RETURN TO ISSUEPREVResearch ArticleNEXT

Enhancing CexZr1–xO2 Activity for Methane Dry Reforming Using Subsurface Ni Dopants

  • Behnam Safavinia
    Behnam Safavinia
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • Yuming Wang
    Yuming Wang
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
    More by Yuming Wang
  • Changyi Jiang
    Changyi Jiang
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • Cameron Roman
    Cameron Roman
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • Pragathi Darapaneni
    Pragathi Darapaneni
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • Jarod Larriviere
    Jarod Larriviere
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • David A. Cullen
    David A. Cullen
    Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, United States
  • Kerry M. Dooley
    Kerry M. Dooley
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
  • , and 
  • James A. Dorman*
    James A. Dorman
    Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States
    *Email: [email protected]
Cite this: ACS Catal. 2020, 10, 7, 4070–4079
Publication Date (Web):March 16, 2020
https://doi.org/10.1021/acscatal.0c00203
Copyright © 2020 American Chemical Society
ACS AuthorChoicewith CC-BYlicense
Article Views
2149
Altmetric
-
Citations
LEARN ABOUT THESE METRICS
PDF (2 MB)
Supporting Info (1)»

Abstract

CeO2–ZrO2 (CZO) nanoparticles (NPs) have applications in many catalytic reactions, such as methane dry reforming, due to their oxygen cycling ability. Ni doping has been shown to improve the catalytic activity and produces active sites for the decomposition of methane. In this work, Ni:CZO NPs were synthesized via a two-step co-precipitation/molten salt synthesis to compare Ni distribution, oxygen vacancy concentration, and catalytic activity relative to a reference state-of-the-art catalyst prepared by a sol–gel-adsorptive deposition technique. To better understand the dispersion of Ni and oxygen vacancy formation in these materials, the Ni concentration, position, and reaction time were varied in the synthesis. X-ray diffraction (XRD) measurements show a homogeneous, cubic phase with little to no segregation of Ni/NiO. Catalytic activity measurements, performed via a differential scanning calorimetry (DSC)/thermogravimetric analysis (TGA) method, displayed a 5-fold increase in the activity per surface area with an order of magnitude decrease in the coking rate for the particles synthesized by the molten salt method. Additionally, this approach resulted in an order of magnitude increase in oxygen vacancies, which is attributed to the high dispersion of Ni2+ ions in the NP core. This ability of controlling the oxygen vacancies in the lattice is expected to impact other such systems, which utilize the substrate redox cyclability to drive conversion via, e.g., a Mars–van Krevelen mechanism.

1. Introduction

ARTICLE SECTIONS
Jump To

One of the major problems faced in Ni-doped CeO2–ZrO2 (CZO) catalysts is sintering and coke formation at dry reforming of methane (DRM) operating conditions, 650–850 °C and 1–10 atm. (1−4) Homogeneously dispersing the active sites over the surface of the catalyst has been shown to prevent coking because of the absence of faceted Ni surfaces. (2,5−10) Alternatively, coke formation is reduced on materials with high oxygen storage capacities (OSCs) due to lattice oxygen interactions with adsorbed carbon moieties. The OSC of CZO is known to be a function of the Zr concentration, peaking at around 30 mol % Zr. (11−18) As such, cerium-based catalysts have been used in a variety of lattice oxygen-mediated catalytic reactions, e.g., three-way exhaust catalysis, (19) dry(tri) reforming of methane, (20,21) water–gas shift reactions, (22) and others. (12) Additionally, other transition metals (Pd, Pt, Rh, Mn, etc.) have been studied in combination with the CZO substrate to lower the reaction onset temperatures, along with other effects. (23−29) However, with all of these materials, there is the synthetic challenge to maximize the OSC and active site dispersion while maintaining catalyst stability with respect to both metal ripening and coking. One relatively unexplored approach is the subsurface incorporation of the Ni dopant to increase internal strain without the formation of faceted islands.
Accordingly, there is a large body of literature on the preparation of CZO nanoparticles (NPs) using various synthetic methods such as sol–gel, (30,31) co-precipitation (CP), (9) flame-made, (32)etc., which yield different oxygen vacancy concentrations despite containing identical compositions. (33−36) To achieve an ideal Ce/Zr ratio of ∼2:1 for maximum OSC, the synthetic conditions must be carefully controlled to ensure local and long-range homogeneity due to the different formation rates of the oxides. (2,6,37) Additionally, post-processing techniques (e.g., strong electrostatic adsorption or incipient wetness impregnation) are often used to activate the catalysts with Ni. However, high temperatures and oxidative/reducing atmospheres can lead to structural changes and reduced OSC. (10,38) Specifically, exposure to reducing atmospheres often leads to the formation of segregated CeO2 and ZrO2, as well as nonuniform distribution of Ni. (10) This process results in larger Ni clusters that can facilitate dendrite formation and rapid deactivation of the catalyst. (9,10,39) Therefore, alternative methods are needed to synthesize activated NPs that are less susceptible to oxide segregation and Ni island formation.
To address this, others have incorporated Ni salts (up to 15 mol %) during the Ce–Zr–O co-precipitation process prior to a high-temperature sintering step, producing Ni cluster sizes of ∼12 nm on the catalyst surface. (2) However, with this method, it is difficult to spatially control the Ni during incorporation in the bulk of the catalyst and to correlate the role of Ni to the OSC of the catalyst. (40) On the other hand, a two-step co-precipitation (CP)/molten salt synthesis (MSS) method has been reported, which facilitates the incorporation of dopants within complex metal oxides (below the surface) during synthesis. (41−45) By precipitating the precursors from an aqueous solution in the initial step, M–O–M bonds are formed, which then undergo reorganization during the high-temperature treatment to solidify the structure. (46,47) The formation of the disorganized initial structure results in enhanced phase stability and facilitates the incorporation of a homogeneous distribution of dopants. (48,49) Modification of this reported two-step process will allow for controlled Ni distribution in CZO catalysts and for the elucidation of the effects of oxygen mobility/concentration and active site location on catalyst stability, activity, and coking.
In this work, Ni:CZO NPs are prepared by a two-step synthesis, CP followed by MSS, to create a highly dispersed disordered fluorite phase with increased oxygen vacancies. This method allows for the production of structure-controlled NPs where the oxygen vacancies can be modified by varying precursor compositions and reaction time. The oxygen vacancy concentration and catalytic activity of the Ni:CZO NPs prepared via CP/MSS were then compared to a more conventional state-of-the-art catalyst prepared via the sol–gel technique. Structural and chemical characterizations were performed on the synthesized NPs to determine the morphology, stoichiometry, and chemical composition prior to any catalytic reaction tests. The percentage of Ni was systematically varied up to 20 mol % to identify the optimum Ni content for catalytic activity in dry reforming of methane (DRM) and oxygen vacancy concentration. Differential scanning calorimetry (DSC)/thermogravimetric analysis (TGA), Raman spectroscopy, and X-ray photoelectron spectroscopy (XPS) were performed to determine the catalytic activity, Ce oxidation state, and oxygen vacancy concentrations of the NPs. The ability to spatially distribute the Ni within the catalyst resulted in a near 5-fold increase in activity per surface area and an order of magnitude decrease in coking, suggesting that subsurface active sites play an important role in the catalytic activity of methane reforming reactions. While this work only includes a single dopant (Ni), this process can be used to control dopant distribution of multiple types. (50,51)

2. Experimental Section

ARTICLE SECTIONS
Jump To

2.1. Materials

Cerium(III) nitrate hexahydrate (Ce(NO3)3·6H2O, Strem Chemicals, 99.9%), ammonium cerium nitrate (NH4)2Ce(NO3)6 (98+%, Alfa), zirconium oxynitrate hydrate (ZrO(NO3)2·xH2O, Beantown Chemical, 99.9%), nickel chloride hexahydrate (NiCl2·6H2O, Beantown Chemical, 99.0%), 28–30% ammonium hydroxide (NH4OH, VWR Chemicals, ACS grade), and urea (CH4N2O, VWR Chemicals, ACS grade) were used for the synthesis. NaNO3 (high purity grade, VWR Chemicals) and KNO3 (ACS grade, VWR Chemicals) were utilized for the molten salt synthesis. Hydrochloric acid (HCl, 36.5–38%, VWR Chemicals) and nitric acid (HNO3, 68–70%, VWR Chemicals) were used for the inductively coupled plasma optical emission spectrometry (ICP-OES) digestion process.

2.2. Preparation of Ni:CZO NPs

Particles were prepared using a two-step, CP/MSS process. (41−44,52) The CP was performed by dissolving a 2:1 Ce/Zr molar ratio of Ce(NO3)3·6H2O/ZrO(NO3)2·xH2O in 400 mL of deionized (DI) water. Ni was incorporated by adding up to 20 mol % of NiCl2·6H2O in the solution. Actual doping concentrations are reported in Table S1. The pH of the solution was adjusted by adding 28% NH4OH dropwise until the pH reached ∼11. The solution was stirred for 2 h, and the resultant powder was collected through vacuum filtration and washed with DI water until the supernatant was neutral. (53) The resultant powder was dried at 90 °C overnight. In the second step, 700 mg of the as-prepared single-source complex precursor was mixed with a eutectic salt mixture (e.g., NaNO3/KNO3 = 1:1, molar ratio), transferred to a porcelain crucible, and heated at 650 °C with a 10 °C/min ramp rate. The reaction was allowed to occur for up to 6 h. The resulting solution was cooled naturally to room temperature, and the salt crystal dissolved in DI water. After the complete dissolution of the salts, the powder was centrifuged and washed five times with DI water to completely remove excess salts. The particles were then collected and dried overnight at 90 °C. For the synthesis of the core–shell NPs, the core was synthesized via the above CP/MSS synthesis and the shell was added by adsorptive deposition. (54,55) The core NPs along with NiCl2·6H2O were added to 20 mL of 0.3 M of urea and was refluxed for 24 h. After filtering and washing the powders, the sample was reduced by 5% H2/95% Ar (300 mL/min) for 6 h at 750 °C. The reference catalyst was prepared by a templated sol–gel method followed by a deposition–precipitation step. (54) The precursors (NH4)2Ce(NO3)6 and ZrO(NO3)2 were dissolved in 96% water/3% methanol/1% tetramethylammonium hydroxide (TMAOH) surfactant (25% in methanol). The pH was adjusted to 10.3 by gradually adding the ammonia solution, and the solution reacted for 2 days at 90 °C. After washing of the sol–gel with DI water, the mixed oxide was dried at 120 °C overnight and calcined in air for 6 h. To deposit the Ni active by adsorptive deposition, the mixed oxide and Ni(NO3)2·6H2O (Aldrich, reagent) were added to 0.3 M urea and reacted for 24 h at 90 °C. After washing and drying (same as above) was performed, the catalyst was reduced by 5% H2/N2 at 750 °C for 6 h.

2.3. Structural Characterization

All structural characterizations were performed prior to the reduction treatment unless noted in the text. The crystal structure was identified via powder X-ray diffraction (XRD) using a PANalytical X-ray diffractometer at 45 kV and 40 mA. The θ – 2θ radial scan was performed over the range 5–70° with a step size of 0.04° and dwell time of 60 s, using Cu Kα (λ = 1.54 Å) as the radiation source. The morphology and size of Ni:CZO NPs were confirmed by high-resolution transmission electron microscopy (HRTEM) using a 200 kV JEOL NEARM electron microscope equipped with double aberration correctors, a dual-energy-loss spectrometer, and a cold field emission gun (FEG) source. The powder sample was dispersed in ethanol and drop-casted on a 300 mesh, lacey carbon grid prior to imaging. Specific surface area (Brunauer–Emmett–Teller, BET) measurements were carried out in a Micromeritics ASAP-2020 porosimeter (three points). The elemental composition of NPs was measured using a PerkinElmer Optima 8000 ICP-OES spectrometer.
Raman spectra were measured using a Renishaw inVia Reflex Raman spectrometer with a 0.1 mW diode laser at an excitation wavelength of 532 nm, exposure time of 0.5 s, spectral resolution of 1 cm–1, and ∼5 μm spot size.
XPS measurements were performed using a Scienta Omicron ESCA 2SR equipped with a monochromatic Al Kα (hν = 1486.6 eV) X-ray source and a hemispherical analyzer with a 128-channel detector, at 1.3 × 10–9 Torr. The Gaussian width of the photon source was 0.5 eV and the focus voltage was 300 V. The XPS spectra were calibrated to the adventitious C 1s peak at 285.7 eV. All peaks were fit (using CasaXPS software (56)) as Gaussians after Shirley background subtraction.
CO chemisorption was performed to verify the active site dispersion (Ni/NiO) on the surface of the catalyst. The catalyst (500 mg) was loaded and pretreated in 5% H2/95% N2 overnight. CO pulse chemisorption (Micromeritics 2700) at various temperatures was measured after purging the surface with N2. Temperature-programmed reduction (TPR) was performed by reduction of preoxidized (500 °C, 1 h, air) samples in the DSC/TGA. The samples were reduced at 10 °C/min from 50 to 1000 °C in 5% H2/N2. They were then reoxidized at 500 °C to check for weight recovery.

2.4. Catalytic Performance Evaluation

The coking rate and catalytic activity were measured using a differential scanning calorimetry (DSC)/thermogravimetric analyzer (TGA) (TA SDT Q600). The catalyst was activated in air overnight at 750 °C. The DRM reaction was performed at 750 °C and 135 mL/min total flow rate (1:1 CH4/CO2, 0.25 atm partial pressure of each, with 0.50 atm N2). The heat flux and change in mass were both measured, the former being proportional to the endothermic heat of the DRM reaction while the latter being proportional to the coking rate. The reforming rate was extracted from the heat effect using an Aspen HYSYS model of the process.
To scale up the DSC/TGA measurement, selected catalysts were tested in a fixed-bed reactor. The fixed-bed reactor is a 1/2 in. stainless steel reactor tube with α-alumina and 0.25 g of the catalyst. The DRM performed in this reactor contains higher partial pressures for CO2 and CH4 (∼0.65 atm each) than that in the DSC/TGA measurements.

3. Results and Discussion

ARTICLE SECTIONS
Jump To

The 2:1 Ce/Zr mixed oxide and a 6.6 mol % Ni:CZO were prepared by CP/MSS with a reaction time of 6 h in air at 650 °C. The diffraction peaks of both powders (Figure 1a) were indexed to cubic (Ce0.69Zr0.31)O2 (ICSD 157416) (57) with no Ni or NiO peaks detected. The lattice parameters of the CZO and 6.6 mol % Ni:CZO (starting from 10 mol % Ni in the initial solution) NPs were calculated to be 5.31 and 5.34 Å, respectively, and are similar to the literature (a = 5.33 Å (57)), with slight expansion upon Ni doping. Crystalline sizes (Scherrer equation) were calculated to be 7.5 nm (CZO) and 8.5 nm (6.6 mol % Ni:CZO), showing that incorporation of Ni slightly increases the size of the NPs (Table 1). The complete XRD results for all CP/MSS samples suggest a homogeneous dispersion of Ni in all samples at the detectability level of XRD (Figure S1). The presence of the signature Ni Lα peak (0.85 keV) from energy-dispersive X-ray (EDX) spectroscopy further confirms the presence of Ni in the 6.6 mol % Ni:CZO NPs (Figure 1b). As with the XRD measurements, the bright-field scanning transmission electron microscopy (BF-STEM) images of the 6.6 mol % Ni-doped sample show a slight change of the (222) lattice spacing (d = 3 Å, Figure 1c). The size distribution is observed in the inset of Figure 1c for the 6.6 mol % Ni. Figure S2 shows the CZO particle size distribution with and without doping, with negligible changes observed, consistent with the crystallite sizes extracted from XRD. The structural characterization indicates that Ni was incorporated into the lattice with no measurable island formation on the surface. As such, different Ni dopant concentrations of up to 20 mol % in solution were employed to study the effect of dopant concentration on the catalytic activity. At higher Ni concentrations, a blue effluent was seen suggesting the Ni was in excess of the solid solubility limit. (58) Excess Ni/NiO is not believed to deposit on the precipitate surface due to extensive washing of the precipitate, which would remove unincorporated Ni–OH. (40) Additionally, the powders were yellow color instead of black, which would be expected with NiO formation on the surface. To quantify the Ni concentration and verify the stoichiometry of the catalyst, ICP-OES was performed (Table S1, Supporting Information). A maximum of 9.3 Ni mol % in the catalyst was determined for the initial 20 mol % Ni in solution. While the ICP-OES cannot determine the position of the Ni, it is expected that these islands would be detectable as elevated coking rates during the DRM reaction.

Figure 1

Figure 1. (a) XRD of CZO and 6.6 mol % Ni:CZO indicating no phase separation for these NPs. (b) EDX confirming the presence of Ni in 6.6 mol % Ni:CZO. (c) BF-STEM image of Ni:CZO (6.6 mol %) with observed d-spacing of 3.01 Å (d222 = 3.05 Å) and HRTEM image in the inset.

Table 1. Effect of Increased Ni Doping (mol %) on Particle Size, Estimated Oxygen Vacancy Concentrations (N), and Catalytic Activitya
samplesizeb (nm)surface area (m2/g)Nc (cm–3)[OIII]d (%)activity (mW/m2)coking rate (mgcoke/(m2·h))reforming rate (mmol/(m2·h))CH4 conversion (%)
CeO2 (sol–gel) 219  7 × 10–3   
CZO (sol–gel) 183  0.01   
CZO7.9386.3 × 10215.320.01   
reference5.1777.9 × 1021 0.121.04 × 10–51.693.4
4.6 mol % Ni8.2389.9 × 10217.850.031.76 × 10–60.870.8
6.6 mol % Ni8.5331.2 × 102211.190.057.27 × 10–71.641.1
9.3 mol % Ni9.2323.8 × 102219.790.133.75 × 10–62.563
core (4.6 mol %)–shell (2 wt %)9.9152.3 × 102217.360.573.4 × 10–68.674.2
a

The activities are normalized to the surface area.

b

Extracted crystalline size from XRD spectra.

c

Estimated oxygen vacancy site concentration, details in Calculation 1, Supporting Information

d

Extracted from deconvoluted O 1s XPS peaks.

The Ni:CZO catalysts doped with Ni during crystallization were compared to a conventionally prepared sol–gel/urea deposition–precipitation (strong electrostatic adsorption) catalyst containing 5 mol % of Ni. (55,59,60) Unlike the reference catalyst, small/no phase segregation was observed for the 9.3 mol % Ni:CZO core (Figure 2a). Additionally, in situ doping of the core did not require a precatalytic reaction reducing treatment to activate and drive the Ni dopant into the structure, which was necessary for the Ni post-deposition. The catalyst activity (the rate of the DRM reaction) and coking rates for the samples run at differential conversion in the DSC/TGA are shown in Table 1. The heating rate is directly proportional to the DRM reaction rate. The DRM rate on a weight basis can be calculated using a process simulator (Aspen HYSYS) from the heat rate. While the reverse water–gas shift (RWGS) is present in the fixed-bed reactor, this side reaction is not considered in the simulation as its nearly thermoneutral nature would not affect the overall heat flux. The reference catalyst exhibited a higher reforming rate on a weight basis. When normalized to the surface area, the 4.6 and 6.6 mol % Ni:CZO showed reduced activity compared to the reference, despite similar Ni concentrations to that of the reference catalyst. However, the 9.3 mol % Ni:CZO catalyst is comparable to the reference catalyst, but, more importantly, the coking rate was 85% less with the subsurface Ni inclusion.

Figure 2

Figure 2. (a) XRD of 9.3 mol % Ni:CZO and the reference catalyst showing greater phase separation for Ni and ZrO2 in the reference catalyst. (b) Raman spectra showing an increase in the area of the LO band for the 9.3 mol % Ni:CZO compared to the reference catalyst.

To test the activity and stability of the 9.3% CP/MSS catalyst at higher partial pressures and conversions, a longer-term (1 day) fixed-bed reactor test at 750 °C was performed (Table 2). The CH4 conversion × GHSV (gas hourly space velocity, an activity metric) of the reference catalyst was ∼8250 vs 6800 mL/(min·g) for the 9.3 mol % Ni:CZO. The ratio of H2/CO of the 9.3 mol % Ni:CZO was 0.44, slightly higher than the reference catalyst (0.42), suggesting a slight decrease, but still considerable, RWGS. (61)
Table 2. Fixed-Bed Reactor Data for the 9.3 mol % Ni:CZO
catalystH2/CO ratioCH4 conversion × GHSV (mL/(min·g))CO2 conversion (%)
reference0.42825029.2
9.3 mol % Ni:CZO0.44680024.4
Incorporation of Ni into the bulk lattice, as would more likely occur with the 9.3 mol % Ni:CZO catalyst, is believed to result in more oxygen vacancies (and higher OSC) due to the difference in the oxidation states of the Ni and Zr. (62) The vacancies act as sites where active oxygens can be stored and released, mitigating carbon polymerization. This effect could be responsible for the decreased coking rate of the 9.3 mol % sample. However, the results also suggest that dopant incorporation during crystal growth may not distribute enough Ni at particle surfaces where it is needed to catalyze the DRM reaction. This lower Ni concentration and an overall decrease in the surface area are responsible for the slightly lower CH4 and CO2 conversions.
To better understand the effect of Ni location on the catalysis, Raman spectroscopy was performed on the reference and 9.3 mol % samples (Figure 2b). Three characteristic regions of interest, around 228, 470, and 615 cm–1, are attributed to vibrations of Zr–O bonds (symmetrical stretching, B2g mode), Ce–O bonds (symmetrical stretching, F2g mode), and a defect-induced (LO) phonon band, respectively. (63) The blue-shift (474–470 cm–1) in the F2g band of the 9.3 mol % Ni is attributed to an increase in the bond length of Ce–O caused by Ni incorporation, consistent with its smaller ionic radius compared to Zr4+. (64) Additionally, the changes in intensities of the F2g and LO bands are indicative of oxygen vacancy incorporation. (65) The oxygen vacancy concentrations were estimated using a correlation length model for the F2g band (Calculation 1, Supporting Information). (66,67) The estimated oxygen vacancy concentration (N) for the 9.3 mol % Ni:CZO catalyst is approximately 5 times more than the reference catalyst (Table 1), suggesting that greater Ni2+ incorporation in the lattice induces more oxygen vacancies. Therefore, it is reasonable to conclude that the reduced coking rates of the CP/MSS catalysts result from the higher oxygen vacancy concentrations characteristic of this synthesis/structure. When comparing two materials of similar Ni concentrations, 4.6 mol % Ni:CZO vs the reference, the increase in