High-Temperature Stable Ni Nanoparticles for the Dry Reforming of

Research Article pubs.acs.org/acscatalysis

High-Temperature Stable Ni Nanoparticles for the Dry Reforming of Methane Katharina Mette,† Stefanie Kühl,† Andrey Tarasov,† Marc G. Willinger,† Jutta Kröhnert,† Sabine Wrabetz,† Annette Trunschke,† Michael Scherzer,† Frank Girgsdies,† Hendrik Düdder,‡ Kevin Kaḧ ler,‡,§ Klaus Friedel Ortega,∥ Martin Muhler,‡ Robert Schlögl,†,§ Malte Behrens,*,†,∥ and Thomas Lunkenbein*,† †

Fritz-Haber-Institut der Max-Planck-Gesellschaft, Department of Inorganic Chemistry, Faradayweg 4-6, 14195 Berlin, Germany Ruhr-Universität Bochum, Lehrstuhl für Technische Chemie, Universitätsstraße 150, 44801 Bochum, Germany § Max-Planck-Institut für Chemische Energiekonversion, Abteilung für Heterogene Reaktionen, Stiftstraße 34-36, 45470 Mülheim an der Ruhr, Germany ∥ Universität Duisburg-Essen, Fakultät Chemie, Anorganische Chemie, Universitätsstraße 7, 45141 Essen, Germany ‡

S Supporting Information *

ABSTRACT: Dry reforming of methane (DRM) has been studied for many years as an attractive option to produce synthesis gas. However, catalyst deactivation by coking over nonprecious-metal catalysts still remains unresolved. Here, we study the influence of structural and compositional properties of nickel catalysts on the catalytic performance and coking propensity in the DRM. A series of bulk catalysts with different Ni contents was synthesized by calcination of hydrotalcite-like precursors NixMg0.67−xAl0.33(OH)2(CO3)0.17·mH2O prepared by constant-pH coprecipitation. The obtained Ni/MgAl oxide catalysts contain Ni nanoparticles with diameters between 7 and 20 nm. High-resolution transmission electron microscopy (HRTEM) revealed a nickel aluminate overgrowth on the Ni particles, which could be confirmed by Fourier transform infrared (FTIR) spectroscopy. In particular, catalysts with low Ni contents (5 mol %) exhibit predominantly oxidic surfaces dominated by Ni2+ and additionally some isolated Ni0 sites. These properties, which are determined by the overgrowth, effectively diminish the formation of coke during the DRM, while the activity is preserved. A large (TEM) and dynamic (microcalorimetry) metallic Ni surface at high Ni contents (50 mol %) causes significant coke formation during the DRM. KEYWORDS: metal−support interaction, dynamic, overgrowth, heterogeneous catalysis, FTIR spectroscopy, microcalorimetry, transmission electron microscopy, carbon deposition temperatures above 870 °C should be used.3 In addition, the industrial process is run under elevated pressure to increase the production capacity. An industrial DRM technology operated under these harsh conditions has led to the development of sophisticated catalysts as a major aspect of research.

1. INTRODUCTION Production of synthesis gas by dry (CO2) reforming of methane (DRM) was first suggested in 1928 by Fischer and Tropsch,1 who were interested in an alternative to coal gasification. In their comparative study, Ni-Al2O3 and CoAl2O3 supported on clay fragments have been identified as the most active materials. Since then, considerable progress has been made in the development of more and more complex catalyst synthesis protocols.2 Today, new methods allow better insights into the role of preparation parameters, leading to better control of the resulting material properties. For an enhanced efficiency of advanced technological processes, such as the DRM, catalytic materials with well-defined properties are important. The challenging DRM process (eq 1) is highly endothermic and, therefore, requires high reaction temperatures (>640 °C3). In addition, coke deposition on the catalyst during DRM causes severe deactivation. Coke originates mainly from two routes: the Boudouard reaction (eq 2) and methane pyrolysis (eq 3). In order to prevent carbon deposition by the exothermic Boudouard reaction thermodynamically, high © XXXX American Chemical Society

CO2 + CH4 → 2CO + 2H 2

ΔH298 = 247 kJ mol−1 (1)

2CO → CO2 + C

CH4 → C + 2H 2

ΔH298 = − 172 kJ mol−1

ΔH298 = 75 kJ mol−1

Most of the group VIII metals suitable catalysts. In particular, the activities and selectivities for a However, economic considerations

(2) (3)

have been identified as noble metals show high carbon-free operation.4 prevent the commercial

Received: June 14, 2016 Revised: August 26, 2016

7238

DOI: 10.1021/acscatal.6b01683 ACS Catal. 2016, 6, 7238−7248

Research Article

ACS Catalysis use of noble metals due to their high cost and limited availability. The development of a Ni-based catalyst is, therefore, desirable for commercial applications. However, one of the major challenges for the use of nickel is their high tendency toward coke formation, which becomes apparent because nickel also efficiently catalyzes the deposition of carbon.5 The preparation of coke-resistant Ni catalysts can be achieved by using a promoter, changing the support, or optimizing the catalyst’s preparation. Furthermore, the synergy of size, morphology, structure, and composition is an important consideration for the design of Ni-based DRM catalysts.6 The proper selection of a suitable support is crucial. In particular, the specific surface area and the acid−base properties of the support can affect the catalytic activity. Since the DRM involves the adsorption and dissociation of acidic CO2, basic supports such as MgO can enhance the ability of CO2 chemisorption, which increases the coke resistance of the catalyst.3 Moreover, a strong interaction between nickel and the support can improve the coke resistance significantly. SMSI (strong metal−support interaction) effects and surface overlayers play a major role in many catalyst systems, as in the industrial Cu/ZnO/Al2O3 methanol catalyst.7,8 Here, a dynamic SMSI effect strengthens the binding of the intermediates and, thereby, increases the activity of the catalyst. There are many experimental studies indicating that also the size of Ni particles has a strong effect on the selectivity during the DRM reaction.9,10 Kim et al.,11 for example, studied the influence of Ni concentration on the DRM activity over a Ni/ NaY catalyst. By varying the Ni content between 1 and 10 wt %, they obtained a maximum conversion of CO2 and CH4 with a 3.3 wt % Ni loading. However, the coke formation increased as the amount of nickel on the support increased, whereas the intrinsic performances were found to be independent of nickel particle size in the low-temperature DRM. Kim et al.12 prepared Ni-alumina aerogel catalysts with various Ni loadings to generate different Ni particle sizes. They found that large Ni particles are prone to grow carbon whiskers and that a minimum diameter of about 7 nm is required for Ni particles to generate filamentous carbon. This is in agreement with the ́ study of Martinez-de la Cruz et al.13 on the particle size limit for carbon filament formation. They observed that Ni-Al-La catalysts with Ni particle sizes below 10 nm showed an absence of filaments. Juan-Juan et al.14 varied the nickel particle size between 6.3 and 7.8 nm by applying different pretreatment procedures on a Ni/Al2O3 catalyst. They found a direct correlation of the mean particle size and the amount of carbon deposited. The activity in DRM at 700 °C, however, was not affected by the pretreatment. In a recent study, Chen et al.15 report on the effect of Ni crystal size on the growth of carbon nanofibers during methane decomposition at 580 °C. An optimal growth rate and yield of carbon nanofibers were achieved on 34 nm Ni crystals; smaller and larger Ni particles exhibited lower growth rates. Luisetto et al.16 observed that the presence of CeAlO3 leads to a suppression of the deposition of graphitic carbon. Despite all the efforts that have been made, the preparation of coke-resistant Ni catalysts for DRM remains still a significant challenge. The major task is to prepare a thermally stable, highly active, and selective material. We have recently shown that a Ni/MgAl oxide catalyst obtained by coprecipitation of a hydrotalcite-like (htl) precursor can at least partly fulfill these requirements, although coke formation is not completely eliminated.17

Catalytic applications of htl precursors as catalysts and catalyst supports have been intensively investigated in recent years, and comprehensive reviews are available.18−20 Their application as precursors for the preparation of Ni catalysts for DRM has been studied by several groups. Takehira and coworkers21,22 as well as Perez-Lopez et al.,23 for instance, presented different synthetic approaches of htl-derived Ni/ MgO/Al2O3 catalysts in the DRM. In the latter report an influence of the catalyst composition and reduction temperature on the catalytic properties of Ni-Mg-Al catalysts prepared by continuous coprecipitation was found.23 These parameters affect the crystallite size and the acid−base character of the surface, leading to differences in the catalytic properties in DRM. For a Ni:Mg:Al molar composition of 55:11:33 the highest resistance to coke deposition and highest activity have been reported, which can be ascribed to Ni particles around 5 nm. The main objective of this work is to study the influence of structural and compositional properties of nickel catalysts in the DRM with respect to catalyst activity, stability, and coke formation. Catalysts with different Ni contents were prepared by coprecipitation, extensively characterized, and studied concerning their DRM and coking performance. We show a detailed characterization of the catalysts in all stages of the preparation as well as after the reaction. To elucidate structure−activity relationships, the surface properties were investigated by CO adsorption using microcalorimetry and IR spectroscopy. Our experimental findings give new insights into the current state of reforming knowledge and coke resistance. Selected results of individual samples presented here have already been reported in previous publications.17,24

2. RESULTS 2.1. Structural Assignment of the Ni/MgAl Oxide Catalysts after Reduction at 1000 °C. A series of nine different Ni/MgAl oxide catalysts were prepared from phasepure, high surface area htl precursors of the general composition (Ni2+,Mg2+)1−xAlx3+(OH)2(CO3)x/2·nH2O (x = 0.33) using a constant-pH co-precipitation technique (Figure S1 in the Supporting Information). Calcination of the dried precursors at 600 °C and subsequent reduction at 1000 °C in flowing hydrogen led to the final catalysts. Here, we focus on exploiting details of the surface structure of the different activated/reduced Ni catalysts. We aim to establish a correlation among surface structure, catalytic behavior, and the formation of carbon deposits. The results are based on various complementary local and integral surface and bulk averaging characterization techniques. Note that the synthesis and structural characterization of the htl precursors and the calcined materials are described in detail elsewhere.17,24−26 The choice of a suitable reduction temperature was based on the temperature-programmed reduction (TPR) profiles (Figure S2 in the Supporting Information) to ensure the complete reduction of nickel oxide for all samples and was determined to be 1000 °C. Phase assignments of the reduced samples were made by XRD measurements (Figure 1). Upon reduction a nanoscopic segregation of the components has taken place. In the vicinity of an oxidic matrix the XRD patterns clearly confirm the presence of metallic Ni (2θ = 52°). The samples with a low Ni loading hardly show metallic Ni. In addition, MgO and a spinel phase can be identified in the reduced samples. However, the most intense reflection at 2θ = 44.5° is superimposed by MgO and the spinel phase. While the oxidic 7239


Research Article

ACS Catalysis

particle size as obtained from TEM ranges between 7 and 9 nm (Table 1). Only the catalyst with the highest Ni content shows substantial sintering, and the average particle size is increased to 20 nm. In addition, high-resolution (HR) (S)TEM ((scanning) transmission electron microscopy) images of Ni2.5-600-1000, Ni5-600-1000, and Ni50-600-1000 (Figures 2 and 3) were

Figure 1. Powder XRD patterns of the mixed oxides after reduction at 1000 °C. The blue bars correspond to Ni (PDF 65-2865), the gray bars to MgO (PDF 65-476), the orange bars to MgAl2O4 (PDF 741132), and the red bars to NiAl2O4 (PDF 10-339). On the right-hand side the resolution is increased at higher angles.

Figure 2. HR-TEM images of Ni2.5-600-1000 (a), Ni5-600-1000 (b), and Ni50-600-1000 (c). Red and green colors represent the overgrowth and Ni particles, respectively (uncolored images are shown in Figures S4 and S5 in the Supporting Information).

components in the catalyst with the highest Ni content (Ni50) are still poorly crystalline, the intensity of the MgO and MgAl2O4 phases rises with decreasing Ni content. The pattern cannot be fitted satisfactorily with a conventional Rietveld fit, because of the complex line shape of the diffraction patterns. From the reflection arising only from metallic Ni at 2θ = 52° a single peak fit could be realized. Although the particle sizes obtained by the Scherrer equation could only be considered as rough estimates, they follow the trend obtained by the particle size estimation from the TEM images (Table 1). Using this equation, a Ni particle size of 13 nm was obtained for reduced Ni50-600-1000, whereas all other reduced samples with the exception of Ni0-600-1000 have Ni particle sizes between 7 and 8 nm. The segregation occurs with preservation of the microstructure and can be seen by the homogeneously dispersed black spheres in the SEM images,17 which is also in line with the observation made by transmission electron microscopy (TEM) imaging (Figure S3 in the Supporting Information). As can be seen from Figure S3 in the Supporting Information, decreasing the Ni content alters the matrix morphology. In agreement with the XRD observations (Figure 1), at lower Ni content the crystallinity of the oxide matrix causes the formation of a large amount of crystalline needles that likely consist of MgO or MgAl2O4. The Ni content has only a minor influence on the Ni particle size. The average

Figure 3. HR-STEM investigation of the overgrowth on Ni nanoparticles reduced at 1000 °C: high-resolution STEM images of Ni2.5-600-1000 (a, c) and Ni5-600-1000 (b, d). Red and green colors represent the overgrowth and Ni particles, respectively (uncolored images are shown in Figure S6 in the Supporting Information).

Table 1. Composition, Particle Sizes, Ni Dispersion, and Interface Ratio (IFR) Determined by TEM and H2 Chemisorption of the Reduced Samples

a

sample label

Ni content/wt %

particle size TEM/nm

Ni SA/m2 gcat−1

dispersion/%

IFR/%

Ni50-600-1000 Ni25-600-1000 Ni15-600-1000 Ni10-600-1000 Ni7.5-600-1000 Ni5-600-1000 Ni2.5-600-1000 Ni1-600-1000 Ni0-600-900

55.4 30.3 18.9 12.9 9.7 6.6 3.3 1.3 0.0

19.4 ± 7.1 7.3 ± 2.0 9.0 ± 3.1 nda nd 9.3 ± 3.7 nd 7.0 ± 4.6

6.0 5.0 4.2 3.2 2.7 3.0 0.6 0.1 0.0

1.6 2.5 3.4 3.7 4.2 6.9 2.9 1.0 0.0

68.7 82.1 70.5 nd nd 37.3 nd 92.2 100.0

nd: not determined. 7240


Research Article

ACS Catalysis

been found by correlating the Ni particle size with the Ni dispersion (Figure 4b). The discrepancy among the Ni surface area, the dispersion, and the particle size can be explained by different degrees of embedment, as are reflected by the interface ratios (IFR) given in Table 1. In brief, the IFR is the part of the particle that is covered by the support phase and therefore not accessible for reactive gases. It can be calculated as the ratio between a theoretical surface area (SAtheo), calculated from the particle size that is determined by TEM, and the metal surface area (SANi), determined by H2 pulse chemisorption.

recorded. The HR-TEM images of Ni2.5-600-1000 and Ni5600-1000 (Figure 2a,b) demonstrate the existence of a crystalline overgrowth on top of the Ni particles. The HRSTEM image of Ni2.5-600-1000 in Figure 3a verifies a rather homogeneous and complete coverage of the nanoparticles, whereas for Ni5-600-1000 a rough and heterogeneous crystalline overgrowth can be observed. For the Ni50-6001000 catalyst (Figure 2c and Figure S4 in the Supporting Information) the overgrowth does not completely cover the nanoparticles and a fraction of the metallic Ni surface is exposed. The overgrowth on the Ni particles of the Ni5-6001000 sample is composed of Ni, Al, and O, as shown by the overlap of the corresponding EDX elemental maps, and might be interpreted as a NiAl2O4 spinel (Figure S7a in the Supporting Information). Moreover, EDX line scans confirm the coexistence of Ni and Al (Figure S7b). The existence of a partially oxidized Ni species after reduction at 1000 °C can also be confirmed by integral sample averaging near-edge X-ray absorption fine structure (NEXAFS) measurements. The NEXAFS spectra of the Ni5-600-1000 fitted with the reference materials are shown in Figure S8a−c in the Supporting Information. A small shoulder remaining in the high-energy feature of the Ni L-edge (Figure S8b) spectra can be observed, which is less pronounced for Ni50-600-800 (Figure S8d−f) and can be interpreted as an incomplete reduction of the oxidic Ni phases and depends inversely on the degree of Ni substitution. The extent of the overgrowth was indirectly quantified by hydrogen pulse chemisorption experiments, which determined the free Ni surface area. As suggested by TEM images, with decreasing Ni content, the Ni metal surface area decreases (Table 1 and Figure 4a). A maximum for Ni5-600-1000 has

IFR = 1 − (SANi/SA theo)

(4)

where SAtheo is calculated from the surface area (ANi) and the spherical particle volume (VNi) and the density of Ni (ρNi = 8.90 g cm−3): SA theo = ANi (wt %)/ρNi VNi

(5)

As Ni5-600-1000 exhibits the lowest IFR, the particles are less embedded in the support and, therefore, possess the highest dispersion despite similar particle sizes. The aforementioned results confirm the presence of oxidic surface species. In order to explore details of surface composition, Ni oxidation state, and Ni dispersion on an integral basis, we performed Fourier transform infrared (FTIR) spectroscopy of CO adsorbed at −196 °C (Figure 5). In general, IR bands above 2100 cm−1 are due to CO adsorbed on coordinately unsaturated cationic species. Band positions at 2100−2000 cm−1 indicate CO coordinated linearly on metal sites (Ni0-CO), whereas vibrational modes below 2000 cm−1 indicate the formation of bridging carbonyls ((Ni0)x-CO).27 In Figure 5 the spectra of adsorbed CO at full coverage (θ = 1) are compared for the reduced catalysts (Figure 5a), as well as for reference compounds (Figure 5b). One intense carbonyl band occurs in all catalysts at 2158 cm−1. The band is ascribed to the carbonyl stretching vibration of CO adsorbed on 5-foldcoordinated Mg2+ ions.28 The lower intensity of the band on Ni50-600-1000 might be related to the lower amount of Mg in this sample or due to notable coverage of the support surface by large Ni particles. The appearance of strong absorption in the range between 1800 and 2100 cm−1 on Ni50-600-1000 indicates the formation of carbonyls on metallic Ni sites, both linear (2020−2065 cm−1) and bridged (

High-Temperature Stable Ni Nanoparticles for the Dry Reforming of

Recommend Documents