Quantitative Conversion of Methanol to Methyl Formate on Graphene-Confined Nano-Oxides.

We demonstrate the nearly quantitative conversion of methanol to methyl formate (MF) with a reliable durability on the reduced-graphene-oxide-confined VTiOx nanoparticles (rGO@VTiO). The rGO@VTiO exhibits superior low-temperature reactivity than the rGO-free VTiO, and the MF yield of 98.8% is even comparable with the noble metal catalysts. Both experiments and simulations demonstrate that the ultrathin rGO shell significantly impacts the shell/core interfacial electronic structure and the surface chemistry of the resultant catalysts, leading to remarkable reactivity in methanol to MF. rGO enhances the dispersion and loading rates of active monomeric/oligomeric VOx. In particular, the electron migration between the rGO shell and oxides core reinforces the acidity of rGO@VTiO in the absence of sulfate acidic sites. Moreover, both in situ NAP-XPS and DRIFTS investigations suggest that the lattice oxygen was involved in the oxidation of methanol and the MF was formed via the hemiacetal mechanism.

Introduction

Methyl formate (MF) is an important building block molecule in synthetic chemistry and a versatile antiseptic and solvent as well (Zeng et al., 2019, Banerjee and Kanan, 2018, Zhang et al., 2019). It is commercially produced by the sodium methoxide-catalyzed methanol carbonylation with CO, which poses severe environmental concerns owing to the toxic and corrosive reagents (Kaichev et al., 2014, Zhang et al., 2019). Thus, there are significant interests in developing more efficient and greener processes for MF synthesis. Direct oxidative coupling of methanol to MF over the supported noble metal or transition metal oxide catalysts is a simple, non-corrosive, and economical process (Whiting et al., 2015, Wang et al., 2013, Wittstock et al., 2010, Zhang et al., 2014a, Zhang et al., 2014b). Among the catalysts, TiO2-supported vanadia (VTiO) was widely studied because it is more available than the noble metal catalysts and more reactive than most of the non-precious catalysts (Zhao et al., 2010a, Zhao et al., 2010b, Kaichev et al., 2014, Liu et al., 2016). However, it is generally difficult to simultaneously achieve high MF selectivity together with high methanol conversion, leading to a low single-pass yield of MF. In view of the deep oxidation of methanol to CO2 as well as the explosion risk, moreover, it is also critical to reduce the reaction temperature. Therefore, further improving the MF yield at lower temperature is highly desired for the VTiO-based catalysts.

Graphene is a 2D atomic crystal with sp2-hybridized carbon, which possesses unique electronic, chemical, and mechanical properties and therefore has been investigated for wide applications including catalysis (Li et al., 2015, Deng et al., 2016, Wang et al., 2019). As a catalyst, graphene takes the advantages of large surface area, excellent mechanical properties, and high thermal/electron conductivities (Deng et al., 2016). In particular, the dislocations, vacancies, edges, impurities, and functional groups of graphene can all promote its catalytic properties (Tang et al., 2019). Graphene-encapsulated metal particles (e.g., Pt) have exhibited superior activities for the reactions like methane oxidation and electrocatalysis of water oxidation (Cui et al., 2018, Deng et al., 2017). With embedding single iron atoms into the graphene matrix, moreover, it offered impressive activity in the selective oxidation of benzene and methane at room temperature (Cui et al., 2018, Deng et al., 2015a). For these graphene-confined metal catalysts, the electrons of metal atoms can penetrate through the graphene shell to promote the catalytic reaction, whereas the graphene shell can protect the inner metals from damage in reaction (Deng et al., 2013a). Potentially, the excellent oxidation activities and low-temperature properties of such graphene@metal catalysts inspire us to develop more efficient catalyst for methanol oxidation to MF. However, the catalyst of graphene-confined metal oxides has not been demonstrated.

Herein, we developed the reduced GO (rGO)-encapsulated VTiOx catalyst (rGO@VTiO) through a facile co-precipitation method. Over these catalysts, not only nearly quantitative conversion of methanol to MF was achieved but also the low-temperature activity was highly improved. Effects of the ultrathin rGO shell on the structure and surface chemistry of the catalyst as well as the reaction mechanism were thoroughly investigated by experiments and computations.

Results

rGO Impacts on the Structure and Texture of Catalysts

The n(rGO)@VTiO catalysts were prepared through co-precipitating the acidic aqueous mixture of VOSO4, TiSO4, H2SO4, and GO suspension with ammonia solution, where n refers to the weight percentage of GO used in the preparation recipes. As control, the sulfated VTiO catalysts were fabricated using the same method without GO addition (Figure 1A). The transmission electron microscope (TEM) images show the microstructure of VTiO and rGO@VTiO. VTiO sample displayed clean and clear particle boundaries with the well-ordered microstructure that can be assigned to the anatase TiO2 (101) (Figure 1B).(Sima et al., 2017) For 0.6(rGO)@VTiO, the VTiO particles were isolated and well encapsulated by the rGO layer, and the cross-sectional view indicated that most of the layer is around 1.5 nm thick (Figures 1C and S1). As the rGO mono-sheet is 0.4–0.7 nm thick (depends on the reduction degree of GO), the number of rGO mono-sheet on 0.6(rGO)@VTiO is 2–4 (Yang et al., 2018, Qi et al., 2017). Similarly, the rGO shells were also observed from other n(rGO)@VTiO samples (n = 0.3, 1.0, 2.0) (Figure S2). Both measurements and simulations have suggested that the electron shuttle across the graphene is limited to <4 carbon layers and the thinner graphene layer can enhance the adsorption and reduction of O2 owing to the promoted electron transfer inside catalyst (Deng et al., 2013b). Therefore, the ultrathin rGO shell with ∼3 carbon layers on 0.6(rGO)@VTiO may contribute to the physicochemical properties of catalyst.

Figure 1

Preparation Procedure of Catalyst and the Structural Analysis of Catalysts

(A–E) (A) Schematic of VTiO and rGO@VTiO preparation; TEM images of VTiO (B) and 0.6(rGO)@VTiO (C); (D) XRD patterns and (E) Raman spectra of n(rGO)@VTiO n = 0 (1), 0.3 (2), 0.6 (3), 1.0 (4), and 2.0 (5). Scale bar is 5 nm.

See also Figures S1–S6 and Table S1.

In addition, rGO increased the surface area of catalysts, in which the Brunauer-Emmett-Teller (BET) specific surface area grew from 46 m2 g−1 of VTiO to 121 m2 g−1 of 0.6(rGO)@VTiO, for example (Table S1). The larger surface area improved the surface dispersion of vanadium species, as evidenced by the elemental analysis. The overall V/Ti ratio of catalysts determined by the inductively coupled plasma atomic emission spectrometry (ICP-AES) slightly increased from 0.24 of VTiO to 0.31 of rGO@VTiO (Table S1), suggesting that rGO improved the loading rate of V species. The near- and sub-surface V/Ti measured by X-ray photoelectron spectroscopy (XPS) and energy dispersive spectrometer (EDS), respectively, were higher than that of the bulk catalysts, indicating the surface enrichment of vanadium species. At the same time, the near-surface V/Ti decreased from 0.41 for VTiO to 0.33–0.36 for rGO@VTiO samples owing to the larger surface area and the higher V dispersion on rGO@VTiO than that on VTiO (Table S1). On the other hand, the elemental mapping of rGO@VTiO revealed that V, Ti, and C were uniformly distributed on the sample (Figure S3).

As proved by XPS, EDS, and CHNS elemental analyzer, interestingly, the sulfur content decreased with the increase of rGO and closed to 0 for the samples of 0.6/1.0/2.0(rGO)@ VTiO (Table S1). It is reasonable that the deposition of SO42− was suppressed during the catalyst precipitation owing to the electrostatic repulsion between GO and SO42− (Qi et al., 2017). Furthermore, the carbonyl groups on GO are enable to react with sulfur-centered anions during the catalyst calcination, resulting in the loss of sulfur (Gu et al., 2017, Li et al., 2016, Zhu et al., 2015). Since the acidity is vital for the catalytic oxidation of methanol to MF, the sulfation is an efficient method to enhance the acidity of VTiO-based catalysts (Zhang et al., 2019). But the sulfate groups are unstable upon the calcination and reaction conditions owing to their weak interactions with VTiO, which is always challenging the durability and reactivity of sulfated VTiO catalysts (Liu et al., 2016). In view of the superior reactivity of rGO@VTiO hereinafter, however, it indicates that their acidities were well improved even in the absence of sulfur, which will be proved by the acidity analysis.

The crystal structure of catalysts was measured by XRD (Figure 1D). Apart from anatase TiO2, the V2O5 crystals were also observed from the pattern of VTiO due to the aggregation of vanadia species, in line with the elemental distribution analysis of catalyst surface. However, V2O5 was undetectable from the 0.3, 0.6, and 1.0(rGO)@VTiO samples. This confirms that rGO promoted the dispersion of the vanadia species and suppressed the growth of crystalline V2O5. For 2.0(rGO)@VTiO, on the other hand, the thickened GO layer may impede the thermal transfer and elevate the real temperature of inner particles during the isothermal calcination of catalyst, leading to the formation of rutile TiO2 (Figure 1D). As references, GO pattern exhibited a typical peak at ∼12° assigned to GO (001), whereas the thermal reduced rGO only displayed a wide peak of graphite (002) at ∼26° (Figure S4) (Yang et al., 2018). It reveals that GO was well reduced to rGO under the conditions of catalyst calcination, which was further supported by XPS and IR results that the C/O ratio increased from 3.2 of GO to 11.0 of rGO and the bands of O-containing groups of rGO significantly weakened (Figures S5 and S6).

The structures of rGO@VTiO and VTiO were checked by Raman spectroscopy (Figure 1E). The bands at 404, 507, 521, and 640 cm−1 were attributed to Ti-O groups of anatase TiO2. For 1.0/2.0(rGO)@VTiO, the additional weak peaks at 243, 446, and 609 cm−1 assigned to rutile TiO2 (Sima et al., 2017), in line with the XRD results. The bands at 284, 303, 697, and 997 cm−1, ascribed to crystalline V2O5 (Zhao et al., 2010b), were gradually weakened with the rGO contents and then slightly increased for 1.0/2.0(rGO)@VTiO. This is consistent with the XRD observations that the optimal amount of rGO suppressed the formation of V2O5. Moreover, a new band around 1,034 cm−1, assigned to the polymeric VO, appeared in rGO@VTiO samples and a broad band at 820 cm−1 assigned to the isolated VO4 tetrahedral increased in intensity for 0.3/0.6(rGO)@VTiO (Zhang et al., 2019, Sima et al., 2017). It confirms that vanadium species mainly exist as crystalline V2O5 in VTiO but partial vanadium transformed into monomeric/polymeric VO for rGO@VTiO owing to the high dispersion. Furthermore, two characteristic peaks around 1,350 and 1,600 cm−1, assigned to the D band (from in-plane imperfections such as defects and heteroatoms of the graphitic lattice of the disordered sp2-hybridized carbon) and G band (from the tangential stretching mode of highly ordered pyrolytic graphite), respectively, were detected from the rGO@VTiO samples, which confirms the existence of rGO in the samples (Figure 1E).

rGO Impacts on the Chemistry of Catalysts

The surface chemistry of the catalysts was measured by XPS. As shown in Figure 2A, the O1s spectrum of VTiO was composed by three peaks centered at 529.7 (lattice oxygen), 531.0 (oxygen of -OH), and 531.9 eV (sulfate oxygen species) (Zhao et al., 2010b). The binding energy of S2p spectrum at 169.1 eV was ascribed to SO42− for VTiO but no sulfur was detectable for 0.6(rGO)@VTiO (Figure 2B), confirming the sulfur loss in 0.6(rGO)@VTiO. Thus, the O1s peak of 0.6(rGO)@VTiO at 532.1 eV is assigned to the oxygen vacancy (OV) rather than SO42− (Figure 2A) (Wan et al., 2018). The state of OVs was further evinced by Raman spectra (Figure 2C). The E mode of TiO2 shifted from 144 cm−1 of VTiO to 151 ± 2 cm−1 for rGO@VTiO owing to the formation of OVs (Wang et al., 2018a). Reasonably, the OV formation is relevant to the interactions between rGO ultrathin-shell and VTiO nanoparticles.

Figure 2

Surface Chemistry of rGO@VTiO-Based Catalysts

(A–E) (A) O1s and (B) S2p1/2 XPS spectra of VTiO and 0.6(rGO)@VTiO; (C) Raman spectra, (D) Ti2p3/2, and (E) V2p3/2 XPS spectra of n(rGO)@VTiO n = 0 (1), 0.3 (2), 0.6 (3), 1.0 (4), and 2.0 (5).

See also Table S2.

The carbonyl groups of GO could interact with the activated O2 molecules and then attack the adjacent sulfur groups (Gu et al., 2017). Instead of O2, it is also fair that the surface lattice oxygen of VTiO interact with the carbonyl groups of GO and then trigger the desulfurization during the catalyst calcination in N2, leading to sulfur loss and OV formation. In principle, the OV formation would reduce the metal valence, as confirmed by Ti2p and V2p spectra. The binding energies of Ti2p3/2 slightly down-shifted from 459.0 eV of VTiO to 458.7 eV of rGO@VTiO (Figure 2D), indicating a small amount of Ti4+ changed to Ti3+ (Kaichev et al., 2014). Similarly, V4+ proportions increased from 21.8% for VTiO to 26.5%–33.6% for rGO@VTiO samples (Figure 2E and Table S2). Since OVs possess high adsorption and activity for O2, it is reasonable to enhance the reactivity in methanol oxidation.

Bifunction of redox and acidity is the key character to promote the partial oxidation of methanol to MF (Zhao et al., 2010a). As shown in Figure 3A, the temperature-programmed desorption of ammonia (NH3-TPD) profiles of VTiO and rGO@VTiO displayed two desorption peaks of NH3 at 430–450 K and 590–690 K, corresponding to the weak and medium-strong acidic sites, respectively (Lu et al., 2011). For VTiO, the residual sulfate groups significantly impact its acidity (Sima et al., 2017). With the increase of rGO, the medium-strong acid gradually up-shifted from 591 K of VTiO to 601–690 K for rGO@VTiO samples. The peak areas of weak and medium-strong acid significantly increased from 808 to 389 of VTiO to 1,884 and 1,324 of 0.6(rGO)@VTiO and then decreased to 1,206 and 625 for 2.0(rGO)@VTiO, respectively (Table S3). This reveals that the moderate rGO improved the acidity, whereas the excessive rGO might cover the acidic sites. 0.6(rGO)@VTiO contained the most acidic sites, which is consistent with its reactivity in the methanol oxidation. The acidity of VTiO and 0.6(rGO)@VTiO was further measured by pyridine adsorption infrared spectroscopy (Figure S7). The bands at 1,612 and 1,452 cm−1 correspond to Lewis acid sites, whereas the band at 1,545 cm−1 is attributed to Brønsted acid sites. The bands around 1,490 and 1,574 cm−1 are ascribed to the overlapping of Brønsted and Lewis acid sites. The enhanced signal intensities from 0.6(rGO)@VTiO confirmed that its acidity was significantly improved with interaction between rGO and VTiO, in line with the NH3-TPD results. As a control, the NH3-TPD results of rGO exhibited negligible acidity (Figure S8). Since it has been confirmed that sulfate groups were almost free in rGO@VTiO, the notably enhanced acidity of rGO@VTiO is evidently contributed by the interactions between rGO and VTiO rather than the sulfate groups and/or the rGO itself.

Figure 3

Acidity and Redox Properties of rGO@VTiO-Based Catalysts

(A) NH3−TPD profiles of n(rGO)@VTiO n = 0 (1), 0.3 (2), 0.6 (3), 1.0 (4), and 2.0 (5).

(B and C) (B) Electron density difference of NH3 adsorption on the interface of rGO@VTiO (C, brown; Ti, blue; O, red; N, light blue; H, white. Dark blue and yellow contours refer to the electron depletion and electron accumulation, respectively. The isosurface levels are set to be 0.001 e/Bohr3) and (C) H2−TPR profiles of samples 1–5.

See also Figures S7 and S8 and Tables S3 and S4.

The effects of rGO on the acidity of rGO@VTiO were further simulated (Figure 3B). The intercalation of NH3 molecule leads to a significant electron migration and accumulation on the graphene surface. The simulation indicates that NH3 adsorption is stronger on rGO@VTiO than on VTiO (−0.89 versus −0.77 eV), which aligns with the NH3-TPD results. In rGO@VTiO, the curled graphene offers electrons to the inner particles and displays a different electron density between the outer/inner surfaces, which further enhances the density of states near the Fermi level of graphene and therefore increases the number of acidic sites (Deng et al., 2013a).

The redox properties of the catalysts were studied by hydrogen temperature-programmed reduction (H2-TPR) (Figure 3C). The reduction temperature downshifted from 754 K of VTiO to 720 K of 0.6(rGO)@VTiO and then increased to 786 K for 2.0(rGO)@VTiO. This corresponds to the different types of VOx species. In a high dispersion state, monomeric VOx groups with three V-O bonds to TiO2 and a terminal partly hydrated V=O bond would be formed (Bulushev et al., 2002). Upon the enrichment of V species, some V-O-Ti bridges rearranged into V-O-V polymerization bridges, which turn to crystalline V2O5 with further increasing V coverage (Sima et al., 2017). In general, the monomeric and polymeric VOx species are easier to be reduced than crystalline V2O5 (Bulushev et al., 2002). Therefore, 0.6(rGO)@VTiO displayed the lowest reduction temperature owing to the high content of monomeric/polymeric VOx, whereas VTiO displayed higher reduction temperature owing to more crystalline V2O5. This agrees well with the XRD, XPS, and Raman measurements. Moreover, the easier reduction of 0.6(rGO)@VTiO is consistent with its low-temperature reactivity in methanol to MF. Meanwhile, the peak area gradually decreased with the increase of rGO in the catalysts (Table S4), as the formation of OVs decrease the H2 consumption. On the other end, the small peaks centered at 860–896 K, ascribed to the reduction of sulfate species (Liu et al., 2016), rapidly decreased in the peak area from 50 for VTiO to <4 for rGO@VTiO (Table S4), which verifies again its sulfur loss.

rGO Impacts on the Methanol Oxidative Coupling

The explosion proof is the principal concern in the real practice of catalytic oxidation (Figure S9). However, the available explosive limits of methanol-oxygen (air) systems were derived from the ambient temperature and pressure, which is imprecise for the real oxidation. Thus, the real explosive limits of methanol-O2-N2 were recorded by an alternating current center-fire method at 423 K and 3 bar (Figure 4A). Correspondingly, the methanol oxidation on our catalysts was conducted under a safe envelope.

Figure 4

Selective Oxidation of Methanol to MF on rGO@VTiO Based-Catalysts

(A–C) (A) The explosibility of O2-MeOH-N2 system (423 K, 3 bar, alternating current 7 kV center fire); temperature dependence of (B) methanol conversion, MF selectivity and yield, and (C) CO2 selectivity on VTiO, rGO, and 0.6(rGO)@VTiO catalysts (GHSV = 3000 mL gcat−1 h−1).

(D) Stability of 0.6(rGO)@ VTiO at 408 K.

(E) Comparison of methanol oxidation to MF on 0.6(rGO)@VTiO with various catalysts.

See also Figures S9 and S10 and Tables S5 and S6.

Methanol oxidation on 0.6(rGO)@VTiO and the reference samples were tested in a micro-fixed-bed reactor at 388 to 423 K (Figure 4B). The optimized feeding conditions, P = 0.1 MPa, N2/O2 = 9, O2/CH3OH = 1.5, and GHSV = 3,000 h−1, were used for the oxidation test (Table S5). With temperature increasing, the methanol conversion (CMeOH) on VTiO increased from 63.2% to 96.2%, whereas the MF selectivity (SMF) exhibited a peak value of 93.0% at 408 K, reflecting the MF yields (YMF) from 47.5% at 388 K to the maximum of 88.7% at 413 K. As a control, rGO showed a poor reactivity that the inferior CMeOH (<36%) and SMF (<27%) derived to YMF < 9% at 388–423 K. In contrast, 0.6(rGO)@VTiO offered superior reactivity for methanol to MF. A YMF of 88.0%, translated from CMeOH of 91.0% and SMF of 96.7%, was obtained even at 388 K, which is 85% higher than that on VTiO (i.e., YMF = 47.5%). This demonstrates an enhanced low-temperature reactivity of rGO@VTiO. Then CMeOH exceeded 99.0% at 403 K and further closed to complete conversion (99.8%) at 413 K. Meanwhile, SMF increased to >99.0% at 403–408 K. Correspondingly, YMF = 98.8% was obtained even at 408 K on 0.6(rGO)@VTiO. On the other hand, the YMF of 0.3/1.0(rGO)@VTiO were higher than that of VTiO but slightly lower than that of 0.6(rGO)@VTiO (Figure S10). It suggests that the appropriate amount of rGO in the catalyst tends to significantly improve its reactivity in methanol to MF.

The CO2 selectivity (SCO2) on the catalysts were shown in Figure 4C. For rGO, CO2 was undetectable at 388–413 K and then slightly increased to 0.5%–1.3% at ≥ 418 K. This indicates that rGO itself is chemically stable with poor activity for methanol conversion, in line with its MF yields. However, it is much easier to generate CO2 on VTiO even at 403 K and the SCO2 reached 8% at 423 K. In contrast, SCO2 was significantly suppressed on 0.6(rGO)@VTiO. No CO2 can be detected until 418 K, implying that rGO restrains the deep oxidation of methanol. The long-term stability of 0.6(rGO)@VTiO was investigated at 408 K (Figure 4D). Both CMeOH and SMF, averaging at 99.7 ± 0.2% and 99.1 ± 0.3%, respectively, remain stable with slight fluctuations during 500-h testing, demonstrating a reliable catalytic stability with high reactivity.

The oxidation of methanol to MF on 0.6(rGO)@VTiO was compared with those on the various catalysts under their optimized conditions (Han et al., 2014, Whiting et al., 2015, Wang et al., 2013, Wittstock et al., 2010, Zhang et al., 2014a, Zhang et al., 2014b), as summarized in Figure 4E and Table S6. Owing to the high reactivity of the noble metal-containing catalysts, methanol can be converted into MF at relatively low temperature. Among the noble metal catalysts, the highest MF yield of ∼90% was obtained on the Au-Pd nanoparticles at 343 K (Wang et al., 2013). TiO2-supported Au-Ag offered the highest MF yield of 77.0% at room temperature in photo-catalysis of methanol (Han et al., 2014). However, the high cost of noble metals limits their application. The non-noble oxide catalysts, like ReOx/CeO2, V2O5/TiO2, and RuOx-ZrO2, are more accessible, but they present much lower reactivity for the methanol conversion even at high temperature. Evidently, 0.6(rGO)@VTiO exhibited higher single-pass MF yield compared with these catalysts including noble metals. Compared with the sulfated VTiO, obviously, 0.6(rGO)@VTiO demonstrated excellent low-temperature activity, methanol conversion, and MF selectivity. In methanol to MF, the over-oxidation of methanol to COx, which normally depends on the reaction temperature, should be avoided as far as possible to increase the usage of methanol. To lower the difficulty of product purification, moreover, it requires high yield of target product. Therefore, rGO@VTiO exhibits the high reactivity at low temperature, which not only suppresses the complete oxidation of methanol but also relieves the challenge of product separation.

In Situ NAP-XPS and IR of Methanol Oxidation on rGO@VTiO

To gain insights into the reaction, 0.6(rGO)@VTiO was measured in situ near atmospheric pressure XPS (NAP-XPS) under the simulated reaction conditions. In pure methanol, V5+ weighted >70% of vanadium at 303–353 K and then V4+% grew at higher temperature (e.g., V4+% = 42.6% at 373 K) (Figure 5A and Table S7). Moreover, a new peak at 515.3 eV ascribed to V3+ appeared at 393 K and more V4+ + V3+ were formed above 393 K, leaving only 26.7% of V5+ at 473 K. The reduction of V5+ to V4+/V3+ corresponds to the chemisorption and oxidation of methanol on the catalyst (Odriozola et al., 1991). As a comparison, the appearance of V3+ on VTiO is 20 K higher (Figure S11 and Table S8). This reveals that rGO@VTiO possesses superior low-temperature activity for methanol adsorption and activation, in line with the real reaction.

Figure 5

In Situ XPS and IR Characterizations of rGO@VTiO Catalysts.

In situ NAP-XPS V2p3/2 (A and B) and C1s (C and D) spectra of 0.6(rGO)@VTiO in methanol (A and C) at 323 to 473 K or in CH3OH (0.4 mbar) -O2 (0–0.6 mbar) at 403 K (B and D), and (E) in situ DRIFTS spectra of the 0.6(rGO)@VTiO in CH3OH-O2 mixture at 1, 323; 2, 353; 3, 373; 4, 403; 5, 423; and 6, 473 K.

See also Figures S11–S13 and Tables S7–S13.

The thermal effects on the activation of O2 over 0.6(rGO)@VTiO and VTiO were checked in methanol + O2 mixture at 303–473 K (Figure S12). The increased V5+% reveals that O2 was activated and replenished on catalysts. However, 0.6(rGO)@VTiO possesses higher V4+% than VTiO (Table S9), indicating that it can activate more O2.

The effects of O2 partial pressure were tested using 0–0.6 mbar O2 mixed with 0.4 mbar methanol at 403 K (Figure 5B and Table S10). With the presence of O2, V3+ rapidly converted into higher valence states and the ratios of V5+/V4+ increased with the O2 pressure. The V4+ fractions of 0.6(rGO)@VTiO were higher than those of VTiO under the same conditions (Figure S13 and Table S11). This confirms that rGO enhanced the redox capacity of catalysts, in line with the H2-TPR results. It has been proposed that O2 can adsorb and be activated on graphene (Deng et al., 2013a), (Deng et al., 2015b) In rGO@VTiO, the electron transfer between rGO shell and nanoparticles may decrease the local work function of graphene surface, which promotes the adsorption and activation of O2 on rGO. Thus, it enhanced the oxygen migration and the availability of the oxygen atom in the bridging V-O-Ti bond and thus improves the redox capability of catalyst (Deng et al., 2013a) (Liu et al., 2009), Therefore, rGO@VTiO possesses superior low-temperature activity in comparison with VTiO owing to the interactions between rGO shell and inner oxides.

Figure 5C displays the corresponding C1s spectra of 0.6(rGO)@VTiO in pure methanol at 323–473 K. The deconvoluted peaks ascribed to graphite C-C (285.1 ± 0.1 eV), methoxy species C-O (286.3 ± 0.1 eV), dioxymethylene (287.5 eV), gas-phase methanol (287.9 ± 0.2 eV), and formate species and MF (288.9 eV), respectively (Romanyshyn et al., 2008). The amount of the methoxy species initially increased with temperature in 303–393 K and then gradually decreased at higher temperatures (Table S12). Simultaneously, the signals of dioxymethylene and formate species increased with temperature. This confirms that the elevated temperature promotes the chemisorbed methanol to form methoxy groups, whereas the further higher temperature triggers the conversion of methoxy to formate species. The peak assigned to dioxymethylene increased rapidly from 17.2% in pure methanol to 28.5% in 0.1 mbar O2 + methanol at 403 K (Figure 5D and Table S13). It reveals that methoxy was rapidly oxidized to dioxymethylene with little O2. Increasing O2 pressure, more MF was detected while dioxymethylene decreased, suggesting O2-rich condition was favorable to form MF (Zhang et al., 2019).

To identify the exact nature of the surface species, in situ diffuse reflectance infrared Fourier transform spectra (DRIFTS) were obtained for 0.6(rGO)@VTiO in methanol-O2 mixture at 323–473 K (Figure 5E). The bands of physiosorbed methanol (i.e., 2,954 and 2,848 cm−1) and ν(OH) of adsorbed water and/or methanol (3,368 cm−1) were observed at 323 K and then the signal intensity was weakened with temperature. At T > 353 K, the signals of the methoxy species at 2,931 cm−1 (νs(CH3) of the symmetric stretching), 2,825 cm−1 (2δs(CH3) of the Fermi resonance), 1,150 cm−1 (ρ(CH3) of the rocking vibration), 1,441 cm−1 (δas(CH3) of the deformation asymmetrical vibrational), and 1,462 cm−1 (δs(CH3) of the deformation symmetrical vibrational) became stronger because the adsorbed methanol dissociated to methoxy groups (Busca et al., 1987). Above 373 K, the bands at 1,584 and 1,562 cm−1 (νas(O–C–O) of formate species), 1,538 and 1,361 cm−1 (νs(O–C–O) of formate species), 1,482 cm−1 (δ(CH2) of dimethoxymethane), as well as 1658, 1,633, and 1,205 cm−1 (ν(C=O) of adsorbed MF) evinced the formation of CH2O∗ and MF (Kaichev et al., 2014) (Whiting et al., 2015). The intensities of MF band at 1,205 cm−1 reached the crest at 403 K and then declined at > 423 K probably owing to the formation of CO2 (2,350 cm−1) (Whiting et al., 2015, Wang et al., 2018b), coinciding with the CO2 formation in the reaction. No formaldehyde was observed, as the consumption of formaldehyde is a fast step.

Reaction Mechanism over rGO@VTiO Catalysts

To better understand the effects of rGO on the CH3OH oxidation, the reaction mechanisms with and without rGO were investigated by theoretical calculation. Figure 6A compared the potential energy profiles of CH3OH oxidation with the intermediate CH2O∗. CH2O∗ formation can be divided into three steps (Figures 6B and 6C). CH3OH is first physiosorbed on the catalyst surface, and the physisorption is slightly stronger on rGO@VTiO than on VTiO (−0.45 eV versus −0.43 eV). The methoxy intermediate (CH3O∗) is then formed by a proton transfer from the OH of the physiosorbed CH3OH to the terminal OH of the V site, leading to chemisorption and the formation of V-OCH3 bond. The energy barrier of this step is considerably lower on rGO@VTiO than on VTiO (0.26 versus 0.37 eV), indicating that rGO can promote the chemisorption of CH3OH. The improved chemisorption of CH3OH can be attributed to the enhanced acidity and OVs of rGO@VTiO, as indicated by NH3-TPD and Raman results. Then, further H transfers from the CH3O∗ to the bridge -O-V, leading to the formation of CH2O∗. Thus, rGO was predicted to promote chemisorption and activation of CH3OH.

Figure 6

Simulation of Methanol Adsorption on Various Samples

(A–C) (A) Potential energy profiles for CH3OH to CH2O∗ on the catalyst models; optimized structures of the intermediate states and transition states for VTiO (B) and rGO@VTiO (C).

(D) Possible reaction network for CH3OH to MF over rGO@VTiO. Color of atoms: C, gray; Ti, blue; V, green; O, red; H, white; OV, yellow.

On combining the experiment and simulation findings, it is justified to explain the methanol oxidation on rGO@VTiO according to the chemical states of V and the roles of rGO. As depicted in Figure 6D, this reaction mainly involves the formation of CH2O∗ and MF. For the surface adsorption (step 1), methanol is easily captured by the acidity sites and OVs on rGO@VTiO (Oviedo et al., 2008). Then one proton of the adsorbed methanol transfers to the terminal OH group of V site to form CH3O∗, leading to chemisorption and the formation of VOCH3 bond (step 2), as it has been confirmed by DRIFTS and density functional theory (DFT) methods. In step 3, H of CH3O∗ transfers to the O site of V–O–Ti forming Ti–OH along with the reduction of V5+ to V4+, as proved by NAP-XPS. Then, CH3O∗ transforms into CH2O∗ with one H atom recombining with V-OH to form H2O (step 4). The consumption of CH2O∗ and the desorption of H2O corresponds to the reduction of V5+ to V4+ and the recovery of OVs. Subsequently, the active sites are re-oxidized from V4+ to V5+ by O2 (Step 5), which has been clearly observed from NAP-XPS. This cycle could be expressed as CH3OH + 1/2 O2→ CH2O∗ + H2O.

CH2O∗ interacts with CH3O∗ on the acidic sites of V−OH to form physiosorbed hemiacetal (CH3OCH2OH) and V−O bond, and then hemiacetal chemisorbs on the terminal V=O bond to form CH3OCH2O−V and HO−V (step 6). The stronger acidity of rGO@VTiO accelerates the formation of hemiacetal. Subsequently, one H of CH3OCH2O- transfers to the V–O–Ti site with generating Ti–OH and the adsorbed MF, and concurrently the vanadium cations are reduced from V5+ to V4+ (step 7). With the desorption of MF, V4+ is further reduced to V3+ (step 8). These have been proved by in situ NAP-XPS and DRIFTS methods. In the end, the V3+/V4+ atoms are re-oxidized to V5+ by O2 and the generated H2O desorbs from the catalyst surface in step 9. The latter cycle is conveyed as CH3OH + CH2O∗ + 1/2 O2 → CH3OCHO + H2O.

Discussion

In summary, the ultrathin rGO layer with well-confined VTiOx nanoparticles has been realized through a facile co-precipitation and thermal reduction method. The resultant rGO@VTiO demonstrated a superior low-temperature reactivity for the methanol oxidation to high-valued methyl formate in comparison with the rGO-free VTiO catalyst. It exhibited nearly quantitative conversion of methanol to MF with a maximum MF yield of 98.8% at 408 K, even comparable with the noble metal-based catalysts. The reliable stability of rGO@VTiO catalyst was proved by a 500-h testing. Both experimental analyses and computational simulations demonstrate that the presence of a small amount of rGO significantly affects the catalysts in the shell-core interface electronic structure, the surface chemistry of acidity and oxygen vacancies, the methanol adsorption/activation capacity, and the methanol oxidation mechanism. The electron migration between the rGO shell and oxides core significantly reinforced the acidity of rGO@VTiO catalyst in the absence of sulfate groups. This provides a facile method to intensify the highly stable acidity of catalysts not only for VTiO-based catalyst but also for wider solid catalysts, in principle.

Limitations of the Study

The catalyst support is moderately toxic.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Gaofeng Zeng (zenggf@sari.ac.cn).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

The published article includes all datasets/code generated or analyzed during this study.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.