Combining Theoretical and Experimental Methods to Probe Confinement within Microporous Solid Acid Catalysts for Alcohol Dehydration

,


■ INTRODUCTION
To fulfill future energy and chemical demands, we must investigate renewable, sustainable feedstocks. 1−23 However, many other fermentation technologies and microbial methods exist that convert lignocellulose into alcohols, particularly ethanol and propanols. 24−31 Specifically, ethene and propene are vital precursors for the polymer, pharmaceutical, and fuel industries.−34 However, these processes are energy intensive and rely on fossil fuel feedstocks, making an alcohol-based dehydration route appealing.
−48 These materials, with active sites located within small pores (less than 8 Å), offer significant control over a reaction.The pores themselves act as molecular sieves, allowing only smaller molecules to access the internal active sites. 49Further once inside, the spatial constraints around the active site can dictate which transition states form and what reaction pathways are followed. 50Finally, once formed, larger products can be trapped within the pores, near the active site, forcing further transformations to occur. 51Diffusion of molecules to and from the active site is therefore an important feature in microporous catalysts, meaning the precise framework topology, pore diameter, and cage sizes within the material heavily influence the reactivity of the system. 48,52,53his has been explored experimentally using a variety of characterization techniques, including one-dimensional (1D) and two-dimensional (2D) infrared spectroscopy 35,54 and solid-state NMR, 55,56 looking at identifying reactive intermediates.Along with the activity, the catalyst lifetime is also highly dependent on diffusion.Coking occurs where larger aromatic byproducts are trapped within the pores, eventually causing deactivation, or in some cases, serve as cocatalysts in the reaction. 57,58It is the desire to control diffusion that has prompted interest in the synthesis of hierarchical materials and the development of increasingly detailed molecular dynamics simulations and combined ab initio and dynamics models to understand diffusion behavior. 59,60This is particularly pertinent when moving toward "real systems," where pore/ cage occupation can play a major role.Recent work by Cnudde et al. using ab initio molecular dynamics highlights this by investigating propene diffusion through a chabazite (SAPO-34) framework, where the free energy of diffusion between cages is lowered as pores become more populated with propene molecules. 48The presence of other molecules within a pore is an important consideration when reproducing experimental systems.This is particularly poignant when considering bio-based feedstocks with significant quantities of water.Very recently, Grifoni et al. also investigated the role of water clusters in a range of zeolite species (GIS, CHA, MFI, and FAU) using metadynamics methods. 47It was shown that water typically resides within zeolite channels, close to the acid site, with limited mobility, and the size of these species is limited in size, typically with more than three water molecules; however, it is noted that the space between water molecules can be occupied by other molecules, allowing for proton transfer to other species.Here, the role of pore confinement was used to explore the transfer of the acidic framework proton to the water cluster, confirming that it played a major role in acid site strength.Similarly, in their work on 1-propanol, Zhi et al. 29 considered the role of water on the dehydration of 1propanol.Here, it was found that water hinders the formation of both propene and dipropyl ether, which was attributed to water-stabilizing adsorbed intermediates, therefore increasing the activation energy for a range of transformations.
Aluminophosphates (AlPOs) are structurally similar to zeolites, sharing many framework topologies, though unlike zeolites that are made of AlO 4 and SiO 4 tetrahedral units, AlPOs are made of alternating AlO 4 and PO 4 units, joined by Al−O−P bonds.AlPOs themselves are comparatively inert, however, the ability to incorporate a range of dopants into the framework leads to the formation of active sites with different functionalities. 61Si 4+ is among the most common dopants in AlPOs, creating silicon-doped aluminophosphates (SAPOs), where Si 4+ can replace a P 5+ framework atom; however, other substitution pathways can occur. 44The resulting charge imbalance is then neutralized by a proton binding to an oxygen atom, adjacent to the Si 4+ dopant, creating a Brønsted acid site (BAS). 44SAPO-34 is the most widely investigated SAPO species due to its industrial implementation in the methanol-to-olefins (MTO) process. 62,63The chabazite (CHA) framework of SAPO-34 promotes the formation of smaller propene and butene species due to the narrow pore diameters (3.8 Å), while the cages (7.3 Å in size) still allow larger aromatic molecules to form, as shown in experimental and theoretical MTO studies. 48,62,63Further, the moderate acidity of SAPO-34 effectively activates the alcohol feedstock while limiting deactivation from the formation of heavier aromatics. 43,48,52This makes it a good candidate for further experimental studies into alcohol dehydration and a useful model system.The Si loading in SAPO-34 is typically quite modest, with less than 10% of framework P atoms substituted by Si, meaning that the active sites are dispersed and can be considered "isolated", with limited interactions between different active sites.Further, unlike common zeolites, such as ZSM-5 or mordenite, 64,65 which have multiple distinct Tsites, SAPO-34 only possesses one crystallographic Al and P site.This makes modeling the active sites in SAPO-34 much simpler, as far fewer combinations of dopant T-site location and specific protonated oxygen need to be considered.This combined with the vast majority of Si isomorphously substituting one framework P site means the active sites in SAPO-34 are not only isolated but also uniform, making it an excellent model system to study theoretical reaction pathways.
In our previous work on ethanol dehydration, we suggested the superior ethene production of SAPO-34, over other SAPO systems, stemmed from its (comparatively) stronger BAS. 66,67nvestigating the catalytic mechanism showed that ethene is not directly formed via the monomolecular dehydration pathway in SAPO-34.Instead, diethyl ether is initially formed through a surface ethoxy species, which then decomposes into ethene, with decomposition being the rate-limiting step, with ethene formation favored at higher temperatures. 66,67This process has repeatedly been seen through theoretical models with ZSM-5 showing ether decomposition to ethene, 68 which has also been seen in the microkinetic models of butanol dehydration by John et al. in Faujasite. 69A preliminary quasielastic neutron scattering study observed ethene and ethanol diffusion through the SAPO-34 pore network, however, diethyl ether diffusion was not observed, suggesting that it was notably slower or just did not occur. 70The latter result raises questions on the suitability of SAPO-34 for larger alcohol systems including propanols and butanols.
Recent theoretical work on ethanol and propanol dehydration in zeolites has begun to tackle whether confinement or acidity is the key feature.−74 Other studies focus on the initial binding of the alcohol to the acid site.Of particular interest is whether η 2 or a η 1 binding occurs, and whether these exist as Eigen or Zundel style complexes, 35 as this will influence which pathway is followed. 75,76Even before this, other studies have questioned the true nature of the active site in zeolites, with recent work by Pfriem et al. focusing on the influence of water clusters on hydronium ion formation and their role in catalytic activity. 30n this work, we present a combined experimental and theoretical study focusing on ethanol, 1-propanol, and 2propanol dehydration using both SAPO-34 and SAPO-5 molecular sieves.While ethene and propene production are vital areas, they will also serve as exemplar reactions to study the possibility of these catalysts being extended to larger alcohols by looking at the influence of subtle modifications to both the size and shape of the pore architecture.SAPO-5 is chosen to contrast the SAPO-34 species as the 1D channels of SAPO-5 are 7.3 Å in diameter, the same size as the cages within SAPO-34. 67This will highlight the influence of the 3.8 Å pore windows of SAPO-34 on the alcohol dehydration.We will combine the catalytic study with molecular dynamics (MD) simulations to explore the influence of molecule size and shape on internal pore diffusion and link this back to the reaction pathways using high-level DFT quantum chemistry calculations, informing microkinetic models on the reaction itself.This will allow us to explore the influence of confinement, and acidity, of the different catalysts, extending our findings to a wider range of potential bioalcohol dehydration catalysts.

■ EXPERIMENTAL METHODS
Alcohol Dehydration Catalysis Methodology.Catalysis was performed using a custom-built flow reactor provided by Cambridge Reactor Design.The reactor comprised of a syringe pump, a mass flow controller, and a reactor with a heater and a control box.A 224 mm quartz reactor tube (4 mm id, 6 mm od) with a 4 mm high frit 80 mm from the base of the tube and a gas inlet 25.8 mm from the top were placed inside a heater jacket.
All catalysis was performed using a fresh 0.3 g of catalyst sieved between 300−500 μm, which, prior to reaction, was dried in the reactor at 400 °C for 1 h under a constant 29.4 mL/min flow of nitrogen.During the reaction, a carrier gas flow of 29.4 mL/min nitrogen and a mixture of 90 wt % alcohol (ethanol/1-propanol/2-propanol) and 10 wt % heptane (internal standard) were flown through the reactor, with liquid flow rates varied to give a WHSV of 1.0, 1.25, 1.50, 1.75, or 2.0 h −1 .Experiments were performed at either 155, 170, 185, or 200 °C for all three alcohols for both SAPO-5 and SAPO-34.Liquid feed flow rates were adjusted, so the same molar quantity of the different alcohols was used.We purposefully limited the study to these three substrates as these to demonstrate the influence of changing alcohol size, while limiting the number of possible products formed, the potential for further reactions and to ensure the reaction occurs in the vapor phase.
After 40 min on stream, 200 μL of vaporized output was injected as a gas into a PerkinElmer Clarus 400 gas chromatogram (GC) with a flame ionization detector with an HP1 cross-linked methylsiloxane (30 m × 0.32 mm × 1 μm) column.All results shown are the average of three consecutive GC injections, therefore limiting the error of our measurements, which is found to be ±3 mol %, in line with the expected GC error.
DFT and DLPNO-CCSD(T):DFT Calculations.Free energies were computed from unit cells of SAPO-34 (hexagonal: a = 13.875Å, c = 15.017Å) and SAPO-5 (hexagonal: a = 13.863Å, c = 16.832Å) with one Brønsted acid site per cell (Figure S1).Geometry optimizations were performed using plane-wave density functional theory (DFT) with a convergence criterion of 0.01 eV/Å.The Perdew− Burke−Ernzerhof (PBE) density functional 77 was employed with the D3 Grimme dispersion correction 78 as implemented in the VASP code using the standard projector-augmented waves (PAWs) 79−81 and a cutoff energy of 400 eV.The Brillouin zone was sampled at the Γ-point only 82 using Gaussian smearing with a width of 0.1 eV.Transition structures were located using automated relaxed potential energy surface scans (ARPESS), 83 and the existence of one imaginary mode connecting the correct minima of the reaction was confirmed.The harmonic approximation was applied to compute entropic contributions to the Gibbs free energy.Vibrational frequencies were derived from the partial Hessian matrix computed using a central finite difference scheme including only the adsorbate, the acid site, and its adjacent Si and Al atoms.Vibrational frequencies below 12 cm −1 were increased to this value because they can lead to inaccurate entropies otherwise. 84Utilizing the hierarchical cluster model approach introduced by Sauer and co-workers, 62,85−90 the electronic energies were refined using domain-based local pair natural orbital coupled cluster (DLPNO-CCSD(T)) 91−93 single-point calculations combined with complete basis set (CBS) extrapolation based on DLPNO-MP2 calculations applied to 46T cluster models of the catalyst (see eqs 1 and 2).The 46T cluster models were cut from the optimized periodic model and saturated with hydrogen atoms. 94,95) where (2)   In this hierarchical cluster model approach, 85−90  ).The CBS extrapolations of Hartree−Fock energies were carried out with the three-point exponential formula 96 with cc-pVXZ (X = D, T, Q).The MP2 correlation was extrapolated using the two-point formula 97 with cc-pVXZ (X = D, T).PBE-D3 calculations on 46T cluster models were performed using the VASP with a vacuum of more than 16 Å separating periodic images of the structure.,104−106 The RIJCOSX (resolution of identity for Coulomb integrals and seminumerical chain-of-sphere integration for HF exchange integrals) approximation 107 with GridX6 was used in RHF calculations. Pats of the calculations were operated using the atomic simulation environment.108 Molecular Dynamics. Moecular dynamics calculations were performed using the COMPASS II forcefield 109 within the Materials Studio package.110 Simulation boxes of 2 × 2 × 5 and 3 × 3 × 3 unit cells were employed for SAPO-5 (AFI) and SAPO-34 (CHA), respectively, with Si substituted for P atoms at random locations. Th temperature used was 428 K.All atoms were free to move during the simulations.Simulations were performed for two million time steps, where each time step was 10 −15 s, giving a total simulation time of 2 nanoseconds.RMS displacement calculations were performed on the final 1 ns of data.
Microkinetic Modeling.Microkinetic modeling was performed using the surfprobe program of the DETCHEM software package 111 at a reference pressure of 1 bar.To ensure convergence of the steady states, coverages are evaluated after 10 4 s.To convert the surface fluxes to TOFs, a surface coverage of 10 −4 mol/cm 2 has been chosen.Further details are given in the Supporting Information (SI).

■ RESULTS AND DISCUSSION
Confirming Catalyst Integrity.As per our previous work, SAPO-34 and SAPO-5 were successfully synthesized. 66,70,112he characterization and physicochemical results (Figures S2− S5 and Tables S1−S3) were in excellent agreement with our previous work, so we refer the reader to these works for a more in-depth discussion of the characterization and acidic properties.Importantly, in both cases, the intended framework (SAPO-34, CHA; SAPO-5, AFI) was the only crystallographic phase detected, showing the phase purity of the systems (Figure S2).Surface areas obtained from nitrogen physisorption (Figure S3) were in line with expected values for these systems and the isotherm shape, and micropore volumes obtained confirmed that both species were primarily microporous (Table S1). 66,70,112 27Al and 31 P NMR confirmed that Al(OP) 4 and P(OAl) 4 were the main signals, as expected (Figures S4A,B). 29Si NMR (Figure S4C) and the metal loadings (Table S2) confirmed that a significant amount of Si(OAl) 4 species were present in SAPO-34, whereas silicon islands had formed within the SAPO-5 framework, as seen by the signal at −97 ppm, attributed to Si(OSi) 2 (OAl) 2 species, likely from 5-and 8-silicon islands, that are known to create BAS. 66,70,112This led to SAPO-34 having a greater number and stronger BAS than SAPO-5, as shown by NH 3 temperatureprogrammed desorption (NH 3 -TPD; Figure S5 and Table S3).Previously, we have seen no evidence of Lewis acidity in these species, in line with other literature reports.Therefore, both frameworks were successfully synthesized.We can assume that due to the difference in framework topologies, SAPO-5 will be a less confined system, with the smaller pore windows of SAPO-34, making it a more confined system for catalysis to occur in.
Ethanol Dehydration.SAPO-5 and SAPO-34 were tested at WHSVs between 1.0 and 2.0 h −1 over a temperature range of 155 to 200 °C (Figures 1 and S6−S8).At the lower temperature range of 155 °C, SAPO-5 achieves modest activity, with conversion ranging between 45−50 mol % (Figure S6A).At this temperature, SAPO-5 is almost exclusively producing diethyl ether (Figure S6B), with a maximum ethene selectivity of 2 mol % (Figures 1A and S6C).Changes in WHSV at this temperature only subtly influence the reactivity (Figures 1A and S6).This is due to the benefits of the increased ethanol concentration being negated by the larger total flow rate and this lower contact time.At 200 °C, SAPO-5 reaches a higher ethanol conversion between 84 and 88 mol % (Figures S6A and S8) due to improved activity from higher temperatures.Diethyl ether is still the primary product (selectivity between 87−91 mol %, Figure S6B), but greater quantities of ethene are also formed (9−13 mol %, Figures 1A,  S6C, and S8).No other products were seen likely due to the mild temperatures used.Thus, increasing the temperature increases the ethene selectivity, in line with our previous observations. 66,67ompared to larger-pored SAPO-5, the smaller-pored SAPO-34 shows increased activity (Figures 1B, S7, and S8).Across the whole temperature and WHSV range, the ethanol conversion remained between 91−95 mol % (Figures S7A and  S8).The ethanol conversion for SAPO-34 remains constant, despite changes in operating parameters, suggesting that this is not a kinetic effect.Recent works suggest that this could be due to our conditions, where SAPO-34 only is approaching equilibrium conversion for ethanol dehydration, 113 and agree with our findings that lower temperature favors diethyl ether formation, before this subsequently forms ethene, and this may also be due to nonoptimized reactor design or catalyst packing within the fixed bed. 114In our previous works, the activation energies of three possible reaction steps were found to be between 66 and 143 kJ/mol, under similar conditions, as those employed here.These values are orders of magnitude higher than what would be expected for pore diffusion processes, suggesting that the system is not pore diffusion limited.Similarly, the estimated Weisz−Prater criterion for our system is 2.7 × 10 −5 (see SI for calculation), which is significantly below 1, signifying that this process is also not mass-transportlimited; thus, we are in a reaction limited regime, and thus, our experiments are kinetically limited with minimal contributions from diffusion limitations.Unlike SAPO-5, changes in WHSV noticeably influence the catalytic behavior of SAPO-34, namely, the product selectivity, where larger WHSV favoring diethyl ether formation and smaller WHSV favoring ethene, something highlighted in the 200 °C data (Figure 1B).In our previous work, we suggest that diethyl ether formation is a second-order process with respect to ethanol; thus, increasing the WHSV leads to greater ethanol concentration and increased diethyl ether formation.We do not see such behavior in SAPO-5, likely due to the larger pore of SAPO-5, facilitating more interactions between ethanol molecules within the pore; thus, diethyl ether formation is already accelerated.
The increased activity of SAPO-34 over SAPO-5 (Figure S8) is likely due to a combination of acidity and confinement factors.NH 3 -TPD confirms that SAPO-34 has almost five times the number of BAS than SAPO-5, but SAPO-34 also has a stronger BAS (Figure S5 and Table S3).Also, one of the fundamental principles of microporous catalysis is that by being confined, lower reaction activation barriers are achieved.This has repeatedly been shown in DFT and combined QM:MM studies, where the van der Waals and dispersion forces from pore confinement lower the activation energy. 34As an extension, it is likely that the larger-pore SAPO-5 will stabilize the ethanol transformation to a lesser extent than the smaller SAPO-34 species.Therefore, there are several factors that will be contributing to the improved activity of SAPO-34 over SAPO-5 (Figure S8).
Focussing on the products formed in SAPO-34 (Figures 1B  and S7B), we see that at low temperatures (155 °C), the diethyl ether selectivity ranges from 92 to 95 mol % for all WHSVs studied, making it the dominant product (Figure S7B).However, increasing the reaction temperature to 200 °C shows an increase in ethene selectivity to between 35− 59 mol % (Figure S7C, again no other products were seen).In our previous work, we attributed this to the sequential reaction pathway of ethanol forming diethyl ether, which then decomposes to ethene, with the latter step being rate limiting, causing a build-up of diethyl ether. 66,67Increasing the temperature increases the rate of ether decomposition, leading to more ethene.We also note that other studies have also come to similar conclusions by considering ethene and diethyl ether formation as parallel competing reactions, both directly forming from ethanol. 39,115The improved ethene selectivity of SAPO-34 compared to SAPO-5 (Figure S8) again is likely the combination of confinement and acidity effects, which we explore through a combined DFT/MD study.
Brønsted acid-catalyzed dehydration pathways were calculated using the hierarchical cluster model approach described in DFT and DLPNO-CCSD(T):DFT calculations section.The considered mechanism for ethanol dehydration according to Figure 2 includes the formation of ethene and diethyl ether from ethanol through concerted and stepwise pathways.
Stepwise pathways begin with the formation of surface ethoxy species (SES) in the first step. 72An analogous mechanism was assumed for dehydration of 1-propanol to dipropyl ether, propene, and surface propoxy species (SPS).
The calculated Gibbs free energy barriers for ethanol dehydration in SAPO-5 and SAPO-34 at 473.15 K and at a reference pressure of 1 bar are shown in Figure 3.The transition structures of the most important elementary steps are depicted above the free energy diagrams.We have chosen here to focus on a SAPO-34 and SAPO-5 system with identical substitution mechanisms, despite this not being the main site in SAPO-5.This was done to limit the number of variables and explore the influence of confinement on the activation energies.
Alcohol dehydration to the ether or alkene starts with adsorption of ethanol at the acid site of the catalyst, with a computed Gibbs free energy of adsorption of −14 kJ/mol in both SAPO-5 and SAPO-34.Hydrogen bonds of the activated complexes to the framework oxygen atoms are more easily formed in SAPO-34 and often shorter than in SAPO-5 due to beneficial confinement.SES formation is a prominent example (Figure 3, gray box), where the water molecule can form two hydrogen bonds of 2.0 Å to the SAPO-34 framework in a chelate-like coordination.A similar two-fold coordination site was not found in SAPO-5, leading to a less stabilised transition state and a higher energy barrier.Unsurprisingly, the transition structures are thus generally less stabilized in the larger SAPO-5 pores.Except for diethyl ether formation from SES (green; 113 kJ/mol), all energy barriers are higher than in SAPO-34, pointing to SAPO-5 being a less active material.Among the adsorbed species (ZOH*), water is more stable on SAPO-34, whereas diethyl ether is more stable on SAPO-5, see also Table S4.
The stepwise reaction paths (Figure 3, gray/green, gray/ black) via SES to ethene or diethyl ether are clearly preferred in SAPO-34, compared to the concerted (direct) paths from adsorbed alcohol (Figure 3, orange, blue).In SAPO-5, however, the higher barrier of 165 kJ/mol for SES formation (gray) makes concerted diethyl ether formation (Figure 3, orange, 153 kJ/mol) the preferred pathway.Ethene would then be formed upon stepwise diethyl ether decomposition via SES.This suggests that the SES is a vital part of the SAPO-34 ethanol dehydration pathway, whereas SAPO-5 likely favors the concerted pathway to the ether.
Comparing the barriers of diethyl ether and ethene formation from SES in SAPO-5 (Figure 3, green and black, respectively) shows that the ether is significantly favored kinetically (113 v 143 kJ/mol).This supports the experimental observation that diethyl ether formation is heavily favored in SAPO-5.Note that the SES-mediated stepwise diethyl ether decomposition to ethene (Figure 3, reverse green, then black) is preferred over the concerted path (red) in both catalysts.Relative to adsorbed diethyl ether (ZOH*DEE), the barriers are 171 kJ/mol in SAPO-5 and 145 kJ/mol in SAPO-34, explaining the low ethene formation in SAPO-5.In contrast, the concerted formation of diethyl ether and ethene from the SES in SAPO-34 (Figure 3, green and black, respectively) shows that ether formation is still favored (119 v 126 kJ/mol), though not as significantly.Thus, the diethyl ether selectivity in SAPO-34 should be more sensitive to ethanol partial pressure than in SAPO-5 at identical conversions, which, for the low pressures applied, could also explain the greater yields of ethene seen (Figures 1 and S8).Using microkinetic modeling, it has been found that for ethanol dehydration in SAPO-34, an increase in temperature from 155 to 200 °C results in an increase of the alkene/ether ratio by a factor of over 100 (see Table S5); this is solidified by the experiment (Figure 1B), which show a drastic increase in ethylene selectivity at higher temperatures.For ethanol dehydration on SAPO-5, a smaller increase of the alkene/ether ratio was calculated (factor 20), which is substantiated by the experimental results in Figure 1A.At higher temperatures, the formation of alkenes is thermodynamically favored.The values calculated here are in good agreement with similar studies, albeit using different materials and reaction temperatures.Alexopoulos et al. calculated an energy barrier of 118 kJ/mol for SES formation at 500 K in ZSM-5, and suggested, as we do, that higher temperatures and lower ethanol partial pressure favor ethene formation. 68Similar work by Roy et al., 116 on an γ-Al 2 O 3 surface, focusing on Lewis acidity found the experimental activation energy for ethanol to ethylene to be 145 kJ/mol, which compares well with our values (Figure 3).
While the calculated reaction energetics (Figure 3, green and black lines) suggest that diethyl ether should decompose to ethene more readily in SAPO-34 than SAPO-5 (see above), we also investigate the mobility of diethyl ether within these frameworks.Molecular dynamics (MD) simulations (Figure 4) show, as expected, that diethyl ether is significantly more mobile (nearly 10 times the RMS displacement) within the larger 7.3 Å pore of the SAPO-5 (AFI) framework (diffusion coefficient of 3.71 × 10 −9 m 2 s −1 ) than within the smaller 3.8 Å pore windows of the SAPO-34 (CHA) framework (2.99 × 10 −10 m 2 s −1 ).This means diethyl ether will reside considerably longer within the CHA system, near active sites allowing far more opportunities for a catalyzed decomposition  reaction to ethene to occur, increasing the ethene selectivity in SAPO-34, while reducing the diethyl ether selectivity.Indeed, we see that this behavior is not exclusive to diethyl ether, as ethanol and ethene both diffuse far more slowly through SAPO-34 (Figure S9; 2.81 × 10 −11 and 1.08 × 10 −9 m 2 s −1 , respectively) than SAPO-5 (Figure S9; 6.19 × 10 −9 and 7.41 × 10 −8 m 2 s −1 ).We note that in SAPO-34, unlike in SAPO-5, ethanol diffuses more slowly than diethyl ether (Figure S9).Despite being smaller than diethyl ether, ethanol will experience greater hydrogen bonding with acid sites due to the hydroxy functional group, which may hinder diffusion through the pore.SAPO-34 has stronger acid sites than SAPO-5, meaning that there would be stronger interactions with ethanol in the SAPO-34 system, anchoring ethanol.Further, the larger pore of SAPO-5 would mean that if ethanol detached from the acid site, it could diffuse away through the channel.However, in SAPO-34, the smaller pore forces ethanol to remain close to the acid site promoting reattachment and thus slower diffusion.In contrast, the weaker hydrogen bonding of SAPO-5 with ethanol and the more spacious pore mean that ethanol can diffuse more freely than diethyl ether.
To better distinguish between the influence of confinement and active site, we now expand our catalytic study to include 1propanol and 2-propanol feedstocks.As ethanol and the propanols (and associated products) are all much smaller than the SAPO-5 pore, the confinement effects should be similar for all species with this catalyst.However, we would expect a greater influence of confinement for ethanol and the propanols in SAPO-34.We note that changing the alcohol feedstock will not alter the active sites in SAPO-5 or SAPO-34; therefore, the effect of acidity should be constant.Thus, changing the alcohol feedstock should highlight just the influences of confinement between the two SAPOs.
1-Propanol Dehydration.At the range of temperatures studied (155−200 °C), SAPO-5 shows similar activity with 1propanol and ethanol (Figures 5A and S10).Specifically, at 155 °C, the 1-propanol conversion is between 50−56 mol %, compared to 74−79 mol % at 200 °C, spanning the same range as the ethanol data (Figures S6A and S10A).As both ethanol and 1-propanol are primary alcohols, the C 3 chain should have a greater positive inductive effect than the C 2 chain, helping to activate the molecule.As before, we see that increasing the reaction temperature increases conversion (Figures 5 and  S10A).
There is significantly more alkene forming in SAPO-5 with the 1-propanol feedstock than when ethanol is used (Figures 1A and 5A).At 155 °C, propene selectivity varies between 23− 30 mol %, rising to between 47−53 mol % at 200 °C, depending on the WHSV used (Figure S10C).However, we note, like the ethanol case (Figure 1), varying the WHSV has little influence on the catalytic performance of SAPO-5.The improved alkene selectivity and reduced ether formation (compared to ethanol; Figures S6 and S10) are likely due to the longer carbon chain lowering the activation energy of propene formation, compared to ethene.It is unclear whether this is due to faster dipropyl ether decomposition or the direct unimolecular dehydration of 1-propanol to propene occurring at a faster rate.To further investigate this, microkinetic modeling has been performed to investigate the influence of the reaction barrier for the green and black pathways of Figure 8 on the alkene/ether ratio; results of this are shown in Table S6.We see that variation of the black barrier has a far greater impact on the alkene/ether ratio than the variation of the green barrier does, showing us a clear preference for the black pathway for production of the alkene.
Despite the variations in temperature, SAPO-34 has lower conversions with 1-propanol (72−94 mol %, Figure S11A) than with ethanol (91−95 mol %, Figure S7A) as the feedstock, despite the 1-propanol expected to be more active for this reaction than ethanol.We reason that this reduced activity is due to confinement effects, where the smaller-pored SAPO-34 may struggle to accommodate larger reaction intermediates and transition states, compared to SAPO-5 (Figures 3 and S12).Despite this reduced activity, SAPO-34 is still more active than SAPO-5, showing greater activity, under identical reaction conditions (Figure S12).Like SAPO-5, SAPO-34 typically favors propene over dipropyl ether.Thus, moving from ethanol to 1-propanol reduces the activity of SAPO-34 but increases total alkene production (Figures S8  and S12).These findings suggest that either diffusion of 1propanol to the active sites is being hindered or is harder to activate within the SAPO-34 pores or a combination thereof.An analogous prediction of the Gibbs free energy barriers for 1-propanol under the same reaction conditions in SAPO-34 is shown in Figure 6.Given the variety of active sites present in the SAPO-5 system, we have not performed an analogous DFT study for SAPO-5; instead, we chose to focus on the influence of changing substrate size, transitioning from ethanol to 1propanol in SAPO-34.
Going from an ethanol to a 1-propanol feedstock in SAPO-34 increases the energy barrier for forming the surface alkoxy species (Figures 3 and 6) from 122 to 133 kJ/mol�the first barrier in the stepwise alcohol dehydration path (147 kJ/mol relative to adsorbed 1-propanol) to propene is now higher than the second one.Roy et al. suggested that on a nonporous γ-Al 2 O 3 surface, where confinement does not play a significant role, 116 the activation energy from ethanol to 1-propanol would decrease from 145 to 141 kJ/mol.This shows that, in SAPO-34, the SPS is less readily formed than the SES, likely due to spatial constraints within the smaller SAPO-34 pore.We have calculated a 31 kJ/mol difference in dispersive interaction for SAPO-5 compared to SAPO-34, which is in line with the fact that SAPO-5 has a larger pore and therefore smaller dispersive interactions. 117Despite this increase in energy, the SPS is still the preferred pathway for alkene and ether formation (Figure 6).The greater energy barrier for 1propanol, compared to ethanol, aligns with the lower catalytic activity seen (Figures 1 and 5).It is noteworthy that the surface alkoxy route (Figures 3 and 6, gray) increases from 122 to 133 kJ/mol for SAPO-34 from ethanol to 1-propanol.Likely this is due to the smaller substrate being less sterically hindered within the smaller-pored SAPO-34, therefore lowering the activation barrier of the process.The concerted alkene formation (Figures 3 and 6, blue) decreases from 166 to 150 kJ/mol.This minor stabilization would be expected as 1propanol has slightly larger dispersion interactions compared to ethanol (we calculated a difference of 9.8 kJ/mol from D3).In addition, the alkene is strongly favored in the equilibrium of dipropyl ether (DPE) and propene.Extrapolating this trend further to bulkier reagents would likely see that the concerted alkene formation plays a greater role, potentially becoming the favored pathway.Dispersion interactions mainly depend on the number of adsorbate atoms (Figure S13), as has also been observed in the literature. 117Our computational findings are in good agreement with previous literature, particularly from Zhi et al., who focussed on 1-propanol dehydration within the ZSM-5 catalyst.At 433 K, they calculated a value of 136 kJ/ mol for propoxide dehydration to propene, whereas our smaller-pored SAPO-34 system has a lower value of 118 kJ/ mol, likely due to a combination of the smaller pore size and higher temperatures studied.
In summary, kinetic selectivity for ether or alkene can be explained by the free energy barriers (green, black) in the stepwise paths.Figures 3 and 6 illustrate the relations between these barriers relative to the adsorbed alcohol.The largest difference is clearly observed for ethanol dehydration in SAPO-5 with a preference of 30 kJ/mol toward diethyl ether.The difference is only 7 kJ/mol for ethanol dehydration in SAPO-34 and 18 kJ/mol for 1-propanol dehydration in the same catalyst.An experimentally observed selectivity toward propene can only be explained if DPE stays sufficiently long in vicinity to the active site, or visits multiple sites, allowing it to decompose according to the thermodynamic equilibrium.
Again, comparing the diffusion of dipropyl ether in the two SAPO systems shows that it diffuses far more rapidly through SAPO-5 (3.79 × 10 −10 m 2 s −1 ) than SAPO-34 (<1.00 × 10 12 m 2 s −1 ), differing by an order of magnitude (Figure 7).This suggests that while the ether to alkene step has the largest energy barrier of the 1-propanol SAPO-34 pathway (Figure 6), there will be greater opportunities for it to occur, whereas the propene can diffuse out of the pores far more rapidly (Figure S14).This will lead to greater selectivity to propene and less selectivity to dipropyl ether.
2-Propanol Dehydration.At lower temperatures (155 °C), SAPO-5 achieves similar levels of conversion with 2propanol (Figure S15), as with 1-propanol (Figure S10) and ethanol (Figure S6).However, at elevated temperatures (200 °C) and a lower WHSV (1.0−1.5 h −1 ), it reaches nearquantitative conversion (Figure S15A).This improved activity at higher temperatures is due to the superior activity of secondary (and tertiary) alcohols, compared to primary  alcohols for dehydration reactions.Again, alcohol conversion drastically increases at higher temperatures, with significant jumps in activity seen between 170 and 200 °C (Figure S15A).
Considering the product distribution of SAPO-5 with 2propanol (Figures 8 and S15B), greater preference is shown for alkene formation, more so than with ethanol and 1-propanol (Figures 8 and S16).This is shown by propene now being the dominant product under all studied conditions, with propene selectivity ranging from 58−78 mol % at 155 °C to above 98 mol % at 200 °C (Figure 8).Thus, here, changing between ethanol and 2-propanol results in increased activity and alkene formation for SAPO-5.Unlike in the ethanol and 1-propanol cases, varying the WHSV significantly affects the catalytic performance (Figure 8A) of SAPO-5.This suggests that the alcohol interactions within the pore are becoming hindered, as we saw in the SAPO-34 system, and thus WHSV, and therefore, alcohol concentrations are beginning to influence activity.
Despite possessing a far greater number and stronger BAS, the activity of SAPO-34 for 2-propanol dehydration is comparable to SAPO-5, and at low temperatures (155 °C), SAPO-5 has become the more active catalyst (Figures 8, S17, and S18).At 155 °C, SAPO-34 has a 2-propanol conversion of between 32−51 mol % (Figure S17A), compared to SAPO-5 (46−56 mol %; Figure S16A).This is a significant finding, showing that SAPO-34 is noticeably less effective at converting 2-propanol than 1-propanol or ethanol, though despite SAPO-34's lower activity with 2-propanol, propene is almost exclusively formed (Figures 8B, S17B, and S18), showing a strong preference for alkene formation.The low quantities of diisopropyl ether suggest the following: (1) If a parallel competitive mechanism is occurring, then ether is not being formed, likely due to the steric constraints of the bulky ether, as suggested above, or, once formed it cannot leave the pores.
(2) If the sequential mechanism is happening, the ether is being formed but is consumed so fast that it is in a pseudosteady state, with no build-up.This would support the idea that the ether becomes trapped in the pores of SAPO-34 and is forced to react further to propene before leaving the pores (product selectivity). 51omparing the behaviors of the SAPO-5 and SAPO-34 systems with 2-propanol as a feedstock, we see that the largerpored SAPO-5 has a higher conversion but lower propene selectivity (Figure S18).Again, comparing the diffusion of the bulkier diisopropyl ether through the SAPO-5 and SAPO-34 pores (Figures 9 and S18; 3.00 × 10 −10 and <1.00 × 10 −12 m 2 s −1 , respectively) confirms our theory above that indeed the ether can diffuse far more readily out of SAPO-5, whereas it is far less mobile in SAPO-34, leading to improved decomposition to propene.We note that the reaction mechanism likely differs between primary (ethanol and 1-propanol) and secondary alcohols (2-propanol); however, the pathway and possible routes are expected to be the same for SAPO-5 and SAPO-34 and will be influenced by the larger substrate, allowing us to draw comparisons from our kinetic findings.Previous work by Gunst et al. 118 has compared the activity of alcohol isomers, namely, 1-butanol and iso-butanol with ZSM-5, concluding that iso-butanol favors direct dehydration to isobutene, while 1-butanol favors ether formation.We envisage that a similar trend, in line with our catalytic data, would apply to the mechanism of 1-propanol and 2-propanol.
Overall, combining our catalysis and quantum chemical and molecular dynamics findings (Table 1), we see a trend where SAPO-34 is becoming increasingly inactive, as the substrates get larger.We believe that this is due to a combination of an increase of the formation barrier for the surface alkoxy species, which is entropy related, slowing down the reaction, and the more hindered pore diffusion of the alcohol substrates, preventing the substrate reaching the active sites (Figure 10).Through this study, we believe we have successfully distinguished between acidity effects (ethanol) and confinement effects (2-propanol) for the two SAPO systems.This will aid future design of alcohol dehydration catalysts for a much wider range of alcohol substrates, including bio-based alcohols.

■ CONCLUSIONS
Alcohol dehydration is an appealing route toward the sustainable production of short-chain alkenes.Designing optimal catalysts for the selective formation of alkenes requires detailed knowledge of the alcohol dehydration pathway, pore diffusion, and the influence of confinement and acidity.To understand these different factors, we have explored the behavior of two SAPO systems, the larger-pored SAPO-5 and the smaller-pored SAPO-34, for the dehydration of ethanol, 1propanol, and 2-propanol.In doing so, the acidity influence is kept constant within the framework structure, while the confinement effects are varied, and we investigated combining experimental catalysis data, with theoretical quantum chemical data to calculate reaction energy barriers and molecular dynamics simulations to explore pore diffusion.
Our catalytic study showed that changing the alcohol substrate had a significant influence on the behaviors of the two systems.When ethanol was used as a substrate, confinement would play a minimal role, and therefore, SAPO-34 proved to be more active than SAPO-5, with greater selectivity to the alkene.Progressing to 1-propanol and then 2propanol, we saw the activity of SAPO-34 drop, eventually matching the activity of SAPO-5.Further, despite the reduced activity, SAPO-34 transitioned to exclusively forming the alkene product, with scarce little ether being formed.These observations were investigated through a DFT study.This concluded that, as the substrate increased in size, the energy barrier for the activation of the alcohol to a surface alkoxy species also increased in SAPO-34, explaining the reduced activity.Compared to ethanol dehydration in SAPO-34, a bypassing of ether formation is likelier during 1-propanol dehydration due to a decreased barrier for the concerted pathway.Thus, the experimentally observed selectivity toward propene can only be explained if the ether is sufficiently immobile to be decomposed toward the thermodynamic equilibrium.
Molecular dynamics repeatedly showed that the ether product diffused far more readily in the SAPO-5 framework than SAPO-34, where it was quite immobile.The increased residency time of the ether, especially compared to the alkene, is likely a key factor in the improved alkene selectivity of SAPO-34.Extending the length of time, the ether spends inside the framework offers more opportunities for ether decomposition into the desired alkene product, which could allow the alkene to be selectively formed at much lower temperatures.
Overall, our approach of combining high-level quantum chemical calculations, MD simulations, and the gathering of detailed catalytic data provided complimentary insights for the design of solid acid catalysts for alcohol dehydration, allowing us to explore the role of confinement separately to the nature of the acid site.In doing so, this will provide unique insights for future catalyst design for sustainable, bio-based processes.

Coordinate files for calculated geometries for high-level quantum chemical calculations (TXT)
Further experimental methodology, characterization, catalysis, molecular dynamics, and high-level quantum chemical results (PDF) the cc-pVDZ basis set was used for E DLPNO-CCSD(T)/DZ 46T of the 46T cluster model.ΔE MP2/CBS 46T represents the difference between MP2based CBS extrapolation (E DLPNO-MP2/CBS 46T ) and MP2/cc-pVDZ calculations for 46T clusters (E DLPNO/DZ 46T

Figure 2 .
Figure 2. Suggested dehydration mechanism for the formation of ethene and diethyl ether (DEE) from ethanol (EtOH).The surface ethoxy species (SES) is the intermediate along the stepwise paths, whereas the concerted paths in orange, blue, and red circumvent SES formation.Adapted from Amsler et al., ref 72, with permission from the Royal Society of Chemistry.

Figure 3 .
Figure 3. Free energy barriers of ethanol dehydration for (A) SAPO-5 and (B) SAPO-34 at 473.15 K and at a reference pressure of 1 bar.Transition structures with distances in Å are depicted above the diagrams, the rate-determining steps (RDS) are extracted below.The values for the barriers in the preferred path analysis were determined using the energy span model that is measured from the lowest adsorbed state to the highest barrier.The atoms are color coded, such that brown = carbon, pink = hydrogen, red = oxygen, yellow = aluminum, blue = silicon.

Figure 6 .
Figure 6.Free energy barriers of 1-propanol (PrOH) dehydration for SAPO-34 at 473.15 K and a reference pressure of 1 bar.Transition structures with distances in Å are depicted above the diagram, the size of the rate-determining step (RDS) is extracted below.The atoms are color coded such that brown = carbon, pink = hydrogen, red = oxygen, yellow = aluminum, blue = silicon.

a
Catalytic data represents the values at 155 °C, with a WHSV of 2 h −1 .Values shown are the composition of the molecules in the outlet stream.b Highest energy values for the lowest energy pathway.c Values after 1000 ps.

Figure 10 .
Figure 10.Schematic showing the proposed mechanism of alcohol dehydration.Confinement will influence the diffusion to and from active sites and the energy barriers, whereas acidity will determine the product selectivity and overall activity.

Table 1 .
Summarizing the Key Findings Relating to Catalysis, Quantum Chemical Calculations, and Molecular Dynamics Simulations catalytic data (mol %) a calculated energy barrier (kJ/mol) b RMS displacement (Å) c