Insight into the Nature of the ZnO_x Promoter during Methanol Synthesis

Abstract

Despite the great commercial relevance of zinc-promoted copper catalysts for methanol synthesis, the nature of the Cu–ZnO_x synergy and the nature of the active Zn-based promoter species under industrially relevant conditions are still a topic of vivid debate. Detailed characterization of the chemical speciation of any promoter under high-pressure working conditions is challenging but specifically hampered by the large fraction of Zn spectator species bound to the oxidic catalyst support. We present the use of weakly interacting graphitic carbon supports as a tool to study the active speciation of the Zn promoter phase that is in close contact with the Cu nanoparticles using time-resolved X-ray absorption spectroscopy under working conditions. Without an oxidic support, much fewer Zn species need to be added for maximum catalyst activity. A 5–15 min exposure to 1 bar H₂ at 543 K only slightly reduces the Zn(II), but exposure for several hours to 20 bar H₂/CO and/or H₂/CO/CO₂ leads to an average Zn oxidation number of +(0.5–0.6), only slightly increasing to +0.8 in a 20 bar H₂/CO₂ feed. This means that most of the added Zn is in a zerovalent oxidation state during methanol synthesis conditions. The Zn average coordination number is 8, showing that this phase is not at the surface but surrounded by other metal atoms (whether Zn or Cu), and indicating that the Zn diffuses into the Cu nanoparticles under reaction conditions. The time scale of this process corresponds to that of the generally observed activation period for these catalysts. These results reveal the speciation of the relevant Zn promoter species under methanol synthesis conditions and, more generally, present the use of weakly interacting graphitic supports as an important strategy to avoid excessive spectator species, thereby allowing us to study the nature of relevant promoter species.

1. Introduction

Methanol synthesis is an important, decades-old industrial process. Nowadays, a coprecipitated Cu/ZnO/Al₂O₃ catalyst is used to hydrogenate CO₂ to methanol in a CO-rich environment. It has been well established that the methanol is predominantly formed from CO₂ rather than from CO. The role of the CO is to supply CO₂via the reaction with water, which also keeps the water level low.¹⁻⁶ Generally accepted is that Cu is the main active component where ZnO_x plays a crucial role in promoting the catalyst activity with about an order of magnitude.^5,7−11 Yet, the exact role of the ZnO_x promoter is still under debate,¹²⁻¹⁴ especially due to a lack of detailed knowledge on the ZnO_x speciation, structure, and its interaction with Cu under the typical methanol synthesis conditions at 473–573 K and 20–100 bar.^15,16

Various hypotheses exist to explain the role of the ZnO_x promotion. It has been suggested that ZnO_x increases the Cu dispersion and thereby the active Cu surface area^17,18 and that the promoter supplies hydrogen to the Cu surface by spillover.^19,20 The oxidation state of ZnO_x can also play a role in the morphological change of small Cu particles due to a varying degree of the Cu–ZnO_x interaction, thereby varying the exposed Cu surface planes.²¹ However, by now it is broadly accepted that the coverage of Cu nanoparticles with partially reduced ZnO is essential for the enhanced methanol production. An open question is still whether the promotion is due to the formation of a ZnO_x layer on the Cu particles,^19,22−24 to the formation and migration of Zn atoms on (or into) the Cu surface,^10,13,25,26 or to the creation of active defects upon Cu–ZnO_x interaction.^19,23,27 Research is typically performed on catalysts supported on metal oxides, which may obscure the active ZnO_x phase by the formation of mixed Zn metal oxides and hence may significantly differ from the relevant speciation and distribution of the active fraction of the ZnO_x promoter.

It is generally accepted that ZnO_x (partially) covers the Cu nanoparticles in reducing conditions. The fractional coverage of Cu with ZnO_x during reaction conditions is mainly influenced by three factors: the feed composition, governing the degree of ZnO_x reduction; the ZnO_x loading; and the Cu particle size. For example, Kuld et al.¹³ showed that by applying various feeds during catalyst activation an optimal Zn coverage over a Cu surface of 0.47 was achieved using a Cu/ZnO/Al₂O₃ catalyst of constant composition during CO/CO₂ hydrogenation at ambient pressure. Yet, contradictory results for the optimal Zn coverage were reported by varying the ZnO_x loading under different reaction conditions.^28,29 Also in a pure H₂/CO₂ feed, an optimal Zn coverage of 0.20, or an atomic Zn/Cu ratio of 1.2–1.6, was reported for Cu/ZnO catalysts.^{19,22,23,30,31} The question remains what the actual state of the ZnO_x is during working conditions in different feeds at high pressure (e.g., H₂/CO feed), syngas enriched with a relevant amount of CO₂ (2–6 vol %¹⁻⁵), or an H₂/CO₂ feed.

Much effort has been devoted to studying the interaction and oxidation state of ZnO_x species in CuZn-based catalysts in the calcined state^29,32−34 and before/after^33,35−39 or during^{14,32,40−43} exposure to reducing atmospheres at (near-)ambient pressures (up to 8 bar). On the basis of those results, it is still inconclusive whether the oxidation state of ZnO_x slightly changes^35,36,40 and whether Cu–Zn alloys are formed^14,38,44 or not.^33,35,41,42 For example, recent studies reported the formation of a Cu–Zn alloy in a Cu/ZnO/Al₂O₃ catalyst during a (CO₂/)H₂ treatment at 15 bar and 533 K,⁴⁵ but this alloy formation was absent for a Cu/ZnO/faujasite catalyst with almost a 1-to-1 ratio of Cu and Zn.⁴⁶ A recently developed, unique tool that allows us to gain insight into the Zn oxidation state and speciation under realistic high-pressure conditions and in the working state is X-ray absorption spectroscopy (XAS). Very recently, Divins et al.²⁴ published an interesting operando study at 20–40 bar in a CO₂-enriched syngas feed using silica and alumina supports, ascribing the active ZnO_x speciation to a distorted ZnO_x phase with a maximum content of 9 at% Zn⁰ atoms but most of the Zn species present as metal oxides.

A major obstacle to study the nature of the active site of the ZnO_x promoter is the strong interaction of the promoter with the oxidic catalyst support, which leads to the formation of a large fraction of Zn spectator species present as formates, oxides, or mixed metal phases.^{24,37,45−49} Hence, the active promoter species represents only a fraction of the Zn species present in the system, and averaged information, such as the Zn oxidation state and coordination number, are not representative for the active ZnO_x promoter species.

We present graphitic carbon as a support with very limited interaction with Cu and ZnO_x.^47,48 In combination with a relatively low ZnO_x loading, it allows us to study specifically the ZnO_x in contact with the Cu nanoparticles during methanol synthesis and its speciation and interaction with the Cu, based on time-resolved XAS experiments under working conditions, also as a function of different feed compositions.

2. Experimental Section

2.1. Catalyst Synthesis

A series of CuZnO_x/C catalysts, with similar Cu weight loadings (8.0 ± 0.4 wt %) but varying Zn/Cu molar ratios, were prepared via incipient wetness impregnation following a published method.⁵ In brief, powdered high-surface-area graphite (TIMREX E-HSAG500, TIMCAL Graphite & Carbon) was dried at ca. 443 K under dynamic vacuum for 1.5 h. The support was impregnated at room temperature under static vacuum to 95% of the total pore volume with an acidified aqueous solution containing 1.8 M copper nitrate (Acros Organics, 99%) and 0–1.8 M zinc nitrate (Sigma-Aldrich, ≥99%). Subsequently, the impregnated support was dried overnight at room temperature under dynamic vacuum and further reduced at 503 K (ramp 2 K min^–1) in a 100 mL min^–1 flow of 20 vol % H₂/N₂ for 2.5 h. After cooling to room temperature, the sample was exposed to a flow of 100 mL min^–1 flow of 5 vol % O₂/N₂ for 1 h, heated to 473 K with a ramp of 1 K min^–1 and oxidized at 473 K in 15 vol % O₂/N₂ for 1 h.

The Cu/C (8.1 wt % Cu), ZnO_x/C (9.9 wt % ZnO), and CuZnO_x/SiO₂ catalysts were synthesized following the same procedure as for the CuZnO_x/C catalysts using the respective metal nitrate(s). A different heat treatment was applied only for the ZnO_x/SiO₂ catalyst (10.0 wt % ZnO): the dried impregnate was heated to 723 K (ramp 2 K min^–1) in a 200 mL min^–1 g_cat^–1 flow of 2 vol % NO/inert for 1 h.⁵⁰ Both SiO₂-based catalysts were supported on silica gel (25–75 μm, Davisil, grade 643, Sigma-Aldrich, ≥99%). All catalysts are named CuZn-X/C or CuZn-X/SiO₂, in which X represents the molar Zn/(Cu + Zn) ratio expressed as percentage and is based on the nominal loading. A commercial Cu/ZnO/Al₂O₃/MgO catalyst from Alfa Aesar, containing a Cu/Zn/Al/Mg ratio of 63.8/24.8/10.1/1.3 wt %, served as a reference.

2.2. Catalyst Characterization

N₂ physisorption isotherms were recorded on a Micromeritics TriStar II Plus apparatus at 77 K. The samples were first dried at 443 K (or at 573 K for the SiO₂ support) under an N₂ flow overnight. The BET surface area was determined according to the IUPAC procedure.⁵¹ A Barrett–Joyner–Halenda (BJH) analysis was applied to obtain pore size distributions, using either a carbon black or Harkins-Jura statistical thickness curve. The single-point total pore volume V_tot was determined at p/p₀ = 0.995. Integration of the differential pore size distribution (derived from the adsorption branch) between 2 and 50 nm yielded the mesoporosity. The micropore volume V_micro was calculated using the t-plot method.

Transmission electron microscopy (TEM) imaging was performed on an FEI Tecnai 20 apparatus, operating at 200 kV. High-angle, annular, dark-field scanning transmission electron microscopy (HAADF-STEM) images were obtained on a Thermo Fisher Scientific Talos F200X apparatus, operating at 200 kV. With the same apparatus, chemical maps were recorded using energy-dispersive X-ray (EDX) detectors. The EM samples for the carbon-supported catalysts were prepared by deposition of an ethanolic dispersion of the catalyst onto holey carbon film-coated Cu or Au grids (Agar, 300 mesh). As adequate TEM measurements on the silica-supported catalysts as such were not possible, they were ultramicrotomed. The catalysts were embedded in a two-component epoxy resin (Struers, EpoFix), which was heated overnight at 333 K and cut in 60–70 nm slices on a Leica Ultracut E. The slices were deposited on the aforementioned Au grids, which were made hydrophilic by glow discharge in a Cressington 208 carbon coater. At least 350 individual particles at various locations within the sample were measured to determine the number-averaged Cu(Zn)O_x particle sizes (d_N) with the standard deviation (s_N) representing the width of the size distribution. These mean sizes were translated into surface-averaged particle sizes (d_S) via, with d_i the i-th particle size and N the total number of measured particles. Only the relevant part of the log-normal distribution (>1% of maximum) was considered for the calculation of the average particle sizes.

Powder X-ray diffractograms were recorded on a Bruker AXS D2 Phaser diffractometer at room temperature with a fixed divergence slit. Samples were irradiated by Co Kα radiation (λ = 1.790 Å) at 30 kV and 10 mA. Not only fresh catalysts, but also used catalysts were analyzed. These were exposed to ambient conditions, separated from the SiC in the reactors, finely ground, and characterized without any further pretreatment.

Temperature-programmed reduction (TPR) profiles were obtained on a Micromeritics AutoChem II 2920 apparatus. The sample (50 mg, <75 μm granulites) was first dried in situ under an Ar flow at 1 L min^–1 g_sam^–1 at 393 K for 30 min. The cooled sample was then exposed to 5 vol % H₂/Ar at the same flow and heated to 873 K with a ramp of 2 K min^–1. The formed H₂O was captured with a dry ice/isopropanol cold trap, and the reduction profiles were recorded with a thermal conductivity detector (TCD). The H₂ reduction profiles of the CuZn-15/C and CuZn-15/SiO₂ catalysts (25–75 μm) were also obtained at a temperature ramp of 5 K min^–1 in a 0.5 L min^–1 g_sam^–1 flow without prior drying to directly compare with the H₂ treatment during XAS.

Time-resolved, operando X-ray absorption spectroscopy (XAS) measurements on simultaneously the Cu (8979 eV) and Zn K-edges (9659 eV) were performed at the SOLEIL synchrotron (ROCK beamline).⁵² Typically, ca. 3.5 mg of catalyst (25–75 μm sieve fraction) was loaded in a quartz capillary (ID 1.5 mm, 50 μm thick), which was tightly glued into a frame connected to gas feed lines. Heating of the capillary was ensured by a hot gas blower (FMD Oxford). After the capillary was leak-checked at 20 bar, XAS data was obtained in He at room temperature. The catalyst was exposed to a 15 mL min^–1 flow of 20 vol % H₂/He and heated to 543 K (ramp 5 K min^–1) with a hold time of 5–15 min at ambient pressure. After the H₂ treatment, the capillary was cooled to 453 K prior to introducing a syngas feed (H₂/CO/He = 60/30/10 vol %) at 15 mL min^–1. Within ca. 100 min, the capillary was pressurized to 20 bar, and subsequently, the temperature was increased to 533 K (ramp 5 K min^–1) and held for 160 min. Only for the CuZn-15/C catalyst was the feed subsequently switched to H₂/CO/CO₂/He = 60/27/3/10 vol %, recording XAS spectra for 160 min, and after that to H₂/CO₂/He = 67.5/22.5/10 vol %. Finally, XAS data were recorded after cooling to room temperature at 20 bar and in the last experienced gas atmosphere (H₂/CO₂/He and H₂/CO/He for the CuZn-15/SiO₂ and CuZn-15/C catalysts, respectively).

During all treatments, XAS spectra were recorded while scanning the X-ray energy from 8.70 to 10.65 keV (20 averaged scans per 10 s) in transmission mode using a Si(111) quick-XAS monochromator. Methanol production and gas compositions were recorded with a mass spectrometer (Cirrus, MKS) at ambient pressure. If the measurement involved CO, a conditioned carbonyl trap was used upstream the capillary to capture metal carbonyl compounds. ZnO (abcr, 99.999%), CuO (Sigma-Aldrich, 99.999%), Cu₂O (Sigma-Aldrich, ≥99.99%), in-house synthesized Zn₂SiO₄ (Figure S27), all mixed with boron nitride (Sigma-Aldrich, 98%), and Cu (6 μm) and Zn (5 μm) foils were used as references, with their spectra being recorded at room temperature under air. A Zn₃₀Cu₇₀ brass measured at the ESRF (LISA beamline (BM 08)) was also used as a reference. The optics and performance at the two beamlines are different, and hence, this might give rise to slight changes in the observed XAS spectra. Data analysis was performed using Athena and Artemis software, as further detailed in supplementary section S4.

2.3. Catalyst Testing

A 16-reactor setup (Flowrence, Avantium) was used to evaluate the catalyst performance for methanol synthesis at 40 bar(g) and 533 K for at least 100 h. The powdered catalysts were pressed, crushed, and sieved into granules of 75–150 μm and were loaded (3–180 mg) in the stainless-steel reactors (ID 2.6 mm). The catalysts were diluted with SiC (212–245 μm fraction, Alfa Aesar, ≥98.8%, 46 grit), resulting in SiC contents between 22 and 88 vol % of the total packed catalyst bed.⁵³ The SiC had been previously calcined at 1073 K for 10 h, washed with 65 wt % HNO₃, rinsed with Milli-Q water until pH 7 was reached, and dried in static air at 393 K overnight. The varying catalyst loadings enabled us to achieve similar CO (+ CO₂) conversions (ca. 10%). The difference in sieve fractions between the catalysts and diluent facilitated postanalysis by EM and XRD.

An in situ reduction was performed in 2.8 mL min^–1 of 5 vol % H₂/N₂ at 523 K for 3 h after which the temperature was lowered to 393 K. The reactors were exposed to a 2.2 mL min^–1 flow of CO₂-free syngas (H₂/CO/He = 60/30/10 vol %) or CO₂-enriched syngas (H₂/CO/CO₂/He = 60/27/3/10 vol %), leading to a flow of 0.2–2.1 L min^–1 g_Cu^–1 and a gas-hourly space velocity (GHSV) of 400–53 200 h^–1. The reactors were pressurized to 40 bar(g), heated to 533 K (ramp 2 K min^–1), and the reaction was run for at least 100 h. Alternatively, the ZnO_x/C and ZnO_x/SiO₂ catalysts were alternately exposed to the predefined H₂/CO/He and H₂/CO/CO₂/He feeds. A triphase carbonyl trap (activated carbon, γ-Al₂O₃, ZnO) was located upstream of the CO feed to remove metal carbonyls and sulfur species. Products were periodically analyzed by online gas chromatography every 15 min. After catalysis, the samples were slowly exposed to air at 393 K. Details on the calculations of activity, selectivity, and stability are given in section S5.

3. Results and Discussion

3.1. Structural Properties of the Catalysts

Figure 1 shows representative electron micrographs, including elemental maps, of CuZn-15/SiO₂ (frames A–C) and CuZn-15/C (frames D–F) catalysts both with 8.1 wt % Cu and 1.8 wt % ZnO (see Figures S2–S5 for other catalysts and zoomed-out micrographs). Note that the number in the catalyst names refers to the Zn/(Cu + Zn) fraction of 15 at%. The silica as support consists of aggregated spheres of ca. 8 nm, whereas the graphitic carbon has a sheet-like morphology of a few nanometers thick. Both materials have a high specific surface area (>260 m² g^–1), and ca. 50–60% of the total pore volume consists of mesopores (Figure S1, Table S1), making these materials suitable supports for model catalyst studies.Frame A shows no clear Cu(Zn)O_x nanoparticles on the silica support, demonstrating that it was challenging to distinguish metal particles on the silica support because of the limited phase contrast. Only by imaging ultramicrotomed slices were we able to obtain a representative HAADF-STEM micrograph with an elemental map (frame B). Cu(Zn)O_x particles of ca. 3–4 nm (bright spots) were observed for the CuZn-15/SiO₂ catalyst, corresponding to mainly Cu species (blue dots) and ZnO_x species (red dots). The distribution of Cu and ZnO_x looked similar after 150 h of catalysis (frame C), which is probably more representative for the catalyst during catalysis.

Figure 1.

Representative EM images of the (A–C) CuZn-15/SiO₂ and (D–F) CuZn-15/C catalysts. Frames A and D involve BF-TEM, and frames B–C and E–F involve HAADF-STEM with an elemental map overlay. Number-averaged Cu(Zn)O_x particle sizes are 3.4 ± 0.8 nm (frames B–C) and 4.2 ± 1.7 nm (frame D) for the fresh CuZn-15/SiO₂ and CuZn-15/C catalysts, respectively. The used catalysts (frames C and F) are after 150 and 100 h of catalysis in an H₂/CO/CO₂ feed, respectively. Please note that the pixel size in frame F is larger (521 pm) than in frames B, C, and E (368 pm). Corresponding zoomed-out images and EDX spectra are shown in Figures S5–S6.

When using a graphitic support (frame D) CuO_x nanoparticles of ca. 4 nm were clearly discernible by TEM and well-distributed (dark spots indicated with white arrows) on the carbon surface (light gray).

The distribution was confirmed by the elemental maps of Cu and Zn species, projected on a HAADF-STEM image (frame E). There was a strong correlation between the location of the Cu nanoparticles and the distribution of the ZnO_x species, both in the fresh CuZn-15/C catalyst and after catalysis (frames E and F). The characteristics of the full series carbon-supported CuZnO_x/C catalysts both in the fresh and used state can be found in the Supporting Information (Table S2) and shows similar Cu(Zn)O_x particle sizes (d_N = 5–9 nm) with varying ZnO_x loadings. Additional structural information includes N₂ physisorption (Figure S1), additional electron microscopy imaging (Figures S2–S5), X-ray diffraction (XRD) analysis (Figure S8), and H₂ reduction profiling (Figure S9). Overall, we showed that in both catalysts well-distributed Cu(Zn)O_x particles of similar size were present and that the relatively thin sheets of graphitic carbon as a model support facilitated the determination of the particle sizes by electron microscopy.

3.2. Influence of the Support and Feed Composition

In this section, we compare the catalytic performance of CuZn-15/SiO₂ and CuZn-15/C catalysts, which were prepared and tested in the same way and have similar Cu(Zn)O_x particle sizes and ZnO_x loadings but only have a different support. Figure 2 shows the methanol formation rate under industrially relevant temperature and pressure as a function of time in an H₂/CO feed as well as in an H₂/CO/CO₂ feed (mimicking industrially relevant conditions¹⁻⁵). Figure S10 gives the CO (+ CO₂) conversion and total activity, and Table S4 provides additional information on the conversion levels, turnover frequencies (TOFs), and Cu(Zn)O_x particle growth during catalysis.

Figure 2.

Methanol formation rate of the CuZn-15/SiO₂ (red circles) and CuZn-15/C (black squares) catalysts in a CO₂-free (open symbols) or -enriched (filled symbols) syngas feed. The data points of the CuZn-15/C catalyst in H₂/CO and H₂/CO/CO₂ are the average over 4 and 2 separate runs, respectively. Conditions: 533 K, 40 bar(g), H₂/CO/He = 60/30/10 vol % or H₂/CO/CO₂/He = 60/27/3/10 vol %.

The TOF for the carbon-supported catalyst (3.9–15.3 × 10^–3 s^–1) was always higher than for the silica-supported catalyst (1.6–3.0 × 10^–3 s^–1) (Table S4). Strikingly, the beneficial effect of CO₂ enrichment of the syngas feed on the methanol formation rate was much larger for the CuZn-15/C catalyst (factor 3.5) than for the CuZn-15/SiO₂ catalyst (factor 1.7) (Figure 2). Upon CO₂ enrichment the methanol selectivity increased from 83 to 99%_C and from 85 to 98%_C after 100 h on stream for, respectively, the CuZn-15/SiO₂ and CuZn-15/C catalysts, in line with earlier published results⁵ and significantly higher than recently reported for CuZnO_x/Al₂O₃.²⁴ In the literature, enhancement factors upon CO₂ enrichment of 2–4 are reported for Cu/ZnO/Al₂O₃ catalysts depending on the reaction conditions,^2,3,47 and differences were also observed between silica- and alumina-supported CuZnO_x particles.²⁴ However, our results, obtained in the same reaction conditions and with similar Cu particles sizes, unequivocally proves that promotion with a given amount of ZnO_x is much more efficient using a carbon than using an oxide support. Under all conditions, the ZnO_x promotion is more effective in the CuZn-15/C catalyst than in the CuZn-15/SiO₂ catalyst, but the effect is especially pronounced with CO₂ enrichment of the feed.

It is known that ZnO_x itself can also act as a methanol synthesis catalyst, albeit with a lower activity than in combination with Cu.^16,54,55 Supported ZnO_x species without Cu were investigated under similar reaction conditions to check if the catalysis by ZnO_x on graphitic carbon contributed significantly. The ZnO_x/SiO₂ and ZnO_x/C catalysts have the same ZnO loading (both 10 wt %) with ZnO_x particle sizes of 7.7 and ca. 4.5 nm, respectively. Figure 3 shows the CO (+ CO₂) conversion versus time on stream in the presence and absence of CO₂ and for both an oxidic and a carbon support.

Figure 3.

CO (+ CO₂) conversion of silica- and carbon-supported ZnO_x (10 wt %) in various syngas compositions. Label “+3% CO₂” in the total feed corresponds to a CO₂/(CO + CO₂) volume fraction of 0.10. Conditions: 533 K, 40 bar(g), H₂/(CO + CO₂)/He = 60/30/10 vol %, 21.9 mL min^–1 g_cat^–1.

Irrespective of the syngas composition, carbon-supported ZnO_x species were much more active than silica-supported ZnO_x species, even when taking the slightly different Zn surface areas into account (Figure S12). EM analysis showed no Zn-based nanoparticles (Figure S4) in the fresh ZnO_x/SiO₂ catalyst (frame B), whereas they were present in the ZnO_x/C catalyst (frame A). During pure CO hydrogenation, the ZnO_x/C catalyst had a significant conversion of ca. 2% (of which ca. 1.3% was methanol (Figure S12)). The activity of both supported ZnO_x species clearly decreased in the presence of CO₂. The conversion level was restored when switching back to an H₂/CO feed. This demonstrates that the negative CO₂ effect on the conversion is not related to, for example, irreversible changes in the catalyst morphology but probably can be attributed to the significant reduction of ZnO in a CO₂-free feed, making it a more efficient methanol synthesis catalyst. Yet, the activity of the supported ZnO_x was too small to explain the overall effects of CO₂-enrichment in methanol synthesis of CuZnO_x-based catalysts. Nevertheless, these results clearly show that an oxide support has a strong interaction with the ZnO_x and leads to a different speciation than for the weaker interacting carbon supports.

3.3. Influence of the ZnO_x Loading on Activity and Stability

For the carbon-supported catalysts, we investigated in more detail the effect of the ZnO_x loading on the activity and stability of supported Cu nanoparticles. Figure 4 shows the specific activity (see Figures S13–S14 for the total activity) as a function of the ZnO_x loading in a syngas atmosphere with and without CO₂ (please note the logarithmic scale) for ca. 5.1 nm Cu nanoparticles supported on either a carbon support (black lines) or an oxide support (red markers). The activity increased when CO₂ was in the feed for all studied ZnO_x-promoted catalysts. After the initiation period, all catalysts had a methanol selectivity of >97%_C in CO₂-enriched syngas. The highest TOF_MeOH values were obtained for carbon-supported catalysts with Zn/(Cu + Zn) molar fractions between 0.15 and 0.25, irrespective of the presence of CO₂ in the feed. These ZnO_x loadings are lower than the well-established optimal loading for the commercially used Cu/ZnO/Al₂O₃/MgO methanol synthesis catalyst as well as for other oxide-based Cu catalysts in literature (Zn/(Cu + Zn) content of 29–47 at%).^{10,13,15,19,22,23,28−31,56}

Figure 4.

Initial methanol turnover frequency (TOF_MeOH) of CuZnO_x/C (black squares) and metal oxide-supported (red circles) catalysts in an H₂/CO (open symbols) or an H₂/CO/CO₂ (filled symbols) feed (at t = 0). “SiO₂” = CuZn-15/SiO₂ catalyst. “com cat” = commercial, coprecipitated Cu/ZnO/Al₂O₃/MgO catalyst (58 wt % Cu, ca. 10 nm CuO particles). Conditions: 533 K, 40 bar(g), H₂/CO/He = 60/30/10 vol % or H₂/CO/CO₂/He = 60/23/7/10 vol %.

Another important factor in catalysis is the stability. In Figure S15, this stability is defined as the ratio between the activity after 100 h and after 50 h on stream. The addition of only 5 at% ZnO_x was sufficient to increase the catalyst stability from 74 ± 8% to 84 ± 3% upon syngas conversion. Further increasing the ZnO_x content to 15–35 at% maximized the stability to 91 ± 2% and 83 ± 3% in an H₂/CO and H₂/CO/CO₂ feed, respectively. In the most heavily promoted CuZn-50/C catalyst, the stability was somewhat lower. The presence of 15–35 at% ZnO_x apparently limited the CuZnO_x particle growth during catalysis as evident from TEM and XRD analysis (Figures S3, S7, and S8) and is in line with the stability improvement for intermediate amounts of ZnO_x. Hence, ZnO_x is not only an activity promoter but also a stability promoter for carbon-supported Cu catalysts.

3.4. Catalyst Evolution during Reduction in H₂

From the literature, it is known that the coverage of the Cu surface with ZnO_x species¹³ and the reduction degree of these ZnO_x species^14,35 are parameters that determine the effectiveness of ZnO_x as a promoter. However, mostly metal oxides are employed to support CuZnO_x particles, which can result in the formation of spectator species such as zinc silicates and aluminates,^11,37,43 hampering the study of the active fraction of the ZnO_x promoter. The presence of the mixed Zn metal oxides may hence obscure the results also of, for example, electron energy loss spectroscopy (EELS) and operando X-ray absorption spectroscopy (XAS) measurements. While EELS is a valuable technique to study the local oxidation state of metals,^56,57 we chose to assess the chemical state of our supported catalysts by XAS because of the small particle sizes and relatively low metal loadings. Hence, our hypothesis was that our use of a carbon support would allow us to much better study the formation, oxidation state, and structure of the relevant ZnO_x promoter by time-resolved, operando XAS at simultaneously the Cu and Zn K-edges at 20 bar and up to 533 K.

A first piece of information about the interaction between CuO_x and ZnO_x species can be derived from the reduction profiles. Figure 5 shows the ex situ H₂ reduction profiles of the CuZn-15/SiO₂ and CuZn-15/C catalysts. The theoretically maximum Cu surface coverage by a monolayer of Zn atoms is 75–95% for these catalysts with 15 at% ZnO_x. The maximum CuO reduction temperature (T_max) as well as the offset temperature for reduction (T_offset) were clearly lower for the CuZn-15/C catalyst than for the CuZn-15/SiO₂ catalyst (T_max of 465 vs 475 K, T_offset of 421 vs 434 K, respectively). Hence, the CuO is more easily reduced on a carbon support than on a silica support. We ascribe this to a stronger interaction of CuO_x with silica.

Figure 5.

Ex situ reduction in 0.5 L min^–1 g_cat^–1 flow of 5 vol % H₂/Ar at 5 K min^–1 in 1 bar, mimicking the conditions used during in situ H₂ treatment monitored by XAS.

The reducibility of the CuZn-15/SiO₂ and CuZn-15/C catalysts was also investigated with in situ XAS under similar conditions as for the ex situ H₂ treatments. Time-resolved X-ray absorption spectra (Figure S16) were analyzed by fitting linear combinations of the macrocrystalline references to extract the Cu oxidation state evolutions (Figure S17), showing that the CuO species in both catalysts were fully reduced to Cu⁰via the formation of Cu⁺ upon an H₂ treatment up to 543 K for 5–15 min. This was confirmed by a more in-depth study using multivariate analysis in which no prior information on the component spectra was imposed but which yielded eigenspectra that corresponded well to the macrocrystalline Cu references (Figure S18). This full reduction of CuO in ZnO_x-promoted CuO nanoparticles has also been reported in the literature.^14,24,42,58

During the H₂ treatment we also studied changes in the ZnO_x oxidation state by in situ XAS. Figure 6 presents the time-resolved, normalized X-ray absorption near edge structures (XANES) and first derivatives at the Zn K-edge before and upon the H₂ treatment. We start with ZnO_x species in the Zn(II) oxidation state for both CuZn-15/SiO₂ (frames A and C) and CuZn-15/C (frames B and D) catalysts, as clear from the comparison to the first derivative of the ZnO reference. Upon heating in an H₂ atmosphere the Zn K-edge shifted to a lower energy (indicated by the arrows), showing that partially reduced ZnO_x was formed in both catalysts. The dominant features were still due to the presence of Zn²⁺, as clear from the peak at 9.6626 keV on the first derivatives, although its intensity had slightly decreased.

Figure 6.

(A,B) Time-resolved, normalized absorption and (C,D) corresponding first derivatives of in situ XANES spectra at the Zn K-edge of the (A,C) CuZn-15/SiO₂ and (B,D) CuZn-15/C catalysts (solid lines). The spectra are depicted in the initial state at 298 K, during a treatment in 20 vol % H₂/He up to 543 K in 1 bar each ca. 5.7 min, and finally in an H₂ atmosphere at 453 K. Dashed lines show the first derivatives of macrocrystalline ZnO, Zn₂SiO₄, and Zn foil references at 298 K.

Interestingly, the CuZn-15/SiO₂ catalyst (frame A) displayed two distinct peaks in the normalized XANES spectra (indicated with α and β). This peak combination has been reported before and is ascribed to the presence of zinc silicates in a single phase such as Zn₂SiO₄.^37,59−62 Yet, the CuZn-15/C catalyst (frame B) only had one single, broad peak (indicated with γ), in line with a ZnO_x phase which was also observed in electron microscopy (Figure 1, frame E). The estimated, average Zn oxidation number (ON) was slightly lower for the CuZn-15/C catalyst than for the CuZn-15/SiO₂ catalyst (+1.3 vs +1.6, see also Table S5). Multivariate analysis on the Zn K-edge is more challenging than for the Cu K-edge, as the XAS signal is lower. Extraction of the components (Figure S20) suggests the presence of three distinct phases for the CuZn-15/SiO₂ catalyst. The eigenspectrum of one of the components resembles that of Zn₂SiO₄ and its contribution is relatively stable throughout the experiment, indicating the presence of a substantial amount of Zn spectator species in the CuZn-15/SiO₂ catalyst. It has to be noted that because of the relatively low Zn loading, these spectator species were not observed by infrared spectroscopy (Figure S26). For the carbon-supported catalyst, a significant contribution of a compound with a relatively high absorption at lower energies is found (Figure S20). The phases do not fully match with the macrocrystalline Zn references, which indicates highly dispersed species of low crystallinity and/or not very well-defined mixed phases. This confirms the impact of the support on the ZnO_x speciation: on an oxidic support, the majority of the Zn species is irreducibly bound to the oxidic support and a fraction of the Zn is bound in silicate species, while on a carbon support a highly dispersed ZnO_x phase with an average Zn oxidation number significantly lower than +2 is present, which might be due to a high defect density in the ZnO (creating oxygen vacancies and a lower average ZnO state) or possibly the intermixing of fully reduced Zn in the compounds.

3.5. Nature of the ZnO_x under Working Conditions

The in situ H₂-treated catalysts were used for high-pressure methanol synthesis by CO, CO/CO₂, and CO₂ hydrogenation. Upon catalysis, no significant changes in the oxidation state and local coordination of the Cu⁰ were detected (Figure S21), in line with results published earlier.^39,45Figure 7 shows the normalized, operando XAS spectra of the Zn K-edge in the XANES region after 160 min in an H₂/CO feed and after 160 min of subsequent H₂/CO/CO₂ feed for the CuZn-15/SiO₂ (frames A and C) and CuZn-15/C (frames B and D) catalysts. The overall results, including CO₂ hydrogenation, are shown in Figures S19 and S21–S24. Note that because of the XAS setup restrictions, the amount of catalyst and hence the conversion was limited (Figure S25). For the CuZn-15/SiO₂ catalyst, the ZnO_x was only slightly further reduced during methanol synthesis (frames A and C) with an estimated Zn ON of +1.1. This was confirmed by a measurement after cooling the catalyst down to room temperature to obtain sharper features (average Zn ON of +1.2, Figure S19). Features that were attributed to zinc silicates were dominant at all stages in the XAS spectra for the CuZn-15/SiO₂ catalyst, as confirmed by multivariate analysis (Figure S20). The fact that only a slight reduction of the Zn(II) is observed when using oxidic supports and that the Zn species strongly interact with the support is in line with earlier reports using oxidic supports.^37,45

Figure 7.

(A,B) Normalized absorption and (C,D) corresponding first derivatives of operando, normalized XANES spectra at the Zn K-edge of the (A,C) CuZn-15/SiO₂ and (B,D) CuZn-15/C catalysts (solid lines). Depicted during H₂/CO (and subsequent H₂/CO/CO₂) conversion at 20 bar and 533 K, each after 160 min. Gas compositions: H₂/CO/He = 60/30/10 vol % and H₂/CO/CO₂/He = 60/27/3/10 vol %. Dashed lines show the initial catalyst state (ZnO), macrocrystalline Zn₂SiO₄ and Zn₃₀Cu₇₀, and Zn foil at 298 K.

Remarkably, in the CuZn-15/C catalyst, a large fraction of metallic Zn was formed during methanol synthesis at 20 bar (Figure 7, frames B and D). This observation was confirmed by multivariate analysis, which showed a resemblance of the independently extracted eigenspectrum of the Zn species to the XAS spectrum of metallic Zn (Figure S20). The average Zn ON was only ca. + 0.6. Assuming that the Zn species are either in the Zn(II) or Zn(0) oxidation state, this means that about 70% of the Zn species was completely reduced. With the addition of 3 vol % CO₂ in the feed, a slightly less-reducing gas atmosphere was created. Nevertheless, the Zn ON decreased further with time to ca. + 0.5 after nearly 3 h in the H₂/CO/CO₂ feed, which is probably rather an effect of time than feed composition. Upon switching to a pure H₂/CO₂ feed, the average Zn ON slightly increased to + (0.6–0.8). An increase is expected in a more oxidizing gas feed, as it is also predicted computationally that there will be a slight dependence of the ON (and hence probably the Zn coverage) on the exact feed composition.¹³ However, even under these conditions, most of the Zn species remain in the fully reduced state under operando conditions. This means that the relatively high Zn ONs in methanol synthesis systems reported until now (in the presence of an oxidic support) can probably be explained by a strong promoter–support interaction, and hence, a large fraction of the Zn promoter species is being chemically bound to the oxidic support (and hence inactive). In contrast, our use of a weakly interacting carbon support allows us to assess an average Zn oxidation state and coordination number that are much more representative of the true nature of the active ZnO_x promoter phase during methanol synthesis.

Several hypotheses have been postulated for the ZnO_x speciation during Cu-catalyzed methanol synthesis. First, the beneficial effect of the ZnO_x promoter was ascribed to the so-called strong metal–support interaction (SMSI) with slightly reduced ZnO having a high affinity for the Cu⁰ metal and partially covering the Cu nanoparticle surface.^35,63 Alternative explanations involve the influence of ZnO_x on the structure of the Cu nanoparticles. For instance, it was proposed that specific steps sites were exceptionally active sites on the Cu⁰ surface and that these step sites were stabilized by Zn⁰ atoms.^10,39,64 Metallic Cu and Zn are quite miscible; up to 33 at% Zn can dissolve in Cu (solid solution) at temperatures between 473 K and the melting point (>1175 K).⁶⁵ Some groups proposed that the active site was related to the decoration of Cu⁰ nanoparticles with Zn⁰ atoms and shallow diffusion of Zn⁰ atoms into the Cu surface.^13,21,44 However, results from ex situ and low-pressure studies have limited value, as it is known that the catalytically active phase dynamically adjusts to the working conditions.⁶⁶ A recent high-pressure operando study, based on oxidic supports, concluded that a distorted ZnO_x layer was the majority phase under working conditions with at most 9% of the Zn being present as Zn⁰ atoms.²⁴ Our experiments clearly show that, if a strong interaction of the Zn species with an oxide support is avoided, a much more truthful picture of the active fraction of the Zn promoter species under high-pressure methanol synthesis conditions is obtained, and that this fraction is clearly reduced to zerovalent Zn upon prolonged methanol synthesis conditions (Figure 7).

Zooming in on the local coordination of the Cu and Zn atoms during high-pressure methanol synthesis, we analyzed the extended X-ray absorption fine structure (EXAFS) region of the XAS data. Figure 8 shows the EXAFS data on the Zn K-edge in R-space for the CuZn-15/SiO₂ (frame A) and CuZn-15/C (frame B) catalysts in the initial state and upon heating in an H₂ atmosphere. The EXAFS-derived R-spaces at the Cu K-edge and the EXAFS fitting parameters are available in Figure S24 and Tables S6–S9, respectively. The initial spectra of both catalysts have a main peak at 1.50 Å in the Fourier transform, which corresponds to first-shell Zn–O bonds such as in ZnO with a bond length of 1.97 Å. The R-space of the CuZn-15/SiO₂ catalyst (frame A) closely resembles that of the Zn₂SiO₄ reference, showing that a majority of the Zn atoms is bound to the oxide support, as reported before.³⁷ No contribution of second-shell Zn–Zn bonds was observed (frames A and B) (which could be expected in crystalline ZnO at 2.91 Å in the nonphase corrected Fourier transform, corresponding to a real bond length of of 3.2 Å⁴¹), indicating the absence of larger ZnO crystallites in both samples and in line with the high ZnO_x dispersion observed by TEM (Figure 1). During in situ H₂ reduction, the Zn–O bond intensity at 1.50 Å apparently decreased for both catalysts, but this was simply due to the increasing measurement temperature³⁵ as the overall peak intensities significantly increased in the spectra taken at room temperature after catalysis (purple lines in frames C and D) compared with the spectra taken during the last stage of catalysis at high temperature. It is important to note that in neither of the catalysts was Zn⁰ formation observed during reduction in atmospheric-pressure H₂ (the Zn–Zn or Zn–Cu bond fingerprint is expected at 2.30 Å in the nonphase corrected Fourier transform, its position is indicated with an unlabeled arrow in the frames). After the in situ H₂ reduction, the first-shell Cu–Cu coordination number (CN) in metallic Cu was ca. 11 (for bulk Cu⁰ this CN is 12), and it remained unchanged for both catalysts, independent of the exact feed. This means that no change in the Cu nanoparticles was observed upon exposure to working conditions: neither a significant fraction of oxidized copper nor the presence of highly dispersed copper. This is in agreement with the fully reduced Cu observed in the spectra at the XANES region.

Figure 8.

Fourier-transformed EXAFS spectra at the Zn K-edge of the (A,C) CuZn-15/SiO₂ and (B,D) CuZn-15/C catalysts (solid lines). (A,B) Depicted during in situ reduction in the initial state at 298 K and in an H₂ atmosphere at 453 K after an H₂ treatment at 1 bar (for conditions, see Figure 6). (C,D) Depicted during H₂/CO (and subsequent H₂/CO/CO₂) conversion at 533 K and 20 bar (for conditions, see Figure 7) and after catalysis. Dashed lines depict the macrocrystalline ZnO, Zn₂SiO₄, Zn₃₀Cu₇₀, and Zn foil references. The unlabeled arrows indicate the position of Zn–Zn or Zn–Cu bond formation.

We continue the EXAFS analysis under operando methanol synthesis conditions by focusing on the Zn local surrounding. Figure 8 (frames C,D) shows selected R-spaces from the EXAFS data on the Zn K-edge for both catalysts (for the complete set, please see Figure S23). For the CuZn-15/SiO₂ catalyst during CO hydrogenation, only a very minor fraction of metallic Zn was observed (signal around 2.30 Å indicated with the arrow in frame C), in line with earlier reports on oxide-supported catalysts.^24,35,63 Nevertheless, a change was observed, as the first-shell Zn–O CN decreased from 4 (as in bulk ZnO and Zn₂SiO₄) to 2.6 ± 0.5 and a very low second-shell Zn–Zn or Zn–Cu (from here onward denoted as Zn–M) CN of 2.2 ± 1.4 was obtained (for bulk Zn⁰ this CN is 12). This indicates a very slight change in the average Zn surroundings, but because of the small changes and the very similar Zn–Zn and Zn–Cu bonding distances, it is not possible to analyze this in detail. Overall, the signal remains dominated by features that are attributed to Zn silicate species, and there is very little difference between the reduced fresh catalyst and that under working conditions.

Interestingly, the CuZn-15/C catalyst (frame D) displayed large changes when switching to working conditions, which was already expected from the zerovalent Zn as evidenced by the XANES analysis (Figure 7, frame B). An average Zn–M bond length of 2.54 Å (close to that of 2.66 Å of the Zn⁰ foil reference) and a quite high Zn–M CN of 6.1 ± 1.3 were obtained. This is a clear supporting evidence for the large fraction of zerovalent Zn species in the active catalysts. Upon prolonged exposure (while slightly enriching the feed with CO₂), the increase in coordination number continues to a Zn⁰ CN of ca. 8. This means that the majority of the Zn promoter species is present in metal nanoparticles. The very similar Zn–Zn and Zn–Cu bonding distances do not allow us to unequivocally derive the nature of these metal nanoparticles. However, the zerovalent Zn is very likely located in Cu–Zn nanoparticles. The high Zn–M coordination number suggests that the Zn does not remain as adatoms or a monolayer on the outside of the Cu particle. The diffusion coefficient of Zn⁰ in Cu⁰ strongly depends on the Cu particle size and temperature (see also Table S10)⁶⁷⁻⁶⁹ but is high enough to support a full distribution of the Zn⁰ throughout the relatively small Cu⁰ particles at the time scale of hours, in line with the XRD pattern of the used CuZn-15/C catalyst showing a small downshift of the Cu⁰ diffraction line and hence suggests CuZn alloy formation (Figure S8, frame D).

Interestingly, the time scale of the formation of highly coordinated zerovalent Zn is quite in line with the generally observed activation period for Cu methanol synthesis catalysts exposed to high-pressure working conditions.^5,9 To our knowledge, no clear explanation for this activation period has so far been reported in academic literature, but our results suggest that the gradual reduction of Zn(II) to active Zn(0) promoter species might be an important factor in this activation.

Figure 9 summarizes the results of our study by depicting the ZnO_x speciation in silica- and carbon-supported Cu catalysts containing 15 at% Zn/(Cu + Zn) after reduction as well as during high-pressure methanol synthesis. On both supports, the Cu²⁺ nanoparticles (depicted in dark blue) were fully reduced to Cu⁰ nanoparticles in 5–15 min exposure to 1 bar H₂ at 543 K. Even during high-pressure methanol synthesis with a H₂/CO₂ feed, no significant subsequent change in the oxidation state of the Cu was observed. Using an oxidic support, which is standard in commercial catalysts and most academic studies, it was difficult to derive detailed information about the speciation of the Zn component (depicted in red) that was active as a promoter. Only slight changes in the average Zn speciation were observed (in line with earlier literature), as the signal was dominated by Zn species that had a strong interaction with the oxidic support (depicted as a layer of ZnSiO_x), and these species remained dominant under all (also methanol synthesis) conditions. In contrast, using a much less strongly interacting carbon support, allowed us to follow the fraction of the Zn species that was closely affiliated with the Cu nanoparticles and hence most likely represents the active Zn promoter species during catalysts. Under methanol synthesis conditions, the relevant ZnO_x phase is in a deeply reduced state with an average Zn oxidation number of only +0.6. Assuming that only Zn²⁺ and Zn⁰ species exist, this means that ca. 70% of the ZnO is fully reduced to Zn⁰. The Zn–M coordination number was as high as 8 during methanol synthesis working conditions, showing that the Zn⁰ is almost fully coordinated with other metal atoms and has likely mostly diffused into the Cu nanoparticles. It is likely that this Zn speciation for the active promoter species is also relevant for the more conventional oxide-supported catalysts, to which much more Zn must be added to reach an optimum promoter effect, which is probably explained by the fact that a large fraction of the added Zn is not active as promoter.

Figure 9.

Schematic representation of the ZnO_x speciation in the (A-C) CuZn-15/SiO₂ and (D-F) CuZn-15/C catalysts, depicted (A,D) in the initial state, (B,E) after reduction, and (C,F) under working conditions at 20 bar and 533 K. The various shades of between blue (Cu) and red (Zn) in the CuZn particles represent the relative extent of Zn⁰ incorporation into the Cu⁰ nanoparticles based on the estimated Zn ONs from the XANES analysis. For frames B and C, separate Cu⁰ nanoparticles may exist next to alloyed CuZn particles.

4. Conclusions

Cu nanoparticulate catalysts on graphitic carbon were prepared, and compared to SiO₂-supported catalysts, to better understand the interaction between the Cu and the Zn-based promoter species and the speciation of Zn acting as a promoter during high pressure methanol synthesis. With a modest amount of ZnO_x promoter, the methanol formation for the CuZnO_x/C catalyst was significantly faster than for a CuZnO_x/SiO₂ catalyst with similar Cu particle size in a pure H₂/CO feed. This difference was even much more pronounced in a CO₂-enriched syngas feed. Importantly, the use of graphitic carbon model supports allowed us to reveal the true speciation of the active fraction of the Zn-based promoter under commercially relevant methanol synthesis conditions. The vast majority of the Zn(II) is reduced all the way to Zn(0) during methanol synthesis at 20 bar. Also the Zn coordination number was high, making it likely that the Zn(0) diffused into the Cu nanoparticles. The characteristic time for this diffusion corresponds to the activation time that is generally observed with this type of catalysts. For the first time this gives direct insight into the nature of the active fraction of the Zn-based promoter in high pressure methanol synthesis, not obscured by the commonly large fraction of Zn species that strongly interacts with an oxidic support, and hence dominates the structural characterization results.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.1c05101.

• Brief descriptions of electron imaging, crystallinity, reducibility, and the porosity of the catalysts; additional information on the catalyst activity and stability; complete set of XAS data; and details on the calculation of catalyst performance (PDF)

Author Contributions

R.D. and P.d.J. conceived and planned the experiments. R.D. measured the performance of the catalysts, analyzed the catalytic data, and drafted the manuscript together with P.d.J. S.T. performed electron microscopy imaging by EDX, and G.T. performed the ultramicrotomy. R.D., L.B., G.T., X.C., and P.d.J. performed the XAS experiments. C.L.F. designed the experimental method and assisted with data interpretation. L.B., G.T., A.v.d.E., F.M., and F.d.G. analyzed the XAS data. P.d.J. supervised the work. All authors contributed to the final manuscript.

The authors declare no competing financial interest.