READ ME File For 'Experimental and computational optimisation of methanol dehydration to dimethyl ether'. Dataset DOI: https://doi.org/10.5258/SOTON/D3366 Date that the file was created: January, 2025 ------------------- GENERAL INFORMATION ------------------- ReadMe Author: Maciej G. Walerowski, University of Southampton [https://orcid.org/0009-0006-4763-8169]. Date of data collection: August 2022 - December 2024. Information about geographic location of data collection: Southampton, UK and Harwell Research Complex, UK. Related projects: -------------------------- SHARING/ACCESS INFORMATION -------------------------- Licenses/restrictions placed on the data, or limitations of reuse: CC BY - This license enables reusers to distribute, remix, adapt, and build upon the material in any medium or format, so long as attribution is given to the creator. The license allows for commercial use. Recommended citation for the data: M. G. Walerowski, S. Kyrimis, M. E. Potter, A. E. Oakley, M. Carravetta, L. Armstrong & R. Raja, Data in support of 'Experimental and computational optimisation of methanol dehydration to dimethyl ether' This dataset supports the publication: Experimental and computational optimisation of methanol dehydration to dimethyl ether AUTHORS: Maciej G. Walerowski, Stylianos Kyrimis, Matthew Potter, Alice Oakley, Marina Carravetta, Lindsay-Marie Armstrong & Robert Raja. TITLE: Experimental and computational optimisation of methanol dehydration to dimethyl ether JOURNAL: Catalysis Science & Technology PAPER DOI IF KNOWN: DOI https://doi.org/10.1039/D5CY00062A Links to other publicly accessible locations of the data: N/A Links/relationships to ancillary or related data sets: N/A -------------------- DATA & FILE OVERVIEW -------------------- This dataset contains: Raw characterisation data > Broken down into 7 folders containing different types of characterisation data (BET, CHN, EDS, PXRD, SEM, ssNMR, & TPD) > Each folder then contains the relevant raw data for all catalysts in the appropriate file format (e.g. txt, csv, or jpg) Raw catalysis data > Contains 5 spreadsheets in the csv file format which contain raw catalysis data for all catalyst at the relevant conditions Raw kinetics data > Contains 2 spreadsheets in the csv file format which contain the raw kinetics data and residence time calculations Raw CFD data > Contains 1 spreadsheet in the csv file format which contains CFD results compared to experimental results Raw DoE results > Contains 1 spreadsheet in the csv file format which contains the CFD-drive DoE results and a correlation matrix obtained using the DoE results -------------------------- METHODOLOGICAL INFORMATION -------------------------- Description of methods used for collection/generation of data: EDS analysis was performed using an Oxford Instruments EDS (backscattered electron – composition (BED-C)) detector which was integrated into a JSM-7200F field emission scanning electron microscope. AZtect software was used for elemental analysis. Powder was loaded directly onto a carbon tape and analysed without any sputter coating. 10 kV acceleration voltage was used with a working distance between 8 and 12 mm. For EDS elemental composition, 2 separate particle-rich regions were selected, in each image, 5 particle zones were scanned. Each particle region was around 100-1000 μm in size which resulted in analysis of a high number of particles. Elemental composition was obtained by averaging those 10 separate scans and the error shows the standard deviation between the scans. PXRD characterisation was performed using a Bruker D2 Phaser instrument. Patterns were obtained using Cu Kα radiation (λ = 1.54184 Å) at 30 kV voltage and 10 mA current using a 0.6 mm slit. Patterns were obtained in the 5-45° 2θ range with 0.01° increments and 0.4 s per step. Whole powder pattern fitting (WPPF) was performed using Rigaku PDXL 2 software to obtain the unit cell parameters. LaB6 was used as the width standard. CIF file for the AEL framework was obtained from the IZA database of zeolite structures. Surface area and porosity characterisation was performed using Micromeritics Tristar II 3020 analyser. N2 was used as the adsorptive, and a liquid N2 bath was utilised. Analysis performed between 0.00 and 0.95 p/p0 (relative pressure). 124 adsorption and 30 desorption points were used to obtain the full physisorption isotherm. BET surface area and pore volume were calculated automatically by the Micromeritics Tristar II 3020 software. Samples (~0.15 g) were thoroughly degassed prior to measurements for a minimum of 21 hours using a Micrometrics Vac Prep 062 system by heating them under vacuum at 120°C, with final pressure of ~100 mTorr. Scanning electron microscopy (SEM) characterisation was performed using a JEOL JSM-7200F field emission scanning electron microscope. 5 kV acceleration voltage was used with 93 μA emission current. Working distance ranged between 6 and 12 mm. Sample was imaged directly on tape without prior sputter coating. For all ssNMR analysis, the samples were thoroughly dried overnight at 180°C under a Schlenk line vacuum. The samples were then transferred under a nitrogen atmosphere into a glovebox. All ssNMR rotors were packed under a nitrogen atmosphere in a glovebox to minimise exposure to moisture. Post-processing of NMR data was done using MestreNova software. For the 27Al and 31P analysis, the weighed samples (~0.03 g) were packed into a 3.2 mm thin wall zirconium rotor. Analysis was performed using an Agilent Varian 600 MHz Premium Shielded spectrometer, with a 14.1 T field strength. OpenVnmrJ software was utilised. The rotors were span at 14,000 Hz at the magic angle and the spectra acquired in triple (1H27Al31P) resonance mode. The 27Al spectra were acquired at 156.46 MHz using a 312500 Hz spectral width, 20 ms acquisition time, 3 s recycle delay, 128 scans, 63 W power, 2000 aX90 amplitude and 1.2 μs pwX90 pulse width. The 31P spectra were acquired at 243.06 MHz using 100000 Hz spectral width, 20 ms acquisition time, 25 s recycle delay, 128 scans, 63 W power, 3800 aX90 amplitude and 3.25 μs pwX90 pulse width. YAG and H3PO4 standards were used to calibrate the 27Al and 31P chemical shift axes, respectively. For the 29Si analysis, the weighed samples (~0.08 g) were packed into a 4 mm zirconium rotor and capped with a KELF rotor/cap. Analysis was performed using a Bruker 600 MHz wide bore Ultrashield spectrometer, with a 14.1 T field strength. Topspin 4.0.7 software was utilised. The rotors were span at 8000 Hz at the magic angle and the spectra acquired in double ( 1H29Si) resonance mode. The 29Si spectra were acquired in direct acquisition mode. Direct acquisition spectra were acquired with proton decoupling, 4 μs pulse width, 200 W power, 10 W decoupling power, 260 scans and 300 s relaxation delay. TMSS standard was used to calibrate the 29Si chemical shift axis. NH3-TPD experiments were performed on Quantachrome ChemBET Pulsar TPR/TPD instrument. 0.2 g of 100 – 425 μm pelletised sample was dried at 550°C for 2 hours under a 24 mL/min He and 6 mL/min O2 flow. Sample was then cooled down to 150°C and put under a 30 mL/min flow of 5% NH3/He, the sample was held at 150°C for 2 hours. Flow was then changed to 30 mL/min He and sample held at 150°C for further 2 hours. System was then heated to 600°C at a rate of 5°C/min and evolution of NH3 as a function of temperature was monitored. The system was then held at 600°C for 1 hour to fully desorb any remaining NH3. CHN analysis was performed by the London Metropolitan University elemental analysis service. The samples were weighed using a Mettler Toledo high precision scale and analysed using a ThermoFlash 2000. Samples analysed in duplicate and the error shows the standard deviation between the two results. Catalysis was performed in a custom built reactor which comprised of a nitrogen cylinder, mass flow and temperature controller, dual syringe pump, heating jacket, round bottom flask and a round bottom flask heater. The SAPO-11 catalysts powders were pelletised at 4 tonnes for 10 s to obtain self-supporting pellets of 2.5 cm diameter which were then crushed and sieved 5 times to obtain catalyst granules in either a 106-300, 300-500 or 500-710 μm range. The catalyst granules (0.150, 0.300 or 0.450 g) were then loaded into a 40 cm quartz reactor tube (0.4 cm i.d, 0.6 cm o.d) and sandwiched between two layers of 1 mm borosilicate beads. Bottom borosilicate bead layer was adjusted to ensure the catalyst bed was located in the isothermal zone of the reactor. The top borosilicate bead layer length was kept constant for all catalysts to ensure reproducible mixing and heating of the reaction gas. Catalysts were activated at 400°C in a 25 mL/min flow of nitrogen for 1 hour before the reaction. The reactor was then cooled to 170°C and once temperature attained, methanol (3, 6, 12 or 18 μL/min, equivalent to WHSV of 0.5, 1, 2 or 3 h-1 when 0.300 g of catalyst was used, respectively) was carried into the reactor using a 25 (or 31, 38 or 41 for WHSV experiments) mL/min stream of nitrogen. The nitrogen flow rate was adjusted for the different WHSV experiments to ensure the combined gas flow and hence residence time was constant across all WHSV experiments. Calculations for this can be found below. The reaction was allowed to proceed for a minimum of 45 minutes and the outlet gas was collected continuously in a heated round bottom flask. The vaporized outlet gas (0.2 mL) was injected into a Perkin Elmer Clarus 480 gas chromatograph with the injector set at 170°C. The volatile species were carried in a He carrier gas through an Elite-5 column (5% diphenyl/95% dimethyl polysiloxane, 30m, 0.25 mm, 0.25 μm) which was located in a 30°C isothermal oven. The hydrocarbon species were quantified using a flame ionisation detector set at 250°C with a total analysis time of 5 minutes. Outlet gas was analysed in triplicate, reactor temperature was then increased by 15°C and the process repeated. Chloroform was used as an external standard to calculate methanol conversion. Each experiment was performed in triplicate, using fresh catalyst on a different day in a randomised order. Standard deviation was calculated between the repeat results and presented as an error bar. For the time-on-stream stability study, the temperature and methanol WHSV were held at 275°C and 3 h-1 throughout the study to accelerate catalyst deactivation. To determine the combined gaseous flow rate of methanol and nitrogen, the gaseous methanol flow rate was firstly calculated. This was estimated by considering the volumetric expansion of methanol during evaporation as outlined in equation S1. The combined gaseous flow rate could then be calculated by summing the individual methanol and nitrogen gas flow rates. For a methanol WHSV of 3 h-1, the combined gaseous flow of methanol and nitrogen could thus be estimated as shown in Equation S2. Combined gas flow rate and hence residence time can thus be kept similar by increasing the nitrogen flow rate to counteract the decrease in methanol gas flow rate as the methanol WHSV is reduced. Methanol dehydration over SAPO-11 was modelled using a first-order rate equation (Equation S3 and S4). The experimentally observed rate constant k, was estimated at a range of temperatures using the open source software Copasi. A genetic algorithm with a population size of 2000 and 100 generations was used for the initial estimation of k, followed by a Levenberg-Marquardt local optimisation to obtain the final value of k. Equation S4 can be rearranged into its non-exponential form as shown in Equation S5. The apparent activation energy and pre-exponential factor for methanol dehydration over SAPO-11 was then calculated from the gradient and intercept of the linear plot obtained by plotting ln(k) vs 1/T. The average activation energy and pre-exponential factor was calculated by averaging the individual values for the four different methanol WHSVs considered. A three-dimensional CFD model was built in Ansys Fluent 19.2.8 Only the catalytic bed was modelled as it was experimentally confirmed that the borosilicate bead layers are chemically unreactive. Catalyst bed was modelled using a cylindrical geometry (0.2 cm radius and a 1.9, 3.6, or 5.4 cm height) and meshed using Ansys meshing to give 700,000 hexahedral computational cells. Thermal equilibrium between solid and fluid phases was assumed. Inlet, wall, and outlet initial temperatures were set to match experimental temperatures and the wall was defined as a non-slip boundary. Pressure across the model was set to atmospheric to match experimental conditions. Inlet was defined as a mass flow inlet with mass fractions defined from the respective WHSV. A k-ω turbulence model was applied with a term introduced in the diffusion model to account for turbulent viscosity. A SIMPLE scheme was used for pressure-velocity coupling with second-order upwind equations used for hydrodynamic terms and for the species scalars.9 The following under relaxation values were used: 0.7 for pressure, 0.3 for momentum and energy and 0.5 for scalars. Simulation proceeded for 4000 iterations or until convergence was reached whichever came first. The residuals were: 5 x 10-4 for all scalars, momentum and continuity, 1 x 10-6 for energy, and 1 x 10-3 for k-ω. Capabilities of the model were extended using user defined functions in which additional parameters were included to replicate the physiochemical phenomena more accurately. Dusty-gas diffusion model was used to describe diffusion through the catalyst bed as per Kyrimis et al. 10 Particles were assumed to be homogenous and spherical with an average diameter of 200 μm. Inertial and viscous resistances were calculated from the Ergun equation.11 Porosity across the bed was modelled using de Klerk’s correlation.12 Full description and derivation of associated conservation equations used for CFD modelling can be found in Kyrimis et al.13 Methanol dehydration over SAPO-11 was modelled using a modified Arrhenius equation which took into account limited active site number as shown in Equations S6-S8. Eq. S7 and S8 describe that if the local concentration of MeOH is greater than the local number of Si active sites, then the overall rate of the reaction will be limited by the number of Si sites. This is sensible as when all Si sites are occupied, the catalyst is fully utilised and any MeOH beyond that will be unreactive. On the contrary, if the local concentration of MeOH is lower than the Si sites, MeOH concentration will determine the magnitude of Zk. Not all Si will generate catalytically active Brønsted acid sites in SAPO-11 as some may form inactive Si islands. Using the correlation between Si loading and Si island formation developed by Grenev et al., 2 based on a 2.7 wt% Si loading for this SAPO-11 catalyst, only 61% of this Si will be catalytically active. This is then considered when calculating the number of active sites [Si]. A three-factorial DoE study employing a single centre point was employed to investigate the influence of reaction temperature (245-275°C), catalyst bed length (2.5-7.5 cm) and methanol WHSV (4-6 h-1) on simulated methanol conversion over SAPO-11. On top of the 8 vertex and 1 centre point cases, 6 additional cases inside the design space were investigated to give 15 unique points. The 15 individual cases were simulated using the CFD model described in section 5 and the results processed using the Sartorius MODDE® 13.1 Pro software suite.14 The results were fitted using a MLR model to give a model with an R2 of 0.88 and RSD of 7%. People involved with sample collection, processing, analysis and/or submission: Maciej G. Walerowski, Stylianos Kyrimis, Matthew Potter, Alice Oakley & Marina Caravetta. -------------------------- DATA-SPECIFIC INFORMATION -------------------------- Number of variables: Catalyst granule size (106-300, 300-500 or 500-710 um), catalyst bed length (1.9, 3.6 or 5.4 cm), methanol WHSV (0.5, 1, 2, or 3 h-1), time (1-6 hours), temperature (170, 185, 200, 215 or 230oC). Variable list, defining any abbreviations, units of measure, codes or symbols used: mL = mililitres, oC = degrees Celsius, SAPO-11 = silicoaluminophosphate-11 EDS = Energy Dispersive Spectroscopy, PXRD = Powder X-ray Diffraction, BET = Brunauer–Emmett–Teller, SEM = Scanning Electron Microscopy, ssNMR = solid state Nuclear Magnetic Resonance, NH3-TPD = Ammonia Temperature Programme Desorption CHN = Carbon Hydrogen Nitrogen, CFD = Computational Fluid Dynamics, DoE = Design of Experiments.