MAPO-18 Catalysts for the Methanol to Olefins Process: Influence of Catalyst Acidity in a High-Pressure Syngas (CO and H2) Environment

The transition from integrated petrochemical complexes toward decentralized chemical plants utilizing distributed feedstocks calls for simpler downstream unit operations. Less separation steps are attractive for future scenarios and provide an opportunity to design the next-generation catalysts, which function efficiently with effluent reactant mixtures. The methanol to olefins (MTO) reaction constitutes the second step in the conversion of CO2, CO, and H2 to light olefins. We present a series of isomorphically substituted zeotype catalysts with the AEI topology (MAPO-18s, M = Si, Mg, Co, or Zn) and demonstrate the superior performance of the M(II)-substituted MAPO-18s in the conversion of MTO when tested at 350 °C and 20 bar with reactive feed mixtures consisting of CH3OH/CO/CO2/H2. Co-feeding high pressure H2 with methanol improved the catalyst activity over time, but simultaneously led to the hydrogenation of olefins (olefin/paraffin ratio < 0.5). Co-feeding H2/CO/CO2/N2 mixtures with methanol revealed an important, hitherto undisclosed effect of CO in hindering the hydrogenation of olefins over the Brønsted acid sites (BAS). This effect was confirmed by dedicated ethene hydrogenation studies in the absence and presence of CO co-feed. Assisted by spectroscopic investigations, we ascribe the favorable performance of M(II)APO-18 under co-feed conditions to the importance of the M(II) heteroatom in altering the polarity of the M–O bond, leading to stronger BAS. Comparing SAPO-18 and MgAPO-18 with BAS concentrations ranging between 0.2 and 0.4 mmol/gcat, the strength of the acidic site and not the density was found to be the main activity descriptor. MgAPO-18 yielded the highest activity and stability upon syngas co-feeding with methanol, demonstrating its potential to be a next-generation MTO catalyst.

The evaluation of the samples' structure is complex due to several factors. (1) The significant phase change of the AEI materials upon calcination, 3 (2) the small domain size and high disorder in all samples (which is intentional to improve their performance as catalysts) and (3) the absence of accurate structure models. The lack of accurate structure models, and the low signal to noise ratio of the data, prevents the effective refinement of known AEI/CHA stacking faults, and the quantification of the AEI/CHA ratio in the samples where large CHA crystallites are present.
Analysis of the PXRD patterns of SAPO-18 by Rietveld refinement against published structures shows a close agreement to the calculated pattern and high phase purity. For the non-calcined sample, peak broadening of certain peaks point to the presence of stacking faults.
Inspection of the PXRD patterns of Co/Mg-AlPO-18 before and after calcination reveals that the material consists of two distinct phases, AEI and CHA. The peak profiles of non-overlapping peaks from the two phases are significantly different, where the CHA peaks are sharper indicating larger diffracting domains. This was confirmed qualitatively by a combined Rietveld/Pawley fit, giving a domain size ratio of CHA/AEI of ≈2:1. Thus, the presented evidence points towards large cuboid CHA crystals, with AEI surface growth, closely resembling previously reported SAPO-18 surface crystallites on cuboid SAPO-34 crystals. 4 In both the Mg-AlPO-18 and Co-AlPO-18 samples, the CHA phase is no longer clearly observed in the PXRD patterns after calcination, indicating a lower thermal stability of this CHA phase than pure AlPO-18. The experiments have been reproduced several times, with varying degree of CHA content before calcination, with no discernable effect on the catalytic performance. Thus, we expect that the active phase in the materials reported herein is AEI in all cases. Product desorption temperature is expected to be inversely proportional to the acidic strength of BAS, since its main contributor is propyl amine cracking temperature, but product diffusion rates may also influence the observed desorption temperature.

Supporting
Supporting Figure 6. IR spectra of MAPOs-18: (a) SAPO-18, (b) MgAPO-18, (c) CoAPO-18, (d) ZnAPO-18 in presence of adsorbed CO at 77K at decreasing coverages. Left panels: OH stretching region (from 3800 to 2600 cm -1 ) of activated samples (dark lines) and in presence of the highest CO coverage (light lines). Right panels: CO stretching region (from 2250 to 2050 cm -1 ) of spectra recorded in presence of decreasing CO equilibrium pressure (from lighter to darker lines).
Left panel illustrates the effect of CO interaction towards the variety of OH groups. Upon interaction with CO in all the samples, most of the P-OH are eroded, with the parallel growth of a band at 3510 cm -1 , confirming the low acidic character of these species. More complex is the evolution of BAS. In particular, BAS signals are downward shifted to 3350 cm -1 for SAPO-18, and to 3280 cm -1 for MgAPO-18 respectively, following the trend observed for the ν(OH) (see main text). In case of CoAPO-18 the BAS species interacting with CO generate a very broad band extending from 3300 to 3000 cm -1 .
Right panels show the corresponding CO stretching range. All the samples are characterized by a strong component due to physiosorbed CO at 2140 cm -1 and a contribution around 2160 cm -1 , associated to CO interacting with the low acidic P-OH. Moreover, the interaction of CO with BAS is observed at 2170 cm -1 and at 2178 cm -1 in case SAPO and MgAPO-18 respectively. In MgAPO-18, a minor contribution is also observed at 2205 cm -1 , corresponding to the interaction of CO with extraframework Mg 2+ cation. As regards the CoAPO-18, a very strong band is centred at 2180 cm -1 . The literature assigns this band to a convolution of different contributions, 5,6 nominally the interaction of CO with Co 2+ Lewis acid centres with different coordination within the framework (tetracoordinated Co 2+ at 2180 cm -1 and defective tricoordinated Co 2+ at 2185 cm -1 ) and a component assigned to tetracoordinated Co 3+ at 2178 cm -1 , due to the oxidation process caused by exposure of the sample to atmospheric moisture. From our experimental data (the evolution of the IR spectra in the OH region) we suggest that the component at 2180 cm -1 contains also the contribution of CO interacting with BAS, as this signal decreases together with the band extending from 3300 to 3000 cm -1 , while the BAS band is being restored. Finally, at low pressures of CO a small band is visible at 2210 cm -1 , generated by the interaction of the probe with a minor fraction of extraframework Co 2+ cations. ZnAPO shows, apart the signal of CO liquid like and CO interacting with P-OH, a component at very high frequency (2210 cm -1 ) suggesting the presence of highly unsaturated Zn 2+ sites.
Supporting Table 2 The IR spectra of all samples are characterized by the components already described in the main text. Notably, the variation of BAS bands intensities, moving from the bottom to the top of the series is not proportional to the increment of the heteroatom contents but it follows the trend evaluated by propylamine-TPD (Table S2). The decreased amount of BAS observed for the sample with the higher amount of Si, indicates that in this sample Si is present aggregated in islands as widely reported in the literature. 7,8 In case of MgAPO-18_a (Figure S12 below); MgAPO-18_b and MgAPO-18_c samples, CO adsorption at low temperature did not evidence the presence of any Mg 2+ counterion: lack of the band at 2205 cm -1 .