Jiyu
Kim
ab,
Kyoung Deok
Kim
ac,
Unho
Jung
*a,
Yongha
Park
a,
Ki Bong
Lee
*b and
Kee Young
Koo
*ad
aHydrogen Research Department, Korea Institute of Energy Research (KIER), 152 Gajeong-ro, Yuseong-gu, Daejeon 34129, Republic of Korea. E-mail: kykoo@kier.re.kr; uhjung@kier.re.kr; Fax: +82-42-860-3739; Tel: +82-42-860-3192 Tel: +82-42-860-3074
bDepartment of Chemical & Biological Engineering, Korea University, 145 Anam-dong, Seongbuk-Gu, Seoul 02841, Republic of Korea. E-mail: kibonglee@korea.ac.kr; Tel: +82-2-3290-4851
cGraduate School of Energy Science and Technology, Chungnam National University(CNU), 99 Daehak-ro, Yuseong-gu, Daejeon 34134, Republic of Korea
dAdvanced Energy and System Engineering, University of Science and Technology (UST), 217 Gajeong-ro, Yuseong-gu, Daejeon 34113, Republic of Korea
First published on 12th January 2024
Unlike most H2 production methods, the decomposition of NH3 does not result in carbon dioxide emission and is therefore classified as clean technology. Thus, NH3 holds great promise for the large-scale transportation and storage of H2, and efficient low-temperature NH3 decomposition catalysts are highly sought after. Herein, we examined the textural properties and NH3 decomposition performances of zeolite 13X-supported Ni catalysts prepared by ion exchange, deposition precipitation, and incipient wetness impregnation. The main surface species were identified as Ni phyllosilicates (ion exchange), NiO + Ni phyllosilicates (deposition precipitation), and NiO (impregnation). Compared to other catalysts, the catalyst produced by deposition precipitation achieved the highest H2 formation rate (22.9 mmol gcat−1 min−1 at 30000 mL gcat−1 h−1, 600 °C) and exhibited a 30–40 °C lower nitrogen desorption temperature. Given that nitrogen desorption is assumed to be the rate-determining step of catalytic NH3 decomposition, this decrease in the desorption temperature was attributed to improved low-temperature performance. Specifically, the excellent performance of the catalyst obtained by deposition precipitation was ascribed to its large specific surface area and strong metal-support interactions due to the high dispersion and uniform deposition of the active Ni metal on the surface and in the pores of the zeolite support.
d = kλ/(βcosθ) | (1) |
Specific surface areas and pore distributions were measured using the Brunauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) methods (BELSORP-MAX, MicrotracBEL Corp.). The sample (100 mg) was heated in a vacuum at 300 °C for 24 h, and N2 sorption isotherms were then measured at −196 °C. Samples were reduced at 700 °C for 1 h in a 10% H2/N2 gas flow before analysis. The distribution of meso- and micropores was determined using BJH and MP plots.
The catalyst reduction temperature and active metal–support interactions were probed using H2-temperature programmed reduction (H2-TPR; BELCAT-B, MicrotracBEL Corp.). The sample (50 mg) was heated in a flow of Ar at 400 °C, cooled to 50 °C, and heated to 1000 °C at a rate of 10 °C min−1 in 10% H2/Ar.
Active metal dispersion was examined using H2 chemisorption (BEL-METAL-3, MicrotracBEL Corp.). The sample (50 mg) was reduced in pure H2 at 700 °C for 1 h and then cooled to 50 °C under Ar. Pulse adsorption was conducted at 50 °C in 10% H2/Ar until saturation. The Ni dispersion state and surface area were estimated from the amount of adsorbed H2, assuming an adsorption stoichiometry of H/Nis (surface nickel atom) = 1.0.
The composition of the catalyst surface and the oxidation states of the constituent elements were examined using X-ray photoelectron spectroscopy (XPS; Sigma Probe, Thermos VG Scientific) in ultrahigh vacuum (10−9 Torr) using an Al Kα (1486.7 eV) X-ray source. The baseline was adjusted using the Shirley background, and all spectra were calibrated using the peak of adventitious carbon at 284.8 eV.
The shape and distribution of Ni particles were examined by transmission electron microscopy (TEM; JEM-F200, JEOL) at 200 kV. The distributions of Al, Si, O, Na, and Ni in the catalyst matrix were probed by energy-dispersive X-ray spectroscopy (EDS). Samples were dispersed in ethanol by ultrasonication, and a drop of the resulting suspension was placed on an ultrathin holey carbon grid and allowed to evaporate. The ultra-microtome (Powertome, RMC) cutting technique was used to examine cross-sections of the Ni/13X catalysts.
NH3-temperature programmed surface reaction and temperature programmed desorption (NH3-TPSR and NH3-TPD, BELCAT-B, MicrotracBEL Corp.) measurements were conducted to study catalyst surface acidity and NH3 decomposition performance. The catalyst was reduced in a flow of 10% H2/Ar at 700 °C for 1 h, cooled to 50 °C in a flow of He (50 sccm), held in an atmosphere of 10% NH3/He at 50 °C for 1 h, maintained in a flow of He (50 sccm) for 1 h to desorb weakly physically adsorbed NH3. And then the catalyst was heated from 50 to 800 °C at a rate of 10 °C min−1 under 10% NH3/He gas and He gas in NH3-TPSR and NH3-TPD, respectively. Residual NH3 (m/z = 17), H2 (m/z = 2), and N2 (m/z = 28) desorption were analyzed by mass spectrometry (BELMASS, microtracBEL Corp.). Pyridine-adsorbed diffuse-reflectance infrared Fourier transform spectroscopy (DRIFTS) was used to analyze the acidic sites on the catalyst surface. The sample was heated at 400 °C for 1 h to remove moisture, treated with 1 vol% pyridine/He at 50 °C for 1 h, and purged with He flow to desorb the weakly bound pyridine. Spectra were recorded using a step size of 4 cm−1 on a Nicolet iS50 FTIR spectrometer (Thermo Fisher Scientific) equipped with a focal plane array detector and a praying mantis DRIFTS attachment (Harrick) with a ZnSe window in the reaction chamber. For each analysis, 128 scans were performed on average.
XNH3 (%) = (Fin − Fout)/Fin | (2) |
H2 formation rate (mmol gcat−1 min−1) = [(Fin/22.4) × XNH3 × 1.5]/mcat | (3) |
k = Aexp(−Ea/RT) | (4) |
TOF (s−1)= (H2 formation rate × mass of catalyst)/(Ni dispersion × Ni contents) | (5) |
Zeolite 13X | Ni/13X-IE | Ni/13X-DP | Ni/13X-IMP | |
---|---|---|---|---|
a Determined by ICP-MS. b Estimated from H2-Chemisorption at 50 °C. c Estimated from XRD. d Estimated from the N2 adsorption at −196 °C. e Estimated from the integration of H2-TPR peaks. | ||||
Ni content (wt%)a | — | 14.5 | 14.8 | 14.2 |
Na content (wt%)a | 4.27 | 1.54 | 2.85 | 4.20 |
Metal dispersion (%)b | — | 0.96 | 2.71 | 0.59 |
NiO crystallite size (nm)c | — | 6.8 | 9.0 | 25.8 |
Ni0 crystallite size (nm)c | — | 27.3 | 21.9 | 36.4 |
Surface area (m2 g−1)d | 919 | 457 | 563 | 493 |
Pore volume (cm3 g−1)d | 0.38 | 0.26 | 0.48 | 0.23 |
Average pore diameter (nm)d | 1.62 | 2.32 | 3.42 | 1.88 |
H2 consumption (mmol g−1)e | — | 1.28 | 1.50 | 2.40 |
Fig. 2 presents the XRD patterns of catalysts recorded after calcination and reduction, in which NiO and Ni peaks corresponded with reference ICDD #44-1159 and #04-0850. The patterns of the calcined samples showed peaks of NiO at 2θ = 37.3°, 43.3°, and 62.9°, corresponding to the (1 0 1), (0 1 2), and (1 1 0) crystal planes. Upon reduction (700 °C, 20% H2/N2), the patterns of all catalysts indicated the conversion of NiO to Ni0. For the calcined Ni/13X-IE catalyst, the characteristic peak of zeolite 13X was not observed due to the formation of Ni phyllosilicate on the surface of zeolite 13X due to the ion exchange of Ni ions. The broad peak at 2θ = 23° corresponds to amorphous silica, and the Ni phyllosilicate characteristic peaks (2θ = 34.1° and 60.8°) were not clearly observed due to low crystallinity.30 The patterns of reduced Ni/13X-IE and Ni/13X-DP were dominated by Ni peaks and featured weakened and broadened zeolite 13X peaks, whereas that of reduced Ni/13X-IMP featured more pronounced zeolite 13X peaks. When fresh zeolite was reduced, the characteristic pattern of zeolite 13X did not change (Fig. S2†). The size of Ni crystallites was calculated using the Ni (2 0 0) peak; it was found to increase in the order of Ni/13X-DP (21.9 nm) < Ni/13X-IE (27.2 nm) < Ni/13X-IMP (36.3 nm). According to previous studies, the impregnation-based loading of active metals onto zeolites decreases metal dispersion, while deposition–precipitation-based loading improves metal dispersion because of strong metal–support interactions.31 Therefore, the loading of Ni by IE and DP mainly affected bonding within the zeolite pore structure and framework, while IMP mainly resulted in the loading of Ni onto the zeolite surface, and negligible change in the zeolite crystal structure was observed after calcination and reduction. Notably, the Ni/13X-DP catalysts exhibited distinct XRD peak changes before and after reduction. For the reduced Ni/13X-DP catalyst, the zeolite 13X peak disappeared and the peak of amorphous silica was observed. This is because during reduction, the Ni–O–Si bonds on the surface of the particles were cleaved, resulting in a Ni metal phase and amorphous SiO2, and the characteristic peaks of zeolite 13X were not observed. In addition, during reduction at 700 °C, the crystals of Ni metal particles grew and the peak intensity increased. To explain this, the XRD analysis of the Ni/13X-DP catalyst as a function of reduction temperature is shown in Fig. S2.† When only fresh zeolite 13X support was reduced at 700 °C, no change in the characteristic peaks was observed. However, for the Ni/13X-DP catalyst, the characteristic peaks of zeolite 13X disappeared and an amorphous silica peak was observed at reduction temperatures ≥600 °C, and the intensity of the crystal peak of Ni metal increased with increasing reduction temperature.32
The catalyst reduction temperature depends on the location (e.g., support surface and pores) of the loaded active metal.33 In particular, the strong interaction between the support and Ni located in zeolite pores results in a high reduction temperature. Fig. 3 presents the H2-TPR profiles of the three catalysts, which are deconvoluted into three peaks (α, β, and γ). The α peak, observed at ≤400 °C, corresponds to the reduction of NiO located on the external surface of the zeolite and the weak interaction with the same.21 The β peak, observed at 400–600 °C, corresponds to the reduction of NiO located within the pores (e.g., zeolite supercages and sodalite cages) and the stronger interaction with the support than NiO located on the zeolite surface. The γ peak, observed at >600 °C, corresponds to the reduction of Ni2+ located in hexagonal prisms.34 Ni located in hexagonal prisms was considered as present in ion-exchangeable positions. α peaks were observed for all catalysts and had the largest relative area in the case of Ni/13X-IMP (Table S1†). High-temperature γ peaks were observed only for Ni/13X-IE and Ni/13X-DP because Ni in these catalysts was loaded onto the sites previously occupied by Na, thus interacting more strongly with framework Al and Si. Thus, in Ni/13X-DP, Ni was uniformly and efficiently dispersed in the zeolite pore structure, engaging in strong metal–support interactions.22 This agrees with the results of H2-chemisorption measurements.
Fig. 4 presents the TEM images of calcined and reduced catalysts, revealing that calcined Ni/13X-IE featured long lamellar Ni particles, while Ni/13X-DP featured both lamellar and spherical particles. Ni/13X-IMP contained only spherical particles. This finding is consistent with the observation of both zeolite and NiO XRD peaks due to the presence of agglomerated NiO particles on the zeolite surface. A lamellar structure was observed on the surface of the Ni/13X-IE catalyst even after reduction, and although a lamellar structure was observed on the surface of the Ni/13X-DP catalyst, the formation of spherical particles of Ni metal due to reduction was more pronounced. For Ni/13X-IMP, the reduction did not change the particle shape but induced Ni particle agglomeration due to sintering at high temperatures. Therefore, in the case of Ni/13X-IMP, the weak interaction between the support and Ni particles led to their agglomeration, which was reflected in the largest Ni crystallite size and the lowest metal dispersion among the catalysts. Interestingly, the simultaneous detection of Si and Ni in Ni/13X-IE, which maintained its morphology after reduction, indicated the presence of Ni phyllosilicate.29,35,36
Fig. 4 TEM images of (a) fresh zeolite 13X, (b–d) calcined Ni/zeolite 13X catalyst, (e) reduced zeolite 13X, and (f–h) reduced Ni/zeolite 13X catalysts. |
Lehman et al. used a template ion-exchange method to load Ni onto MCM-41 and showed that lamellar Ni(OH)2 or Ni phyllosilicates were present on the surface at or above a certain Ni loading.29 In general, Ni(OH)2 decomposes into cubic NiO crystallites above 197 °C, whereas Ni phyllosilicates are thermally stable. Therefore, the lamellar structures observed on the surface of Ni/13X-IE after reduction at 700 °C were identified as Ni phyllosilicates.37 This conclusion is consistent with the H2-TPR profile of Ni/13X-IE, where the reduction peak was mainly observed at 650 °C. The composition and dispersion of Ni particles on the calcined catalyst surface were examined using EDS (Fig. 5). The lamellar particles observed on the surfaces of Ni/13X-IE and Ni/13X-DP were found to contain Ni. Furthermore, Ni particle agglomeration induced by sintering was observed on the surface of Ni/13X-IMP. Notably, in the case of Ni/13X-DP, lamellar and spherical Ni particles coexisted on the support surface, which agreed with the wide reduction temperature range of this catalyst. To confirm the distribution of Ni deposited inside the pores of the zeolite 13X support, we obtained cross-sectional TEM images of the Ni/13X-DP catalyst with ultra-microtome pretreatment (Fig. S4†). EDS mapping indicated that NiO particles were uniformly distributed inside the pores along with lamella-structured Ni phyllosilicate particles on the surface of the calcined Ni/13X-DP catalyst. This indicates that Ni can be distributed through ion exchange and pore internal deposition. After reduction, Ni metal particles coexisted on the surface and inside the pores of the zeolite 13X support. This is consistent with the H2-TPR results, which exhibited a high reduction temperature distribution of Ni particles with stronger interactions than the zeolite surface. Fig. 6 shows the X-ray photoelectron spectra of reduced catalysts, revealing the presence of Ni 2p3/2 peaks at ∼852 (Ni0), ∼853 (NiO), ∼855 (Ni2+), and 860 eV (broad shake-up satellite peak). The NiO peak was observed only for Ni/13X-DP and Ni/13X-IMP, indicating the partial oxidation of surface Ni during analysis.38 For the Ni/13X-IE catalyst, a significant Ni2+ peak was observed due to the strong chemical bonding and stable structure of Ni–O–Si present on the surface. This is consistent with the H2-TPR results, where a high-temperature reduction peak was observed for the Ni/13X-IE catalyst due to the Ni–O–Si bonds of Ni phyllosilicate formed by ion exchange.32 In the Si 2p spectra, a peak due to a Si–O–Al bonding structure within the zeolite was observed at 102.9 eV for all catalysts. In addition to the Si–O–Al peak, the Si 2p spectrum of Ni/13-IE featured the Ni–O–Si peak of Ni phyllosilicate at 101.6 eV.39 Similarly, to TEM analysis, this finding suggested that the preservation of lamellar particles in Ni/13X-IE, even after reduction at 700 °C, was due to Ni phyllosilicate formation.15,25,40Fig. 7 presents the pyridine-adsorbed DRIFT spectra. Bands corresponding to the pyridine adsorbed on Lewis acid sites (LASs) 19b and 8a were observed at 1441–1453 and 1591–1607 cm−1, respectively. The peak at 1441 cm−1 is due to the pyridine adsorbed on the Al3+ ions adjacent to the Na+ cations of the zeolites, and the same peak as in zeolite 13X was observed in Ni/13X-IMP. Moreover, the peak at 1446 cm−1 observed for Ni/13X-IE was attributed to a band shift caused by the incorporation of Ni2+ through ion exchange with Na+.41 This finding is consistent with the results of ICP-MS and H2-TPR measurements, as some Na+ was ion exchanged for Ni2+ during precipitation. Notably, a peak at 1453 cm−1 was observed for Ni/13X-DP, which was attributed to the presence of LASs due to Al3+ ions present in the tetrahedral sites of the zeolite framework.42 In addition, for Ni/13X-IE and Ni/13X-DP, the 1607 cm−1 band was more intense than the 1591 cm−1 band in the 1591–1606 cm−1 region. Peaks at 1453 and 1606 cm−1 were attributed to LASs caused by the presence of Al3+.43 The band at 1488 cm−1 observed for all samples reflected the presence of both Brønsted-acidic sites (BASs) and LASs. For Ni/13X-IE and DP, surface acidity changes such as a shift in LASs and an increase in the peak intensity of BASs were observed as Ni was ionically exchanged with Na.44,45
Table S2† summarizes the NH3 decomposition performances of previously reported catalysts, revealing that the H2 formation rate of Ni/13X-DP exceeded that of the 15 wt% Ni/MRM-600 catalyst with a similar Ni content (18.4 mmol gcat−1 min−1) and was similar to that of the 25 wt% Ni/rGO catalyst with a higher Ni content (24.8 mmol gcat−1 min−1).47,48 The H2 formation rates of the Ni/SiO2 and Ni/SiO2-AEH catalysts with a SiO2 support were 11.4 and 16 mmol gcat−1 min−1, respectively, which were lower than those of the prepared Ni/13X catalyst. The TOF (s−1) values were calculated using the dispersion of Ni particles determined by H2-chemisorption. The TOF of the prepared catalysts increased in the order of Ni/13X-DP < Ni/13X-IE < Ni/13X-IMP as the dispersion of Ni particles decreased (Table 2). However, the Ni/13X-DP catalyst exhibited high NH3 conversion activity, which contradicted the catalytic activity order based on TOF. Considering the results of previous studies performed under the same reaction conditions (WHSV = 30000 mL gcat−1 h−1, 600 °C), the TOF of 10 wt% Ni/SiO2 (metal dispersion = 0.9%) prepared by the precipitation method49 and that of 9 wt% Ni/BN (metal dispersion = 1.1%)50 were 12.4 and 16.0 s−1, respectively, which were higher than those of the Ni/13X-DP catalyst (metal dispersion = 2.71%); however, the NH3 conversions of these catalysts (36.4% and 48.1%, respectively) were lower. In addition, previous studies51,52 reported that the catalytic activity improved while the TOF tended to decrease with Ni loading. Therefore, the Ni metal dispersion has a major effect on the NH3 conversion. In particular, Ni/13X-DP had a dispersion more than double those of Ni/SiO2 and Ni/BN, indicating that the even distribution of Ni active sites on the support surface improved the NH3 conversion. Fig. 8b shows the NH3 conversions obtained at 550 °C and different WHSVs, revealing that conversion decreased as the WHSV increased from 6000 to 45000 mL gcat−1 h−1, following the order of Ni/13X-DP > Ni/13X-IMP > Ni/13X-IE even at high WHSVs. Fig. 8c shows the Arrhenius plot for NH3 decomposition over Ni/zeolite 13X catalysts. The effect of the catalyst preparation method on the apparent activation energy of NH3 decomposition was determined in the NH3 conversion range of 10–20% at a WHSV of 30000 mL gcat−1 h−1. Ni/13X-DP featured the lowest apparent activation energy (57.0 kJ mol−1), suggesting that the preparation method significantly enhanced catalytic activity. This finding is consistent with the low NH3 decomposition and N2 desorption temperatures of Ni/13X-DP observed by NH3-TPSR.53,54 Therefore, because of its low apparent activation energy and high N2 recombination and desorption rates, Ni/13X-DP exhibited the best low-temperature catalytic activity. To evaluate the stability of the prepared catalysts, we performed NH3 decomposition for 30 h at WHSV = 30000 mL gcat−1 h−1 and 550 °C (Fig. 9). Ni/13X-IE and Ni/13X-DP showed stable activity with no decrease in initial conversion (26.6% and 36.5%, respectively) after 30 h. In the case of Ni/13X-IMP, the initial conversion of 31.8% decreased by 2.6% after 30 h, which was attributed to the low dispersion of Ni metal and its weak interaction with the support.
Footnote |
† Electronic supplementary information (ESI) available: Additional XRD, H2-TPR, TEM-EDS, NH3-TPSR-MS, and NH3-TPD-MS data pertaining to zeolite 13X-based catalysts and performance comparison with previously reported catalysts. See DOI: https://doi.org/10.1039/d3se01426f |
This journal is © The Royal Society of Chemistry 2024 |