PCCP PAPER. Dynamic Article Links. Cite this: Phys. Chem. Chem. Phys., 2012, 14,

Transcription

1 PCCP Dynamic Article Links Cite this: Phys. Chem. Chem. Phys., 2012, 14, PAPER Mechanistic investigations on dimethyl carbonate formation by oxidative carbonylation of methanol over a CuY zeolite: an operando SSITKA/DRIFTS/MS studyw Jana Engeldinger, Manfred Richter and Ursula Bentrup* Received 24th October 2011, Accepted 31st October 2011 DOI: /c1cp23361k The simultaneous combination of steady state isotopic transient kinetic analysis (SSITKA) with diffuse reflectance Fourier transform spectroscopy (DRIFTS) and mass spectrometric (MS) analysis was applied to study the oxidative carbonylation of methanol (MeOH) to dimethyl carbonate (DMC) on a CuY zeolite catalyst prepared by incipient-wetness impregnation of commercial zeolite NH 4 Y. The interaction of the catalyst with different reactants and reactant mixtures (O 2, CO, CO/O 2, MeOH/O 2, MeOH/CO, and MeOH/CO/O 2 ) was studied in detail using 16 O 2 / 18 O 2 as well as 12 CO/ 13 CO containing gas mixtures. DMC is produced via a monodentate monomethyl carbonate (MMC) species as intermediate which is formed by the concerted action of adsorbed methoxide and CO with gas phase MeOH. Adsorbed bidentate MMC species were found to be inactive. Lattice oxygen supplied by CuO x species is involved in the formation of MMC. Gas phase oxygen is needed to re-oxidize the catalyst but favours also the oxidation of CO to CO 2 and unselective oxidation reactions of MeOH to methyl formate, dimethoxymethane, and CO 2. The appropriate choice of reaction temperature and of the oxygen content in the reactant gas mixture was found to be indispensable for reaching high DMC selectivities. 1. Introduction Focus on production of dimethyl carbonate (DMC) results from the increasing market for this biodegradable and nontoxic chemical with wide applicability, e.g. as a fuel additive, as a phosgene alternative in polycarbonate and isocyanate syntheses, and as a methylation agent replacing dimethyl sulfate and methyl iodide. The capacity of the current corrosive liquidphase oxidative carbonylation of methanol (MeOH) in the presence of Cu(I) chloride cannot satisfy the increasing demand. 1 Therefore, a better DMC production process, preferably a vapour phase process, is needed. Cu-containing zeolites, in particular CuY zeolites, have been found to be attractive catalysts which promote the oxidative carbonylation of MeOH to DMC in the gas phase. 1 7 Based on mainly spectroscopic studies the established view on the mechanism of the heterogeneously catalyzed oxidative carbonylation of MeOH in the gas phase involves the direct insertion of CO into the C O bond of surface methoxides. 2 6 Leibniz-Institut fu r Katalyse e.v. an der Universita t Rostock (LIKAT), Albert-Einstein-Str. 29a, D Rostock, Germany. ursula.bentrup@catalysis.de; Fax: ; Tel: w Electronic supplementary information (ESI) available: Details of the experimental setup and additional results from DRIFTS and MS analysis. See DOI: /c1cp23361k Carbomethoxide species 2 4 or monomethyl carbonate (MMC) 5,6 are discussed as intermediates which give DMC by the consecutive reaction with MeOH. Alternatively, it is proposed that DMC formation can also be accomplished by methylation of surface carbonates, formed from CO/O 2 via oxidation on Cu(II) cations, fixed at ion exchange positions of zeolitic supports, favourably faujasites. 7 The specific role of adsorbed carbonate-like and formate species in DMC formation was also reasoned from in situ FTIR spectroscopic studies. 8 Concerning the methoxide formation different positions are represented in the literature. While the necessary participation of gaseous oxygen was stated by King, 2 Zhang and Bell, 6 as well as by Anderson and Root, 4 it was found by Engeldinger et al. 8 that methoxide formation proceeds on preferable Cu(I) Lewis sites without additional supply of oxygen. The nature of the intermediate species formed by the subsequent insertion of CO into the methoxide was also controversially discussed. According to Zhang and Bell 6 as well as Engeldinger et al. 8 a monomethyl carbonate species is formed as intermediate whereas King 2 discussed the formation of a carbomethoxide. Insofar it is not clear in which step of the catalytic cycle oxygen is involved. Furthermore, the possibly additional oxidation of Cu(I) to Cu(II) and, consequently, the role of Cu(II) species remained unconsidered. 2,6 This might be explained by the use of solid-state ion-exchanged CuY catalysts in which the This journal is c the Owner Societies 2012 Phys. Chem. Chem. Phys., 2012, 14,

2 presence of stable Cu(I) cations was assumed. However, impregnated CuY catalysts containing Cu(II) besides Cu(I) species react in the same way implicating a specific role of Cu(II), too. 8 To get more insights into the catalytic reaction mechanism the simultaneous combination of infrared spectroscopic and mass spectrometric analyses under SSITKA (steady state isotopic transient kinetic analysis) 9 conditions using a single reactor might be a powerful tool The SSITKA technique consists in replacing one of the reactants by its isotopomer during the reaction while the exchange of the labelled reaction product is simultaneously followed by mass spectrometry (MS) and the surface species by infrared spectroscopy. For this purpose diffuse reflectance Fourier transform spectroscopy (DRIFTS) has been proved to be a suitable technique. In contrast to self-supporting wafers used in infrared transmission spectroscopy the DRIFTS reaction cell works like a fixed-bed flow reactor in which external mass transport limitations can be minimized by utilizing catalyst powders with an appropriately selected particle size. Indeed, it has to be considered that contrary to transmission FTIR DRIFTS is a more surface sensitive method which allows the probing of the sample typically less than 200 mm in depth. 11 Therefore a real quantitative analysis is restricted. Nevertheless, the SSITKA/ DRIFTS/MS approach is suited for elucidating the role of the different surface species present on catalyst surfaces in operando. Using this technique it is possible to discriminate between active and spectator species by comparison of the isotopic exchange rate of the surface species, followed by DRIFTS, with the rate of exchange of gas phase species measured by MS. We present here for the first time a SSITKA/DRIFTS/MS study of the oxidative carbonylation of MeOH to DMC on a CuY zeolite as catalyst. The aim was to elucidate the role of oxygen in this reaction and its relevance in terms of selective and unselective reaction pathways, and to follow the way of CO insertion during carbonylation. For this purpose the interaction of the catalyst with the different reactants and reactant mixtures was studied in detail using 16 O 2 / 18 O 2 as well as 12 CO/ 13 CO containing feeds, respectively. For the experiments a CuY zeolite with a Cu content of 13 wt% Cu was selected. 2. Experimental The CuY zeolite was prepared by incipient-wetness-impregnation of a commercial NH 4 Y zeolite (Zeolyst International, Si/Al = 2.6) with an aqueous solution of Cu(NO 3 ) 2 3H 2 O (Sigma- Aldrich). After impregnation the samples were dried at 120 1C overnight and calcined at 400 1C in air overnight. The Cu content of the CuY sample determined by ICP-OES amounts to 13 wt% (denotation in the following: 13CuY). The Cu loading was proved to be high enough to compensate inherent Brønsted acidity of the zeolite Y, and, moreover to enable the formation of agglomerated oxidic copper species. The formation of CuO x could be verified by UV-vis spectroscopy, and crystalline CuO was detectable by XRD, too. A commercial DRIFTS reaction cell (Harrick) implemented into a BioRad FTS-60A FTIR spectrometer was used. The cell practically acts as a fixed-bed flow reactor. The gas outlet of the reaction cell was connected to an OmniStar quadrupole mass spectrometer (Pfeiffer Vacuum). For the SSITKA experiments with 16 O 2 / 18 O 2 and 12 CO/ 13 CO different gas dosing systems were used the detailed flow diagrams of which are shown in Fig. S1a, and S1b (ESIw). The following gases and gas mixtures were used: 5 vol% 12 CO/He, 5 vol% 16 O 2 /He, and 1 vol% Ne/He (Air Liquide), 13 CO (pure, Linde) and 5 vol% 18 O 2 /He (Linde). MeOH was dosed using a saturator (14 1C) with He or CO/He (cf. Fig. S1a, b, ESIw). The general feed composition was 5.1 vol% MeOH/2.5 vol% CO/1.2 vol% O 2 balanced with He. In the experiments with Ne as a marker the mixture additionally contained 0.2 vol% Ne. The switching from the normal to the isotopic labelled gas mixture was done by a four-way valve maintaining a constant flow rate of 25 ml min 1. Prior to the reaction at 150 1C or 130 1C, the catalyst (ca. 50 mg in a particle fraction of mm) was pre-treated by heating in flowing He up to 400 1C. After recording a DRIFT spectrum of the pre-treated catalyst at the chosen reaction temperature, the respective gas mixture was dosed and the reaction was simultaneously monitored by DRIFTS and MS in dependence on time. If the steady state was reached it was switched from the normal non-labelled gas mixture to the respective isotopomer-containing feed. Generally, difference DRIFT spectra were evaluated and obtained by subtracting the spectrum of the pre-treated catalyst from the respective adsorbate spectra. 3. Results and discussion 3.1 Interaction of the CuY catalyst with 16 O 2 / 18 O 2 At first it was checked whether the oxygen of the zeolite lattice or the CuO x agglomerates exchanges with gaseous oxygen. The experiment was carried out at a reaction temperature of 150 1C. If 16 O 2 is replaced by 18 O 2 under steady state conditions a simultaneous increase of the MS signals of 18 O 2 and the tracer Ne was observed 30 s after switching whereas the MS signal intensity of 16 O 2 decreases in parallel (Fig. S2, ESIw). Because no 16 O 18 O was detected an exchange between the lattice and gas phase oxygen can be excluded. It is known from the literature that an exchange of zeolite lattice oxygen against gas phase oxygen proceeds only at relatively high temperatures (ca C). These temperatures are obviously necessary because of the weak interaction between lattice oxygen and gas phase oxygen at comparable low temperatures. 13 However, the presence of transition or noble metal cations decreases the exchange temperature. 13,14 Thus, investigations of transition metal containing Na-ZSM-5 catalysts showed that introduced Cu(II) cations lower the oxygen exchange temperature from 400 1C to 240 1C. 13 There are no hints from the literature that an exchange at 150 1C would be possible which is in accordance with our experimental results. 3.2 Interaction of the CuY catalyst with 12 CO/ 13 CO After switching from the 12 CO-containing to the 13 CO-containing feed the DRIFT spectra shown in Fig. 1 were obtained. The simultaneously recorded MS signal intensities of 12 CO and 13 CO 2184 Phys. Chem. Chem. Phys., 2012, 14, This journal is c the Owner Societies 2012

3 Fig. 1 DRIFT spectra of adsorbed CO on 13CuY at 150 1C after 30 min exposure to 2.5 vol% 12 CO/He and subsequent switching to 2.5 vol% 13 CO/He. exposure the catalyst to 12 CO/ 16 O 2 and 12 CO/ 18 O 2, respectively, revealed that comparable data were obtained. This means that the catalyst is able to form CO 2 without a marked participation of gas phase oxygen. Boccuzzi and Chiorino 18 studied the CO oxidation on Cu supported oxides and found two independent pathways for CO 2 formation. The first one comprises the reaction of gas phase oxygen with adsorbed CO according to an Eley Rideal mechanism, whereas in the second one CO is oxidized by surface lattice oxygen. On our CuY sample CO 2 is formed with but also without gas phase oxygen in comparable amounts. This clearly points to the participation of lattice oxygen as long as available, and subsequent re-oxidation by gas phase oxygen according to the Mars-van-Krevelen mechanism. In the case of an Eley Rideal mechanism a rapid formation of C 16 O 18 O should occur, but this was not observed in our experiments. For the sake of completeness it has to be mentioned that carbonate species adsorbed at the catalyst surface were additionally observed. Due to the broadness of the respective bands in the range cm 1 the possible incorporation of 18 O cannot definitely be evaluated. Because the position and intensity of the Cu(I) CO band remained unaffected the carbonate species are obviously located at Cu(II) sites. Fig. 2 MS signal intensities of (a) 12 CO/ 13 CO, Ne and (b) 12 CO 2 / 13 CO 2, Ne versus time. Switching from 2.5 vol% 12 CO/He to 2.5 vol% 13 CO/He at time = 0. as well as 12 CO 2 and 13 CO 2 in dependence on time are shown in Fig. 2. The bands at 2160/2144/2112 cm 1 obtained after 30 min exposure to the 12 CO-containing feed (Fig. 1) are assigned to Cu(I) 12 CO modes of Cu(I) carbonyls at different sites. 8 Switching to the 13 CO-containing feed a rapid intensity decrease of these bands is observed accompanied by the appearance of new ones at 2110/2097/2062 cm 1 (Fig. 1; Fig. S3, ESIw). During the exchange of isotopomers the reduced mass of the molecule changes but bond strength and molecule geometry persist. Thus, the expected band shifts can be calculated 16,17 and amount to approximately 48 cm 1 for n 12 CO - n 13 CO. This correlates well with the experimental results (cf. Fig. 1). For the sake of completeness also the MS signal of the tracer Ne is shown in Fig. 2. The slight time displacement of the 13 CO signal compared to the Ne signal (Fig. 2a) points to adsorption effects of CO. A low amount of 13 CO 2 could also be detected in the gas phase after switching to 13 CO (Fig. 2b) where the observed time shift of the 13 CO 2 signal goes along with that of 13 CO. Obviously, CO is partially oxidized by participation of lattice oxygen of CuO x species. 3.3 Interaction of the CuY catalyst with CO/O 2 using 16 O 2 / 18 O 2 During experiments with 12 CO/ 16 O 2 and 12 CO/ 18 O 2 mainly C 16 O 2 is formed in the gas phase, only low signal intensities were observed for C 16 O 18 O. The comparison of the signal intensities of fragment m/z =44(C 16 O 2 ) measured during 3.4 Interaction of the CuY catalyst with MeOH/O 2 using 16 O 2 / 18 O 2 The simultaneously measured DRIFT spectra and MS signal intensities of oxygen and different products after switching from the MeOH/ 16 O 2 to the MeOH/ 18 O 2 feed are shown in Fig. 3 and 4. After reaching a steady state with MeOH/ 16 O 2 several adsorbate bands are seen in the DRIFT spectrum (Fig. 3a). The band around 1704 cm 1 is assigned to a ncqo vibration and results from adsorbed methyl formate (MF). The adsorption of pure MF on 13CuY at 150 1C was checked in a separate experiment on a self-supporting wafer of the 13CuY sample in the transmission mode showing a typical sharp band at 1714 cm 1. In the DRIFT spectra this band is broader and appears at lower wavenumbers. The band at 1460 cm 1 is characteristic for adsorbed methoxide species, 19 the respective CH valence vibrations were observed at 2959/2847 cm 1 (not shown). The bands at 1590 cm 1 and 1369 cm 1 indicate the formation of formate species, while the band at 1643 cm 1 can be assigned to adsorbed monomethyl carbonate (MMC) species. 8,20 After switching to the MeOH/ 18 O 2 feed, the spectra do not change in principle, only the formate band at around 1590 cm 1 shifts to lower wavenumbers. This effect can be seen more clearly by comparing the spectra after 210 min MeOH/ 16 O 2 and 240 min MeOH/ 18 O 2 feed, respectively (Fig. 3b). Because of the broadness of the formate band a detailed analysis of the different components is difficult. Nevertheless, the exchange of 16 O 2 against 18 O 2 affects the position and shape of the formate band while the intensity slightly increases. In Fig. 4 the respective MS signal intensities of O 2,CO 2, MF, and dimethoxymethane (DMM) are shown. DMM is only detectable in the gas phase. The formation of C 16 O 18 O proceeds very quickly while the signal of C 16 O 16 O loses This journal is c the Owner Societies 2012 Phys. Chem. Chem. Phys., 2012, 14,

4 Fig. 3 (a) DRIFT spectra of 13CuY at 150 1C measured after 210 min exposure to 5.1 vol% MeOH/1.2 vol% 16 O 2 /He and subsequent switching to 5.1 vol% MeOH/1.2 vol% 18 O 2 /He; (b) Comparison of the spectra obtained after 240 min MeOH/ 16 O 2 and 240 min exposure to MeOH/ 18 O 2. Fig. 4 MS signal intensities of oxygen, MF, CO 2, and DMM versus time. Switching from 5.1 vol% MeOH/1.2 vol% 16 O 2 /He to 5.1 vol% MeOH/ 1.2 vol% 18 O 2 /He at time = 0. intensity in the inverse manner. Whereas the intensity of 16/18 MF rises continuously, the formation of 16/16 DMM remains unaffected and increases slightly. Only very low intensities of 16/18 DMM are observed. The different reaction pathways of methanol oxidation are well known 19,21 and can be divided into two principal kinds of reactions: 21 (i) oxidation reactions which need oxygen (molecular or supplied by the catalyst) and (ii) dehydration reactions which do not need oxygen (cf. Scheme 1). According to this scheme it is obvious that DMM as well as MF formation requires formaldehyde as an intermediate which is produced in the first step by oxidation of methanol. Considering the fact that 16/16 DMM is continuously produced in low amounts despite dosing the MeOH/ 18 O 2 feed it can be concluded that lattice oxygen is involved in the formation of formaldehyde while 18 O is not inserted in formaldehyde but rather in the formed water (cf. Scheme 1) as observed by MS. Gas phase oxygen participates mainly in the next oxidation Scheme 1 Oxidation and dehydration reactions of MeOH. steps of formaldehyde leading to 16/18 MF and C 16 O 18 O which are detected very quickly after switching from the MeOH/ 16 O 2 to MeOH/ 18 O 2 feed. This is in agreement with the DRIFTS results showing the continuous shift of the formate band at 1590 cm 1 to lower wavenumbers (cf. Fig. 3). Obviously, the formate species partly reacts with MeOH to form MF while the other ones remain adsorbed at the catalyst surface Phys. Chem. Chem. Phys., 2012, 14, This journal is c the Owner Societies 2012

5 With progressive reaction in the MeOH/ 18 O 2 feed low but increasing MS signal intensities of 16/18 DMM are observed (Fig. 4). DMM can be perceived as acetal formed by the reaction of formaldehyde with MeOH. It is known that such acetals can undergo hydrolysis reaction in the presence of water which leads to the back-formation of aldehyde and MeOH. 22 It was demonstrated by Stasiuk et al. 23 using 18 O-labelled aldehydes that the carbonyl carbon oxygen bond is cleaved by hydrolysis with water because H 2 18 O was detected. In reverse, 18 O-labelled aldehydes are formed by hydrolysis of the respective acetals with H 2 18 O. Thus, if H 2 18 O is involved in the hydrolysis of DMM then isotopic labelled formaldehyde (HCH 18 O) should be formed. Consequently, if labelled formaldehyde undergoes reaction with MeOH also isotopic labelled DMM is produced. And in conclusion, if labelled formaldehyde is further oxidized to formate by 18 O 2 also C 18 O 18 O can be formed as can be seen in Fig Interaction of the CuY catalyst with MeOH/CO using 12 CO/ 13 CO The simultaneously measured DRIFT spectra and MS signal intensities of CO, CO 2, and DMC after switching from the MeOH/ 12 CO to the MeOH/ 13 CO feed are displayed in Fig. 5. After 120 min exposure the catalyst to MeOH/ 12 CO feed bands at 2120, 1735, 1639, 1602, and 1477/1452 cm 1 can be observed in the DRIFT spectrum (Fig. 5a). The band at 2120 cm 1 is assigned to ncu(i) 12 CO the position of which, however, is shifted to lower wavenumbers in comparison to the adsorption of 12 CO alone (cf. Section 3.2 and Fig. 1). This is caused by the additional adsorption of methoxide species at the same Cu(I) site as already described. 6,8 The band at 1477/1452 cm 1 is assigned to adsorbed methoxide species, 19 the band at 1602 cm 1 indicates the formation of formate species, while the band at 1639 cm 1 results from adsorbed MMC species. 8,20 Compared to the interaction of the catalyst with MeOH/O 2 the formation of formate species is suppressed as expected. The band at 1735 cm 1 results from adsorbed DMC as has been checked by separate experiments probing the thermal stability of adsorbed DMC on 13CuY. After 5 min exposure of the catalyst to the MeOH/ 13 CO feed a new band at 1689 cm 1 appears whereas the Cu(I) 12 CO band vanishes and a new one at 2072 cm 1 becomes dominant. The latter effect is caused by the exchange of Cu(I) 12 CO - Cu(I) 13 CO (cf. Section 3.2). The appearance of the band at 1689 cm 1 goes along with decreasing intensity of the band at 1735 cm 1 which results from adsorbed DMC. The band shift amounts to 46 cm 1 and points to the 12 C/ 13 C exchange of the carbonylic carbon in DMC. Isotopic labelled DMC is also observed in the gas phase as can bee seen from the respective MS signal intensities (Fig. 5b). Whereas the MS signal intensity of the fragment m/z =90( 12/12/12 DMC) decreases with time the intensity of the fragment m/z =91( 12/12/13 DMC) increases. As expected no signal of the fragment m/z =92( 12/13/13 DMC) was observed. The temporal change of signal intensities 12/12/12 DMC/ 12/12/13 DMC proceeds slowly and not simultaneously which is obviously caused by overlapping processes of adsorption/desorption and isotopic exchange. In contrast, decreasing and increasing of MS signal intensities of the fragments m/z = 44 ( 12 CO 2 ) and m/z = 45 ( 13 CO 2 ) are comparable. The formation of 12 CO 2 is due to the CO oxidation by lattice oxygen of CuO x species as already described (cf. Section 3.2) and partly caused by the total oxidation of MeOH which proceeds on CuO containing samples very quickly already at 150 1C. 24,25 DRIFTS and MS results reveal that CO is needed for DMC formation whereas no additional oxygen is apparently required. Accordingly, the question arises on the way DMC is formed and on the intermediates involved. King 2 discussed the formation of a carbomethoxide species as intermediate whereas Zhang and Bell 6 argue a MMC species. From own investigations an intermediate MMC species was assumed, too. 8 This assumption is based on the appearance of a band at around 1640 cm 1 and another one at 1344 cm 1 which was assigned to adsorbed MMC. A band located at Fig. 5 (a) DRIFT spectra of 13CuY at 150 1C obtained after 120 min exposure to 5.1 vol% MeOH/2.5 vol% 12 CO/He and subsequent switching to 5.1 vol% MeOH/2.5 vol% 13 CO/He; (b) simultaneously measured MS signal intensities of 13 CO, CO 2, and DMC versus time (switching from 12 CO/MeOH to 13 CO/MeOH at time = 0). This journal is c the Owner Societies 2012 Phys. Chem. Chem. Phys., 2012, 14,

6 1639 cm 1 is also clearly seen in Fig. 5a. However, looking at the intensity changes of this band no alterations (except for low deviations) are observable by switching from MeOH/ 12 CO to the MeOH/ 13 CO feed. This clearly indicates that this species is strongly adsorbed and does not act as an intermediate. For the sake of completeness it has to be mentioned that this species is formed whenever MeOH is present in the feed. Because MeOH is also partly oxidized to CO 2 we explained the formation of this species by an interaction of CO 2 with adsorbed methoxide according to Lamotte et al. 20 The respective band positions on Al 2 O 3 were found at 1623, 1476, and 1370 cm 1 attributed to n as COO, dch 3, and n s COO modes, respectively. Zhang and Bell 6 described bands of adsorbed MMC at 1656, 1499, 1354 cm 1 assigned to ncqo, dch 3, and n s COO modes, respectively. With regard to DMC formation by interaction of MeOH and CO on 13CuY the following mechanistic steps based on the described experimental results are proposed (Scheme 2). MeOH is adsorbed as methoxide species together with CO at the same Cu site. If adjacent lattice oxygen is available an oxidation step can take place leading to the adsorbed monodentate-like MMC (I) species. Otherwise, the formation of another adsorbed species MMC (II) with a bidentate-like structure is probable. Because of the charge distribution in the latter complex MMC (II) the appearance of two bands (n as /n s ) of the O C O grouping at lower wavenumbers compared to the ncqo is expected. For this reason the band at 1639 cm 1 observed in our experiments is assigned to MMC (II). The MMC (II) species seems to be the more stable one because it is observed in all cases when MeOH is adsorbed on 13CuY independently from the presence of further components like CO or gaseous oxygen. There are no experimental hints for the formation of adsorbed MMC (I) species on 13CuY from the experiments described above. For the subsequent reaction with methanol MMC (I) seems to be more reactive than the stable MMC (II) species. Taking into account that the reaction is an electron donor acceptor interaction between methoxy oxygen of MeOH (via single electron pairs, after the H atom has interacted with surface oxygen forming a hydroxyl group), and the electron-deficient MMC carbon atom, such interaction is only possible in the case of monodentate species as demonstrated in Scheme 2. This mechanistic approach is comparable to the mechanism described by Lamotte et al. 26 for the reaction of different structured methoxy species on ThO 2 and CeO 2 with CO 2.It was found that only monodentate methoxide species are able to interact with CO 2. Following the reaction steps in Scheme 2 the concerted action of adsorbed methoxide and CO with gas phase MeOH leads to the formation of DMC via MMC as intermediate. The required oxidation step for MMC formation proceeds under participation of lattice oxygen supplied by CuO x species of the catalyst. In this step the catalyst is partially reduced whereat gas phase oxygen is used for the re-oxidation according to the Mars-van-Krevelen mechanism. 3.6 Interaction of the CuY catalyst with MeOH/CO/O 2 using 12 CO/ 13 CO In Fig. 6 the DRIFT spectra obtained after the reaction of 13CuY with MeOH/ 12 CO/O 2 and switching to MeOH/ 13 CO/O 2 at 150 1C are compared with those obtained after reaction with MeOH/ 12 CO and switching to MeOH/ 13 CO. For the sake of clarity only the spectra after the steady state of the respective reactions are displayed. From the spectroscopic point of view, comparable reaction characteristics are evident (Fig. 6). The same typical bands are observable characterizing adsorbed formate (1595/1602 cm 1 ), MMC (1639 cm 1 ), methoxide (1476/1452/1464/1456 cm 1 ), and 12/12/12 DMC/ 12/12/13 DMC (1735/1689 cm 1 ). However, looking at the band intensities of DMC, a lower intensity is observed when oxygen is present in the feed. It is also seen that the band intensities of the ncu(i) 12 CO/nCu(I) 13 CO bands at 2116/2072 cm 1 differ distinctly. They are lower in the presence of gas-phase oxygen which is caused by an essentially higher percentage of CO 2 mainly resulting from CO oxidation Scheme 2 Proposed mechanism for DMC formation. Fig. 6 DRIFT spectra of 13CuY at 150 1C obtained after 120 min exposure to 5.1 vol% MeOH/2.5 vol% 12 CO/He and 5.1 vol% MeOH/ 2.5 vol% 12 CO/1.2 vol% O 2 /He, respectively; and after 120 min subsequent exposure to the respective 13 CO containing feeds Phys. Chem. Chem. Phys., 2012, 14, This journal is c the Owner Societies 2012

7 Fig. 7 DRIFT spectra obtained after 120 min exposure of the fresh 13CuY catalyst at 150 1C to 5.1 vol% MeOH/2.5 vol% 13 CO/1.2 vol% O 2 /He and 5.1 vol% MeOH/2.5 vol% 12 CO/1.2 vol% O 2 /He, respectively. (cf. Section 3.3). It is evident that CO oxidation is preferred in the presence of oxygen. These findings are also reflected by comparison of the respective MS signal intensities of CO 2, MF, and DMC (Fig. S4, ESIw). For a possible improvement of the band assignment in terms of the MMC species an additional experiment was carried out starting with the MeOH/ 13 CO/O 2 feed without a pre-reaction with MeOH/ 12 CO/O 2. In Fig. 7 the DRIFT spectra are compared which were obtained after exposure of the respective fresh 13CuY catalyst to MeOH/ 13 CO/O 2 or to MeOH/ 12 CO/O 2. It is conspicuous that the intensity ratios of the bands at 1642 cm 1 and 1595 cm 1 change depending on the feed composition. Normally, the band at 1595 cm 1 is assigned to the n as COO mode of formate species, but, the intensity of this band is abnormally strong when the fresh catalyst is exposed to the 13 CO-containing feed (Fig. 7). Taking into account that formate species contribute to the whole intensity of the band at 1595 cm 1 (the percentage can be estimated by analyzing the band intensity in the spectrum obtained after exposure to the 12 CO-containing feed) a distinct intensity remains. Furthermore, comparing the bands at 1642 cm 1 after exposure of the catalyst to the different feeds, a lower intensity is observed after treatment with the 13 CO-containing feed. This suggests that both bands (1642/1595 cm 1 ) belong to each other. The band shift amounts to 47 cm 1 which is in the range of the expected shift for noco mode by 12 C/ 13 C exchange. Thus, it can be concluded that the band at 1642 cm 1 stems from a species which is formed by the participation of CO. Consequently, this band can be assigned free of doubt now to the n as COO mode of MMC (II) species as already assumed (cf. Section 3.5). The respective n s COO mode is expected to be around 1460 cm 1, but the analysis of this spectral region is difficult due to overlapping methoxide bands. The reaction was also investigated at different temperatures (Fig. 8). The spectra obtained after the respective reactions with MeOH/ 12 CO/O 2 and MeOH/ 13 CO/O 2 at 130 1C and 150 1C are displayed in Fig. 8a. The MS signal intensities of MF, DMM, and DMC recorded during exposure of the 13CuY catalyst to the MeOH/ 12 CO/O 2 feed are shown in Fig. 8b. In accordance with the observation in the DRIFT spectra where adsorbed formate species were only detected in the spectra measured at 150 1C the MS analysis also reveals an essentially higher amount of MF and DMM. This result points to a favoured oxidation of MeOH at higher temperatures leading to oxidation products formed via intermediate formaldehyde (cf. Scheme 1). The amounts of DMC obtained at 130 1C and 150 1C differ not so much, but the DMC selectivity is essentially higher at 130 1C. Additionally, it has to be considered that DMC is adsorbed to a higher extent at the Fig. 8 (a) DRIFT spectra of 13CuY obtained after 120 min exposure to 5.1 vol% MeOH/2.5 vol% 12 CO/1.2 vol% O 2 /He at 130 and 150 1C, respectively; and after 60 min subsequent exposure to the respective 13 CO containing feeds; (b) simultaneously measured MS signal intensities of MF, DMM, and DMC versus time obtained during exposure of the catalyst to MeOH/ 12 CO/O 2. This journal is c the Owner Societies 2012 Phys. Chem. Chem. Phys., 2012, 14,

8 Fig. 9 MS signal intensities of CO 2 (m/z = 44, 45), DMC (m/z = 59) and MF + DMC (m/z =60)versus time measured after switching from the 5.1 vol% MeOH/2.5 vol% 12 CO/1.2 vol% O 2 /Hefeedtotherespective 13 CO-containing feed (switching at time = 0) at 130 1C and 150 1C. catalyst surface at 130 1C (Fig. 8a) which is observable by comparison of the respective band intensities at 1735/1685 cm 1 ( 12/12/12 DMC/ 12/12/13 DMC). The MS signal intensities of CO 2, MF, and DMC recorded at 130 1C and 150 1C after switching from the MeOH/ 12 CO/O 2 to the MeOH/ 13 CO/O 2 feed are shown in Fig. 9. Because of the fact that the fragment m/z = 60 belongs to 12/12 MF as well as to 12/12/13 DMC only the overall intensities are given. The stronger oxidation power of the catalyst at higher temperatures is obvious comparing the signal intensities of CO 2 and MF/DMC. No 13 C-containing DMM and MF were detected. Thus, the slow increase of the 12/12 MF + 12/12/13 DMC signal is caused by the formation of isotopic labelled DMC. 4. Conclusions Operando SSITKA/DRIFTS/MS measurements were applied to study the elementary steps involved in the DMC formation by gas phase oxidative carbonylation of MeOH on a highloaded CuY zeolite catalyst. This operando technique was proved to be a suitable tool to discriminate between active and non-active reaction intermediates as well as selective and unselective reaction pathways. The simultaneous inspection of the species adsorbed at the catalyst surface (DRIFTS) and those occurred in the gas phase (detected by MS) enables a comprehensive insight into the catalytic reaction mechanism. In the case of such complex reaction it was proved to be advantageous to study the interaction of the different reactants with the catalyst separately. Only in this way it is possible to elucidate the specific role of the several reactants and their impact on the respective reaction. In the following the main findings are summarized: - An exchange ( 16 O/ 18 O) between lattice oxygen (both of the zeolite and oxidic Cu species) and gas phase oxygen on 13CuY at temperatures up to 150 1C can be excluded. CO is partially oxidized to CO 2 on 13CuY whereat CO 2 is formed with but also without gas phase oxygen in comparable amounts. This clearly points to the participation of lattice oxygen of oxidic Cu species and following re-oxidation by gas phase oxygen according to the Mars-van-Krevelen mechanism. - By interaction of the catalyst with CO/O 2 the additional formation of adsorbed carbonate species located at Cu(II) sites existing in 13CuY is favoured. Because of the low intensities and broadness of the respective surface carbonate bands it was not possible to clarify if 18 O is incorporated into these carbonate species, and to which extent such carbonate species can take part in DMC formation. Further experiments including additional FTIR transmission measurements are planned to elucidate their role. - Methoxide species formed by dissociative adsorption of MeOH without additional gas phase oxygen and CO are adsorbed at the same Cu(I) site. During exposure the catalysts to MeOH/O 2 adsorbed methoxide, formate, and MMC species were observed while CO 2, MF and DMM were detected in the gas phase. The 16 O 2 / 18 O 2 exchange experiments revealed that oxygen is involved in MF and CO 2 formation but to an essentially lower extent in the formation of DMM. - CO is needed for DMC formation whereat no additional oxygen is apparently required. As illustrated in Scheme 2 DMC is formed via a monodentate monomethyl carbonate (MMC) species as intermediate which is formed by the concerted action of adsorbed methoxide and CO with gas phase MeOH. The required oxidation step for the MMC formation initially proceeds under participation of lattice oxygen supplied by CuO x species of the catalyst. Because the catalyst is partially reduced in this step gas phase oxygen is used to re-oxidize the catalyst according to the Mars-van-Krevelen mechanism. Only monodentate-like MMC (I) species are active in DMC formation because bidentate-like MMC (II) species were found to be strongly adsorbed at the catalyst surface. Due to its high reactivity MMC (I) species cannot be detected by DRIFTS. - Reaction temperatures lower than 150 1C enhance the desired DMC formation. This means that the activation energy for DMC formation should be lower than for MF and DMM formation. Furthermore, MeOH oxidation via intermediate formaldehyde to MF and DMM is obviously favoured on CuO x agglomerates at high oxygen content in the feed. Thus, a low oxygen percentage in the feed is advantageous for DMC formation. The tuning of the reaction temperature and oxygen content in the feed is indispensable for reaching high DMC selectivities at an optimum Cu loading of zeolite Y (n Si /n Al ca. 3) between wt%. Acknowledgements The authors thank the Deutsche Forschungsgemeinschaft (DFG) for the financial support (grant No. RI 829/7-2), and Dr Michael Bartoszek (LIKAT) for helpful discussions concerning MS analysis. Notes and references 1 G. Rebmann, V. Keller, M. J. Ledoux and N. Keller, Green Chem., 2008, 10, S. T. King, J. Catal., 1996, 33, S. T. King, Catal. Today, 1997, 33, Phys. Chem. Chem. Phys., 2012, 14, This journal is c the Owner Societies 2012

9 4 S. A. Anderson and T. W. Root, J. Mol. Catal. A: Chem., 2004, 220, Y. Zhang, I. J. Drake, D. N. Briggs and A. T. Bell, J. Catal., 2006, 244, Y. Zhang and A. T. Bell, J. Catal., 2008, 255, M. Richter, M. J. G. Fait, R. Eckelt, E. Schreier, M. Schneider, M.-M. Pohl and R. Fricke, Appl. Catal., B, 2007, 73, J. Engeldinger, C. Domke, M. Richter and U. Bentrup, Appl. Catal., A, 2010, 382, S. L. Shannon and J. G. Goodwin Jr., Chem. Rev., 1995, 95, A. Goguet, F. C. Meunier, D. Tibiletti, J. P. Breen and R. Burch, J. Phys. Chem. B, 2004, 108, F. Meunier, Catal. Today, 2010, 155, F. Meunier, Catalysis, 2010, 22, Y.-F. Chang, G. A. Somorjai and H. Heinemann, J. Catal., 1995, 154, J. Nováková and L. Brabec, J. Catal., 1997, 166, N. Mizuno, H. Mori, K. Mineo and M. Iwamoto, J. Chem. Phys. B, 1999, 103, J. Weidlein, U. Müller and K. Dehnicke, Schwingungsspektroskopie, Thieme, Stuttgart-New York, 1982, p E. E. Ortelli, J. M. Weigel and A. Wokaun, Catal. Lett., 1998, 54, F. Boccuzzi and A. Chiorino, J. Phys. Chem., 1996, 100, G. Busca, Catal. Today, 1996, 27, J. Lamotte, O. Saur, J.-C. Lavalley, G. Busca, P. F. Rossi and V. Lorenzielli, J. Chem. Soc., Faraday Trans. 1, 1986, 82, J. M. Tatibouët, Appl. Catal., A, 1997, 148, E. H. Cordes and H. G. Bull, Chem. Rev., 1974, 74, F. Stasiuk, W. A. Sheppard and A. N. Bourns, Can. J. Chem., 1956, 34, A. Lawson and S. J. Thomson, J. Chem. Soc., 1964, G. Busca, J. Mol. Catal., 1987, 43, J. Lamotte, V. Moravek, M. Bensitel and J. C. Lavalley, React. Kinet. Catal. Lett., 1988, 36, 113. This journal is c the Owner Societies 2012 Phys. Chem. Chem. Phys., 2012, 14,