Haifeng
Gao
a,
Zhiwei
Yao
*a,
Yan
Shi
a,
Renren
Jia
b,
Feixue
Liang
a,
Yue
Sun
a,
Wei
Mao
a and
Haiyan
Wang
a
aCollege of Chemistry, Chemical Engineering and Environmental Engineering, Liaoning Shihua University, Fushun, 113001, P.R. China. E-mail: mezhiwei@163.com
bPetroChina No. 3 Refinery of FuShun Petrochemical Company, Fushun, 113001, P.R. China
First published on 25th October 2017
In this work, we propose a facile, fast and large-scale synthesis of β-Mo2C catalysts (about 100 grams). The current approach uses a simple solid mixture of (NH4)6Mo7O24·4H2O and carbon nanotubes (CNTs) as a precursor and employs a fast heating rate method without holding time. Moreover, there was no need for passivating the carbide products. A series of β-Mo2C/CNT catalysts (denoted as MoC-5, -15, -30 and -60) were successfully prepared from the precursors with the Mo content ranging from 5 to 60 wt% and their catalytic activities for dry reforming of methane (DRM) were evaluated. CH4-TPSR and CO2-TPO over the fresh samples proved that the abilities of CH4 dissociation and oxidation resistance of β-Mo2C were greatly enhanced by compositing with CNTs. It was also found that these abilities were dependent on the Mo content, the particle size and the interaction between β-Mo2C and CNTs. Among these β-Mo2C/CNT catalysts, MoC-30 showed the best catalytic stability for DRM, attributed to its high activity for CH4 dissociation and high resistance to oxidation. Additionally, it is worth noting that the catalytic stability of monometallic MoC-30 was found to be superior to that of bimetallic Ni/β-Mo2C (recently known to be an efficient non-noble metal catalyst for DRM).
β-Mo2C + 5CO2 = 2MoO2 + 6CO, | (1) |
2MoO2 + 5CH4 = β-Mo2C + 4CO + 10H2. | (2) |
Therefore, β-Mo2C catalysts have been studied extensively for DRM in recent years.10–22 However, it is worth noting that monometallic β-Mo2C catalysts can exhibit stable activity only at relatively high pressures.10,21 At atmospheric pressure, the rapid deactivation of these carbides occurred due to their oxidation by CO2.10,17,22 The consequence is the transformation of the β-Mo2C phase into the ineffective MoO2 phase. Therefore, it was reasonable to deduce that at atmospheric pressure the reaction of CO2 with β-Mo2C (reaction (1)) was more favourable than the reaction of CH4 with MoO2 (reaction (2)), which led to the final oxidation of β-Mo2C catalysts.
Subsequently, bimetallic Ni(Co)/β-Mo2C carbides were found to exhibit stable DRM activity at atmospheric pressure due to the synergic effect between the two metals.23–30 In bimetallic carbide systems, the role of Ni(Co) was to conduct the dissociation of CH4, while the activation of CO2 took place on β-Mo2C. It is worth pointing out that the molar ratio of Ni(Co) and Mo was very important for stabilizing the bimetallic carbide systems. If the ratio was higher, the CH4 dissociation was enhanced, which led to the formation of coke. Conversely, the lower ratio would facilitate the oxidation of β-Mo2C by CO2. An appropriate molar ratio of Ni(Co) and Mo was required to match the conversion rates of reactants CH4 and CO2, and then a catalytic oxidation–recarburization cycle (reactions (1) and (2)) can be well established. Unfortunately, the appropriate Ni(Co)/Mo ratio was highly dependent on reaction conditions. As for the Ni/β-Mo2C system with the Ni/Mo ratios from 1/10 to 1/1, the catalyst stability sequence was reported as Ni/β-Mo2C (1/2) > Ni/β-Mo2C (1/1) > Ni/β-Mo2C (1/3) at 800 °C with WHSV = 12000 cm3 g−1 h−1,28 while the catalyst stability followed the sequence Ni/β-Mo2C (1/10) ≈ Ni/β-Mo2C (1/5) > Ni/β-Mo2C (1/3) > Ni/β-Mo2C (1/2) at 850 °C with GHSV = 3800 h−1.23 It was hence highly desirable to develop a new design for a β-Mo2C catalyst that was resistant to coking and oxidation without the addition of a second metal for atmospheric DRM reaction.
In the case of the preparation method of β-Mo2C catalysts, there are two general methods, namely the temperature programmed reduction (TPR) of Mo oxide precursors using mixtures of hydrocarbons/hydrogen31 and the carbothermal (hydrogen) reduction of Mo oxide precursors using various solid-state carbon sources.22,32,33 However, the two methods mentioned above normally involve complicated steps as follows: (i) the dissolution of (NH4)6Mo7O24·4H2O; (ii) the drying and calcination of Mo precursors under air; (iii) the reduction of Mo oxide precursors by reducing gas and (iv) the passivation of the fresh β-Mo2C sample at low concentrations of O2 (typically <1%). On the other hand, the reduction processes usually involved a key stage using a low heating rate (e.g. <1 °C min−1) and involved a holding time (e.g. 2 h) at the final temperature. Furthermore, only a small amount (a few grams) of β-Mo2C catalysts can be produced. Thus, developing a facile method for large-scale synthesis was vital to promote the industrialization of β-Mo2C catalysts.
Herein, a facile, fast and large-scale synthesis of β-Mo2C catalysts was demonstrated and their catalytic activities were evaluated for DRM. The current approach was conducted just with a simple solid mixture of (NH4)6Mo7O24·4H2O and carbon nanotubes (CNTs) at an elevated temperature under an inert atmosphere and employed a fast heating rate (heating rate = 10 °C min−1) method without holding time and there was no need for passivating the as-prepared samples. In view of the simple preparation of precursors and solid–solid reaction without significant gas diffusion limitations, about 100 grams of β-Mo2C sample can be obtained with one reaction. More significantly, this type of monometallic β-Mo2C/CNT catalyst exhibited an even higher catalytic stability than the bimetallic Ni/β-Mo2C catalyst (recently known to be the most efficient carbide catalyst for atmospheric DRM reaction).
In this study, β-Mo2C was prepared simply in two steps. Firstly, the precursors for β-Mo2C were prepared by a mechanical mixing method through co-grinding (NH4)6Mo7O24·4H2O and carbon nanotubes (CNTs) with different Mo contents. Secondly, the precursor (∼110 g) was heated in a quartz reactor with an inner diameter of 80 mm from room temperature (RT) to 850 °C at a rate of 10 °C min−1 under an Ar flow (300 ml min−1), followed by cooling to RT under Ar. A series of products were obtained from the precursors with Mo contents of 5, 15, 30, 60 and 80 wt%, designated as MoC-c (c = 5, 15, 30, 60 and 80). In addition, the bulk β-Mo2C and bimetallic Ni/β-Mo2C (Ni/Mo = 1/2) were prepared according to previous studies,27,28 and the former was designated as MoC-100.
Table 1 lists the textural properties of various samples obtained in this study. It was clear that the β-Mo2C/CNT composites showed different microstructures in comparison with bulk β-Mo2C. Furthermore, SBET, DP and VP were found to vary linearly with the Mo content. This suggested that the interaction between β-Mo2C and CNTs was dependent on the Mo content. Accordingly, the oxidation resistance of β-Mo2C might change with the Mo content.
Sample | Mo content (wt%) | S BET (m2 g−1) | D P (nm) | V P (cm3 g−1) |
---|---|---|---|---|
S BET = BET surface area; DP = average pore diameter; VP = total pore volume. | ||||
CNTs | 0 | 99.7 | 59.24 | 1.48 |
MoC-5 | 5 | 96.8 | 57.16 | 1.24 |
MoC-15 | 15 | 80.7 | 55.46 | 0.97 |
MoC-30 | 30 | 66.0 | 53.57 | 0.81 |
MoC-60 | 60 | 43.6 | 52.68 | 0.67 |
MoC-100 | 100 | 21.9 | 7.62 | 0.04 |
In order to further investigate the effect of the Mo content on the oxidation resistance of β-Mo2C in the carbided 5–60 wt% Mo/CNT samples, XPS analyses are conducted on the selected MoC-5, MoC-30 and MoC-60 samples. Fig. 3 shows the spectra of Mo 3d levels for MoC-5, MoC-30 and MoC-60 samples. By means of curve fitting, the corresponding distribution and concentration of Mo species are listed in Table 2. As shown in Fig. 3 and Table 2, the MoC-5, MoC-30 and MoC-60 samples displayed three surface Mo species. The peaks at a binding energy of 228.4 eV were assigned to the Mo 3d5/2 of Mo2+ (carbide),36,37 and the peaks at Mo 3d5/2 binding energies of 229.0–229.2 and 232.4–232.7 eV were due to Moδ+ (oxycarbide)38,39 and Mo6+ (oxide),39 respectively. These results indicated that more or less Mo carbides had been oxidized to Mo oxides or oxycarbides on the surfaces. It can be seen from Table 2 that the concentration of Mo2+ carbide (Mo6+ oxide) first increased (decreased) and then decreased (increased) with increasing Mo content from 5 to 60 wt%. Based on the data listed in Table 2, the oxidation resistance of β-Mo2C ranked in the order of MoC-30 > MoC-5 ≈ MoC-60.
Sample | Mo content (wt%) | Mo 3d5/2 (eV) | ||
---|---|---|---|---|
Mo2+ carbide | Moδ+ oxycarbide | Mo6+ oxide | ||
Binding energies (eV) and concentration (%) of Mo 3d5/2 species. | ||||
MoC-5 | 5 | 228.4 (29%) | 229.2 (30%) | 232.7 (41%) |
MoC-30 | 30 | 228.4 (40%) | 229.0 (31%) | 232.5 (29%) |
MoC-60 | 60 | 228.4 (32%) | 229.1 (25%) | 232.4 (43%) |
It was believed that the particle size effect and the interaction between particles and carbon materials were important factors for the tuning of the redox properties of particles.40 It can therefore be deduced that the two factors could affect the oxidation resistance of β-Mo2C. Fig. 4 shows the TEM images of the MoC-5, MoC-30 and MoC-60 samples. It was clear that irregular-shaped nanoparticles were well dispersed on CNTs. The insets in Fig. 4 clearly show the crystal lattice of β-Mo2C (100), which proved the existence of β-Mo2C in the corresponding samples. The size distributions of these nanoparticles in the MoC-5, MoC-30 and MoC-60 samples were determined by measuring the particle size directly from the TEM images (Fig. 4). It can be observed from Fig. 5 that the average particle size of β-Mo2C varied with the Mo content. It was 16.6 nm for MoC-5, 23.0 nm for MoC-30, and 56.5 nm for MoC-60, respectively. It was usually accepted that the bigger particles exhibited higher oxidation resistance due to their lower reactivity.40 Therefore, the oxidation resistance of β-Mo2C should follow the order of MoC-60 > MoC-30 > MoC-5 on the basis of the particle size effect. On the other hand, in order to gain insights into the interaction between β-Mo2C and CNTs, the carburization processes of Mo/CNT precursors were investigated by mass spectrometry (MS). In view of the fact that the Mo/CNT precursors were prepared by a mechanical mixing method, it was reasonable to deduce that the stronger the interaction between Mo and CNTs, the easier the formation of β-Mo2C was achieved via carbothermal reduction.
![]() | ||
Fig. 4 TEM images of (a) MoC-5, (b) MoC-30 and (c) MoC-60 samples. The insets show the β-Mo2C (100) crystal lattices. |
As shown in Fig. 6, the temperature for COx evolution increased with the Mo content, indicating that the easiness in the formation of β-Mo2C ranked in the order of MoC-5 > MoC-30 > MoC-60. In other words, the interaction between β-Mo2C and CNTs should follow the same order. The ensemble of the TEM and MS results allowed us to deduce that the MoC-30 probably possessed more appropriate microstructures than MoC-5 and MoC-60, which showed that the β-Mo2C could display higher oxidation resistance. The deduction was in good agreement with the result of XPS (Table 2).
![]() | ||
Fig. 7 Lifetime study of the Mo2C/CNT catalysts with various Mo contents in DRM (T = 850 °C, P = 1 atm, CH4/CO2 = 1, WHSV = 18![]() |
Subsequently, the effect of temperature on the DRM performance over the MoC-30 catalyst was further investigated and the results are shown in Fig. 8. It can be seen that the DRM performance of MoC-30 was greatly dependent on temperature. The catalytic stability was very low at 800 °C, however, it can be enhanced when the temperature rose above 800 °C. It was indicated that 850 °C was a relatively appropriate reaction temperature, at which the oxidation–recarburization cycle (β-Mo2C ⇄ MoO2) can be maintained for a longer time.
![]() | ||
Fig. 8 Dependence of catalytic performance on temperature over the MoC-30 catalyst (P = 1 atm, CH4/CO2 = 1, WHSV = 20![]() |
Finally, the effect of WHSV on the DRM performance over the MoC-30 catalyst was investigated (see Fig. 9). It was clear that WHSV had a vital effect on the DRM performance over the MoC-30 catalyst. On the one hand, the CH4 conversion, CO2 conversion and H2 selectivity increased but the carbon balance decreased with decreasing WHSV before rapid deactivation. On the other hand, it can be observed that the MoC-30 catalysts suffered from rapid deactivation with higher WHSV values. In the case of 24000 cm3 g−1 h−1, the CH4 conversion, CO2 conversion and H2 selectivity decreased respectively from 78%, 88% and 81% to 11%, 21% and 55% within a period of 5 h. As for 18
000 cm3 g−1 h−1, they decreased respectively from 85%, 93% and 82% to 11%, 23% and 50% over a run of 14 h. Fortunately, as the WHSV was decreased to 9000 cm3 g−1 h−1, the MoC-30 catalyst showed a stable performance (83% CH4 conversion, 92% CO2 conversion and 82% H2 selectivity) throughout the test period of 30 h. At the same time, the catalytic performance of the MoC-30 catalyst for DRM was compared with that of the Ni/β-Mo2C catalyst (Ni/Mo = 1/2 was the optimized molar ratio28) under similar conditions. The results clearly indicated that even the monometallic MoC-30 catalyst showed both higher activity and stability than the bimetallic Ni/β-Mo2C catalyst. The next part of this paper was devoted to the characterization of these used MoC-30 catalysts, in order to gain insights into the deactivation mechanism of this new β-Mo2C/CNT catalyst for DRM.
![]() | ||
Fig. 9 The effect of WHSV on the DRM performance over MoC-30 and comparison of catalytic performance of MoC-30 and Ni/β-Mo2C (CH4/CO2 = 1, WHSV = 9000–24![]() |
Fig. 11 shows the Raman spectra of MoC-30 used in DRM with a WHSV of 9000 and 18000 cm3 g−1 h−1 as well as fresh MoC-30 for comparison. In the case of fresh MoC-30, there were two intense bands at ∼1342 and ∼1574 cm−1 due respectively to the disorder band (D-band) and tangential band (G-band) of the carbon species.24 Compared with the fresh MoC-30, the MoC-30-18000 exhibited very weak peaks attributed to the D- and G-bands, but the MoC-30-9000 also clearly showed strong peaks with a similar intensity. The results indicated that the majority of the carbon species on the MoC-30-18000 were removed during the DRM reaction, which suggested that it had suffered from oxidation.
![]() | ||
Fig. 11 Raman spectra of MoC-30 catalysts used in DRM with a WHSV of 9000 and 18![]() |
Finally, the morphologies of MoC-30 used in DRM with a WHSV of 9000 and 18000 cm3 g−1 h−1 were investigated by TEM analysis (Fig. 12). Comparing the TEM images of MoC-30-18000 and MoC-30-9000 samples with that of the fresh MoC-30 sample (see Fig. 4 and 12), there were almost no differences in the shape and size of β-Mo2C particles except that very large MoO2 particles were observed on the MoC-30-18000 sample. It can be concluded from the XRD, Raman and TEM results that the MoC-30-18000 sample was indeed oxidized but there was no sintering of β-Mo2C particles in the DRM reaction. Therefore, it was reasonable to deduce that the deactivation of MoC-30-18000 was just due to the bulk oxidation of β-Mo2C. It was suggested that the oxidation of β-Mo2C by CO2 was still more favourable than the carburization of MoO2 by CH4 over MoC-30 under a high WHSV of 18
000 cm3 g−1 h−1. In other words, a high WHSV meant high oxidant-to-catalyst exposure, which accounted for a short lifetime of the MoC-30 catalyst. When the WHSV decreased to 9000 cm3 g−1 h−1, the rate of β-Mo2C oxidation can be matched with that of MoO2 carburization and no obvious structural change of MoC-30 was observed by various characterization techniques (Fig. 10–12), which made maintaining a long-time stable activity of β-Mo2C possible (see Fig. 9).
![]() | ||
Fig. 12 TEM images of MoC-30 catalysts used in DRM with a WHSV of (a) 18![]() |
![]() | ||
Fig. 13 Variations of the areas of CO2 and CO chromatographic peaks (normalized, taking into account the relative molar response of the TCD) during CO2-TPO. |
Shown in Fig. 14 are the variations of the areas of CH4 and H2 chromatographic peaks during CH4-TPSR of MoC-5, MoC-30, MoC-60, MoC-100 and CNTs. It was generally accepted that the amount of CH4 consumption and H2 formation increased with increasing temperature, which was due to catalytic cracking of CH4 on these samples. In addition, over MoC-100 there was a consumption peak of CH4 observed at 750 °C accompanied by the formation peak of H2, which should be a result of reactivity of CH4 with the uncarburized Mo species, as suggested before.27,28,42 Based on the peak areas of CH4 and H2, it can be deduced that the β-Mo2C on CNTs was more favorable for the dissociation of CH4 than bulk β-Mo2C and the CH4 dissociation ability of Mo2C/CNTs followed the order: MoC-30 > MoC-60 > MoC-5.
![]() | ||
Fig. 14 Variations of the areas of CH4 and H2 chromatographic peaks (normalized, taking into account the relative molar response of the TCD) during CH4-TPSR. |
In order to give a supplementary explanation for why the MoC-30 showed a better catalytic stability at lower WHSV (see Fig. 9), the conversion rates of CH4 and CO2 at different WHSV values were determined by CH4-TPSR and CO2-TPO studies. It can be seen from Table 3 that although R1 and R2 increased with the increase of WHSV value, the ratio of R1 to R2 decreased with WHSV. The results gave proof that high oxidant-to-catalyst exposure was obtained at a high WHSV, leading to a short lifetime of the MoC-30 catalyst.
WHSV (cm3 g−1 h−1) | R 1 (mol g−1 s−1) | R 2 (mol g−1 s−1) | R 1/R2 |
---|---|---|---|
R 1 = CH4 conversion rate, R2 = CO2 conversion rate. | |||
15![]() |
9.24 × 10−7 | 3.54 × 10−6 | 0.26 |
30![]() |
1.51 × 10−6 | 6.73 × 10−6 | 0.22 |
Correlating the CO2-TPO, CH4-TPSR, XRD and XPS results with the catalytic performance data allowed us to deduce that it was possible to enhance the CH4 dissociation ability and oxidation resistance of β-Mo2C by compositing with CNTs. Among these Mo2C/CNT catalysts, the β-Mo2C in MoC-30 showed the highest CH4 dissociation ability and oxidation resistance, which can greatly promote the oxidation–recarburization cycle. Thus, a highly stable DRM performance can be obtained on the MoC-30 catalyst.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7qi00532f |
This journal is © the Partner Organisations 2018 |