The Activity of Ultrafine Cu Clusters Encapsulated in Nano-Zeolite for Selective Hydrogenation of CO₂ to Methanol

Abstract

Narrowly dispersed ultrafine Cu clusters of sizes smaller than 2.0 nm have been encapsulated in nanosized silicalite-1 zeolite through direct crystallization in the presence of Cu(en)₂²⁺ complex ions as the metal precursor. The growing silicalite-1 crystals are rich in vacancy defects and connectivity defects on the grain boundaries, where the terminating silanols promote the decomposition of Cu(en)₂²⁺, thus the deposition of ultrafine Cu species. The obtained composite material as a model catalyst is active for CO₂ activation and hydrogenation to methanol. The preliminary in situ FTIR study recognizes a series of surface-adsorbed carbonyl, formyl, carbonate, and formate species when the material is exposed to CO₂ and H₂. Among others, the adsorbed formate decays most rapidly upon cofeeding CO₂ and H₂, implying that the most probable pathway toward methanol formation over this material is via the formate-mediated mechanism.

1. Introduction

Regarding the current energy transition from fossil fuels to renewables, resource-utilization of captured CO₂ is indispensable in its remediation. Catalytic hydrogenation of CO₂ to methanol combines CO₂ and H₂ into a liquid form. When H₂ is made through water electrolysis using the erratic renewable energy, e.g., solar or wind, the process stores this energy in CO₂. The produced methanol is called “liquid sunshine” [1,2]. The technology using Cu-based catalysts for the CO₂-to-methanol process has been intensively studied [3,4,5,6,7] and demonstrated in large scales for a fairly long period [8], and it is closing in on the maturity of widespread exploitations.

Taking the George Olah CO₂-to-methanol plant of CRI in Iceland as an example, running at the scale >4000 MT/year since 2012, it provides sound data to demonstrate that over 90% CO₂ emission is saved per unit energy product. However, economically, the technology cannot yet compete with the traditional syngas-to-methanol process because the space-time yield is insufficient due to thermodynamic limitations. Furthermore, the 1 mol equivalent water produced in the CO₂ hydrogenation reaction poses as a “poison” that sinters Cu and related active species, raising the operation costs [8]. Therefore, new catalysts of higher activities that are able to establish the reaction equilibrium at a higher space velocity, as well as better stabilities that can resist water poisoning, are highly demanded.

Reducing the size of metal particles is a way to enhance their catalytic activities by exposing larger fractions of atomic structures to the surfaces [9,10]. For a fundamental study of Cu-based catalysts for CO₂ hydrogenation, we propose to isolate the question on Cu particle size from other parameters, such as the interface of Cu/ZnO, the alloying of Cu/Zn, and the metal–support interactions [11,12], and to study at first the adsorption and activation of CO₂ on ultrafine Cu clusters. For this purpose, we have prepared a model catalyst with Cu clusters of sizes smaller than 2 nm entrapped in a nanosized pure-silica MFI-type zeolite (silicalite-1), which is denoted Cu@NS, by a direct crystallization method using a soluble Cu²⁺ precursor. Employing the microporous structure of zeolites as a scaffold to confine the metal allows ultrafine clusters and nanoparticles to be protected from sintering [13,14], which poses a common reason for deactivation of metal catalysts [15]. Furthermore, the microporous environments may provide further functions such as molecular-level microreactors, shape-selectivity, etc., in addition to the fixation effects [16,17,18].

Specifically for CO₂ hydrogenation, enhanced activity of Cu clusters with reduced sizes has been proven experimentally and through model simulations. For example, the researchers at Argonne Laboratory prepared Cu₃, Cu₄, and Cu₂₀ clusters deposited on an amorphous Al₂O₃ support using a particular cluster-mass selection method and demonstrated that Cu₄/Al₂O₃ has significantly higher activity in the CO₂-to-methanol reaction than the other two clusters [19]. Zhang et al. modeled Cu clusters from Cu₁₃ through Cu₇₉ to bigger nanoparticles and computed the binding strengths of reactants and reaction intermediates, as well as the activation barriers for the elementary reaction steps of CO₂ hydrogenation. They concluded that the Cu₁₉ cluster exhibits the highest CO₂ hydrogenation activity, which can be ascribed to moderate CO₂ coverage and a low CO₂ dissociation barrier [20]. The obvious discrepancy between the experiments and the theoretical calculations implies that the cluster size effect must be separated from other experimental variables such as cluster morphologies, surface structures, electronic charges, and metal–support charge transfers. Further explorations from both sides are required to approach an agreement on the most efficient size of Cu particles.

As for Cu-containing zeolite materials, Cu²⁺ ion-exchanged SSZ-13 (CHA-type) catalysts have been commercialized for selective catalytic reduction of NO_x pollutants (SCR-DeNO_x) from diesel engine exhausts [21]. Cu²⁺-exchanged SSZ-13, ZSM-5, and mordenite zeolites are also studied for the catalytic activity for methane oxidation-to-methanol in aqueous solutions [22,23]. Unlike the fact that Cu²⁺ ions that can be readily introduced into zeolite via ion exchange, the encapsulation of Cu nanoparticles and clusters in the intracrystalline micropores is challenging. The traditional impregnation method usually yields metal particles in a larger nanometer range (>2 nm) both inside the crystals on defective voids and outside on the external surfaces [16,17,24]. Ding et al. developed a procedure to grow Na-Beta zeolite using CuO-impregnated nanosized Beta as seeds and obtained entrapped Cu particles of 2–5 nm in Cu@Beta. They demonstrated that those Cu nanoparticles in the confined environment in the micropores have a higher ethanol selectivity because the desorption of C₁ intermediates is retarded [25]. Cui et al. used a bimetallic MOF material, CuZn-HKUST-1 nanoparticles, as sacrificial Cu and Zn source in the synthesis of ZSM-5 zeolite and obtained Cu/ZnO nanoparticles of ca. 2 nm entrapped in the zeolite, Cu/ZnO@ZSM-5, which has significantly higher activity than the impregnated reference materials [26].

A rather generally applicable strategy for the encapsulation of smaller metal clusters in zeolite micropores was introduced by Iglesia and coworkers [27,28]. Zeolites of various framework topologies were directly crystallized in the hydrogels with an additional metal precursor, which is in the form of an amino- or organoamino-metal ion or cationic metal complex with other types of ligands. Other researchers have widely adopted the method, and ultrafine noble metal clusters of sizes smaller than 2 nm confined in zeolites of SOD, LTA, *BEA, MFI, and other framework types have been synthesized [13,29].

The critical point in this procedure is that the speed of the metal-precursor hydrolysis has to be synchronized with the speed of the zeolite crystallization so that the precipitated metal-hydroxy-oxide clusters are entrapped in the zeolite micropores, rather than them growing into separated and bigger particles. Still, applying this method to create a zeolite-encapsulated Cu cluster is challenging since, at higher pH in aqueous systems where zeolites usually crystallize, oxo-cupric species tend to precipitate much too rapidly. A proper Cu precursor, which exhibits a synchronized decomposition/precipitation speed with the zeolite crystallization, is hard to find.

In the present paper, we chose the pure-silica polymorph silicalite-1 of the MFI-type zeolite instead of its aluminosilicate counterpart ZSM-5 to make a composite material Cu@zeolite. This has the advantage of circling out the contribution of acid sites in the catalytic processes and making the investigation focus on the entrapped Cu clusters. Furthermore, we used the recipe to make the host silicalite-1 material in the format of nanosized spheres, which further eliminates the possible diffusion limitation arising from the zeolite host. As the metal precursor, we chose Cu(en)₂(NO₃)₂. Experiments show that the lifetime of the Cu(en)₂²⁺ ions in the zeolite synthesis gel is synchronized with the growing speed of the nano-zeolite, leading to the deposition of Cu species in the crystallized zeolite.

Taking Cu@NS (NS stands for nanosized silicalite-1) as a model catalyst, we attempted to identify the surface-adsorbed carbonyl, formyl, carbonate, and formate species using in situ FTIR, when the material is exposed to CO₂ and to the reactive gas mixture with CO₂ and three times H₂. The proceeding of CO₂ activation can follow different pathways, due to its various postures of adsorption on the metal surfaces, and the consecutive polarization and bonding to H atoms. One can distinguish between two major types of reaction pathways leading to the methanol product by recognizing different reaction intermediates, i.e., the formate pathway, when the carbon atom of CO₂ bonds to H at first; and the reversed water-gas shift pathway, when one O atom bonds to H at first [11,30,31]. We monitored the accumulations and evolutions of the adsorbed intermediates in flowing CO₂ and in the reactive gas mixture at room temperature and at 150 °C. Above the cumulated H₂O signals and baseline distortions, we attempted to track down the progress of the reactions along the formate pathway and reversed water-gas shift pathway and figure out which pathway is preferential over Cu@NS. The durability/lifetime of the model catalyst has temporarily not been specifically studied yet in the present paper.

2. Results

2.1. The Presence of Cu in the Synthesized Cu@NS Crystals

The method of the synthesis of nanosized silicalite-1 was established by Mintova et al. and verified by several research groups. It reliably produces spherical silicalite-1 particles with a narrow size distribution between 90 and 100 nm in diameter [32]. Adding Cu(en)₂(NO₃)₂ in the initial gels (

Table 1. Compositions of the synthetic batches, the crystallization conditions, and Cu contents in the products.

Sample	Initial Gel Composition (Molar Ratio)					Hydrothermal Condition		Product Cu Content
	SiO2	TPAOH	Cu(NO3)2	en	H2O	Temp (°C)	t (h)	(wt.%)
NS, nano-silicalite-1	1	0.3	0	0	19	100	30	0
Cu1.1@NS	1	0.3	0.014	0.13	19	100	48	1.10
Cu2.1@NS	1	0.3	0.028	0.25	19	100	54	2.10

) and adjusting the crystallization time accordingly, ultrafine Cu clusters of <2 nm are dispersed in the zeolite nanospheres at metal loadings up to 2 wt.%.

The XRD patterns of the as-synthesized nanospheres of silicalite-1 (NS) and Cu@NS in Figure 1A all belong to phase-pure MFI zeolite. There is no trace of any other reflection peaks that can be attributed to Cu compounds. For example, CuO has diffraction peaks at 2θ = 35.6, 38.8, and 48.9°, which are absent in the patterns of Figure 1A. After the calcination at 550 °C in air, the materials remain phase-pure silicalite-1. The XRD patterns in Figure 1B show the usual changes in the relative intensities of the reflections. The material, both before and after the calcination, shows monoclinic symmetry. No new reflections appear. ICP detects that the Cu@NS materials contain around 1.10 and 2.10 wt.% Cu. Copper dispersed across the nanosized silicalite-1 host is present as noncrystalline species since no peaks of a copper phase were observed in the XRD patterns.

Figure 2 shows the SEM images of Cu_1.1@NS and Cu_2.1@NS. Similar to the pristine silicalite-1 nanospheres of diameters in 90–100 nm [32], the particles of both Cu@NS possess nearly spherical shapes with rough surfaces. The particles seem to be agglomerates of smaller primary crystals. The particles of Cu_1.1@NS are slightly ellipsoidal with dimensions of ca. 180 × 60 × 160 nm. Cu_2.1@NS are rather spherical with diameters of 240 ± 10 nm. In both cases, the agglomerate particles show nearly uniform sizes and shapes. With increasing amounts of Cu(en)₂(NO₃)₂ precursor added to the initial gels, the time of crystallization has prolonged; consequently, the particles sizes are enlarged.

TEM images of calcined Cu_1.1@NS and Cu_2.1@NS (Figure 3) confirm that the zeolite nanospheres are indeed agglomerates of smaller crystallites. The sizes of the primary crystallites are in the range of 20–30 nm, with mostly round shapes. It can be envisaged that the grain boundaries are rich in not-fully connected SiO···HOSi linkages and point defects, where the silanols react with Cu(en)₂²⁺ and allow the voids to accommodate ultrafine Cu deposits. STEM-HAADF images (Figure 3) of both samples reveal Cu-containing particles distributed across the silicalite-1 nanospheres. The sizes of these particles in the two samples are not identical: in Cu_1.1@NS, the mean size is ca. 2 nm; in Cu_2.1@NS, it is around 1 nm.

FTIR and ²⁹Si MAS NMR spectra further confirm that the defective framework and associated siloxyl/silanol structures provide the forces for Cu deposition. Figure 4A shows the FTIR spectra in the -OH vibration region of calcined NS and Cu@NS that are outgassed at 350 °C for 1 h. For NS, two bands are resolved at 3740 and 3510 cm⁻¹. The band at 3740 cm⁻¹ belongs to isolated silanol species that locate on the external surfaces of the particles. The broad band at 3510 cm⁻¹ corresponds to H-bonded -OH groups that locate internally in the zeolite at various defect sites, e.g., triad and tetrad silanol nests at point defects, SiO···HOSi clusters, etc. [33,34,35]. In the Cu@NS materials, the intensities of the band at 3510 cm⁻¹ decline with increasing Cu loading. The internal -OH groups are consumed upon Cu loading. It means that these various kinds of internal -OH groups react with Cu(en)₂²⁺, causing the decomposition and deposition of Cu_a(OH)_x(H₂O)_yO_z particles. The complexity of the -OH groups has been translated into the different sizes of the deposits.

In ²⁹Si MAS NMR (Figure 4B) of NS the Q⁴ (Si-(OSi)₄), signals are resolvable in two groups at −112.6 and −114.5 ppm. The Q³ (-Si-(OSi)₃) peak at −102.4 ppm, which corresponds to the vacancy defects, covers 10.0% of total Si sites. In the spectra of Cu@NS, the Q³ portion decreased with increasing Cu loading: 6.2% for Cu_1.1@NS and 5.2% for Cu_2.1@NS. The Q⁴ resolution deteriorates with increasing Cu loading, as well. In both spectra, the −112.6 and −114.5 ppm bands are less resolved, and an additional shoulder at −116.4 ppm becomes detectable. It implies that the vacancy defects are healed to some extent when Cu species squeeze into the voids, and at least a portion of Q³ (-Si-(OSi)₃) becomes a new type of Q⁴ (-Cu-O-Si-(OSi)₃) that causes torsions in the framework, thus ruining the Q⁴ resolution.

The deposited oxo-Cu species are of a complicated nature rather than a defined molecular or atomic form. Figure 4C shows the temperature-programed reduction process of Cu@NS in diluted 10% H₂/Ar flow. H₂ consumption peaks are observed between 100 to 500 °C, and they are quite different for Cu@NS of different Cu loadings. The deposition of Cu_a(OH)_x(H₂O)_yO_z species to the defective zeolite voids has been a random process, resulting in compounds with very different stoichiometry of diverse sizes. The DR UV-Vis spectra (Figure 4D) of the Cu particles indicate a broad distribution of sizes, from a few atoms (the shoulders of 300–350 nm) to nanoparticles (the broad band from 500 nm upwards in wavelength). The uncontrolled deposition explains why the Cu particle sizes in the two Cu@NS are not identical.

2.2. The Observed CO₂ Conversion Intermediates and Their Extinguishments

The results of the catalytic tests in the fixed-bed microreactor are shown in Figure 5. Both Cu@NS samples display CO₂ hydrogenation activities. Methanol and CO are the observed hydrogenation products. However, at the tested temperature between 200 and 300 °C and the feed-flow conditions, both Cu@NS could not reach the equilibrium conversion [11] because of the too-low Cu contents in comparison to the Cu/ZnO/Al₂O₃ catalysts. On the other side, two positive conclusions can still be deducted from the data: (1) Cu_2.1@NS that have smaller Cu particles are superior to Cu_1.1@NS with a bigger Cu size. Downsizing Cu particles from 2 to 1 nm displays a positive effect in terms of both CO₂ conversion and methanol yields. (2) In calculating the space-time yields of methanol, Cu_1.1@NS already comes in a range comparable with the benchmark test of the commercial Cu/Zn/Al₂O₃ catalyst, which spans 0.1–0.8 g·h⁻¹·g_cat⁻¹ [4], and Cu_2.1@NS behaves in a way better than Cu_1.1@NS. With the open possibilities for further improvements, such as introducing further components to boost CO₂ conversion, zeolite-encapsulated ultrafine Cu particles are potentially viable catalysts for the CO₂-to-methanol processes.

Time-resolved in situ FTIR spectra of CO₂ adsorption at room temperature over Cu_2.1@NS prereduced at 400 °C (Figure 6A) reveal accumulations of several species: significantly recognized is a band at 1614 cm⁻¹, which may contain the asymmetric stretching of the O-C-O group of adsorbed bidentate CO₂ and formate on Cu surfaces, i.e., Cu-O-C-O-Cu and Cu-O-C(H)-O-Cu species [36,37,38], in addition to the bending vibration of H₂O molecules [39,40]. In addition, a tiny peak at 1746 cm⁻¹ for the C-O stretching of surface formyl, Cu-C(H)-O [41], is detected. Another peak at 2073 cm⁻¹ is barely detectable for the stretching of surface carbonyl, Cu-C-O [30,36,42]. These absorption bands give hints to the initial reactions with the surface H atoms and the break of one C-O bond:

CO₂* + H* -> HCOO* + *

HCOO* + * -> CO* + OH*

CO* + H* -> HCO* + *

The strong bands from 3000 to 3500 cm⁻¹ are the zeolite silanols and the H₂O molecules formed in the surface reactions, which form H-bonds in the internal channels of the zeolite. The peak at 3740 cm⁻¹ is the surface silanol, which is gaining intensity with the H₂O formation, as well. Due to the intense water signals, as well as the baseline distortions, above-spectral attributions to formate, formyl, and carbonyl are ambiguous. Figure 6B shows the spectra of the same reduced Cu2.1@NS in CO₂ flow at 150 °C. At this condition, all signals that can be related with adsorbed species become weaker, while the H-bonded H₂O in the zeolite (3000–3500 cm⁻¹) has been significantly removed. Comfortingly, the multicomponent band at 1614 cm⁻¹ (carbonate, formate, and water) still exists, as does the formyl peak at 1746 cm⁻¹; carbonyl at 2073 cm⁻¹ becomes even slightly more recognizable.

Figure 6C shows the reactions further proceed when CO₂ and H₂ are cofed. The spectral changes observed are quite complex, and the following tentative interpretations are proposed, awaiting further amendments when more data are acquired with related materials. First, the above-detected three intermediates exhibit slight blue shifts and changes of their relative intensities. The bidentate formate and H₂O moves to 1638 cm⁻¹ and becomes much weaker in intensity because stronger adsorbed species remain adsorbed, while weaker adsorbed ones are converted. Related to the bidentate formate, the C-H stretching of the formate group at 2856 cm⁻¹ is most intensive [36,41]. At the same time, the formyl shifts to 1753 cm⁻¹ and become even less cumulated in intensity. The carbonyl shifts to 2103 cm⁻¹ and seems to accumulate in larger quantity under the reactive feed flow, i.e., it converts slower than formyl and formate. Second, some other intermediate species become observed: the predominant one is unidentate carbonate, which is seen as the symmetric stretching at 1392 cm⁻¹ and asymmetric stretching at 1475 cm⁻¹ [30,36,41]; the formic acid that appears at 1753 cm⁻¹ exhibits very weak C=O stretching vibrations [30,41]. Third, a number of methoxy-related surface species become significant: they are at 2920 cm⁻¹ for the C-H stretching of the methoxy and at 1450 cm⁻¹ for the C-H bending [30,38,43]. The 1121 cm⁻¹ band belongs to the zeolite framework modification. The negative bands at 3740 and 3000 cm⁻¹ correspond to water desorption during the course of the reaction. At 150 °C (Figure 6D), the formate- and carbonyl-related species are generally weaker but mostly still exist. Furthermore, a siloxyl -Si-OH band grows around 3450 cm⁻¹ due to the continuous formation of H₂O. Considering the two major reaction pathways of CO₂ to methanol, i.e., the formate mechanism and the reversed water–gas shift mechanism in Scheme 1 [31], adsorbed intermediates of both pathways have been detected in the FTIR spectra, implying that the reaction takes both pathways. However, based on the comparison of accumulation speeds of these species, it is apparent that the formate mechanism predominates at the Cu@NS model catalyst.

3. Materials and Methods

3.1. Materials

Tetraethyl-orthosilicate and ethylenediamine were obtained from Sinopharm Chemical Reagent Co., Ltd., Shanghai, China. Cu(NO₃)₂·3H₂O was obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. Shanghai, China. Tetra-n-propylammonium hydroxide was purchased from Zhejiang Kente Catalytic Materials Co., Ltd. Zhejiang, China.

3.2. Synthesis

Nanosized silicalite-1 and composite materials Cu@silicalite-1 were synthesized via the direct crystallization following the gel recipe, 1 SiO₂: 0.3 TPAOH: 0–0.04 Cu(en)₂(NO₃)₂: 19 H₂O, a recipe adapted from the established method for the reliable crystallization of nanosized silicalite-1 [32]. The silica source was tetraethyl-orthosilicate (TEOS); TPAOH stands for tetra-n-propylammonium hydroxide. For the preparation of the gels, Cu(NO₃)₂ was first dissolved in H₂O with a great excess amount of ethylenediamine (en), then TPAOH (25% aq.) was mixed. TEOS was then added, and the mixture was stirred at RT overnight to form a blue-colored and transparent gel. Table 1 lists the gel compositions with the various Cu(en)₂(NO₃)₂ contents. The hydrothermal crystallization was performed in Teflon-lined autoclaves at 100 °C for 30–54 h. After the crystallization, the autoclaves were quenched in cold water. Solid products were recovered by filtration and washed with water 3 times. Calcination was performed by heating in air at 1 °C/min to 550 °C, and the end product was kept for 6 h.

3.3. Characterization

Powder X-ray diffraction was performed on a Rigaku Smart Lab 9 KW diffractometer (Tokyo, Japan) using Cu Kα radiation at 40 kV and 150 mA. A 1D detector was used to collect diffraction data in a continuous scan mode in theta–theta geometry. Zeolite powders were pressed on glass plate sample holders.

SEM pictures of powder samples were taken on a JEOL JSM 7900F Scanning Electron Microscope (Tokyo, Japan) equipped with a field emission gun operating at 1–2 kV. The samples were dusted on carbon sheets and inspected without coating.

TEM, STEM images, and corresponding EDS were taken on a JEOL JEM-F200 electron microscope. Before analysis, the samples were suspended in ethanol and dispersed onto ultrathin carbon films supported on 200 mesh Au grids.

ICP element analysis for Cu in Cu@NS was carried out on iCAP Q, Thermo, Waltham, MA, USA.

FTIR spectra were recorded using a Bruker Vertex 70V Spectrometer (Billerica, MA, USA) at a spectral resolution of 4 cm⁻¹. The equipment, including the sample chamber, was evacuated to below 1 hPa. The in situ FTIR studies were performed on self-sustaining pellets of a density of 15 mg/cm², which are enclosed in a gas-tight and heated cell with CaF₂ windows. The samples were purged in 30 mL/min Ar at 350 °C for 1 h prior to spectroscopic recording.

For probing the surfaces with CO₂, or CO₂ + H₂, the sample was reduced at 400 °C in a flow of 4% H₂/Ar. After cooling down to room temperature, as well as to 150 °C, background spectra were recorded. Then, a flow of CO₂ or CO₂/3H₂ was introduced into the cell for 30 min, along with spectral recording in 1 min time intervals.

Diffuse reflectance UV-Vis spectra of reduced Cu@NS were taken using a Rigaku UV-Vis spectrometer equipped with a Kubelka-Munk integration sphere. The powder samples were filled in a quartz cuvette and measured against BaSO₄.

Temperature-programmed reduction tests of Cu@NS were carried out in a quartz tubular reactor (homemade). All catalysts were pretreated in Ar at 350 °C for 60 min. After cooling to room temperature, the experiment was started in 10% H₂/Ar flow with a heating rate of 10 °C/min.

3.4. Catalytic Test

The catalytic performance test was carried out in a stainless steel-armed quartz tube fixed-bed microreactor. An amount of 0.5 g catalyst pellets of 60–100 mesh was packed in the reactor. Prior to a test, the catalyst was prereduced in 100 mL/min mixture of 10% H₂/Ar at 400 °C for 2 h. The CO₂ hydrogenation reaction was carried out in the temperature range of 220–300 °C, and CO₂/3H₂ flow at GHSV = 2000 h⁻¹ at 30 bar. The gas effluents were analyzed using an online GC (Agilent 8890, Santa Clara, CA, USA). All postreactor lines and valves were heated to 140 °C to prevent the condensation of the products. All carbon-containing components, including CO₂, CO, and methanol, were calibrated and quantified using external standards.

4. Conclusions

Ultrafine Cu clusters of sizes smaller than 2 nm in loading amounts up to ca. 2 wt.% have been encapsulated in the nanospheres of silicalite-1 through direct crystallization using Cu(en)₂²⁺ as the metal precursor. The nano-zeolite particles are rich in connectivity defects and vacancy defects on the grain boundaries. The silanols and nested silanols at these defect positions react with Cu(en)₂²⁺, promote the decomposition of the Cu precursor and lead to the deposition of the ultrafine cupro-hydroxy-oxides particles at the grain boundaries that can be converted into ultrafine metallic Cu particles. The complexity of the zeolite defects results in different sizes of the deposited Cu cluster.

As a model catalyst, Cu@NS produces methanol and CO in CO₂ hydrogenation reactions. The methanol space-time yield is comparable with the benchmark test using a commercial Cu/ZnO/Al₂O₃ catalyst. Preliminary in situ FTIR has detected adsorbed intermediates of both the formate mechanism and reversed water-gas shift mechanism, implying that the reactions take both pathways. However, based on the different accumulation speeds of these species, it is apparent that the formate mechanism predominates at the Cu@NS model catalyst.