Copper-zeolites Prepared by Solid-state Ion Exchange - Characterization and Evaluation for the Direct Conversion of Methane to Methanol

Direct conversion of methane to methanol (MTM) over Cu-zeolites is a so-called “dream reaction” for the chemical industry. There is still a lot that can be done in order to optimize the reaction by e.g. achieving a deeper understanding of the reaction mechanism and the nature of the Cu-sites. In this study, we investigated a solid-state ion exchange method to incorporate CuI ions into zeolites (MOR, BEA, ZSM-5 and FAU), as a more scalable technique. The solid-state ion exchange led to a Cu/Al ration of about 0.8, however with a heterogeneous distribution of Cu. Regardless, Fourier transform-infrared spectroscopy still revealed that most Brønsted acid sites were exchanged in all four samples. Further, CH4-temperature programmed reaction experiments showed that some Cu-sites formed were reactive towards CH4, with CuI-MOR and CuI-FAU having the largest CH4 consumption. Ultimately, the CuI-zeolites were tested in the MTM reaction and proved capable of producing methanol, even without the presence of Brønsted sites. A MOR with lower Cu/Al ratio (0.30) was also tested for comparison, and as this sample obtained a much higher productivity than the CuI-MOR with high Cu-loading (0.10 vs. 0.03 molMeOH/molCu), it was demonstrated that some fine-tuning is necessary to obtain the active Cu sites for methane activation.

to methanol ratio of two-to-one [13,14]. This ratio is based on the active site being a dicopper active site (Cu-O-Cu or Cu-O-O-Cu), or the combination of two monomeric sites (Cu-OH) [11,15]. It has been found that both the zeolite framework and placement of Cu have large impacts on the formation of the active sites and as a consequence, the activity towards methanol formation [12]. Therefore, finding new materials and synthesis methods is still very important in the attempts to optimize the MTM process. In literature, the Cu loading into zeolites is mostly performed via liquid ion exchange or incipient wetness methods with Cu II salts, both giving well-dispersed Cu-zeolites, however only in lab scale quantities [4,16]. For industrial application, it would be much more beneficial to use a simpler exchange method like solid-state ion exchange (SSIE). SSIE is also relevant for the selective catalytic reduction of NO x with NH 3 , and to that end, Vennestrøm and Shwan et al. have previously investigated a novel method to perform SSIE with CuO/Cu 2 O in an atmosphere of NO/NH 3 [17,18]. Therein, they suggest that the Cu-oxo species are formed at low temperatures due to the formation of mobile [Cu(NH 3 ) 2 ] + complexes.
In this study, we have prepared Cu I exchanged zeolites via SSIE with Cu I Cl (Cu I -BEA, -FAU, -MOR and -ZSM-5) and tested the materials in the methane to methanol reaction. Le et al. have previously shown that SSIE with Cu II (acac) 2 instead of Cu I Cl over mordenites lead to a higher yield of methanol [19]. However, when exchanging Cu I ions instead of Cu II ions into the zeolites, it is theoretically possible to obtain a Cu/Al = 1, thus maximizing the possible Cu-oxo sites. A similar protocol using SSIE with Cu I Cl to exchange H-mordenites for the MTM reaction was previously applied by Bozbag et al., however, they did not explore the possibility of fully exchanging the materials, and used Cu/Al ratios around 0.35 [20]. The SSIE method discussed in this study have also been used to synthesise Cu I -exchanged FAU zeolites that have been investigated as a catalyst for C-C bond Glaser-type homocoupling reactions by Kuhn et al. [21]. Furthermore, the Cu I -FAU zeolite has been shown to be applicable as a catalyst in several other types of organic synthesis [22][23][24][25][26]. This truly shows the adversity of Cu-sites in zeolites. To fully study the effect of this SSIE synthesis method for the MTM reaction, the samples were subjected to elemental analysis, and investigated with CO-adsorption coupled with Fourier Transform-Infrared (FT-IR) spectroscopy to understand if the increased amount of Cu led to changes in the Cu speciation. Furthermore, CH 4 -TPR experiments, shown to be a helpful tool in optimizing the MTM reaction both in terms of setting the reaction parameters as well as understanding material properties, were performed [27,28]. Finally, the samples were tested in the methane to methanol reaction and compared to previously obtained results with the same reaction protocol in order to assess their performance. The target with this investigation is to understand whether it is possible to form methanol from methane when most Brønsted sites are exchanged with Cu, or if this alters the properties of the Cu-sites to an extent where C-H activation is no longer possible. Further, we also aim to clarify the importance of minimizing the formation of Cu agglomerates and nanoparticles to obtain a high methanol productivity.

Solid-state Cu-exchange
The four zeolite materials used as starting material in this study were commercially available and received in the NH 4 -form. The four zeolite frameworks were FAU (USY) zeolite (Zeolyst International (CBV500), Si/Al = 2.9), MOR zeolite (CBV21A, Si/Al = 10.3), BEA zeolite (Zeochem International (ZEOCAT PB/H), Si/Al = 12.5-17.5) and ZSM-5 zeolite (Zeolyst International (CBV5020), Si/Al = 25). The H-form was obtained by calcining the NH 4 -form for 12 h at 550 °C in static air. To obtain the expected Cu I -exchanged forms of the samples, a solid-state ion exchange was performed by mixing 1 g of dry zeolite with a calculated amount of CuCl (Aldrich, > 99% purity, melting point: 430 °C) to obtain a specific Cu/Al ratio, where the Cu exchange degree was about 80%. The samples were named Cu I -FAU, Cu I -MOR, Cu I -BEA, and Cu I -ZSM-5. All samples were made with an intended Cu/Al ratio ≈ 0.8. One Cu-MOR was made with a lower Cu/Al ratio (~0.30) for comparison. This was named 0.30Cu I -MOR. To limit the exposure to moisture, the mixture was rapidly ground with a mortar and pestle before being loaded into a tubular reactor and connected to a flow setup. The powder mixture was heated to 350 °C (1 °C/min) and held isothermally for 3 d. N 2 was used as a purging gas (40 ml/min) to remove the gaseous HCl formed during the reaction.

Standard characterization
Microwave Plasma Atomic Emission Spectroscopy (MP-AES) was used for elemental analysis of the sample. Before analysis, 20 mg of sample was dissolved in 1 mL 15% hydrofluoric acid (HF) in a Teflon liner for 1 h. Safety note: The use of hydrofluoric acid requires special precautions. To verify that all of the solid material was dissolved, a light was shone through the thin wall of the Teflon liner. After quenching excess fluoride anions by adding an appropriate amount of 5 wt% H 3 BO 3 , the solution was diluted with distilled water (50 mL). An Agilent 4100 MP-AES instrument was used to analyze the solution, and the amount of Si, Al and Cu in µmol/g was quantified with external calibration curves based on commercial elemental standards.
Scanning Electron Microscopy (SEM) combined with backscattered electron (BSE) imaging was used to investigate the samples for Cu nanoparticles before and after activity testing. A Hitachi SU8230 microscope was used to obtain the images using an acceleration voltage of 1 kV and a 10 µA current. On the same instrument, Energy Dispersive X-ray (EDX) was used as a secondary elemental analysis to determine the ratio of Cl in the sample in relation to the Cu content before and after testing. The distribution of Cu on the sample was investigated with EDX mapping.
Thermogravimetric Analysis (TGA) was used to measure the water content (in %) of the samples. For this, a Netzsch STA 449 F Jupiter system was applied to measure the sample weight. The materials were heated to 300 °C with a ramp rate of 2 °C/min in a flow of synthetic air (20 ml/min O 2 and 40 ml/min N 2 ).
X-ray Diffraction (XRD) patterns of the tested materials were obtained using a Bruker D8 Discovery diffractometer with a Cu Kα radiation (λ = 1.5418 Å). The materials were grinded, and evenly dispersed to make a flat surface on a glass plate sample holder. The experiments were run in a 2θ range of 2-50° with a step size of 0.02 °/s at ambient conditions. N 2 -physisorption isotherms was used to determine the specific surface area of the samples. The measurements were conducted on a BELSorp Maxi volumetric gas adsorption instrument (MicroTrac MRB) at − 196 °C. The samples were pretreated by evacuation at 300 °C for 10 h. The specific surface areas were calculated by using the BET approximation in a p/p 0 range relevant for microporous materials as defined by Llewellyn et al. [29].

CO-adsorption with Fourier transform-infrared (FT-IR) spectroscopy
For the CO-adsorption FT-IR experiments, the samples were pressed (2-3 tons) into thin wafers (~11.3 mg/cm 2 ) and carefully placed into gold envelopes. Then the gold envelope was placed into a T-shaped cell with KBr windows. The cell was connected to vacuum and pretreated at 450 °C for 90 min after a ramp of 5 °C/min from room temperature (RT). After cooling back down to RT, the setup was placed into a Bruker Vertex 80 FT-IR instrument and incremental doses of CO was sent onto the sample until reaching an equilibrium pressure of ~8 mbar. A spectrum was collected after each dose.

Activity measurements in the methane to methanol reaction
All samples were tested in the methane to methanol reaction. 100 mg of sample (sieve fraction: 425 − 250 μm) were loaded into a linear quartz reactor (i.d. = 6 mm) and then connected to a gas manifold controlled by Mass Flow Controllers (MFC). The gas feeds were sent stepwise onto the sample, controlled by stop-and 4-way valves. Water was connected to the gas line through a bubbler immersed in an oil bath kept at 45 °C. All gas lines were heated to ~140 °C to avoid any condensation in the lines. The effluent was analyzed with an online Pfeiffer Mass Spectrometer after the reactor. All gas flows were adjusted to 15 mL/min, using a flow meter. In a flow of 100 O 2 , the sample was heated to 500 °C (5 °C/min) and kept there for 8 h. Then the temperature was reduced to 200 °C, and 100 He was used to flush out excess O 2 for 1 h. After that, a flow with 100 CH 4 , was sent onto the sample for 3 h. Another round with He flush for 1 h was performed, before a 10 H 2 O saturated 10 Ne/He stream was sent onto the sample for 2 h. The mass fragments of MeOH (m/z = 31), DME (m/z = 46) and CO 2 (m/z = 44) was quantified using calibration curves from obtained using a calibration gas mixture with known amounts of all the components. To obtain the accurate yield, the sample weight was corrected by withdrawing the water content of the samples (measured with TGA). The total yield (µmol/g zeolite ) of MeOH is given as the amount of MeOH + 2 ~ DME. In the text this will be referred to as total amount of methanol.

CH 4 -Temperature programmed reaction (TPR)
For the CH 4 -TPR experiments, the same setup as for testing was used. After pretreating the samples in O 2 in the same manner as above, the temperature was lowered to 105 °C, and He was sent onto the sample to flush away all excess O 2 for 1 h. Then 15 mL/min 5% CH 4 /He was sent onto the sample for 20 min isothermally, before a ramp up to 550 °C with 5 °C/min was started. The temperature was recorded simultaneously with a thermocouple placed in a thin quartz sheet on top of the sample bed. The output of the reactor was analyzed with an online MS.

Basic characterization by MP-AES, EDX, SEM, XRD and N 2 -physisorption
Solid state ion exchange (SSIE) is industrially a much simpler synthesis process than various solution-based exchange procedures. Furthermore, by performing the SSIE with Cu I lower Al content than indicated by the manufacturer in this sample. A full report on the elemental analysis from MP-AES is presented in Table 1. Figure 1 depicts the SEM and backscattered electron images obtained for the four samples. In Cu I -BEA and Cu I -ZSM-5, Cu nanoparticles can be observed as bright spots in the backscattered electron images, as pointed out by the red circles. In the regular SEM images of Cu I -MOR and Cu I -FAU, even larger agglomerates of Cu are observed. This was also corroborated with EDX mapping, which showed big agglomerates of Cu outside the framework (Fig.  S1). These images indicate a very heterogeneous distribution of Cu in the materials and demonstrates the possible drawback of SSIE as exchange procedure. salts, it would be possible to obtain a higher exchange degree of Cu in the zeolite, possibly leading to a larger concentration of copper in close proximity. Their close proximity could make the formation of active sites for C-H activation more feasible. To investigate the resulting Cu-zeolites after SSIE, MP-AES and EDX was used for elemental analysis, N 2 -physisorption was applied to determine the specific surface area, and SEM and backscattered images was used to search for Cu outside the frameworks. A more extensive characterization on the effect of the SSIE has been performed on the Cu I -exchanged FAU sample previously by Kuhn et al. [21]. For three of the zeolites, the SSIE yielded Cu-loaded zeolites with a molar Cu/Al ratio ≈ 0.8. Cu I -BEA ended up over-exchanged with a Cu/Al = 2, due to the much by larger agglomerates present around the pores. Therefore, even though the exchange led to a Cu/Al ratio of 0.8, it is not likely that all the Cu present in the samples will contribute to C-H bond activation.

CO-adsorption with FT-IR spectroscopy
The effects of the very high exchange degree achieved for the four zeolites (Cu/Al ratio of 0.8) on Cu speciation were studied using FT-IR spectroscopy. Figure 2 depicts the ν(OH) stretching region of the spectra at RT after pretreatment at 450 °C in vacuum. O-H stretching bands connected to Brønsted sites (Si(-OH)Al) should be found in the region between 3700 and 3550 cm -1 . The lack of bands in this region indicates that all Brønsted sites are covered by Cu, except a broad band in the Cu I -FAU sample. A comparison between the parent MOR and the Cu-exchanged MOR for this region can be found in the SI (Fig. S3). The bands at ~3750 cm -1 are related to silanols (Si-OH) in the samples. The non-existing Brønsted site vibrations should be noted, as it has been previously suggested that Brønsted sites are necessary to produce methanol, as shown with the help of FT-IR and NMR spectroscopy by Dyballa et al. as well as Sushkevich et al. [30,31]. The effect of the lacking Brønsted sites on the performance of the materials will be discussed in-depth later in this paper. Dosing CO at RT onto the pretreated materials can determine if the high Cu/Al ratio has an impact on the nature of In addition to the heterogeneous distribution of Cu, elemental analysis with EDX revealed that a significant fraction of Cl was found in all samples as a remnant from the SSIE.
The specific surface area of all samples were obtained using N 2 -physisorption, and compared to the specific surface area of the H-form (Table S1). It is apparent that there is a reduction in surface area (> 10%) for most of the samples after the Cu exchange. This loss in surface area could indicate that the formation of nanoparticles and the large excess of Cu contributes to a partial pore blockage for reactants. This is substantiated by the specific surface area of the MOR sample with a smaller Cu/Al ratio, 0.30Cu I -MOR being much higher than for the higher Cu-loaded Cu I -MOR sample.
XRD of the as-synthesized FAU has previously shown that the framework is intact after the Cu exchange and HCl release [21]. Herein, we have measured the XRD pattern of the tested samples. The patterns are presented in Fig. S2. The patterns obtained indicate that 0.30Cu I -MOR, Cu I -MOR, Cu I -ZSM-5 and Cu I -FAU has kept the structural integrity after testing. However, it should be noted that impurities of both CuCl 2 and CuO was present in the highly exchanged samples, as evidenced by some additional peaks at 2θ = 16, 32 and 40°.
The basic characterization of the solid-state ion exchanged materials reveal that there is a heterogeneous distribution of Cu in the samples and possible pore blockage  (Fig. S4). By first comparing the high and low loaded CuCl exchanged samples, it is evident the highly exchanged sample also have more available Cu I to form monocarbonyl adducts, as the band intensity for monocarbonyl is higher in the Cu I -MOR sample compared to 0.30Cu I -MOR. However, comparing the symmetric dicarbonyl band at 2178 cm -1 , the two samples show approximately the same intensity, indicating that the same amount of Cu I -(CO) 2 is formed on the two samples. It is important to note that these values are not normalized to Cu content, indicating that 0.30Cu I -MOR has a higher ratio of Cu sites available for dicarbonyl formation. Even so, when comparing both samples to the LIE sample published previously by Pappas et al., it is evident that the LIE sample has markedly higher bands for the dicarbonyl adducts, even though it has the same Cu content as 0.30Cu I -MOR. Higher intensities of the symmetric and asymmetric stretch for Cu I -(CO) 2 have been previously attributed to Cu-zeolites with a larger concentration of low-coordination Cu sites, more likely to be able to activate the C-H bond in methane [40]. This indicates that the SSIE method used for the Cu I -zeolites herein, yields fewer Cu sites that are in a coordination state and position preferred for reaction with methane, compared to LIE. This could partially be due to less accessibility to the sites because of nanoparticle formation blocking the pores as observed with the reduction in specific surface area for all the zeolites. Another possibility is that some of the Cu sites are coordinated to Cl, thus limiting the coordination environment available for CO and only allowing for monocarbonyl species to form. The latter can also be corroborated by the observation of CuCl 2 nanoparticles with XRD after testing.

Investigating the redox properties with CH 4 -TPR
To further explore the propensity of the samples for reaction with methane, CH 4 -TPR experiments were used. This technique probes the redox properties of the materials. FT-IR indicated the lack of Brønsted acid sites in the material, while CO adsorption showed that only some of the Cu sites were available to form dicarbonyl adducts. Such adducts have been linked to efficient methane activation [41]. With CH 4 -TPR, it is possible to determine if these properties affect the reactivity of materials towards methane. Figure 4 shows the CH 4 consumption (a) and the CO 2 production (b) against temperature during CH 4 -TPR over the four materials highly exchanged materials. In Fig. 4 (c) and (d), Cu I -MOR is compared to 0.30Cu I -MOR.
The CH 4 consumption trace obtained with this method should be considered to be only semi-quantiative, due to the quite small changes of the intensity of the methane m/z fragments. We will therefore focus on the most distinct the Cu-sites and their accessibility. CO will only form stable adducts with Cu I ions and not Cu II ions at RT. Therefore, ideal conditions would be to ensure that all Cu ions in available, redox active sites, are oxidation state 1+, and hence able to react with CO. To do so, the Cu I -exchanged samples were pretreated at 450 °C under vacuum prior to the COadsorption measurements, as this has been shown to induce "self-reduction" of Cu II to Cu I in zeolites [32][33][34][35][36].
In Fig. 3, IR absorption spectra of all four samples during CO adsorption is shown. The background subtracted spectra have been normalized to the sample weight. The last spectra in all four series is recorded at ~8 mbar equilibrium CO pressure. These precautions ensure a fair comparison when discussing the band intensities of the four samples. As the samples are exposed to incremental doses of CO, the bands typical for mono-and eventually dicarbonyl adducts on Cu I sites will be formed [37]. The band(s) related to the formation of monocarbonyl species are highlighted with a grey background. This appears as a growing band at 2158 and 2157 cm -1 for Cu I -MOR and Cu I -ZSM-5, while for Cu I -FAU and Cu I -BEA, the monocarbonyl species form two distinct bands at 2161/2147 and 2159/2151 cm -1 . These two separate bands have previously been correlated in literature to two structurally distinct Cu-sites [37,38]. Cu I -BEA seems to have fewer accessible Cu-sites, as shown by the lower intensity for the monocarbonyl bands for this material than the corresponding band(s) for the other zeolites. This is also corroborated by the appearance of the band at 2143 cm -1 at high coverages for the Cu I -BEA sample, since this is the vibrational frequency of gaseous CO, indicating that all possible adsorption sites are already covered. We postulate that this is related to the low Al concentration, and hence high Cu ion loading on this sample, relative to available exchange sites, resulting in the formation of copper agglomerates.
As more CO is dosed onto the samples, a band starts appearing in the 2178-2182 cm -1 region. This is the symmetric stretch of Cu I -(CO) 2 , and is a clear indication that dicarbonyl species are forming. For all samples, the asymmetric stretch of Cu I -(CO) 2 is also present at lower wavenumbers (~2149-2152 cm -1 ), although, this is distorted by a broad tail, indicating that another band is also appearing in that area. Indeed, there seems to be a band/shoulder at around 2135 cm -1 , which is likely due to the presence of Cu 2 O nanoparticles formed during the exchange procedure [37]. The intensity of the symmetric and asymmetric stretch for Cu I -(CO) 2 for the Cu I -MOR sample was further compared to the same vibrational frequencies from CO on the 0.30Cu I -MOR sample as well as a liquid ion exchanged (LIE) Cu-MOR (Cu/Al = 0.28) with the same Si/Al ratio as the MOR investigated herein, previously published by Pappas et al. [39]. A figure comparing the three samples can be the same temperature, whereas the methane consumption occurs at considerably higher temperature for FAU. Further, a considerably larger quantity of methane is consumed over Cu I -MOR and Cu I -FAU compared to the other two differences that can be observed between the samples in this discussion. First, as observed in Fig. 4 (a), it is noteworthy that the onset of CH 4 consumption for the highly exchanged MOR, BEA and ZSM-5 seems to start around Fig. 3 FT-IR spectra showing the ν(CO) stretching region of the four Cu I -zeolites during CO adsorption. The spectra have been collected with increasing CO pressure. All spectra are normalized to their framework overtone and background subtracted using the spectrum taken after pretreatment at 450 °C in vacuum. The last spectrum in all graphs is obtained after an equilibrium pressure of ca. 8 mbar was achieved. The grey area gives the last spectrum taken before the total CO pressure was large enough for dicarbonyl adducts to develop When inspecting the CO 2 production (Fig. 4 (b) and (d)), it should be remarked that 0.30Cu I -MOR, Cu I -MOR and Cu I -FAU produce more CO 2 than the other zeolites. The CO 2 production is also shifted to higher temperatures for FAU, in line with the higher temperature onset for CH 4 consumption. Sushkevich and van Bokhoven correlated the temperature difference for Cu-FAU to a higher fraction of monomeric Cu active sites, which require a higher temperature to activate methane. Further, it can be observed that the production of CO 2 occurs as two peaks for the highly exchanged samples, especially visible for Cu I -MOR. Over 0.30Cu I -MOR, on the other hand, there is one main peak with just a broad shoulder towards higher temperatures. For materials. Sorting the materials based on the temperature of their maximum consumption rate gives Cu I -ZSM-5 < Cu I -MOR < Cu I -BEA < < Cu I -FAU. This order is in line with the results previously reported by Sushkevich and van Bokhoven for CH 4 consumption on ion-exchanged Cu-zeolites [28]. When comparing the highly exchanged Cu I -MOR to the 0.30Cu I -MOR sample though (Fig. 4 (c)), there appears to be a significant difference in the temperature for maximum consumption rate among the two. This earlier methane consumption may suggest that 0.30Cu I -MOR has a higher reactivity towards methane than Cu I -MOR, likely coming from more accessible Cu x O y -sites.

Performance studies in the methane to methanol reaction
The propensity of Cu I -zeolites (Cu I -ZSM-5, Cu I -FAU, Cu I -BEA, Cu I -MOR) towards producing methanol from methane was investigated by performing a well-established stepwise reaction protocol [7]. The test results on the four Cu I -zeolites are presented in Fig. 5 below. The yield is presented in bar columns as µmol of total amount of methanol per gram of zeolite. The squared points represent the selectivity to the total amount of methanol. The unwanted byproduct of the reaction is CO 2 . It should be noted that CH 3 Cl is also observed as a product from the reaction but has not been included in the quantified yield or selectivity. As mentioned previously, the Cl is a remnant from the SSIE, and from EDX analysis after testing, it was observed that a large fraction of the Cl in the framework disappeared after one test, as shown in Table S2.
Excitingly, all four samples show the ability to produce methanol, although only Cu I -ZSM-5 and Cu I -MOR show a selectivity towards methanol similar to that previously reported for ion-exchanged zeolites [7,11]. The appreciable yields of methanol are somewhat surprising, considering that the FT-IR study revealed very few Brønsted sites in these materials, which have been suggested to be needed for methanol formation. Thus, the results presented here might the highly exchanged sample, it does appear as if the onset of CO 2 release occurs before appreciable methane consumption. However, as pointed out above, the true starting point for the uptake of CH 4 is very difficult to determine precisely. The first CO 2 peak appears at around 320 °C for all samples. As this is the main peak in 0.30Cu I -MOR, the sample with the least Cu agglomerates and nanoparticles, it is likely that this is the temperature corresponding to the products formed from selective C-H activation at the active Cu-sites. Alternatively, it could be suggested that since there is a shift of methane consumption to higher temperatures for the Cu I -MOR vs. 0.30Cu I -MOR, the second CO 2 peak for the highly exchanged samples, which overlays well with their CH 4 consumption, is actually the one linked to CH 4 oxidation over the active Cu-sites for these species. If this is the case, it could indicate that the higher Cu-loading has led to more monomeric Cu active sites, also for the other zeolites, and not only FAU. A similar effect with methane consumption being at higher temperatures due to different framework properties was also shown comparing two different CHA frameworks (SAPO-34 vs. SSZ-13) in a previous publication by Kvande et al. [27]. Another possibility is that the higher amount of Brønsted sites in 0.30Cu I -MOR aids in the reactivity towards methane, and that this is the reason behind the clear difference in CH 4 consumption temperature between the two MOR we observe herein, and not necessarily the formation of different types of Cu-oxo sites. Since it has been shown previously that Brønsted sites are Activity data for the methane to methanol reaction over the Cu I -zeolites. The left axis gives the yield in µmol of total amount of methanol per gram of zeolite. On the right axis is the selectivity. In the reaction protocol, the materials were exposed to the following gas flows ( To investigate if we could obtain a higher fraction of active sites, and hence increase the productivity by reducing the Cu exchange degree, a Cu I -MOR with a Cu/Al ≈ 0.30 was synthesized via SSIE. Indeed, this material exhibits a higher methanol productivity (0.10 mol methanol /mol Cu ) as well as a higher yield of methanol, as shown in Fig. 5. The increase can likely be attributed to the formation of less nanoparticles and Cu agglomerates. Alternatively, based on the CH 4 -TPR results discussed above, one could also argue that the increased productivity is related to a higher fraction of Brønsted sites present in this sample compared to the higher Cu loaded samples. However, to unravel this question would require a more extensive study where measures are taken to minimize the presence of unwanted Cu species. Interestingly, the productivity of 0.30Cu I -MOR is within the range of the productivities obtained by ) on most of the materials with Si/Al = 11. However, LIE of Cu on the H-form still generates Cu-MOR with 0.25 mol methanol /mol Cu at comparable Cu-content, outperforming the 0.30Cu I -MOR reported here.
The 0.30Cu I -MOR sample was also tested over three repeated reaction cycles to see if the removal of Cl through CH 3 Cl formation had an impact on the methanol yield by potentially freeing up Cu-sites for C-H bond activation. Only a small increase was observed (from 38 to 42 µmol/g, see Fig S5), even though a qualitative analysis of the amount of CH 3 Cl formed revealed a significant reduction in the CH 3 Cl production from cycle to cycle (Fig S6). Previously it has been shown that repeating the test over several cycles tends to lead to a minor increase in methanol productivity [7]. Based on this, it seems that the removal of Cl from the material does not lead to an increase in methanol productivity for the samples.

Conclusion
Herein an alternative approach to liquid ion exchange for introducing Cu into zeolites has been investigated in the context of selective partial methane oxidation. Extensive characterization of the Cu-zeolites suggests the solid state ion exchange (SSIE) method reported here leads to a significant fraction of Cu nanoparticles and agglomerates. This is believed to hinder the ability of the materials to effectively activate methane. However, the performance data do suggest that methanol is produced on these materials, possibly without the need of Brønsted sites, indicating that methoxy intermediates can be stabilized on other surface sites. To improve on the low methanol yields however, it is necessary with the presence of more Cu-oxo species and possibly Brønsted sites, as well as higher accessibility to all of these sites, which could be achieved with a more fine-tuned suggest that some Cu-species can be capable of selectively converting methane to methanol without facilitation by Brønsted sites. However, formation of some Brønsted sites during the reaction protocol cannot be ruled out. Cu I -ZSM-5 and Cu I -MOR also give the highest yield of methanol of the four samples, with Cu I -MOR clearly outperforming the others, in line with what has been observed previously for ion exchanged materials at these reaction conditions [4,28]. These results are also very much in line with Cu I -MOR being the sample that was able to consume most methane at the lowest temperature area during TPR (Fig. 4). However, Cu-FAU also has a high consumption of methane in TPR, but this occurs at higher temperature. We note that Sushkevich and van Bokhoven were able to show that Cu-FAU yielded much higher amounts of methanol when the methane activation was performed at higher temperatures [28]. Interestingly, in the same study they obtained a higher yield for BEA compared to ZSM-5, which is opposite to the results presented herein. However, this difference could be due to several unknown factors like a difference in methane loading pressure between the two studies, or it could be the higher Si/Al for BEA in this study, as well as the large over-exchange of Cu, causing fewer accessible active sites as observed by CO-adsorption with FT-IR.
To place the yields obtained in this study into perspective, the Cu I -MOR can be compared to a previous publication by Dyballa et al. [4]. There, two different commercial mordenites (Si/Al = 7 and 11) were studied, starting with both H-and Na-form. A comparison between two different Cu-loading procedures was performed, i.e. liquid ion exchange (LIE) and incipient wetness impregnation (denoted SSIE in the paper). This thorough study showed that Cu exchanged onto the H-form generated more productive materials than when the ion exchange was carried out starting with the Na-form. In addition, LIE led to better Cu-mordenites with better methanol productivity than incipient wetness impregnation. The Cu I -MOR used in the present study is based on the commercial MOR, CBV21A, which is the same parent material as the material with Si/ Al = 11 in the paper by Dyballa et al. [4]. When comparing the total methanol yield obtained over the Cu I -MOR (38 µmol/g) to the yields obtained previously (16-169 µmol/g), it is apparent that the total methanol yield of Cu I -MOR is comparable to those achieved for Cu-zeolites obtained with LIE over the Na-form. However, the large fraction of inactive Cu as presented above results in an appreciably lower methanol productivity per copper for the Cu I -MOR (0.03 mol methanol /mol Cu ) compared to those presented by Dyballa et al. (> 0.06). From these initial results, it is suggested that a precise method is necessary to obtain a high fraction of active sites.
technique. As a proof of concept, we showed that the introduction of a lower amount of Cu via SSIE does result in the formation of a higher fraction of active and accessible Cuoxo species, more Brønsted sites and a subsequently higher methanol yield. Therefore, we believe that a SSIE method, with some reduction in the amount of Cu exchanged, and a change in Cu-source that does not lead to Cl-based byproducts, could be a viable option in the future for efficient and scalable synthesis of Cu-zeolites for the methane to methanol reaction. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.