Research ArticleChemistry

Freezing copper as a noble metal–like catalyst for preliminary hydrogenation

See allHide authors and affiliations

Science Advances  21 Dec 2018:
Vol. 4, no. 12, eaau3275
DOI: 10.1126/sciadv.aau3275

Abstract

The control of product distribution in a multistep catalytic selective hydrogenation reaction is challenging. For instance, the deep hydrogenation of dimethyl oxalate (DMO) is inclined to proceed over Cu/SiO2 catalysts because of inevitable coexistence of Cu+ and Cu0, leading to hard acquisition of the preliminary hydrogenation product, methyl glycolate (MG). Here, the oriented DMO hydrogenation into MG is achieved over the sputtering (SP) Cu/SiO2 catalysts with a selectivity of more than 87% via freezing Cu in a zero-valence state. Our density functional theory calculation results revealed that Cu0 is the active site of the preliminary hydrogenation step, selectively converting DMO to MG via •H addition, while Cu+ is a key factor for deep hydrogenation. The prominent Coster-Kronig transition enhancement is observed over SP-Cu/SiO2 from Auger spectra, indicating that the electron density of inner shells in Cu atoms is enhanced by high-energy argon plasma bombardment during the SP process. Thus, the “penetration effect” of outermost electrons could also be enhanced, making these Cu nanoparticles exhibit high oxidation resistance ability and present noble metal–like behaviors as Au or Ag. Therefore, the regulation of Cu chemical properties by changing the electron structure is a feasible strategy to control the hydrogenation products, inspiring the rational design of selective hydrogenation catalysts.

INTRODUCTION

The clean utilization of coal resources to synthesize high value–added chemicals is greatly desired with the increasing energy and environmental problems (1). One of the successful industrial applications is the synthesis of alcohols from coal via syngas (CO and H2), which contains the coupling of CO with nitrite esters to dimethyl oxalate (DMO) and the sequential hydrogenation of DMO to alcohols (2). Typically, the tandem hydrogenation of DMO can obtain three main products, including methyl glycolate (MG), ethylene glycol (EG), and ethanol (EO). Now, copper-based catalysts have been intensively investigated (36), and very high yields of EG (or EO) and rational insights have been achieved in previous work (79). In Cu-based catalysts, the balance of Cu0 and Cu+ is indispensable for deep hydrogenation to EG or EO from DMO. MG, as a preliminary DMO hydrogenation product, is an essential intermediate with higher commercial price than EG and EO for the synthesis of pharmaceutical products, fine chemicals, and perfumes. However, MG is difficult to be gained via copper catalysts, because the thermodynamic constant of the second hydrogenation step is two orders of magnitude larger than that of the first hydrogenation step (10). Therefore, for tandem hydrogenation reactions, the development of an efficient catalyst to control and regulate the target products is still a great challenge for both academia and industry.

In general, the synthesis of MG via DMO needs a moderate reaction condition and a catalyst with a relatively weak hydrogenation property, such as Ag-based and Au–Ag–based noble metal catalysts (1012), and nonsilica-supported Cu catalysts, such as hydroxyapatite and activated carbon (13, 14). Silver surfaces, unlike copper, generally lack affinity toward H2 dissociation because of the filled d band and contribute to MG production (15). However, the application of catalysts with high loading of noble metals is undesired. The highly efficient silver catalyst also suffers from by-products due to its activity in chemoselective hydrogenation of both C=C and C=O bonds (10). By comparison, copper-based catalysts are still a good alternative in aspect of product regulation if their hydrogenation capability can be controlled to a moderate level since the copper sites are very selective toward hydrogenation of C=O bonds rather than the hydrogenolysis of C=C bonds (16).

In this study, freezing Cu species of a Cu/SiO2 catalyst in a metallic state with low reducibility is achieved via argon plasma sputtering (SP) from a copper target by a self-made SP. The sputtered Cu atoms are distributed on SiO2 uniformly with the continuous hexagonal rotation and mechanical vibration in the polygonal barrel-SP process (1719). This process can be directly used in the DMO hydrogenation reaction without any reduction pretreatment and is hard to be oxidized even at a high temperature under oxidizing agent–containing atmosphere. The SP-Cu/SiO2 catalyst exhibits high selectivity toward MG in a wide range of temperatures, attributed to its relatively lower reducibility than the conventional ammonia evaporation (AE)–made one. The catalytic behavior of sputtered Cu on product distribution of DMO hydrogenation is very similar to that of noble metals (Ag or Au), resulting from the electron structure change in the bombardment of high-energy argon plasma. The present study is expected to demonstrate the noble metal–like property of sputtered Cu and resolve the catalytic mechanism of Cu0 and Cu+ species in preliminary and deep hydrogenation processes, exploring the potential applications of sputtered Cu in substitution of noble metal catalysts.

RESULTS

Chemical state of copper over SP- and AE-made Cu/SiO2 catalysts

We carried out x-ray diffraction (XRD) characterizations to understand the nature of copper species and their variations during a DMO hydrogenation reaction (Fig. 1A). As reported before, the weak and broad diffraction peaks at ca. 31.2° and 35.8° suggest the presence of copper phyllosilicate on fresh AE-Cu/SiO2 (20). This copper compounds need to be reduced into metallic Cu or Cu2O before reaction. Then, Cu2O species become predominant during the reaction. For the reduced AE catalyst, it is difficult to distinguish the copper state from the weak XRD diffraction peaks since Cu is well dispersed or small. Therefore, we also provided the extended x-ray absorption fine structure (EXAFS) results (Fig. 1B). It gives further evidence on the coexistence of metallic Cu and Cu2O species in which the latter is the primary phase, as compared to reference Cu and Cu2O. For SP-Cu/SiO2, only metallic Cu species are detected in the as-prepared sample, and they keep stable during the reaction. That indicates that the metallic Cu can be formed and dispersed on the support in the process of SP preparation without any reduction treatments, which is also proved by temperature-programmed reduction (TPR) in Fig. 1C. A large amount of H2 is consumed at around 240°C when Cu precursors are reduced into metallic Cu or Cu2O for AE-Cu/SiO2. However, only a small H2 consumption peak derived from the passivation of active Cu species can be observed on SP-Cu/SiO2 during reduction from 100° to 500°C. As reflected in Fig. 1B, the SP-used catalyst shows only Cu–Cu coordination in metallic Cu, suggesting that copper is “frozen” in a metallic state, which is well consistent with XRD analysis.

Fig. 1 Chemical state characterizations of SP- and AE-made Cu/SiO2 catalysts.

(A) XRD patterns of fresh and used Cu/SiO2 samples. a.u., arbitrary units. (B) EXAFS spectra of used SP and reduced AE Cu/SiO2 samples. FT, Fourier transform. (C) H2-TPR results of used SP-Cu/SiO2 and reduced AE-Cu/SiO2 samples. (D) In situ DRIFTS results of SP- and AE-made Cu/SiO2 samples. (E) TEM image of fresh SP-Cu/SiO2 and (F) TEM image of fresh AE-Cu/SiO2.

The in situ diffuse reflectance infrared (IR) Fourier transform spectra (DRIFTS) results in Fig. 1D show strong adsorption of CO on the Cu+ and Cu0 species of AE-Cu/SiO2, corresponding to the IR absorbance peaks at about 2127 and 2102 cm−1, respectively (13). In comparison, no CO adsorption is observed on the SP-Cu/SiO2 sample, neither on Cu0 nor on Cu+, indicating that the interaction between CO and Cu species is very weak. Detail CO adsorption processes can be seen in fig. S1. It reveals that the chemical behavior of Cu0 species on SP-Cu/SiO2 and AE-Cu/SiO2 samples is completely different. The former exhibits nonreactivity in supplying electrons, which is anomalous for Cu and more approximate to noble metals. The particle size effects on the adsorption property can be excluded since both Cu particles are centered about 3 to 5 nm with the similar copper loading amount and highly dispersed on the silica support as reflected from transmission electron microscope (TEM) images in Fig. 1 (E and F). It is noteworthy that the Brunauer-Emmett-Teller (BET) surface area is slightly decreased by only 5.6% after physical SP of Cu nanoparticles on silica as compared with the raw silica support, whereas it is remarkably increased from 283 to 401 m2 g−1 after the AE process. These findings suggest that the copper deposition process on silica is completely different despite the similar dispersion of Cu nanoparticles. Assisted by a powerful Ar plasma stream, metallic copper nanoparticles are shocked down and homogeneously dispersed on the silica surface with physical polygonal rotation and mechanical vibration. This is a physical deposition process without destroying the pore properties of silica. By contrast, the layered copper phyllosilicates [Cu2Si2O5(OH)2] are formed as a key precursor in the chemical AE, dispersing Cu nanoparticles in silica, enlarging the surface area and providing stable Cu+ in DMO hydrogenation (7).

Electron structure information of copper

We carried out x-ray photoelectron spectroscopy (XPS) surface analysis to identify the chemical valence and the electron structure of copper. The binding energy (BE) peak at ca. 932 to 933 eV in Cu 2p spectra (Fig. 2A) is generally attributed to the presence of Cu+ and/or Cu0 species on the catalyst surface. The absence of ca. 933.5 eV to Cu2+ and the characteristic satellite peaks (940 to 945 eV) demonstrates that the copper species in the fresh SP and reduced AE catalysts are of low valence (<2) and that they cannot be oxidized to CuO during the reactions (21). In detail, the Cu 2p3/2 peaks of AE-made catalysts appear at a BE of 933.1 eV, while that of SP-made catalysts shift toward a lower BE of 931.8 eV, suggesting that copper is in a lower valence in the latter.

Fig. 2 Electron structure information of copper.

(A) Cu 2p XPS spectra. (B) Cu L3M45M45 XAES spectra. Before XPS and XAES experiments, Ar ionic SP was preperformed with a beam energy of 3 keV and a SP rate of 20 Å min−1 for 7.5 min to clean the surface. (C) Schematic of enhanced CK transition and electron penetration effect for SP-made Cu/SiO2 prepared with high-energy argon plasma bombardment. EAuger, Auger electrons; Ep, incident photon energy.

The detailed Cu state of low valence can be precisely distinguished by L3M45M45 x-ray–excited Auger electron spectroscopy (XAES) spectra (Fig. 2B). According to literatures, the asymmetric and broad Auger peak of AE-made supported Cu/SiO2 catalysts can be deconvoluted into two symmetrical peaks centered at 914.8 and 918.2 eV, corresponding to Cu+ and Cu0, respectively (13, 22, 23). The peak position and their intensity are shown in table S1. About 56% Cu species in the reduced AE catalyst is Cu+. The increasing content of Cu+ from 56.0 to 78.1% during the reaction suggests that about half of Cu0 is oxidized to Cu+ by DMO in the catalytic hydrogenation reaction.

In Fig. 2B, the Auger peaks for sputtered Cu catalysts can be split into several Cu terms, primarily at ca. 917 and 919 eV accompanied by small characteristic peaks at 913 to 917 and 921.7 eV. The detailed peak information is shown in table S2. They are very close to the one of pure Cu bulk in fig. S2 rather than the supported AE-Cu catalysts, demonstrating that Cu is primarily in the metallic state in the case of SP catalysts. It is noteworthy that the kinetic energy of primary Cu0 Auger peak for AE catalyst is 0.8 eV lower than that of Cu bulk or sputtered copper, probably resulting from the stronger interaction between AE-Cu and silica. The sputtered copper nanoparticles are anchored on the surface of silica and interact with silica in a weak physical force. That is a universal phenomenon in our previous SP-made catalysts (18, 19, 24).

In general, the Cu Auger transition is produced from a separate L3 (2p3/2) core hole decay via the Auger process involving two M45 (3d) electrons resulting in a final 3d8 configuration (25). In detail, according to the work by Pauly et al. (25), the primary five final-state terms are observed in Cu bulk spectra as shown in fig. S3 and table S2. The peaks at 919 and 922 eV are defined as 1G and 3F, respectively, which split from L–S coupling corresponding to two 3d holes in metallic copper. These two terms are ascribed to “normal” Cu, and their Auger contribution to the measured L3M45M45 transition is 81.9%. The other three Auger peaks in the range of 913 to 917 eV are defined as 2F*, Sum*, and 4F*, which are characteristic features of the L2L3M45 Coster-Kronig (CK) transition. From the schematic in Fig. 2C, L2L3M45 CK transition will occur if the electron energy level of Auger ionization hole (2p1/2) and filling hole (2p3/2) is in the same core level (2p), leaving an “extra” hole in the M45 level for the initial state of normal L3M45M45 transition (26, 27). Thus, the extra Auger vacancy satellite transition corresponds to a 3d7 configuration after the common L3M45M45 process, which is shifted to lower kinetic energy due to the Coulomb interaction between this M45 spectator vacancy and the Auger electron (25). The contribution of all extra structures to the total Cu contribution (62.9%) in the SP-Cu/SiO2 catalyst is much higher than those of pure Cu bulk (18.1%), suggesting that the CK transition is obviously enhanced. Besides, the composition of various copper terms in the sputtered Cu catalyst remains almost stable even after DMO hydrogenation at a high temperature (>280°C). The slightly promoted ratio of extra Cu0 from 62.9 to 65.5% may originate from the formation of a few Cu+ species with an overlapped Auger peak at ca. 914.8 eV.

The enhanced CK transition phenomenon is probably attributed to the increment of the electron density of inner shells in Cu atoms by high-energy argon plasma bombardment during the SP process. In general, the single electron in the 4s orbit has a chance to arrive at inner shells owing to its lower energy level than those in the 3d orbit, which is called as “penetration effect.” As described in Fig. 2C, the penetration effect could also be enhanced when the electron density of inner shells increases in the assistance of the continual plasma bombardment. In this case, the outermost electron in SP-Cu is more difficult to escape from Cu atom, making these Cu nanoparticles exhibit a higher oxidation resistance ability than AE-Cu and Cu bulk, and thus may present noble metal–like behaviors.

Oxidation resistance ability and noble metal–like property

The N2O molecule is well known in oxidation of Cu0 to Cu+ and usually used in the surface area measurement of Cu0 species (28). Here, N2O temperature-programmed oxidation (N2O-TPO) is conducted to investigate the oxidation behavior of metallic Cu in SP- and AE-made Cu/SiO2 samples via monitoring the released N2 signal, as shown in Fig. 3A. It can be found that the N2 signal is obviously detected immediately after introducing N2O into the system. In addition, this oxidation process quickly finishes in about 10 min. That reveals that the Cu0 species in AE-Cu/SiO2 are very active and can be easily oxidized by N2O even at room temperature. This is in accordance with the oxidation of Cu0 to Cu+ by ester or products during a catalytic reaction (29, 30). In contrast, most of the metallic Cu cannot be oxidized by N2O at room temperature. Only a small amount of Cu on the surface of SP-Cu/SiO2 can be oxidized, which may correspond to those active Cu species reduced in H2-TPR measurement (Fig. 1C). It is observed that the Cu0 species could react with N2O when raising temperature. However, the oxidation process passes through about 40 min until the temperature reaches 200°C. It suggests that the metallic Cu prepared by the SP method is more difficult to be oxidized in an oxidation environment, very similar to that expressed on stable Au or Ag nanoparticles. It may be responsible to the enhanced penetration effect, where the isolated electron in a 4s orbit penetrates inside the 2p orbit, leaving the filled 3d orbit as the outermost layer. That makes Cu difficult to lose electrons and be oxidized.

Fig. 3 Oxidation resistance ability and noble metal–like property for sputtered copper.

(A) N2O-TPO results of SP- and AE-made Cu/SiO2 catalysts, where the top-left panel shows a N2 release at 25°C with time on stream and the top-right panel shows a N2 release with temperature programming from 25° to 250°C. (B) UV-vis adsorption spectra of SP- and AE-made Cu/SiO2 catalysts. The inset shows magnification of the marked area. The Cu powder with a purity of 99% is also tested for comparison. The vertical dashed line marks the position of peak surface plasmon absorbance. (C) Photographs of various catalysts.

In general, the free electrons in noble metals (especially d electrons in silver and gold) are free to travel through the material (31). Light in resonance with the surface plasmon oscillation causes the free electrons in the metal to oscillate. As the wave front of the light passes, the electron density in the particle is polarized to one surface and oscillates in resonance with the light’s frequency, causing a standing oscillation (31). This phenomenon is called surface plasmon resonance (SPR) and exclusive in noble metals, such as Ag and Au, because of their air-stable nature (32). The SPR is reported to be also detected in Cu nanoparticles sputtered at NaCl substrates and is determined by an absorption peak at ca. 570 nm in ultraviolet-visible (UV-vis) spectroscopy (33). Here, an absorption peak at 563 nm is detected in UV-vis spectroscopy in Fig. 3B over the used SP-Cu/SiO2 catalyst, while no obvious adsorption peak signals can be observed on both reduced and used AE-Cu/SiO2 catalysts. The fresh SP-Cu/SiO2 catalyst shows an obscure adsorption peak at the same wavelength as the used one, probably owing to its smaller Cu particle size than the latter (34). We also find that pure Cu powders are difficult to exhibit the SPR phenomenon. These findings reveal that Cu nanoparticles made by the SP method exhibit different properties from ordinary supported Cu nanoparticles or pure Cu powders with respect to electron structure and oxidation reduction capability. It can also get some clues from the picture of these samples in Fig. 3C. For the SP-Cu/SiO2 catalysts, the color changed from dark brown to red, which certified the existence of metallic copper after the reaction. These Cu particles, which have weak interactions with supports, exhibit properties like pure Cu. By comparison, the one made by AE shows black after the reaction as we generally observe on supported Cu nanoparticles.

Catalytic performance in DMO hydrogenation

The conversion and product distribution of DMO hydrogenation reaction are shown in Fig. 4 (A and B). The process of DMO hydrogenation is a typical temperature-controlled reaction as the hydrogenation ability of Cu catalyst is very sensitive to temperature (14). For the AE-Cu/SiO2 catalyst, the DMO hydrogenation as a tandem reaction first generate the partial hydrogenation product MG and immediately convert to EG at a low temperature range of 170° to 230°C. After those deep hydrogenation products, alcohols are generated at a high temperature above 230°C. With the temperature decreasing from 230° to 170°C, EG selectivity is gradually promoted with a maximum value of 96% at 240°C. But if the temperature further decreases to 175°C, then both EG selectivity and DMO conversion will quickly drop to 77 and 71%, respectively, in parallel with an enhanced MG selectivity to 21%. MG will rapidly become prominent in the final product when keeping a very low temperature below 175°C for several hours (fig. S3A). This reflects a typical inactivation for DMO hydrogenation into EG.

Fig. 4 Catalytic performance over various Cu/SiO2 catalysts.

Conversion (conv.) and product distribution of DMO hydrogenation reaction over AE-Cu/SiO2 (A) and SP-Cu-SiO2 (B) catalysts. (C) and (D) show product distributions of the tested catalysts at 240° and 280°C, respectively. Selec., selectivity. (E) Stability tests of AE- and SP-made Cu/SiO2 catalysts.

Very surprisingly, the maximum selectivity of MG over the SP catalyst reaches as high as 87% with a DMO conversion of 29% at 250°C. The selectivity of MG continuously stays above 50% even when the temperature reaches 280°C. The product distribution in Fig. 4 (C and D) shows that deep hydrogenation happens on the AE catalyst with EO and low-carbon alcohols (C3-4 alcohol) as primary products at 240°C. It seems that the EG and EO as the deep hydrogenation products still cannot climb to the maximum selectivity even at 280°C. In addition, the selectivity of MG is compared at a similar low conversion using AE-Cu/SiO2 to that of the SP-made catalyst. In a low conversion of ca. 20% over the AE catalyst at below 175°C, the MG selectivity can reach 86%, which is very close to the optimal value of 87% over the SP catalyst at 250°C. However, the similar performance is only achieved during the inactivation process at a low temperature for DMO hydrogenation with a rapidly decreasing activity. In general, the superiority of the SP catalysts is apparent in terms of a very wide temperature range (230° to 290°C) for producing MG.

From the previous analysis, Cu shows different states in DMO hydrogenation in two catalysts. For the SP-Cu/SiO2, the metallic Cu exhibits an extreme oxidation resistance property even exposing to DMO at high temperature, while most Cu0 are easily oxidized to Cu+ species over the conventional AE-Cu/SiO2. Another interesting result is shown when using fresh as-prepared SP-Cu/SiO2 as an efficient catalyst without any reduction (Fig. 4C). It also exhibits a high MG selectivity of 79% with a DMO conversion of near 15%, which is close to that of the reduced catalyst. The similar product distribution results from a large amount of frozen metallic Cu in the as-prepared SP catalyst. Therefore, it is of great importance to maintain the stability of initial metallic Cu0 for preliminary hydrogenation.

We investigate stability in DMO hydrogenation for these catalysts. In a time on stream of near 30 hours (Fig. 4E), two catalysts show relative stable catalytic performance. With the same reaction conditions, the DMO conversion and MG selectivity on the AE catalyst remain at ca. 100% and below 2%, while those on the fresh SP one remain at 16% and near 80%, respectively. The comparison of BET surface area, crystallite size, and loading amount before and after the reactions are listed in table S3. All the BET surface areas for the used catalysts are slightly decreased if compared with the fresh one. The physically sputtered Cu is anchored on surface of the silica support with a weaker interaction than the AE catalyst, resulting in an inevitable growth for crystallite size after exposure to a higher temperature of 290°C. However, the results of time on stream reaction show that the catalytic performance is stable after near 30 hours, suggesting that particle size is not the main factor that affects the catalytic stability. In the future work, the introduction of a structural promoter may be a useful pathway to avoid the excessive growth of copper nanoparticles.

Furthermore, the TPO–mass spectrometry (MS) experiments (fig. S3B) for the used SP and AE catalysts after stability tests are supplemented to investigate the possible coke effect. The TPO-MS indicates a notable comparison between two used catalysts. The AE catalyst exhibits a CO2 release at 200° to 300°C and 400° to 500°C accompanied by a water release in a wider temperature range, suggesting possible adsorption of carbon species on the surface. However, no obvious CO2 and water MS signals are detected over the SP catalyst. Running in a high selectivity toward MG usually causes a quick and irreversible inactivation on Cu/SiO2 catalysts. But for the sputtered catalyst, no coke deposition and well stability are observed, which make it distinct from conventional catalysts. The frozen copper in zero valence with the absence of Cu+ in catalytic reactions results in no adsorption and activation process of C=O or C–O species on the surface, which is one possible reason for no coke formation.

To illustrate the importance of initial copper state, we use two physically mixed catalysts (PM-Cu/SiO2, CuO and SiO2 with reduction treatment; PM-Cu+/SiO2, Cu2O and SiO2 without reduction) in DMO hydrogenation and characterize their chemical state at different stages using XRD and TPR (fig. S4). Both catalysts exhibit wider reduction temperature than that for the AE catalyst, ranging from 220° to 450°C (fig. S4C). In addition, the reduction peaks of PM-Cu/SiO2 and PM-Cu+/SiO2 shift toward higher temperatures of more than 300°C as compared with the AE and SP catalysts, because of the poor dispersion and large copper particles (more than 45 nm) on physically mixed catalysts.

For the case of PM-Cu/SiO2, a pure Cu0 phase without Cu+ is shown in the initial reaction stage (fig. S4A) owing to the absence of a Cu+ stabilizer (such as copper phyllosilicate in the precursor of AE-Cu/SiO2) and the strong interaction between copper and silica, which make it distinct from the AE catalyst. Because it suffers from the poor dispersion of copper, the maximum MG selectivity over the PM catalyst is only ca. 60%, which is lower than that over the SP catalyst. However, with increasing temperature, part of unstable Cu0 species in the PM catalyst is further oxidized by DMO, resulting in the coexistence of a large amount of Cu0 and few Cu+ observed from the used XRD pattern, EG and EO selectivities are gradually enhanced as reaction temperature increases, and MG selectivity drops to ca. 30% at 290°C (fig. S4D). For the case of PM-Cu+/SiO2, we choose Cu2O as an initial state of copper without reduction. In the starting stage of reactions at 220° to 230°C, Cu+ species are active sites, transforming MG into the deep hydrogenation product EG, with a selectivity of 67.8%. The absence of Cu0 results in a low MG selectivity (ca. 20%), far less than that of more than 85% over the SP catalyst. With the increasing temperature and time on stream, unstable Cu+ species are further reduced to Cu0 by H2 in feedstock (fig. S4B) since there is no strong interaction between copper and silica, similar to the PM-Cu/SiO2 catalyst. Although the maximum of MG selectivity reaches 84.8%, it is not stable if temperature further increases. The MG yield markedly falls down to a value as low as below 10% at 280°C (fig. S4E), which is much faster than the SP-Cu/SiO2 and PM-Cu/SiO2 catalysts.

In comparison, the selectivity of MG over the SP catalyst continuously still stays above 50% even if the temperature reaches 280°C. The product distribution exhibits more stable than two PM catalysts from a high temperature range from 230° to 290°C, which provides a wider temperature range for producing MG than two PM catalysts. Therefore, it is crucial to stabilize copper into a zero valence for producing MG with a high selectivity in a wide temperature range.

Distinct product separation has been achieved through modifying the metal property, which is very similar to Ag-based or Au–Ag–based catalysts. The major products of Ag/SiO2 are MG and EG at 220° and 280°C, respectively (10). The copper nanoparticles in the SP-Cu/SiO2 catalyst can be successfully frozen in a stable metallic state without oxidation in Cu+ by oxygen in air and DMO in catalytic reactions. Therefore, the sputtered Cu has a potentiality in substitution of noble metals in hydrogenation reactions.

In copper-based catalysts, Cu0 is the active site and primarily responsible for the hydrogenation activity, while Cu+ facilitates the conversion of IMs (35, 36). Cu+ is the key factor for hydrogenation of DMO to EG or EO by enhancing the activation of the C=O group in DMO (35). In contrast, it is also reported that the addition of Ag in Cu-based catalysts shows a positive effect on the DMO-to-MG reaction due to the enhancement of proportion of Cu+ in Cu (37, 38). Moreover, nonsilica supports, such as hydroxyapatite and activated carbon, are used to increase the molar ratio of Cu+/Cu0 and constrain hydrogenation activity of copper in MG production (13, 14). It seems that the catalytic function of Cu+ and Cu0 species in different DMO hydrogenation steps is still unclear. Cu+ species are inevitable in coexistence with Cu0 during the reaction in conventional Cu/SiO2 catalysts due to the relatively high Cu reducibility, which causes problems in deeply understanding the hydrogenation functions of Cu species.

Reaction pathways for frozen Cu0 in DMO hydrogenation

For further understanding the catalytic mechanisms, we performed density functional theory (DFT) calculations to simulate the hydrogenation processes for Cu0-based catalysis. In general, •H was formed at Cu0 active sites in the initial stage (35), which mainly cause the subsequent hydrogenation reaction of organic compounds via the addition of •H to the molecular sites with the highest electron density. Therefore, •H may add to the ketonic and etheric O atoms of DMO and MG to generate MG and EG, respectively. As shown in Fig. 5, we calculated four possible pathways for the •H-initiated atmospheric reactions. The computed Gibbs free energy changes (ΔG), enthalpy changes (ΔH), and activation free energies (ΔG) for the possible reaction pathways with selective optimal structures are shown in Fig. 6 and listed in table S4. The detailed structures and bond lengths of the reactant complexes (RCs), transition states (TSs), IMs, and products for the •H addition reactions of DMO to generate MG are also shown in fig. S5. As indicated by the negative ΔG and ΔH values, all the steps of the four pathways are spontaneous and thermodynamically favorable. The highest ΔG values for pathways A and B are 19.9 and 19.2 kcal/mol, respectively, indicating that MG can be produced via the •H addition at ketonic and etheric O atoms of DMO, while the highest ΔG values for pathways C and D (26.2 and 24.7 kcal/mol) are much higher than that for pathways A and B. Thus, it is difficult for MG to further react with •H to form EG, which is consistent with the experimental results that high selectivity for MG can be obtained over the SP-Cu/SiO2 catalyst.

Fig. 5 DFT calculation and schematic of reaction pathway.

Pathways A and B and pathways C and D for the H-initiated atmospheric reaction for DMO to generate MG and for MG to generate EG, respectively.

Fig. 6 Molecule-level free energy surface in four reaction pathways.

Profiles of free energy surface (FES) along with optimal structures and bond lengths of the RCs, TSs, IMs, and products in (A) pathways A and B and (B) pathways C and D for the H-initiated atmospheric reaction for DMO to generate MG and for MG to generate EG, respectively.

For pathway A, the distance of O6–H15 was reduced from 4.36 Å in RCA1 to 1.56 Å in TSA1 and to 0.98 Å in IMA1, falling in the range of an O−H single bond length. The calculated atomic charges (q) and spin densities (ρ) show an obvious charge transfer from •H to DMO, and C1-centered radical (ρ = 0.52 in IMA1) is formed. In the following step, another •H is added to C1, forming methyl 2-hydroxy-2-methoxyacetate. This is a biradical reaction without free energy barriers. As the third •H gradually approached O7, the distance of O7–H17 was reduced from 3.1 Å in RCA2 to 1.23 Å in TSA2 and to 0.97 Å in IMA2, falling in the range of O−H single bond length. Simultaneously, the bond length of C1–O7 was elongated from 1.41 Å in RCA2 to 1.57 Å in TSA2 and to 3.33 Å in IMA2, indicating the rupture of the C–O bond. During this reaction step, electron transfer also occurred, and C1 shows radical character in TSA2 and IMA2. Last, MG is formed with the addition of •H. For pathway B, the reaction process is similar except that •H is added to O7 first and then to O6.

DISCUSSION

In summary, SP Cu atoms on silica by argon plasma bombardment can freeze Cu species in a metallic state, obviously changing their chemical properties. Prominent CK transition enhancement is observed from the Auger spectra, indicating the change of electron structure during the SP process. The SP-made Cu/SiO2 catalyst becomes inactive in CO adsorption and exhibits oxidization resistance performance, presenting noble metal–like behaviors. It is probably attributed to the increment of the electron density of inner shells in Cu atoms, leading to the enhancement of the penetration effect of outermost electrons that makes Cu atoms hard to lose electrons. In comparison, conventional AE-Cu nanoparticles are easy to be oxidized into Cu+ by oxygen in air or DMO in catalytic reactions. The selectivity of MG as a preliminary hydrogenation product continuously stays above 50% from 240° to 280°C on the SP-Cu/SiO2 catalyst, while deep hydrogenation happens on the AE-made catalyst with EO and low-carbon alcohols as primary products at the same temperature range. MG can be easily generated in the reaction of DMO with •H, which is catalyzed by Cu0 active sites, but it cannot further catalyze the reaction between MG and •H. It can be supposed that the frozen Cu0 is crucial for the preliminary hydrogenation step, while Cu+ is the key factor for deep hydrogenation. Foreseeably, the sputtered Cu species with some changes in electron structure and chemical properties have a potentiality in substitution of noble metals in hydrogenation reactions.

MATERIALS AND METHODS

Catalyst preparation

A pure Cu target (purity > 99.9%; 50 mm by 100 mm; Toshima Co. Ltd.) was used to deposit Cu atoms on porous silica pellets CARiACT Q-10. The self-made polygonal rotation SP apparatus is described in fig. S6. After the target was presputtered for 0.5 hours, certain amount of the pretreated SiO2 pellets was loaded into the cavity barrel. Then, the vacuum chamber was evacuated to 9.9 × 10−4 Pa, followed by introducing an Ar (99.995%) flow of 29 ml min−1 into the chamber until the pressure reached 2.0 Pa. The SP experiment started with the input power of 450 W. The hexagonal barrel was rotated at 3.5 rpm and vibrated mechanically to mix the support and deposited metal atoms uniformly. After SP for 3.5 hours, approximately 17 weight % (wt %) of Cu were loaded. Thereafter, a 1.0% O2 (balanced by nitrogen) flow was gradually introduced into the cavity barrel to recover ordinary pressure. Last, the catalyst after SP was pressed and crushed into granules of 20 to 40 mesh before catalytic reaction. The as-prepared catalyst was denoted as SP-Cu/SiO2.

One hundred five milliliters of standard solution (0.3 M) of Cu(NO3)2·3H2O in deionized water was mixed with 10 ml of 25 wt % ammonia aqueous solution and stirred for 30 min to produce uniform copper ammonia complex solution. Subsequently, porous silica pellets CARiACT Q-10 (Fuji Silysia Chemical Ltd.; specific surface area, 283 m2 g−1; mean pore diameter, 10 nm) were slowly added to the copper ammonia complex solution and stirred for 2 hours at room temperature. The suspension was heated in a water bath preheated to 40°C and aged for 4 hours. Afterward, the water bath was continuously heated to 90°C, allowing for the evaporation of ammonia, as well as the consequent deposition of copper species on silica. The evaporation process was terminated as the pH value of the suspension decreased to 6 to 7 (ca. 3 hours). The obtained precipitates were filtrated and washed with deionized water and dried at 120°C for 8 hours. The solid powder was then calcined at 450°C in air for 4 hours. The final catalyst was denoted as AE-Cu/SiO2.

The physically mixed catalysts “PM-Cu/SiO2 and PM-Cu+/SiO2” were prepared by mixing porous silica CARiACT Q-10 and CuO or Cu2O powders with a mortar, respectively. The catalyst was pressed and crushed into granules of 20 to 40 mesh before reduction and catalytic reaction.

Catalyst characterization

XRD spectra were obtained on a PANalytical X’pert Pro diffractometer equipped with Cu Kα (40 kV, 20 mA) irradiation. TEM images were obtained using a JEOL JEM-2100 (120 kV) microscope.

The x-ray absorption data of the reduced AE-Cu/SiO2 and used SP-Cu/SiO2 sample at the Cu K-edge were recorded at room temperature in transmission mode at beam line BL14W1 (39) of the Shanghai Synchrotron Radiation Facility (SSRF) in China.

In situ DRIFTS were conducted on a Bruker TENSOR27 Fourier transform IR spectrometer with a diffuse reflectance attachment and an MCT (mercury cadmium telluride) detector. A ZnSe window was used for the in situ IR cell. The absorbance spectra were collected for 32 scans with a resolution of 2 cm−1. Before CO adsorption, a catalyst of about 0.015 g was heated to 250°C with a N2 flow (99.99%) of 30 ml min−1. After sweeping for 1 hour, a pure H2 flow (99.99%) of 30 ml min−1 was introduced instead of N2 to reduce the catalyst for 1 hour. Subsequently, pure He was flowed into the cell for 30 min, followed by a vacuum desorption for 1 hour to remove H2 residual in cell and adsorbed on the catalyst, followed by cooling down of the temperature to room temperature. After keeping for 0.5 hours, the background spectra under this condition were recorded. Then, a pure CO flow (99.99%) of 30 ml min−1 was introduced into the IR cell for 0.5 hours. After adsorption, a N2 flow of 50 ml min−1 was used to sweep for 2 hours, followed by obtaining desorption spectra of CO desorption.

Hydrogen TPR experiments were performed with the self-made TPR system. Forty-milligram samples were pretreated at 150°C for 1 hour in an Ar flow before the following test. The samples were heated in a 5% H2 (Ar balance, 30 ml min−1) flow from 25° to 500°C with a heating rate of 10°C min−1. Thermal conductivity detector signals were recorded in this process.

XPS and XAES were performed under an ultrahigh vacuum (8 × 10−10 Pa) using a Thermo Fisher Scientific (ESCALAB 250Xi) x-ray photoelectron spectrometer with a monochromatic Al Kα source (hν, 1253.6eV). It operated with a pass energy of 40 eV. Before each test, Ar ionic SP was performed with a beam energy of 3 keV and a SP rate of 20 Å min−1 for 7.5 min. The collected BEs were calibrated using the C 1s peak at 284.8 eV as the reference.

N2O-TPO experiments were performed with the self-made TPO system. First, 40 mg of samples were prereduced at 250°C with 99.99% hydrogen (30 ml min−1). Then, they were purged with Ar flow (30 ml min−1) for 1 hour and cooled down to room temperature. Two mass spectrum signals of 28 (N2) and 44 (N2O) were monitored by an on-line mass spectrometer (Omnisorp Corporation). After the base line getting stable in He, the mixture of 5% N2O (He balance, 30 ml min−1) was introduced instead of He. This process was carried out for 1 hour. Last, the samples were heated in the same atmosphere from 25° to 250°C with a heating rate of 5°C min−1.

The UV-vis spectrometer (Lambda 950) was used to analyze the adsorption range of prepared Cu-based nanoparticles. The adsorption signals were recorded from 200- to 800-nm wavelengths. Metallic Cu powder (99%) was also tested for comparison.

The Cu loading was determined by scanning electron microscope–energy dispersive spectrometer. The specific surface area was analyzed by nitrogen adsorption/desorption at 77 K using the BET method (Autosorb iQ2 by Quantachrome).

Evaluation of DMO hydrogenation reactions

The catalytic performance in DMO hydrogenation was conducted with a fixed-bed reactor. Briefly, 0.5 g of the catalyst (20 to 40 meshes) was packed into a stainless-steel tubular reactor with the thermocouple inserted into the catalyst bed. Catalyst reduction was performed at 250°C for 4 hours with a ramping rate of 2°C min−1 from room temperature in a pure H2 flow. After cooling to the reaction temperature, 8 wt % DMO (purity > 99%) solution in methanol was fed into the fixed-bed reactor with a H2/DMO molar ratio of 150. The reaction pressure was controlled to 3.0 MPa. The liquid hourly space velocity of DMO was set as 0.5 hours−1. The feeding was continuous, and the reaction temperature was ranged from 170° to 290°C. For each temperature, the catalyst was stabilized at least for 3 hours. For the stability test, the temperature was fixed at 240°C. The reaction products were collected with an ice trap and analyzed on an offline gas chromatograph with a flame ionization detector.

The used SP and AE catalyst for characterizations was derived after a continuous reaction with a wide temperature range. For the AE-used catalyst, the continuous temperature range was 170° to 280°C. For the SP-used catalyst, the continuous temperature range was 240° to 290°C.

DFT calculation

The DFT calculations were performed with the Gaussian 09 program suite, using the B3LYP hybrid meta exchange-correlation functional in conjunction with the 6-31+G (d,p) basis set. Frequency calculations were performed to determine the character of stationary points. TSs were characterized with one imaginary vibrational frequency. Intrinsic reaction coordinate analysis was executed to verify that each TS uniquely connected the designated reactants with the products. The profiles of the FES were depicted using relative free energies to the reactants. In FES, the free energies for each species were evaluated at the same atomic type and number. The values of Gibbs free energy and enthalpy were corrected by zero-point energy and thermal energy at 298 K. The Gibbs free energy changes (ΔG) and enthalpy changes (ΔH) were calculated for each reaction step by their changes from RC to IM.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/12/eaau3275/DC1

Fig. S1. In situ DRIFTS results of AE- and SP-made Cu/SiO2 samples.

Fig. S2. Cu L3M45M45 XAES spectra of Cu bulk.

Fig. S3. Catalytic performance at low temperature and coke analysis.

Fig. S4. Characterizations and reaction results for the PM catalysts.

Fig. S5. The detailed structures and bond lengths of the reactant complexes, transition states and products in the reaction of DMO hydrogenation to MG.

Fig. S6. Schematic representation of the Cu SP apparatus.

Table S1. Peak fitting results of Cu 2p and Cu LMM Auger spectra of tested samples.

Table S2. Final-state term (asterisk denotes the Auger vacancy satellite structure) and Cu L3M45M45 peak fitting results of Cu bulk, SP fresh, and SP-used samples.

Table S3. Physical properties of Cu/SiO2 catalysts made by different methods.

Table S4. Computed Gibbs free energy changes (ΔG, kcal mol−1), enthalpy changes (ΔH, kcal mol−1), and activation free energies (ΔG, kcal mol−1) for the H-initiated reaction of DMO and MG.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: This paper is dedicated to the 70th anniversary of Dalian Institute of Chemical Physics, Chinese Academy of Sciences. We would like to thank M. Tan, G. Liu, A. Taguchi, and T. Abe for assistance in the physical sputtering experiment, as well as X. Tong for help in stability tests. We appreciate beamline BL14W1 (SSRF) for providing the beam time to derive XAFS results. Funding: J.S. thanks the financial support of Foundation of State Key Laboratory of High-Efficiency Coal Utilization and Green Chemical Engineering (grant no. 2016-04), the Hundred-Talent Program of Dalian Institute of Chemical Physics, and the Youth Innovation Promotion Association of the Chinese Academy of Sciences (2018214). Author contributions: J.S. and J.Y. conceived the research, performed characterizations, analyzed data, and wrote the manuscript. F.M. and Y.S. evaluated catalysts. X.W. executed the DFT calculation and discussed the concept. Q.M. and N.T. helped with the catalyst preparation and characterizations. All the authors contributed to analysis and discussion on the data. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Navigate This Article