Mengze
Xu‡
a,
Juan A.
Lopez-Ruiz‡
a,
Nickolas W.
Riedel‡
a,
Robert S.
Weber‡
a,
Mark E.
Bowden
b,
Libor
Kovarik
b,
Changle
Jiang
c,
Jianli
Hu
c and
Robert A.
Dagle‡
*a
aInstitute for Integrated Catalysis, Pacific Northwest National Laboratory, Richland, WA 99354, USA. E-mail: Robert.Dagle@pnnl.gov
bEnvironmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA 99354, USA
cDepartment of Chemical & Biomedical Engineering, West Virginia University, Morgantown, WV 26506, USA
First published on 6th March 2023
Thermocatalytic decomposition (TCD) of methane produces CO2-free hydrogen and valuable co-product, solid carbon. In this study, a series of NiCu/CNT catalysts, prepared with varying Ni/Cu metal ratios and synthesis methods, were evaluated for methane TCD performance at various temperatures. Catalysts before and after reaction, and properties of the carbon product, were characterized to identify activity–structure relationships. At 550 °C, a 10 wt% Ni/CNT catalyst was active; however, it deactivated within 1 h of reaction at >600 °C. The addition of Cu increased its stability. At temperatures above 650 °C, only the catalyst with Cu loadings above 10 wt% remained active and stable. Catalyst characterization revealed that changes in i) Ni/Cu ratio, ii) metal particle size, and iii) operating temperature are key factors for TCD activity, stability, and carbon coproduct morphology. The carbon co-product is mainly composed of multiwalled carbon nanotubes (MWCNTs) whose morphology changes with Ni/Cu ratio and reaction temperature.
Thermocatalytic decomposition (TCD) of methane (CH4) is an alternative process for converting CH4 into hydrogen (H2) and forms solid carbon as co-product.6,7 TCD requires less energy input than the net reaction of SMR, because TCD is modestly less endothermic (37.5 kJ mol−1 H2vs. 41.3 kJ mol−1 H2).2 Although TCD consumes twice as much CH4 as SMR per unit of H2 produced, it offers a path to generating H2 without incurring the production of any CO2. Thus, it converts a fossil fuel (methane) into H2 that could be employed without increasing emissions of greenhouse gases.
Ni-based catalysts have been extensively investigated for the TCD reaction3,8–10 because they exhibit higher activity (i.e. H2 or carbon yield) than other transition metals such as Fe and Co, and produce carbon fibers or carbon nanotubes (500–700 °C).1,3 A major challenge for Ni-based catalysts is their fast deactivation especially above 600 °C.2,4,11 However, the TCD reaction is thermodynamically favored at higher temperature. Therefore, to enable broader application of TCD, it is critical to identify a catalyst that is stable and active at relatively high operation temperature. In our prior publication we showed, for a series of supported Ni catalysts, how Ni particle size affects both activity and stability.12 We found that methane TCD turnover increases with Ni particle size. Further, larger Ni particles (e.g., >20 nm) are selective toward the formation of CNTs, while small Ni particles (e.g., <10 nm) are selective toward the formation of graphitic carbon layers. The formation of graphitic carbon layers block access to Ni active sites, thus deactivating the catalyst more quickly than when CNTs are produced. Additionally, the catalyst deactivation observed with time-on-stream was found to be due to the fragmentation of Ni particles into smaller particles followed by their encapsulation with graphitic carbon layers.12 Thus, while larger particle sizes were found to enhance stability, deactivation is still problematic.
Use of a secondary metal as a dopant is one approach that can improve catalytic stability.6,13,14 Various bimetallic catalysts have been explored, such as NiCo, NiFe, NiPd, NiMo and NiCu.15–25 Enhanced catalytic stability is generally attributed to alloy formation.9,26 Addition of Pd to Ni-based catalysts, for example, NiPd/Al2O3, can improve catalytic activity and stability because Pd is also active for the TCD reaction.22 Studies of NiCu systems have also been reported.27–31 The effects of Cu can be qualitatively described as follows: 1) doping Cu dilutes the low index sites of Ni (1 0 0 and 1 1 0) that are active for dissociation of CH4 and promotes the formation of (111) sites that are active for carbon filament formation;9,32,33 2) the presence of Cu affects the morphology of produced carbon filaments. For instance, Echegoyen et al.27 prepared Ni/MgO catalysts with different stability by using co-precipitation, impregnation and fusion methods. In all the preparation methods, addition of Cu promoted the production of H2 (yield around 80 vol% close to the thermodynamic limit for 8 h on stream). NiCu/MgO catalysts produced a more ordered graphitic carbon than did the Ni/MgO catalysts. Similarly, preferential formation of larger carbon nanofibers with a broader distribution of diameters was observed over the NiCu/SiO2 catalysts when compared to Ni/SiO2 catalysts.28 González et al.34 studied the effect of adding Cu to unsupported Ni nanoparticles. They inferred that Cu inhibited the sintering of Ni particles and induced the formation of CNTs with a narrow diameter distribution.
In addition to improving catalytic performance at high operation temperatures, commercialization of TCD process also requires facile separation of carbon products during catalyst regeneration. Developing bimetallic catalysts on carbonaceous support promises to enable a cyclic reaction-regeneration process: carbon materials harvested after reaction can be used as the support for successive catalysts.35 In our previous work with NiPd catalysts supported on CNTs, we identified alloys with optimized metal ratios that yielded excellent catalytic performance at reaction of 600 °C and that showed the feasibility of catalyst regeneration.35 Here we explore the use of cheaper Cu to develop NiCu catalysts on a carbonaceous support.
Studies of NiCu catalysts for methane TCD have reported how changes in the NiCu particles during rection as well as the operating temperature play a role in catalyst deactivation.30,31,36–40 However, carefully controlled studies evaluating the effect of systematic changes in catalyst properties on catalyst performance, and especially the properties of the carbon co-product, are lacking. It remains a challenge to rationally design NiCu catalysts for TCD because of the many variables to consider: support materials, catalyst composition, synthesis method, and reaction conditions.23,24,27,28,31,41–47 In the study of a NiCu/SiO2 catalyst at elevated temperatures (600–780 °C) by Li et al.,39 with an increase of temperature, catalyst activity increased but eventually dropped quickly with time on stream at temperatures beyond 700–720 °C. At a lower temperature (<700 °C), it was speculated that CH4 dissociation rate might be slower than carbon diffusion and carbon deposition rates. The unbalanced rates cause catalyst deactivation. Fragmentation and phase separation on NiCu particles occurred during the growth of CNTs, which caused further enrichment of carbon on the particle surface. Similar observations were also reported over NiCu/CNT catalysts with different metal ratios by Lua and co-workers.30,31 Ni separation from the NiCu alloy particles mainly resulted in the deactivation at lower temperatures (<700 °C), whereas fragmentation of the alloy particles contributed to the deactivation with time on stream at higher temperature (>700 °C). In a recent study of NiCu catalysts on different supports (MgO or Al2O3), Cu doping was found to affect the lattice constants and particle size distribution of a solid solution NixCu(1−x) which affect catalytic activity and stability.24
The quality of the carbon products is also critical to the commercialization of the TCD process. Techno-economic assessments emphasize that the cost of H2 production depends strongly on the recovery and sale of solid carbon products.48 Examples of valuable carbon products are graphitic carbon, carbon nanotubes and carbon nanofibers.48,49 Formation of the carbon product is also affected by catalyst properties (e.g., composition, support, crystalline structure, size of active metal particles) and operating conditions (e.g., temperature, pressure, composition of gas feedstock).9,24,50 For instance, a mixture of carbon-encapsulated metal particles, carbon filaments with different diameters and bamboo-shaped carbon were produced on a NiCu/Al2O3 catalyst.25 Selective formation of the bamboo-shaped carbon tubes was achieved under certain conditions (Cu loading, reaction temperature, composition of gas feedstock). Ultimately, the desirable carbon properties (e.g., ID/IG ratio) would be dictated by the end use application(s). However, we lack quantitative evaluations the relationship of carbon properties to catalyst design parameters.
Here, we investigate the role of Cu in 1) catalytic performance (stability and activity) and 2) properties of carbon co-products. In particular, we evaluated the role of Ni/Cu ratio, reaction temperature, and synthesis protocol. We characterized the fresh and spent catalyst via X-ray diffraction (XRD), Raman Spectroscopy, and Scanning Transmission Electron Microscopy (STEM) to correlate trends in catalytic performance with changes in catalyst properties during reaction.
The sequential-impregnated catalysts followed the same protocol as the co-impregnated catalysts (0.771 g Ni(NO3)2·6H2O, 0.0570 g Cu(NO3)2·2.5H2O), but only impregnating one metal at the time and heat treating after each impregnation. The catalyst was first impregnated with Ni followed by Cu impregnation and is denoted as Cu1Ni/HCNT (SI).
Nitrogen (N2) physisorption of the fresh and spent catalysts was conducted on a Quadrasorb EVO/SI Gas Sorption System from Quantachrome Instruments at 77 K. Samples were degassed at 150 °C under vacuum for 12 h. Surface areas were determined using the 5-point Brunauer–Emmett–Teller (BET) method from the adsorption data in the relative pressure range of 0.05–0.3. Metal loadings were determined by inductively coupled plasma optical emission spectrometry (ICP-OES).
XRD patterns were collected using a Rigaku SmartLab SE Bragg–Brentano diffractometer, equipped with a fixed Cu anode operated at 40 kV and 44 mA and a D/Tex Ultra 250 1-dimensional detector. Patterns were collected with a variable divergence slit between 2 and 100° (2θ) at intervals of 0.01°. The composition present, lattice parameters, and crystallite sizes of the crystalline components were determined by Rietveld fitting between 30 and 100° (2θ) using Topas v6 (Bruker AXS) as discussed elsewhere.53 Because of the presence of NiCu alloys containing a range of compositions, we acknowledge that this method could underestimate the crystallite size. The compositions of the metallic phases were estimated from their refined cubic lattice parameters by linear interpolation between Ni (a = 3.5238 Å) and Cu (a = 3.615 Å).
A Micromeritics AutoChem 2920 instrument was used to conduct temperature programmed oxidation (TPO). The samples were first loaded and pretreated at 120 °C for 120 min under He, and then heated to 800 °C at a ramp rate of 5 °C min−1 under 5 vol% O2 in He.
Raman spectra were recorded on a Renishaw InVia Raman microscope with a 532 nm excitation wavelength at 10 mW laser power. Each spectrum was averaged over three scans to characterize the solid carbon co-produced by CH4 TCD.
A FEI Titan 80–300 High-resolution transmission electron microscope (HRTEM) microscope operated at 300 kV and equipped with a CEOS GmbH double-hexapole aberration corrector for the probe-forming lens, energy-dispersive X-ray (EDX) spectroscopy detector was used to determine the morphology of solid carbon co-products, metal particle size, and element distribution before and after reaction. Metal particle size and composition distributions were calculated from HRTEM images by sampling an average of 100 particles.
CH4 conversion, XCH4, was calculated based on the amount of CH4 reacted shown in eqn (1):
![]() | (1) |
Carbon yield YC(t) and the rate of deposition of carbon were calculated as the accumulated weight of carbon per mass of the catalyst based on the CH4 conversion. Typically, the reactions were run for over 14 h or until we achieved total carbon yields of 3.5 gcarbon/gcatalyst (i.e., >80% of the carbon in the spent catalysts was carbon co-product). That accumulation ensured that the characterization of the spent catalyst was representative of the carbon co-product instead of the starting carbon support.
The mole balance of the results reported in this work was between 95 and 100% and was calculated using eqn (2):
![]() | (2) |
![]() | ||
Fig. 1 CH4 conversions at time on stream for NiCux/CNT (x = 0, 0.6, 1, 2, 5, 10, 15) catalysts prepared by solvothermal method at reaction temperature of 600 °C under 30 cm3 min−1 30 vol% CH4 in N2. The background activity of the CNT support was <0.2% CH4 conversion. The carbon yield and carbon deposition rate as a function of time on stream can be found in Fig. S2.† The lines represent a decaying exponential fit and the fitting parameters are shown in Table 1. |
To approximately quantify the deactivation of the catalysts, we borrowed two ideas from the literature. First, we assumed that the carbon accumulated with time on stream, θ, raised to a small power, viz. C ∝ θ0.5, as postulated by Voorhies for catalytic cracking.54 Second, we assumed that the catalyst deactivation could be represented through a poisoning factor, Φ, that would multiply the initial rate of conversion, X0:
X(θ) = X0 × Φ(C(θ)) | (3) |
Φ(C) = exp(−kC) | (4) |
X(θ) = X0 × exp(−kθ0.5) | (5) |
Catalyst | Cu mol fraction | X 0 (%) | k (h-0.5) | θ (h) | Actual carbon yield (gC/gcat) | Predicted carbon yield (gC/gcat) |
---|---|---|---|---|---|---|
Ni/CNT (ST) | 0 | 135 | 3.09 | 0.87 | 0.282 | 0.251 |
NiCu0.6/CNT (ST) | 0.045 | 97 | 1.17 | 5.00 | 0.957 | 1.02 |
NiCu1/CNT (ST) | 0.081 | 63 | 0.309 | 5.00 | 2.52 | 2.61 |
NiCu2/CNT (ST) | 0.142 | 48 | 0.333 | 5.00 | 1.89 | 1.88 |
NiCu5/CNT (ST) | 0.35 | 41 | 0.120 | 5.00 | 2.23 | 2.24 |
NiCu10/CNT(ST) | 0.455 | 47 | 0.110 | 5.00 | 2.59 | 2.63 |
NiCu15/CNT (ST) | 0.553 | 32 | 0.0561 | 5.00 | 1.93 | 1.94 |
NiCu1/HCNT (IW) | 0.086 | 53 | 0.760 | 5.00 | 1.21 | 1.19 |
NiCu1/HCNT (CI) | 0.079 | 60 | 1.20 | 4.00 | 0.725 | 0.731 |
Cu1Ni/CNT (SI) | 0.083 | 54 | 0.0800 | 5.00 | 3.15 | 3.13 |
![]() | ||
Fig. 2 Activity of NiCu1/CNT prepared by different synthesis methods as a function of time on stream at reaction temperature of 600 °C under 30 cm3 min−1 30 vol% CH4 in N2. GHSV ≈ 3000 h−1. The background activity of the raw CNT was <0.2% CH4 conversion. The carbon yield and carbon deposition rate as a function of time on stream can be found in Fig. S2.† The lines represent a decaying exponential fit and the fitting parameters are shown in Table 1. |
The properties of the Ni and Cu metal in freshly reduced catalysts were investigated with XRD (Fig. S1†) to understand the roles that Cu played in the TCD performance and results are summarized in Table S2.† The freshly reduced NiCu catalysts are mainly composed of reduced NiCu alloy nanoparticles. The catalysts with Ni/Cu mass ratios >5 had larger nanoparticles (e.g., 11–14 nm) compared to the catalyst with Ni/Cu mass ratio <2 (e.g., 7.3 to 8.6 nm). In our previous work we showed the crystallite size of monometallic Ni nanoparticle is the dominant factor for TCD activity,53 which also coincides with the activity trends discussed in this work. That is, catalysts with high Ni/Cu mass ratios (i.e., >5) yield metal nanoparticles with larger particle size and higher TCD activity. On the other hand, catalysts with low Ni/Cu mass ratios (i.e., <2) yield smaller metal nanoparticles with lower TCD activity. This relationship between TCD activity and NiCu crystallite size is also consistent with previous reports by Pinilla et al.24 The Ni/Cu ratios derived from XRD are different from the ones obtained via ICP specially at high Ni/Cu ratios (i.e., >5); we infer that the XRD analysis might be excluding some particles from the analysis or underestimating the metal compositions. However, we speculate that the enhanced TCD stability of the catalysts with <2 Ni/Cu ratio might be a direct result of the higher Cu loadings used in the catalysts.
The spent catalysts were also analyzed via XRD to elucidate the role of Cu on catalyst stability and the results are sown in Table S2.† Overall, we observed changes in both crystallite size and re-distribution of metal compositions. All but one of the catalysts showed a loss of Ni on the alloy (i.e., Ni/Cu ratio decreased), most likely resulting from the selective Ni migration from the metal particles to the carbon co-product. The segregation of Cu and Ni from the different Ni/Cu alloy nanoparticles at the reaction temperatures is consistent with the solubility gap regions of the NiCu phase diagram.56 Out of the 7 different Ni/Cu catalyst evaluated, only NiCu1 underwent particle fragmentation as evidenced by the 15% decrease in average crystallite size. The other six catalysts underwent metal sintering as evidenced by the average crystallite size increase between 7% and 100%, Table S2.† Hence, we speculate that the changes in TCD catalytic performance depicted in Fig. 1 are a result of both the changes in Ni/Cu ratio and average crystallite size of the active site, which is also consistent with previous works by Shen et al. and Adeeva et al.30,31
XRD analysis of the spent catalysts revealed an even larger variation in Ni/Cu ratios (19 to 49). The catalysts synthesized by ST and SI methods had the lowest Ni/Cu ratios of 19 (each), while the catalysts prepared by IW and CI had significantly higher Ni/Cu ratios of 32 and 49 (Table S3†). The ST and SI catalysts also had the lowest deactivation rate constants of 0.42 and 0.08. In contrast, the catalysts prepared by IW and CI had higher deactivation rate constants of 0.76 and 1.2 (Table 1). Thus, catalysts composed of bimetallic NiCu metal nanoparticles with significantly higher Cu contents (smaller Ni/Cu ratios) were more stable. Further, while XRD analysis showed that metal particle size increased after reaction for three of the four catalysts, the catalyst prepared by SI had the largest particle size (15.1 nm) and also exhibited the largest increase in growth (from 8.4 nm). This combination of large particle size and low Ni/Cu ratio, compared to the other catalysts, could explain its superior stability. Taken together, the XRD analysis of the spent catalysts reveals that the particle size and Ni/Cu ratio changed with the synthesis method, and this can directly explain differences in catalytic stability.
![]() | ||
Fig. 3 Activity of Ni/CNT (ST), NiCu1/CNT (ST), and NiCu15/CNT (ST) as a function of time on stream at reaction temperatures of 550–700 °C under 30 cm3 min−1 30 vol% CH4 in N2. The background activity of the raw CNT was <0.2% CH4 conversion and was ignored in the curve fitting. The fitting parameters can be found in Table S4.† |
Given the assumed functional form for the deactivation function, the mass of carbon at any time on stream, C(θ), is easily estimated by integrating eqn (5) multiplied by the inlet hourly mass flow rate of carbon, Cfeed and the initial, fitted conversion, X0
![]() | (6) |
![]() | (7) |
The predicted amount of carbon co-product accumulated (i.e., carbon yield) closely tracked the actual amount of carbon measured in the reactor (Fig. 4), lending credence to the chosen functional form. On the basis of the adequate fits of both the instantaneous and the integral conversion, we have calculated the projected carbon yield extrapolated the accumulation of carbon co-product to θ = ∞ normalized by the weight of catalysts:
![]() | (8) |
![]() | ||
Fig. 4 Parity plot of carbon yield calculated as the predicted or actual accumulated carbon co-product normalized by the weight of catalyst used at the time on stream shown in Table 1. |
Those values exhibit a maximum that depends on both the operating temperature and catalyst composition, represented by the mol fraction of Cu (Fig. 5). These results are consistent with previous reports detailing the effect of Cu addition on Ni TCD activity.27–31 The carbon accumulation shows that there is an optimum in Ni/Cu ratio at each operating temperature.
![]() | ||
Fig. 5 Projected carbon yield calculated as projected carbon co-product accumulation of carbon at infinite residence divided by the weight of catalyst used according to eqn (8) as a function of the catalyst composition and operating temperature for catalysts prepared by the solvothermal (ST) method. |
The spent catalysts were analyzed via XRD to elucidate the role of Cu on catalyst stability. The results show that metal particle restructuring might have caused catalyst deactivation at the different reaction temperatures. For example, as shown in Table S3† the metal particles in Ni/CNT remained small at 9.2 nm when operating at 550 °C but sintered to larger metal particle sizes (14.6 to 19.4 nm) when the reaction temperature increased, suggesting that a cause of deactivation was metal sintering. NiCu1/CNT maintained small metal particle sizes and TCD activity at 550 and 600 °C, which is consistent with the small change in metal particle size and Ni/Cu ratio observed with respect to the fresh material. However, NiCu1 deactivated at 650 °C in less than 1 h while the metal particles sintered and segregated Cu into a secondary Cu-rich alloy. At 700 °C, NiCu1 was not active for TCD and the composition and particle size remained similar to that of the fresh catalyst, suggesting that the catalyst deactivated before metal restructuring occurred.
We speculate that the deactivation at higher temperatures might be caused by selective formation of graphitic carbon and subsequent plugging of the active site as we previously reported.57 Interestingly, NiCu15/CNT catalyst had similar metal particle size (17.2 to 21.4 nm) at 550, 600, and 650 °C for which the catalyst was active and stable; however, the Ni/Cu ratio of the metal particles changed with reaction condition. These results suggest there was a preferential segregation of Ni out of the metal particle at 550 °C by the change in Ni/Cu ratio with respect of the fresh catalyst (0.133 and 0.790 respectively). At 600 and 650 °C, the Ni/Cu ratio increased (0.254 and 0.418) suggesting that the segregation of Ni was less at higher reaction temperatures. At 700 °C, the Ni/Cu ratio of the metal particles was similar to that of the fresh catalyst (0.676 and 0.790), further corroborating that the Ni segregation is less at higher temperatures; however, the metal particle size only stabilized to 10.4 nm (from 8.60 nm on the fresh catalyst) as opposed to the larger metal particle size observed at the lowest temperatures. We speculate that the slow deactivation at 700 °C was caused by the poisoning of active sites before they could stabilize to the preferential particle morphology (i.e., <0.254 Ni/Cu ratio and >17 nm). Hence, these results suggest that the role of Cu is to stabilize large metal particles (>17 nm); however, increasing the reaction temperatures cause metal migration and changes the Ni/Cu ratio below the required to stabilize the large metal particles. The properties of the carbon co-product form might also play a role on the catalyst stability, which is explored in the following section.
Fig. 6 shows the STEM images of selected samples after reaction at 600 °C revealing the selective formation of CNT as carbon co-product as well as the morphology and size of the metal particles. Overall, all the fresh catalysts are composed of a wide range of metal particles between 10 and 20 nm as XRD suggested; however, we found that the spent catalyst was composed of larger >50 nm (as well as <20 nm) particles for the three different compositions, which suggests that metal sintering took place during the TCD reaction. The XRD analysis summarized in Table S2† also showed metal particle sintering but did not capture the formation of the larger metal nanoparticles. We speculate that the larger metal nanoparticles observed by STEM are domains of multiple smaller crystals, which explains why the XRD analysis did not fully capture them. Fig. S6 and S5† are EDS maps of the catalyst after reaction and show that Ni and Cu remain in the metal particle regardless of the metal particle size; however, there appear to be changes in the Ni/Cu ratios of the spent materials and formation of Ni-rich and Cu-rich particles as revealed by the XRD analysis. More importantly, the HAADF STEM micrographs of the spent catalysts also reveal the selective formation of CNTs as the main solid co-product on the large metal nanoparticles. Fig. 7 shows the differences in morphologies of CNT produced by the different catalysts highlighting the role that Ni/Cu ratio played.
![]() | ||
Fig. 6 High-angle annular dark-field (HAADF) imaging using a scanning transmission electron microscope (STEM) of selected catalyst before (fresh) and after reaction (spent) at 600 °C under 30 cm3 min−1 30 vol% CH4 in N2. Associated elemental maps obtained with Energy-Dispersive Spectroscopy (EDS) can be found in Fig. S6 and S7.† |
![]() | ||
Fig. 7 Scanning transmission electron microscope (STEM) images of selected catalyst after reaction (spent) at a–c) 600 °C and d) 700 °C under 30 cm3 min−1 30 vol% CH4 in N2. Additional STEM images can be found in Fig. S8 and S9.† |
The addition of 1 wt% Cu (i.e., NiCu1/CNT) produced multiwall CNT carbon co-product, which had a wall diameter ≈5 nm. Further increasing the Cu loading resulted in the formation of larger Ni/Cu metal particles, which generated large CNT with wall thickness >10 nm. Fig. S8 depicts more STEM images of spent NiCu15/CNT at 600 °C. The changes in carbon co-product morphology depicted in this work as a function of metal particle size is consistent with previous reports.25
As we previously showed,35 Raman spectroscopy can be used to assess the carbon quality using the three main bands: a) D-band (1340 cm-1) associated with defects in the graphitic lattice,59,60 2) G-band (1580 cm−1) associated with ordered carbon,59,60 and 3) G'-band (or 2D band, 2700 cm−1) associated with interactions between stacked graphene layers that can be used to distinguish between single-wall CNTs (SWCNTs) or multi-wall CNTs (MWCNTs).59,60,61Fig. 8 shows the ID/IG and IG′/IG ratios of the spent catalysts compared to the pristine support (MWCNT Support), pristine support run under reaction conditions, and fresh catalysts. The results reveal a direct correlation with the Ni/Cu ratio and the ID/IG and IG′/IG ratios of the generated CNT. For example, the ID/IG ratio was similar for the as received MWCNT (1.11) and MWCNT after exposed to reaction conditions (0.993) as well as the fresh catalysts (1.05 to 1.0); however, the spent catalysts showed a different ID/IG ratio suggesting that the CNT generated during TCD had different properties to that of the starting support. With the exception of Ni/CNT (ST) and NiCu1/HCNT (CI), the rest of the catalysts had a similar final carbon deposited per gram of catalyst (>4 gcarbon co-product/gcarbon support); hence, the changes in ID/IG and IG′/IG ratios as a function of Ni/Cu composition can be attributed to changes in the morphology of the deposited carbon co-product. Ni/CNT was only active for 1 h at 600 °C and there was little carbon deposited (∼1 gcarbon co-product/gcarbon support), hence the Raman features is similar to that of the CNT support. As the Ni/Cu ratio decreases (i.e., higher Cu loadings), the ID/IG ratio increases suggesting that the produced carbon co-product has a higher defect density, which we speculate is caused by the larger diameter and wall thickness of the CNT co-product. The IG′/IG ratio decreases with the increase in Cu loading (i.e., decrease in Ni/Cu ratio) which is consistent with the presence of multiwall CNT (multiwall CNT) with higher number of walls. The Raman observations are consistent with the (HAADF) STEM images which showed the formation of multiwall CNT with wider diameter (and wall thickness), Fig. 6 and 7. The larger diameter multiwall CNT were formed due to the restructuring of metal into larger domains and crystallite, which is also consistent with the XRD analysis.
The amount of Cu addition to Ni can therefore be directly correlated to both catalyst stability and quality of the carbon as evidenced by Raman spectroscopy. Fig. 9 compares both the ID/IG ratio of the carbon product and the deactivation rate after reaction at 600 °C as a function of Cu mol fraction. A small increase in Cu mol fraction (e.g. 0.081; NiCu1/CNT) significantly reduced the deactivation rate constant from 2.77 to 0.42 h−0.5 (85% decrease), while increasing the ID/IG ratio from 1.00 to 1.23 (23% increase). However, further increase in Cu loading had less effect on deactivation rate relative to the ID/IG ratio. For example, by nearly doubling the Cu mol fraction from 0.081 to 0.142 little change in deactivation rate was observed (0.48 versus 0.42). However, the ID/IG ratio increased from 1.23 to 1.41. The ID/IG ratio reached a plateau of ∼1.9 with Cu mol fractions between 0.35 and 0.55.
![]() | ||
Fig. 9 Deactivation rate constant and Raman ID/IG ratio of resulting carbon product for NiCux/CNT (x = 0, 0.6, 1, 2, 5, 10, 15) catalysts prepared by solvothermal method at reaction temperature of 600 °C under 30 cm3 min−1 30 vol% CH4 in N2. The deactivation rate constants and ID/IG ratios are taken from Table 1 and Fig. 8, respectively. |
As shown in Table S5† the crystallite size (10.4 and 8.6 nm) and Ni/Cu molar ratio (0.676 and 0.790) of the spent NiCu15/CNT remained similar to that of the fresh catalyst suggesting that a fraction of the metal nanoparticles deactivated before restructuring into the larger crystallite needed to catalyze the stable TCD reaction. HAADF-STEM images of the NiCu15/CNT catalyst run at 700 °C confirms that the spent catalyst still had sections of intact ≈10 nm NiCu metal particle, Fig. S7.† This is also consistent with our previous work showing that >15 nm particles are needed to catalyze the TCD reaction and selective CNT formation; however, ≤10 nm metal particles deactivate due to the preferential formation of graphitic layers (and metal particle blockage) as opposed to CNT.12Fig. 7 and S7b† confirmed that >50 nm nanoparticles were still formed at 700 °C which explained the initial TCD. STEM images of the spent NiCu15/CNT at 700 °C reveal that the carbon co-product formed on the >50 nm metal particles was primarily multiwall CNT with wall thickness >14 nm. Hence, Cu serves dual purposes for the TCD reaction; 1) modifies the carbon co-product morphology to multiwall CNT and 2) causes the metal particles to restructure into larger metal particles, which are more stable for TCD at higher temperatures and selective towards the formation of multiwall CNT.
Characterization of the catalysts after reaction revealed that the addition of Cu to Ni results in the stabilization of larger metal nanoparticles, which are more stable for TCD at higher reaction temperatures and more selective towards CNT growth. In the absence of Cu, bamboo-shaped CNT was the main morphology observed and the addition of Cu resulted in a change in CNT morphology to multiwall CNT. In all cases, we observed the active site restructuring by the preferential segregation of Cu out of the NiCu alloy (i.e., the Ni/Cu ratio increases) and change in metal particle size. At low Cu loadings (i.e., high Ni/Cu ratio) the catalyst deactivated at reaction temperatures >600 °C due to the encapsulation of the metal active sites before restructuring and loss of Cu. At high Cu loadings (i.e., low Ni/Cu ratio), reaction temperatures between 550 and 650 °C caused metal particle restructuring into <17 nm particles with higher Ni/Cu loading compared to the fresh catalyst, which resulted in the preferential multiwall CNT formation. Operation at 700 °C caused encapsulation of metal particles before being able to restructure, resulting in catalyst deactivation.
Hence, this work highlights how catalyst composition and operation conditions can be used to optimize catalyst stability and yield different carbon co-product morphologies generated during methane TCD.
Footnotes |
† Electronic supplementary information (ESI) available: Characterization results for the fresh and spent catalysts with different Ni/Cu ratio. Characterization results for the fresh and spent catalysts prepared with different synthesis methods. Characterization results of spent catalysts with different Ni/Cu ratios run at different temperatures. See DOI: https://doi.org/10.1039/d2cy01782b |
‡ Equal contributors. |
This journal is © The Royal Society of Chemistry 2023 |