Roles of alkaline-earth metals and electron density of Al³⁺ in octanol dehydration for linear alpha-olefin production

Abstract

Linear alpha-olefins (LAOs), containing a terminal double bond at their carbon chain, are widely used as raw materials in the production of polyolefins, lubricants, alcohol-based detergents, and alpha-olefin sulfonates. One common method for synthesizing LAOs is alcohol dehydration, for example, conversion of 1-octanol to 1-octene. However, this reaction often produces double-bond isomers, increasing production costs due to the need for further purification. Therefore, catalysts that enhance alcohol dehydration while minimizing isomer formation are needed. This study proposes alkaline-earth metal (AEM)-impregnated Al₂O₃ catalysts to improve 1-octene purity. Our approach focuses on understanding how AEM impregnation alters the electron density of Al₂O₃ and, consequently, affects the catalytic activity. Among the catalysts tested, Ba-impregnated Al₂O₃ demonstrated the highest 1-octene purity, which is attributed to the increased electron density at Al³⁺ sites, facilitated by active electron transfer from the AEM to Al³⁺. Density functional theory calculations further reveal that this electron density increase reduces the energy of 1-octene re-adsorption, limiting its subsequent isomerization. These results highlight the relationship between the electronic modulation of the Al³⁺ active site and reaction mechanism, which can contribute to the development of more efficient catalysts in the petrochemical industry.

---

Introduction

Metal oxides are widely used as catalysts and catalyst supports. Their textural characteristics, such as crystal structure,^1–3 surface area,^4–6 and acid/base properties, affect their catalytic performance.^7–9 In particular, the acid/base properties of metal oxides like CaO, LaTi_1−xMg_xO_3+δ, and zeolites influence product selectivity in various reactions, including poly(ethylene terephthalate) pyrolysis,¹⁰ oxidative coupling of methane,¹¹ and n-octane hydroisomerization,¹² used in the production of raw chemicals. When used as a support, metal oxides can enhance catalytic activity by modulating their interactions with transition metals such as Ni,¹³ Co,¹⁴ and Cu.¹⁵ Coke deposition typically occurs at the acidic sites of metal oxides, compromising catalyst stability, particularly in C₁-gas reforming processes. Catalyst stability can be improved by increasing the basicity of metal oxide supports.^16–18

In alcohol dehydration, water is removed from alcohols to form olefins and ethers. Acidic metal oxides like zeolites,^19,20 TiO₂,²¹ WO_x,²² and Al₂O₃ have been investigated as dehydration catalysts. The acid/base properties of the catalyst directly influence performance, as acidic sites act as active sites for alcohol dehydration. Al₂O₃ is the most widely used due to its high activity, mild acidity, and high 1-olefin selectivity.^23–27 For light alcohols such as ethanol and propanol, the position of the double bond is generally unaltered. However, linear alpha-olefins (LAOs) and linear internal olefins (LIOs) with different positions of the double bond can be obtained during the dehydration of long-chain alcohols (>C4). LAOs, which have a double bond at the terminal position, serve as essential feedstocks in the production of polyolefins, lubricants, alcohol-based detergents, and alpha-olefin sulfonates.²⁸ LAOs can be synthesized via alcohol dehydration using Sasol's 1-heptene upgrading²⁹ and biomass upgrading.³⁰ However, these methods often generate LIOs with similar boiling points and molecular geometries, complicating separation and purification. Thus, suppressing LIO formation is essential for improving alcohol dehydration efficiency.^31–33

Dehydration of 1-octanol yields 1-octene, a valuable LAO used extensively in the petrochemical industry as a co-monomer of linear low-density polyethylene, combined with 1-butene and 1-hexene. Among the various 1-octene production methods, Al₂O₃-catalyzed 1-octanol dehydration has gained attention due to its high C8 selectivity and oxygenate upgrading.^30,34 The mechanism underlying alpha-olefin formation during alcohol dehydration remains not fully understood. Previous studies have proposed E1-type mechanisms involving the existence of alkoxy species as intermediates and E2-type elimination mechanisms consisting of the participation of β-H in the reaction.^35–38 For alumina catalysts, the E2 mechanism is generally favored over the E1 mechanism. In the E1-type elimination, the intermediate carbocation may undergo double-bond isomerization, leading to high isomer selectivity. Conversely, in the E2 mechanisms, β-H positioned adjacent to the hydroxyl group favors the formation of alpha-olefin during primary alcohol dehydration. Thus, Al₂O₃, which promotes E2 elimination, typically results in higher alpha-olefin selectivity. In this E2 pathway, the hydroxyl group adsorbs onto the acid site of Al³⁺ on Al₂O₃, and then, β-H is attracted to the O atom of the metal oxide (Scheme 1). Subsequently, the β-H and hydroxyl group form water, which is eliminated to yield the olefin product. This mechanism is especially favorable for producing LAOs when using weakly acidic catalysts like Al₂O₃.³⁹

Scheme 1 1-Octanol dehydration catalyzed by Al₂O₃via an E2-type mechanism.⁴⁰

However, the LAO formed can be re-adsorbed onto the acidic sites of the catalyst and rehydrated to form a secondary alcohol.²⁵ Further dehydration of this secondary alcohol results in a mixture of LAO and LIO products, such as 2-olefins. During 1-octanol dehydration, multiple LIOs such as cis-2-octene, trans-2-octene, trans-3-octene, and trans-4-octene are simultaneously formed due to the long carbon chain of 1-octanol. The prevalence of LIOs is further driven by the thermodynamic stability of internal double bonds. Various studies have sought to improve LAO selectivity using various metal oxide catalysts. Davis et al.⁴⁰ reported a relationship between the basicity of metal oxide catalysts such as Al₂O₃, ZrO, In₂O₃, SiO₂, MgO, and CaO, and the distribution of dehydration products, specifically, the formation of the Hofmann product (1-alkenes). Kawk et al.⁴¹ explored the regioselectivity of 2-butanol dehydration and showed that the stabilization energy of the transition state depends on van der Waals interactions between the adsorbed alcohol and the catalyst surface.

Several previous studies have employed density functional theory (DFT) calculations to investigate the mechanisms and kinetics of alcohol dehydration, aiming to establish a correlation between the acid/base properties and catalytic activity. Roy et al.⁴² reported an E2-type mechanism with a lower energy barrier (134.7 or 153.0 kJ mol⁻¹) than the corresponding E1-type elimination (290.3 kJ mol⁻¹). The Al³⁺ site of Al₂O₃ facilitates OH group removal, while surface O ions interact with the β-H. The higher acidity of Al₂O₃, compared to TiO₂ and ZrO₂, led to a lower overall energy barrier, as the adsorption energy of 1-alcohol is influenced by the acid–base properties of the catalyst.⁴³ Furthermore, double-bond isomer selectivity is affected by the surface properties of the catalyst. During alcohol dehydration over Al₂O₃, isomer formation occurs via a “dehydration–hydration–dehydration” pathway.²⁵ The iso-alcohol formed through 1-olefin hydration features its hydroxyl group within the carbon chain; thus, the location of the resulting double bond depends on the position of the β-H atom involved in the elimination. Dabbagh et al.^44,45 conducted a DFT study on 2-butanol dehydration and discovered that the double bond position of the olefin product is governed by the position of the β-H participating in the reaction. The activation was shown to depend on the distance between the β-H atom and the basic surface sites in the transition state.^46,47 Consequently, the anti-Saytzeff effect was observed, where the Hofmann product (LAO or the cis-isomer) was favored over the Saytzeff product (trans-isomer). Most existing studies have focused on short-chain alcohols, such as ethanol, isopropanol, and butanol. Although many DFT studies have been performed on the reaction mechanism and activation energies (E_as), the role of olefin–catalyst interactions in determining product selectivity remains unexplored. In particular, the impact of olefin adsorption sites on double-bond isomer formation has been investigated mainly in the context of n-paraffin/olefin separation^48–50 using zeolite-based catalysts with Brønsted acidic sites. However, the surface properties of the catalyst and the produced olefins have not been determined.

This study examines the effect of catalyst basicity on the activity and selectivity of 1-octanol dehydration by impregnating a γ-Al₂O₃ catalyst with various alkaline-earth metals (AEMs), including Mg, Ca, Sr, and Ba. We further computed changes in surface electron density of γ-Al₂O₃ following AEM incorporation and correlated the 1-octene adsorption energy with product selectivity using DFT calculations.

Experimental

Catalyst preparation

Commercial Al₂O₃ spheres (diameter = 1 mm, Sasol) were used to prepare the catalysts. Al₂O₃ was pre-calcined in air at 500 °C for 6 h. Elemental impurities in the Al₂O₃ included 0.020% Si, 0.015% Fe, 0.015% Ti, and 0.002% Na. AEM nitrates, M(NO₃)₂ (M = Mg, Ca, Sr, and Ba), with 99% purity, were obtained from Sigma-Aldrich. Incipient wetness impregnation was used to introduce 1 mol% AEM into the calcined Al₂O₃ surface, followed by a second calcination at 500 °C for 4 h in air.

Catalyst characterization

The metal content of the AEM-impregnated catalysts was determined by inductively coupled plasma mass spectrometry (ICP-MS; ELAN DRC II, Perkin-Elmer). Each sample was pretreated with a solution of 7 mL HNO₃ (70%) and 3 mL HCl (35%), and then heated at 200 °C for 30 min in a microwave system (Milestone, Ethos 1). Physical properties of the catalysts were analyzed using X-ray diffraction (XRD) and N₂ adsorption–desorption measurements. Detailed analysis conditions are described in the SI. Acid–base properties of the catalysts were evaluated by temperature-programmed desorption (TPD) of NH₃ and CO₂ using a BELCAT-B (BEL Japan Inc.). Each catalyst (50 mg) was loaded into a U-type quartz tube and preheated with 5 vol% O₂/He gas at 300 °C for 3 h to remove water and surface contaminants. Then, probe gas adsorption was performed at 50 °C for 30 min using 10 vol% probe gas balanced in He. The cell was then purged with He to eliminate weakly physisorbed species. The temperature of the cell was increased from 50 to 500 °C at a heating rate of 5 °C min⁻¹ under a He flow. A thermal conductivity detector was used to monitor the desorbed gases. Diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy was used to identify the acidic and basic sites of the prepared catalysts using pyridine (Py) and CO₂ as probe molecules, respectively. DRIFT spectra were recorded using a Nicolet iS50 FTIR spectrometer (Thermo Fisher Scientific) equipped with a mercury–cadmium–telluride detector and a Praying Mantis DRIFTS diffuse reflectance attachment (Harrick) with a ZnSe window. Spectra were recorded at 4 cm⁻¹ resolution and averaged over 128 scans. Each catalyst (50 mg) was loaded into the DRIFT cells and pretreated at 400 °C for 2 h with 5 vol% O₂/He to remove residual water and hydrocarbons. Py adsorption was conducted at 100 °C for 1 h using 1 vol% Py in He (total flow: 60 mL min⁻¹). An analogous experiment using 10 vol% CO₂ in He was performed to determine CO₂ adsorption. Finally, the samples were purged with He (60 mL min⁻¹) to desorb weakly bound probe gas molecules, and the spectra were recorded under He at 100 °C for 30 min. The number of acidic and basic sites was quantified based on the TPD profiles. Electronic states of the AEM-impregnated Al₂O₃ catalysts were measured by X-ray photoelectron spectroscopy (XPS) using a Sigma Probe (Thermo VG Scientific) equipped with an Al Kα X-ray source (200 W). The binding energies were calibrated using the C 1s peak of the C–C bond at 284.8 eV. Spectral fitting was performed using a Gaussian–Lorentzian peak shape with 30% Lorentzian contribution after subtraction of a smart-type background. The samples were investigated using analytical transmission electron microscopy (TEM, JEM-ARM200F NEOARM, JEOL Ltd) coupled with dual-type energy-dispersive X-ray spectroscopy (EDS, JED-2300T, JEOL Ltd) and TEM (Spectra Ultra, Thermo Fisher Scientific) coupled with electron energy-loss spectroscopy (EELS, GIF Continuum, Gatan Ltd).

Catalytic reaction test

A fixed-bed reactor system (Fig. S1) consisting of a quartz tube (diameter: 9.5 mm and length: 522 mm) was used to evaluate catalytic activity. The catalyst layer and total packed bed length measured 21 and 41 mm, respectively. The catalyst (0.6 g) was loaded at the center of an electric furnace, sandwiched between two layers of SiC (1 g, 1 mm, Goodfellow Corp.). The SiC layers were placed at the top and bottom of the catalyst bed to vaporize and evenly distribute the feed. A thermocouple was positioned at the center of the catalyst bed to control the reaction temperature. The reaction temperature was increased at a rate of 5 °C min⁻¹ under a N₂ flow. After the N₂ gas flow was turned off, a preheated feed (300 °C) was introduced using a tube pump (LEPP 150L, LAB SCiTECH). Three different feedstocks, 1-octanol (99%, Sigma-Aldrich), 1-octene (98%, Sigma-Aldrich), and 2-octanol (97%, Sigma-Aldrich), were used in each experiment. Catalytic activity was evaluated over 3 h under atmospheric pressure in the temperature range of 300 and 400 °C, with a liquid hourly space velocity of 7 h⁻¹ and a gas hourly space velocity of 996 h⁻¹. All feedstocks (1-octanol, 1-octene, and 2-octanol) were heated to 300 °C in a preheater packed with metal foam to enhance vaporization, and the reaction products were subsequently collected using a chiller maintained at 4 °C. The products were analyzed using a gas chromatography-flame ionization detector (Agilent 7980-B) equipped with a 30 m HP-5 column (Agilent 19091J-413). Catalytic activities—including 1-octanol conversion, 1-octene yield and purity, selectivity for 1-octene, selectivity for dioctyl ether (DOE), and cis/trans ratio—were calculated using the equations described in the SI with an error margin of <3.5%.⁵¹

Computational methods

Spin-polarized DFT calculations were performed using the Vienna ab initio simulation package (version 6.3.2).^52,53 The projector augmented-wave method was applied, and the plane-wave cutoff was set to 500 eV.^54,55 A total of 3, 6, 10, 10, 10, and 10 valence electrons were employed for Al, O, Mg, Ca, Sr, and Ba atoms, respectively. The revised Perdew–Burke–Ernzerhof functional was adopted as the exchange–correlation functional⁵⁶ within the generalized gradient approximation with D3 corrections. Structural optimization was performed using the conjugate gradient method.⁵⁷ Convergence criteria were set to 10⁻⁶ eV for the total electronic energy and 0.02 eV Å⁻¹ for the maximum force. The Brillouin zone was sampled using a Γ-centered k-point mesh: 6 × 4 × 4 for the bulk Al₂O₃ structure and 3 × 3 × 1 for the slab. A 20 Å vacuum distance along the z-axis was applied in all slab model calculations to eliminate image–image interactions under periodic boundary conditions. The adsorption energy of 1-octene on the catalyst surface, represented as E_1-octene*, is defined as follows:

ΔE_1-octene* = E_1-octene* − (E_* + E_1-octene)

where * denotes the catalyst surface, and E_*, E_1-octene*, and E_1-octene represent the energies for a bare catalyst surface, a 1-octene-adsorbed catalyst surface, and an isolated 1-octene molecule, respectively.

Charge density difference (Δρ) was computed using VASPKIT⁵⁸ using the following equation:

Δρ = ρ_1-octene* − (ρ_* + ρ_1-octene)

where ρ_1-octene*, ρ_*, and ρ_1-octene represent the total charge density of the adsorption structure, bare catalyst surface, and isolated 1-octene molecule after geometry optimization. Crystal Orbital Hamilton Population (COHP) analyses were conducted using the LOBSTER package.^59,60

Results and discussion

Acid–base properties of catalysts

To evaluate the microstructural changes induced by AEM impregnation on γ-Al₂O₃, we conducted scanning TEM-high-angle annular dark field (STEM-HAADF) analysis for all catalysts. STEM-HAADF images of AEM-impregnated catalysts (Mg/Al₂O₃, Ca/Al₂O₃, Sr/Al₂O₃, and Ba/Al₂O₃) showed consistent morphological features across the series (Fig. S2). High-resolution imaging confirmed that AEMs were present as nanoclusters and single atoms on the Al₂O₃ surface, with Ba/Al₂O₃ presented as a representative example (Fig. 1(a)). EDS mapping demonstrated uniform dispersion of AEM species across the Al₂O₃ surface for all samples; representative mapping data for Ba/Al₂O₃ mapping are shown in Fig. 1(b). XRD analysis (Fig. 1(c)) confirmed that all samples retained the characteristic γ-Al₂O₃ crystal structure following AEM impregnation, with no detectable phase transitions. These consistent diffraction patterns across all samples indicate preservation of the Al₂O₃ support structure. To confirm the chemical interaction between Ba and O, STEM-EELS analysis was performed on Ba/Al₂O₃ (Fig. S3). The EELS spectra of the Ba M₄₅ and O K edges reveal that Ba remains in the Ba–O structural state with the O atoms on the Al₂O₃ surface, confirming the formation of interfacial Ba–O bonds. The systematic analysis of catalyst properties (Table 1) reveals that all AEM-impregnated catalysts exhibited similar surface areas (160–164 m² g⁻¹) and mean pore sizes (10.8–11.0 nm) to pristine Al₂O₃, confirming that the observed catalytic changes are primarily driven by electronic modifications rather than structural alterations.

Fig. 1 (a) STEM-HAADF images of Ba/Al₂O₃, (b) elemental mapping of Ba, Al, and O of Ba/Al₂O₃, (c) XRD patterns of Al₂O₃, Mg/Al₂O₃, Ca/Al₂O₃, Sr/Al₂O₃, and Ba/Al₂O₃, (d) Py-DRIFT and (e) CO₂-DRIFT spectra, (f) Bader charge difference (Δ|q|) of Al in all models, and electronegativities of AEMs. Charge density differences of (g) Mg/Al₂O₃, (h) Ca/Al₂O₃, (i) Sr/Al₂O₃, and (j) Ba/Al₂O₃. The iso-surface level was set to 0.0025 e bohr⁻³. Al, O, Mg, Ca, Sr, and Ba are colored sky blue, red, orange, purple, green, and deep blue, respectively. Yellow and cyan in the iso-surfaces represent charge accumulation and depletion, respectively.

Table 1 Characteristics of AEM-impregnated Al₂O₃ catalysts

Catalyst	Metal contents		NH3 desorption^b (μmol g^−1)	CO2 desorption^b (μmol g^−1)	Surface area^c (m^2 g^−1)	Mean pore size^c (nm)
	(wt%)^a	(mol%)
a Confirmed by ICP-MS. b Calculated by integrating the results of TPD. c Estimated from N2 adsorption–desorption experiments performed at −196 °C.
Al2O3	—	—	249	18	162.24	10.98
Mg/Al2O3	0.23	0.96	232	19	162.70	11.02
Ca/Al2O3	0.46	1.17	221	25	160.79	11.03
Sr/Al2O3	0.87	1.01	206	28	162.80	10.85
Ba/Al2O3	1.36	1.01	195	33	163.94	10.82

---

To understand the electronic influence of AEM incorporation, we analyzed the surface chemistry of the catalysts using XPS. On the Al₂O₃ surface, Al³⁺ and lattice O ions serve as acidic and basic sites, respectively.^61–63 All XPS spectra were aligned to the C–C bond peak at 284.8 eV in the C 1s spectra (Fig. S4a). A single dominant Al 2p peak was observed at ∼74.3 eV for all samples, with a slight decrease in BE as the ionic radius of the AEM increased. This negative shift is attributed to the electron transfer from the incorporated AEM ions to Al centers, indicating a modified electronic environment and the Lewis basicity of the Al₂O₃ surface, which can significantly affect catalytic activity.

The electronic modifications observed via XPS were further correlated with acid–base properties, as revealed by probe molecule studies. The Py-DRIFT spectrum (Fig. 1(d) and Table S1) of Al₂O₃ exhibited peaks at 1621, 1612, and 1596 cm⁻¹. These peaks were classified according to the acid strength of the Lewis acid site (LAS), which was composed of several unsaturated Al³⁺ sites.⁶⁴ The 1621 cm⁻¹ peak arises from Py coordination to tetrahedral Al³⁺ in Al₂O₃, forming a strong LAS through interaction with the Py 8a orbital.⁶⁵ The peaks at 1612 and 1596 cm⁻¹ arising from Py and unsaturated Al³⁺ were also observed, representing medium and weak LASs, respectively.⁶⁶ Among these, the strong LAS showed the highest sensitivity to AEM impregnation.⁶⁷ Notably, the intensity of this peak corresponding to the strong LAS significantly decreased with increasing atomic radius of the AEM.

Complementing the acid site characterization, CO₂-DRIFT spectroscopy was performed to investigate the basic characteristics of the catalysts (Fig. 1(e)). The CO₂-DRIFT spectrum of Al₂O₃ displayed two peaks at 1661 and 1435 cm⁻¹, attributed to bidentate CO₂ groups.^65,68 Upon AEM impregnation, the intensity of the unidentate CO₂ characteristic peak (1586 cm⁻¹) increased with a direct correlation to the period of the AEM impregnation. When the CO₂ interacts with the strong basic O sites of the metal oxides, monodentate CO₂ species appear, which serve as an indicator of strong basic sites.⁶⁹ The increase in the basicities of the impregnated Al₂O₃ catalysts aligns well with previously reported basicities of corresponding metal oxides.^70,71

Quantitative validation of these qualitative trends observed in DRIFT spectroscopy was attained via TPD measurements using NH₃ and CO₂ (Table 1 and Fig. S6). The Al₂O₃ exhibited the highest acidity, with an NH₃ desorption value of 249 μmol g⁻¹, while AEM-impregnated catalysts showed progressively lower NH₃ desorption values as the atomic radius of the AEM increased: Mg/Al₂O₃ (232 μmol g⁻¹) > Ca/Al₂O₃ (221 μmol g⁻¹) > Sr/Al₂O₃ (206 μmol g⁻¹) > Ba/Al₂O₃ (195 μmol g⁻¹), representing a 22% reduction from pure Al₂O₃ to Ba/Al₂O₃. Conversely, CO₂ desorption increased across the series as follows: Al₂O₃ (18 μmol g⁻¹) < Mg/Al₂O₃ (19 μmol g⁻¹) < Ca/Al₂O₃ (25 μmol g⁻¹) < Sr/Al₂O₃ (28 μmol g⁻¹) < Ba/Al₂O₃ (33 μmol g⁻¹), with Ba/Al₂O₃ showing an 83% increase compared to pure Al₂O₃. This inverse relationship between acidity and basicity supports an electron transfer mechanism. According to the hard and soft acid–base theory, the electron density increases with the atomic radius of the AEM,⁶⁸ facilitating electron transfer from the AEM to Al ions and thus enhancing the basicity of the catalyst surface while diminishing its acidity.

These trends at the atomic level are further elucidated by DFT calculations. The γ-Al₂O₃ structure proposed by Digne et al. was adopted to model the AEM/Al₂O₃ surfaces. Each AEM (Mg, Ca, Sr, and Ba) was individually introduced onto the γ-Al₂O₃ (110) surface, since the (110) surface is the experimentally dominant and most stable facet.^64,72–74 To focus on the electronic structure changes when AEMs are impregnated and present on the Al₂O₃ surface in the form of AEM oxides, we introduced a single AEM atom on top of the Al₂O₃ surface. Five potential surface sites for AEM incorporation were considered, classifying Al atoms into tri- and tetra-coordinated species (Al₃c and Al₄c) and O into two top layers (O₃c and O₂c) and one subsurface layer (O₃c_sub) (Fig. S7(a)). Among these, the O₃c_sub site was thermodynamically favorable for all AEMs, as indicated by formation energy calculations (Fig. S7(b)). Geometry optimization of the AEM-doped surfaces revealed that AEMs pushed and penetrated the lattice Al ions of Al₂O₃, forming the AEM/oxide configuration (Fig. S8). Bader charge analysis confirmed that the number of electrons in the surrounding Al atoms increased with the atomic radius of the AEM (Table S2). This trend was attributed to the lower electronegativities of heavier AEMs, which readily lose electrons compared to Al with a Pauling electronegativity of 1.61, thereby donating charge to surrounding Al atoms after lattice penetration (Fig. 1(f)). The visualized charge density difference (Fig. 1(g–j)) for Mg/Al₂O₃, Ca/Al₂O₃, Sr/Al₂O₃, and Ba/Al₂O₃ clearly illustrates this electron donation. Furthermore, the average Al 2p core-level shifts (CLS) were calculated (Fig. S9), and the average CLS systematically increases in magnitude from Mg to Ca, Sr, and Ba, tracking the degree of charge transfer to adjacent Al sites. The average CLS values exhibit excellent quantitative agreement with the experimental XPS peak positions (R² = 0.937). As the atomic radius of AEMs increases from Mg to Ba, it is easier for the AEM to lose electrons and donate them to the surrounding Al. This confirms that the AEM acts as an electron donor, increasing the Lewis basicity of the electron-rich AEM/Al₂O₃ surface as compared to that of the bare Al₂O₃ surface.

1-Octanol dehydration

To validate the influence of the systematically tuned acid–base properties described in Fig. 1, we evaluated the catalytic performance of AEM-impregnated Al₂O₃ catalysts in 1-octanol dehydration (Fig. 2, Tables S3 and S4). 1-Octanol conversion depended primarily on the acidity of the catalyst, particularly the presence of strong LASs, in agreement with previous studies.^51,67 However, the purity of 1-octene, represented by the 1-octene : octene ratio, was affected by the basicity of the catalyst. Catalysts with lower acidities (Al₂O₃ > Mg/Al₂O₃ > Ca/Al₂O₃ > Sr/Al₂O₃ > Ba/Al₂O₃) exhibited lower 1-octanol conversion. DOE, formed by the reaction of two 1-octanol molecules, was obtained with high selectivity because of the high fraction of unreacted 1-octanol in the reactor, owing to the lower alcohol conversion.²⁵ Ethers (including diethyl ether) analogous to DOE decompose into alcohols and olefins at 300–400 °C, similar to alcohol dehydration.

According to Phung et al.,⁷⁵ ethers adsorb onto acidic sites of the catalyst, like alcohols, and the conversion increases with the density of acidic sites. Consequently, AEM-impregnated catalysts with low acidity showed high DOE and low 1-octene selectivity at 300 °C. Nevertheless, at 400 °C, all catalysts exhibited >2% DOE selectivity and <90% 1-octanol conversion. 1-Octene purity, which affects downstream separation, decreased with increasing 1-octanol conversion. Notably, the highly acidic Al₂O₃ catalyst produced a lower 1-octene purity (80.9%) than the Mg/Al₂O₃ catalyst (88.2%) at a similar conversion rate at 300 °C. High catalyst acidity influenced the formation of the double-bond isomer of 1-octene and 1-octanol conversion, leading to lower 1-octene purity. Comparison of the cis/trans ratio, which indicates the ratio of cis- to trans-isomers among the double-bond isomer products, showed that catalysts with AEM additives exhibited higher cis/trans ratios than bare Al₂O₃. Specifically, catalysts impregnated with highly basic AEMs, such as Ba/Al₂O₃, showed negligible trans-isomer formation even at 350 °C. These findings suggest that AEM addition enhances the anti-Saytzeff effect, thereby suppressing trans-isomer production. Fig. 3 illustrates the product distributions across various catalysts at different reaction temperatures. Al₂O₃ showed the highest isomer fraction (∼20%) at 300 °C and an even greater fraction (67.8%) at 400 °C, which minimized the difference between 1-octanol conversions of AEM-impregnated catalysts. Isomer formation proceeded via the re-adsorption of 1-octene by Al³⁺, rather than olefin isomerization. Al₂O₃, mainly composed of LASs, produced double-bond isomers via a “dehydration–hydration–dehydration” sequence involving 1-olefin re-adsorption.^25,36 Akizuki et al.⁷⁶ investigated the relationships between catalyst acid type and selectivity in the reactions of 1-octene and 2-octanol and revealed that cis/trans ratios were affected by the catalyst properties. The E_a of 1-octene hydration at LASs (263 ± 17 kJ mol⁻¹) was lower than that of 1-octene isomerization (304 ± 38 kJ mol⁻¹).

Fig. 2 Effect of AEM impregnation on (a) 1-octanol conversion, (b) 1-octene selectivity, (c) 1-octene yield, (d) 1-octene purity, (e) DOE selectivity, and (f) cis/trans isomer ratio of Al₂O₃ in 1-octanol dehydration. The errors are listed in Table S3.

Fig. 3 Product distribution during 1-octanol dehydration over (a) Al₂O₃, (b) Mg/Al₂O₃, (c) Ca/Al₂O₃, (d) Sr/Al₂O₃, and (e) Ba/Al₂O₃ at various reaction temperatures.

The mechanism of 1-octanol dehydration proceeds through the following steps (Scheme 1). First, 1-octanol is adsorbed onto the Al³⁺ LASs of the catalyst. This adsorption is primarily governed by strong LASs. Then, the β-H on the terminal carbon is adsorbed on the O ions of the catalyst, enabling a concerted β-H elimination mechanism.³⁶ During this step, the β-H atom combines with the hydroxyl group of 1-alcohol, forming water and an olefin product. However, the generated 1-olefin is re-adsorbed onto the Al³⁺ sites. Upon re-adsorption, activation of the π bond of the 1-olefin allows the hydroxyl group to recombine with the C chain.⁷⁷ According to Markovnikov's rule,^78,79 hydration preferentially produces secondary or tertiary alcohol products, primarily producing 2-octanol.

Subsequently, 2-octanol undergoes a dehydration step similar to that of 1-octanol via adsorption onto the acidic sites of the catalyst, forming water and an olefin product. As the hydroxyl group is located on the second carbon, elimination involving β-H atoms on adjacent carbons leads to the formation of either 1-octene (via removal of the H on C1) or 2-octene (via removal of the H on C3).³⁶ At elevated temperatures, the proportion of internal isomers increases due to higher thermodynamic stability relative to 1-octene. Nevertheless, isomer formation was reduced when AEM-impregnated catalysts were used, even at high temperatures. This reduction can be attributed to the weakened acidity of Al³⁺ sites, which reduces the re-adsorption of 1-octene.

1-Octene re-adsorption and mechanistic insights

To understand the reduced isomer formation observed during 1-octanol dehydration, the catalytic activities of the AEM-impregnated Al₂O₃ catalysts were examined using 1-octene as the feedstock. This allowed examination of individual steps in the proposed dehydration–hydration–dehydration mechanism (Fig. 4 and Table S5). The majority of 1-octene was converted into olefin isomers without DOE formation (<0.1 mol%). However, product distributions varied with reaction temperature. Conversion of 1-octene was mainly governed by catalyst acidity (Fig. 4(d)) because the catalyst acidic sites served as active sites for alcohol dehydration and also played roles in 1-octene re-adsorption. The Lewis acidity of the Al₂O₃ surface was primarily influenced by the electronic state of Al³⁺, as shown in Fig. 1. As the electron charge of the AEM increased, electron density on Al³⁺ sites also increased, weakening their Lewis acidity and reducing their affinity for 1-octene adsorption. A critical mechanistic insight emerged from the analysis of the cis/trans isomer ratios, which provided definitive evidence for the proposed dehydration–hydration–dehydration pathway. Al₂O₃ catalysts exhibited trans-isomer ratios of 73.6–84.5% (Fig. 4(e–g)), and the trans-isomer ratio decreased with increasing catalyst basicity. Highly basic Ba/Al₂O₃ showed trans-isomer ratios of 11.9–21.3%.

Conversely, cis-isomer ratios ranged between 10.7 and 20.7% across all catalysts, and small variations were observed between the cis-isomer ratios of the catalysts when compared with the case of the trans-isomer ratios. This was attributed to the mechanism of isomerization. Unlike 1-octanol dehydration, 1-octene isomerization mainly occurs in the absence of water. Since the trans-isomer exhibits a more linear and geometrically stable structure than the cis-isomer, 1-octene preferentially isomerizes to the trans-isomer, with a cis/trans ratio of <1 for all catalysts. However, our experimental results for 1-octanol dehydration revealed cis/trans ratios >1 under all conditions except for Al₂O₃ at 400 °C. This fundamental difference confirms that the isomers were generated via “dehydration–hydration–dehydration” instead of direct 1-octene isomerization.

Fig. 4 Catalytic performance of AEM-impregnated Al₂O₃ in 1-octene re-adsorption: (a) 1-octanol conversion, (b) 1-octene purity, (c) cis/trans ratio, and (d) correlation between 1-octene conversion and acid density evaluated at (e) 300 °C, (f) 350 °C, and (g) 400 °C.

The experimental results demonstrated that AEM impregnation significantly reduced 1-octene re-adsorption, thereby improving selectivity (Fig. 4). To explore the relationship between the reduced 1-octene re-adsorption and the increasing atomic radius of AEMs on the Al₂O₃ surface, DFT calculations were conducted on 1-octene adsorption using various AEM-impregnated surface models. A comprehensive COHP analysis was performed to investigate the interaction between surface Al atoms and the C atoms of 1-octene, facilitating a comparative assessment of 1-octene adsorption on diverse AEM-modified models (Fig. 5(a)). In each projected COHP (−pCOHP) plot, the upper and lower parts represent antibonding and bonding states, respectively. As the ionic radius of the AEM increased, the high-energy window considerably shifted toward the Fermi level. This shift suggests a weakened interaction between 1-octene and the Al₂O₃ surface, attributable to the reduced positive charge of the Al₂O₃ surface, and is consistent with prior research highlighting enhanced olefin adsorption on surfaces with pronounced positive charges.^80,81 Al₂O₃ surfaces impregnated with Ca, Sr, and Ba exhibited significant antibonding states below the Fermi level, suggesting weaker 1-octene adsorption, which is contrary to pristine Al₂O₃. The presence of occupied antibonding states below the Fermi level hinders adsorption by lowering the adsorption energy, thereby promoting desorption of 1-octene. Thus, the upward shift of bonding states and occupancy of antibonding states in AEM-impregnated Al₂O₃ surfaces led to weaker adsorptions. In agreement with the COHP analysis, the calculated adsorption energies of 1-octene decreased with increasing AEM radius. The adsorption energy of 1-octene on bare Al₂O₃ was −1.22 eV, whereas that on Ba-impregnated Al₂O₃ was −0.68 eV. This weaker binding on Ba-impregnated Al₂O₃ facilitates the easier release of 1-octene compared to the pristine Al₂O₃ surface. Additionally, the lower adsorption energy correlated with an elongated Al–C bond, while the C1–C2 bond in 1-octene remained at 1.34 Å, consistent with typical C=C bond lengths (Table S5). A correlation was observed between the integrated pCOHP (IpCOHP) and 1-octene adsorption energy across various AEM-impregnated surfaces (Fig. 5(c)). Since higher IpCOHP indicates stronger binding of 1-octene to the catalyst surface, the reduced bond strength in AEM-impregnated Al₂O₃ further supports the change in 1-octene conversion. Therefore, in the absence of AEMs, Al₂O₃ should undergo 1-octene re-adsorption, resulting in high 1-octene conversion. Conversely, Ba-impregnated Al₂O₃, with the lowest adsorption energy (−0.68 eV), exhibited the weakest re-adsorption tendency. In summary, modulating the electronic properties of Al sites via AEM impregnation reduces 1-octene re-adsorption, thereby enhancing selectivity by directly altering its conversion dynamics.

Fig. 5 (a) −pCOHP between Al in Al₂O₃ and C in 1-octene for Al₂O₃ and AEM-impregnated Al₂O₃, (b) 1-octene adsorption energy, and (c) correlation between the adsorption energy of 1-octene and IpCOHPs of all models.

2-Octanol dehydration

The experimental evidence suggested a re-adsorption mechanism, but the complete reaction pathway required further validation. To confirm that re-adsorbed 1-octene indeed proceeds through 2-octanol formation, direct 2-octanol dehydration was investigated (Fig. 6 and Table S6). Unlike with 1-octanol, all catalysts showed high 2-octanol conversion even at low reaction temperatures (300–350 °C). However, 1-octene purity was significantly lower. At 400 °C, Al₂O₃ yielded only 8.4% 1-octene from 2-octanol, compared to 29.2% when starting from 1-octanol. According to the Saytzeff rule, elimination reactions favor more substituted alkenes due to their higher thermodynamic stability. In alcohol dehydration, the β-H predominantly occurs at the more substituted carbon (typically the third carbon), enhancing selectivity for internal (trans) isomers. However, the highly basic Ba/Al₂O₃ catalyst produced the highest 1-octene purity (30.6%), followed by Sr/Al₂O₃ ≒ Ca/Al₂O₃ > Mg/Al₂O₃ > Al₂O₃, which is consistent with the 1-octanol dehydration results. Product distributions of the reactions catalyzed by Al₂O₃, Ba/Al₂O₃, and other catalysts varied depending on the catalytic properties (Fig. 6(d) and (e), others in Fig. S10). Alkene selectivity in alcohol dehydration is affected by the stereoselectivity of isomers.^40,41 1-Octene re-adsorption, a vital step in isomerization, is determined by surface basicity and change in the electron density. Dabbagh et al.⁴⁵ reported that 2-butanol dehydration over γ-Al₂O₃ proceeds via surface adsorption of 2-butanol on the Al₂O₃ catalyst surface, where a repulsive interaction existed between the methyl group of 2-butanol and the Al₂O₃ surface. The adsorbed cis-isomer was more stable because it was relatively less influenced by these repulsive interactions. This stereochemical preference in 1-octene selectivity influenced the selection of O sites for β-H adsorption. The E_a for cis-2-butene formation (84.32 kcal mol⁻¹) was lower than that of trans-2-butene (86.64 kcal mol⁻¹). These results validate the proposed re-adsorption pathway and confirm that re-adsorbed 1-octene tends to form internal isomers, thereby reducing 1-octene purity, which is undesirable. The consistent AEM effects observed in 1-octanol and 2-octanol dehydration reactions further support a unified mechanism where AEM-induced re-adsorption suppression governs selectivity enhancement.

Fig. 6 Catalytic performance of AEM-impregnated Al₂O₃ in 2-octanol dehydration: (a) 1-octene selectivity, (b) 1-octene purity, (c) cis/trans ratio, (d) product distribution for Al₂O₃, and (e) product distribution for Ba/Al₂O₃.

Conclusions

This study systematically investigated the dehydration of 1-octanol over AEM-modified Al₂O₃ catalysts to produce high-purity 1-octene. Through integrated experimental and theoretical approaches, the underlying cause of low 1-octene purity was elucidated, and a pathway to overcome this limitation via electronic structure modification was demonstrated.

AEM incorporation altered the electronic properties of Al₂O₃ surfaces. Py-FTIR spectra showed a decrease in the peak intensity of LASs with the increasing AEM ionic radius (for Mg, Ca, Sr, and Ba), while CO₂-FTIR spectra revealed a corresponding increase in strongly basic sites, exhibiting characteristic unidentate CO₂ adsorption peaks. These acid/base property changes were attributed to the electron density redistribution at Al³⁺ sites. DFT calculations confirmed electron transfer from AEMs to Al³⁺ upon AEM incorporation, with the extent of electron transfer increasing with the AEM ionic radius, as confirmed by XPS analysis. The primary cause of low 1-octene purity was identified as a “dehydration–hydration–dehydration” mechanism. This mechanism was experimentally validated using 1-octene as a probe molecule and direct 2-octanol dehydration studies, revealing distinct cis/trans selectivity patterns. The electronic structure modifications directly translated to improved catalytic performance. While pure Al₂O₃, despite its high acidity, produced low-purity 1-octene due to extensive isomer formation, AEM addition enhanced catalyst basicity and 1-octene purity. AEMs with larger ionic radii reduced the positive charge density of Al³⁺ sites more effectively, leading to lower 1-octene adsorption energy and minimizing isomer formation. Ba/Al₂O₃ achieved the optimal combination of electronic properties, demonstrating the highest 1-octene purity through effective re-adsorption inhibition.

Overall, this study provides valuable insights into the development of efficient catalysts for the production of high-purity 1-octene, a valuable material used in the petrochemical industry. However, it only examined the effects of individual model compounds on the formation of high-purity 1-octene. Hence, further studies are required to evaluate the influences of co-existing water generated in the reaction on the production of high-purity 1-octene.

Author contributions

Young-eun Kim: conceptualization, investigation, writing – original draft. Jaekyum Kim: DFT calculations, visualization, writing – original draft. Unho Jung: methodology, data curation, validation. Yongha Park: data curation, formal analysis. Min Hye Youn: funding acquisition, investigation. Dong Hyun Chun: project administration, validation. Byeong-Seon An: formal analysis, investigation. Jung Kyu Kim: DFT calculations, validation. Byung-Hyun Kim: DFT calculations, writing – review & editing, supervision. Ki Bong Lee: methodology, supervision. Kee Young Koo: conceptualization, writing – review & editing, supervision.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: additional characterization results (STEM-HAADF, EELS, XPS, BET, and TPD), DFT-calculated data, and raw experimental data from the catalytic reaction tests. See DOI: https://doi.org/10.1039/d5ta06534h.

Acknowledgements

This work was supported by the Research and Development Program of the Korea Institute of Energy Research (KIER) (No. C5-2436) and the Next-generation CCU technology advancement project through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (RS-2025-02219162).

References

1. S. Ishikawa, Y. Yamada, N. Kashio, N. Noda, K. Shimoda, M. Hayashi, T. Murayama and W. Ueda, ACS Catal., 2021, 11, 10294–10307.
2. Y. Zhou, L. Liu, G. Li and C. Hu, ACS Catal., 2021, 11, 7099–7113.
3. T. Arumuganathan, A. Srinivasarao, T. V. Kumar and S. K. Das, J. Chem. Sci., 2008, 120, 95–103.
4. S. Özkar, Appl. Surf. Sci., 2009, 256, 1272–1277.
5. D. Chan, S. Tischer, J. Heck, C. Diehm and O. Deutschmann, Appl. Catal., B, 2014, 156–157, 153–165.
6. X. Zhang, F. Hou, H. Li, Y. Yang, Y. Wang, N. Liu and Y. Yang, Microporous Mesoporous Mater., 2018, 259, 211–219.
7. M. A. Cortés-Jácome, C. Angeles-Chavez, E. López-Salinas, J. Navarrete, P. Toribio and J. A. Toledo, Appl. Catal., A, 2007, 318, 178–189.
8. C. K. S. Choong, L. Huang, Z. Zhong, J. Lin, L. Hong and L. Chen, Appl. Catal., A, 2011, 407, 155–162.
9. M. Kitano, E. Wada, K. Nakajima, S. Hayashi, S. Miyazaki, H. Kobayashi and M. Hara, Chem. Mater., 2013, 25, 385–393.
10. S. Kumagai, R. Yamasaki, T. Kameda, Y. Saito, A. W. atanabe, C. Watanabe, N. Teramae and T. Yoshioka, Chem. Eng. J., 2018, 332, 169–173.
11. L. B. Lopes, L. H. Vieira, J. M. Assaf and E. M. Assaf, Catal. Sci. Technol., 2021, 11, 283–296.
12. W. Zhang and P. G. Smirniotis, J. Catal., 1999, 182, 400–416.
13. L. Fu, Z. Liu, X. Li, Y. Li, H. Yang and Y. Liu, Fuel, 2022, 322.
14. C. Hernandez Mejia, T. W. van Deelen and K. P. de Jong, Nat. Commun., 2018, 9, 4459.
15. X. Yang, Z. Zhang, W. Liu, T. Liang, D. Dang and X. Tian, Chem. Cat. Chem, 2021, 13, 3846–3856.
16. A. Chatla, F. Abu-Rub, A. V. Prakash, G. Ibrahim and N. O. Elbashir, Fuel, 2022, 308.
17. X. Gao, Z. Wang, Q. Huang, M. Jiang, S. Askari, N. Dewangan and S. Kawi, Catal. Today, 2022, 402, 88–103.
18. H. U. Hambali, A. A. Jalil, A. A. Abdulrasheed, T. J. Siang, T. A. T. Abdullah, A. Ahmad and D. V. N. Vo, Int. J. Energy Res., 2020, 44, 5696–5712.
19. H. Hasanudin, W. R. Asri, L. Andini, F. Riyanti, A. Mara, F. Hadiah and Z. Fanani, ACS Omega, 2022, 7, 38923–38932.
20. H. Choi, D. H. Kim, Y. K. Park and J. K. Jeon, Bull. Korean Chem. Soc., 2015, 36, 1974–1979.
21. J. Ballesteros-Soberanas, L. D. Ellis and J. W. Medlin, ACS Catal., 2019, 9, 7808–7816.
22. J. Macht, C. D. Baertsch, M. May-Lozano, S. L. Soled, Y. Wang and E. Iglesia, J. Catal., 2004, 227, 479–491.
23. R. Wang, Z. Zhao, P. Gao, K. Chen, Z. Gan, Q. Fu and G. Hou, J. Phys. Chem. C, 2022, 126, 10073–10080.
24. J. H. Kwak, D. Mei, C. H. F. Peden, R. Rousseau and J. Szanyi, Catal. Lett., 2011, 141, 649–655.
25. R. J. Nel and A. de Klerk, Ind. Eng. Chem. Res., 2009, 48, 5230–5238.
26. C. R. Narayana, S. Srinivasan, A. K. Datye, R. Gorte and A. Biaglow, J. Catal., 1992, 138, 659–674.
27. B. H. Davis, G. B. Freeman, J. C. Watters and D. J. Collins, J. Phys. Chem. C, 1980, 84, 55–56.
28. F. S. Golub', V. A. Bolotov and V. N. Parmon, Catal. Ind., 2021, 13, 203–215.
29. L. M. Farrell, Linear Alpha Olefins, Nexant, 2019.
30. J. Rorrer, S. Pindi, F. D. Toste and A. T. Bell, ChemSusChem, 2018, 11, 3104–3111.
31. C. Casas, C. Fité, M. Iborra, J. Tejero and F. Cunill, J. Chem. Eng. Data, 2013, 58, 741–748.
32. R. Bringué, J. Tejero, M. Iborra, C. Fité, J. F. Izquierdo and F. Cunill, J. Chem. Eng. Data, 2008, 53, 2854–2860.
33. M. Pérez-Maciá, R. Bringué, M. Iborra, J. Tejero and F. Cunill, Chem. Eng. Res. Des., 2015, 102, 186–195.
34. A. de Klerk, Top. Catal., 2013, 57, 715–722.
35. R. J. Nel and A. de Klerk, Ind. Eng. Chem. Res., 2007, 46, 3558–3565.
36. B. C. Shi and B. H. Davis, J. Catal., 1995, 157, 359–367.
37. J. Lee, E. J. Jang and J. H. Kwak, J. Catal., 2017, 345, 135–148.
38. T. K. Phung, A. Lagazzo, M. Á. Rivero Crespo, V. Sánchez Escribano and G. Busca, J. Catal., 2014, 311, 102–113.
39. P. Berteau, S. Ceckiewicz and B. Delmon, Appl. Catal., 1987, 31, 361–383.
40. B. H. Davis, Stud. Surf. Sci. Catal., 1985, 21, 309–318.
41. J. H. Kwak, R. Rousseau, D. Mei, C. H. F. Peden and J. Szanyi, Chem. Cat. Chem, 2011, 3, 1557–1561.
42. S. Roy, G. Mpourmpakis, D.-Y. Hong, D. G. Vlachos, A. Bhan and R. J. Gorte, ACS Catal., 2012, 2, 1846–1853.
43. P. Kostestkyy, J. Yu, R. J. Gorte and G. Mpourmpakis, Catal. Sci. Technol., 2014, 4, 3861–3869.
44. H. A. Dabbagh, K. Taban and M. Zamani, J. Mol. Catal. A: Chem., 2010, 326, 55–68.
45. H. A. Dabbagh, M. Zamani and B. H. Davis, J. Mol. Catal. A: Chem., 2010, 333, 54–68.
46. H. Knozinger, H. Buhl and K. Kochloefl, J. Catal., 1972, 24, 57–68.
47. K. Larmier, C. Chizallet, N. Cadran, S. Maury, J. Abboud, A.-F. Lamic-Humblot, E. Marceau and H. Lauron-Pernot, ACS Catal., 2015, 5, 4423–4437.
48. A. Bhan, Y. V. Joshi, W. N. Delgass and K. T. Thomson, J. Phys. Chem. B, 2003, 107, 10476–10487.
49. D.-e. Jiang, B. G. Sumpter and S. Dai, Langmuir, 2006, 22, 5716–5722.
50. R. Yang, X. Chen, J. Ma, Y. Gao, Y. Wang and G. Luo, Nano Res., 2022, 15, 5322–5330.
51. Y.-e. Kim, H. B. Im, U. H. Jung, J. C. Park, M. H. Youn, H.-D. Jeong, D.-W. Lee, G. B. Rhim, D. H. Chun and K. B. Lee, Fuel, 2019, 256, 115957.
52. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
53. G. Kresse and J. Furthmüller, Phys. Rev. B:Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
54. P. E. Blöchl, Phys. Rev. B:Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
55. G. Kresse and D. Joubert, Phys. Rev. B:Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
56. J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou and K. Burke, Phys. Rev. Lett., 2008, 100, 136406.
57. M. P. Teter, M. C. Payne and D. C. Allan, Phys. Rev. B:Condens. Matter Mater. Phys., 1989, 40, 12255–12263.
58. V. Wang, N. Xu, J.-C. Liu, G. Tang and W.-T. Geng, Comput. Phys. Commun., 2021, 267, 108033.
59. V. L. Deringer, A. L. Tchougréeff and R. Dronskowski, J. Phys. Chem. A, 2011, 115, 5461–5466.
60. R. Nelson, C. Ertural, J. George, V. L. Deringer, G. Hautier and R. Dronskowski, J. Comput. Chem., 2020, 41, 1931–1940.
61. M. Chen, W. Lu, H. Zhu, L. Gong, Z. Zhao and Y. Ding, Ind. Eng. Chem. Res., 2020, 59, 4388–4396.
62. G. Busca, Chem. Rev., 2010, 110, 2217–2249.
63. G. Busca, Catal. Today, 2014, 226, 2–13.
64. M. Digne, P. Sautet, P. Raybaud, P. Euzen and H. Toulhoat, J. Catal., 2004, 226, 54–68.
65. C. Morterra and G. Magnacca, Catal. Today, 1996, 27, 497–532.
66. J. H. Kwak, J. Hu, A. Lukaski, D. H. Kim, J. Szanyi and C. H. Peden, J. Phys. Chem. C, 2008, 112, 9486–9492.
67. Y.-E. Kim, U. Jung, D. Song, H. B. Im, J. C. Park, M. Hye Youn, H.-D. Jeong, G. B. Rhim, D. H. Chun, D.-W. Lee, K. B. Lee and K. Y. Koo, Fuel, 2020, 281, 118791.
68. D. Cornu, H. Guesmi, J.-M. Krafft and H. Lauron-Pernot, J. Phys. Chem. C, 2012, 116, 6645–6654.
69. F. Prinetto, G. Ghiotti, R. Durand and D. Tichit, J. Phys. Chem. B, 2000, 104, 11117–11126.
70. J. A. Duffy, Geochim. Cosmochim. Acta, 1993, 57, 3961–3970.
71. R. C. Deka, R. K. Roy and K. Hirao, Chem. Phys. Lett., 2000, 332, 576–582.
72. M. Digne, P. Sautet, P. Raybaud, P. Euzen and H. Toulhoat, J. Catal., 2002, 211, 1–5.
73. G. Xu, J. Ma, G. He, Y. Yu and H. He, Appl. Catal., B, 2017, 207, 60–71.
74. G. Xu, H. Wang, Y. Yu and H. He, J. Catal., 2021, 395, 1–9.
75. T. K. Phung and G. Busca, Chem. Eng. J., 2015, 272, 92–101.
76. M. Akizuki and Y. Oshima, J. Supercrit. Fluids, 2017, 123, 76–81.
77. S. Saba, D. D. Clarke, C. Iwanoski and T. Lobasso, J. Chem. Educ., 2010, 87, 1238–1241.
78. D. Kalló and R. M. Mihályi, Appl. Catal., A, 1995, 121, 45–56.
79. C. M. Jensen and W. C. Trogler, Science, 1986, 233, 1069–1071.
80. J. H. Lee, H.-K. Lee, K. Kim, G. B. Rhim, M. H. Youn, H. Jeong, J. H. Park, D. H. Chun, B.-H. Kim and J. C. Park, Mater. Adv., 2021, 2, 1050–1058.
81. R. Faiz and K. Li, Chem. Eng. Sci., 2012, 73, 261–284.

---

Footnote

† These authors equally contributed to this manuscript.

---