH) suggested by Wang et al. 29 could not explain the observations noted above, indicating that the active site for the reaction could be the terminal B-O-B bridge of h-BN. Based on the above facts, CH bond activation could occur on bridging oxygen sites to generate chemisorbed methoxy and hydroxyl groups (similar to the first step of ODH) which then desorbed to generate methanol. It is remarkable how the C2-C4 alkane ODH catalyst (h-BN) 51-53 is also active for methane and ethane oxygenation reactions under supercritical conditions.
In most heterogeneous supercritical and subcritical thermal catalysis literature reports, methanol yield (methane-based) stays around 1% 23,54 which showcases the uniqueness of the current study. The maximum carbon-based yield of 2% and oxygen-based yield of 17% under subcritical thermal catalysis have not been reported to date.

Conclusions
In summary, a metal-free thermal catalytic partial oxidation of methane to methanol was developed using h-BN under near supercritical acetonitrile to lead to a yield of 17% (oxygen-based) of methanol. Furthermore, an extension of this method to ethane (C2) leads to 3.6% and 4.5% (oxygen-based) yields of methanol and ethanol respectively.