Tailoring amorphous alumina catalysts with enriched five-coordinated aluminum via pH control

Abstract

Amorphous alumina is known to contain a high fraction of five-coordinated aluminum species (Al(V)), which are recognized as active sites and serve as anchoring sites for noble metals. In this study, we aimed to synthesize highly active amorphous alumina enriched in Al(V) by controlling the pH during the gelation of aluminum hydroxide using buffers with different concentrations (0.1, 1, 10, and 100 mM). After dehydration, the resulting alumina was characterized and evaluated for VOC decomposition. While the Al(V) fraction increased with buffer concentration (42–53% at 650 °C), no direct correlation was observed between activity and the parameter Al(V) ratio in terms of specific surface area. Instead, deconvolution of the ²⁷Al NMR spectra revealed that Al(V) consists of sharp (trigonal bipyramidal) and broad (square pyramidal) components, the latter exhibiting a strong correlation with VOC decomposition when specific surface area was taken into consideration (R² = 0.77). Experimental results and insights from previous research suggest that high pH suppresses hydroxide precipitation, yielding small amorphous particles prone to aggregation, whereas low pH promotes nitrate incorporation into the lattice, leading to structural disorder and site blocking on pyramidal Al(V). Among the tested conditions, the 1 mM buffer produced amorphous alumina with an optimal balance of surface area and Al(V) broad content, achieving for the first time 100% VOC decomposition using pure amorphous alumina. These findings highlight pH control as a promising strategy for tailoring active Al(V) sites in alumina catalysts.

---

Introduction

Amorphous alumina (a-Al₂O₃) has attracted sustained attention due to its widespread usage as a dielectric-insulator material,^2,3 corrosion protective coating,^4,5 and catalytic support.^6–9 Unlike crystalline alumina polymorphs, which are dominated by tetrahedral (Al(IV)) and octahedral (Al(VI)) coordination environments such as α-Al₂O₃, the amorphous phase is structurally distinguished by the presence of a significant fraction of five-coordinated aluminum (Al(V)). Solid-state MAS NMR is the most appropriate analysis technique to evaluate coordination in non-crystalline or nano-crystalline alumina.^10–13 Gleizes and his coworkers summarized the synthesis methods for massive, thin films, and nanoparticles, showing that the fractions of Al(IV), Al(V), and Al(VI) vary depending on the synthesis method.¹⁵ NMR studies have consistently revealed that Al(V) species account for up to 30–40% of the local environments in vapor-deposited or chemically grown amorphous alumina films, while Al(VI) remains a minor component. This unique coordination distribution is now interpreted as a structural fingerprint of the amorphous state.

Active Al(V) is known as “Super 5” and is a very important surface Al species acting as the active site for dehydration and anchoring site for loading catalysts^19,20 and for active metals such as Pt, Pd, Ru (ref. 21–23) due to its Lewis acidity and adsorption affinity by unsaturated coordination and structural distortion.²⁴ Ruthenium catalysts supported by alumina enriched with Al(V) exhibited fairly high specific activity (TOF = 5180 h⁻¹) in hydrogenation of benzene into cyclohexane, although commercial γ-Al₂O₃-supported Ru showed a lower TOF of 417 h⁻¹.²³ The distorted coordination environment of Al(V) generates undercoordinated oxygen species and enhances the acidity of the surface, thereby providing active sites for reactions such as hydrocarbon activation, dehydration, and selective oxidation. Mechanism estimation on active Al(V) in γ-Al₂O₃ for ethanol dehydration and dehydrogenation was conducted. Ethanol adsorbs on Lewis acid sites of Al(V) and forms surface ethoxy species through O–H bond dissociation assisted by adjacent basic oxygen atoms. Subsequently, acetaldehyde is produced via β-H transfer from the ethoxy intermediate to lattice oxygen, while ethylene can form through dehydration on stronger acid sites. This acid and base dual-site mechanism demonstrates that γ-Al₂O₃ enables both dehydration and dehydrogenation depending on the local surface structure.²⁵

Here, it is widely recognized that the concentration of fivefold-coordinated aluminum sites in alumina can be tuned primarily by adjusting the heat treatment conditions.^26–28 When aluminum hydroxide is heated up to about 650 °C, dehydration and dehydroxylation processes occur, giving rise to coordinatively unsaturated Al five-coordinated sites. These five-coordinated species are metastable and, upon further heating to 750 °C or higher, they tend to transform into more stable four- or sixfold-coordinated forms.²⁶ Physical vapor deposition (PVD) and atomic layer deposition (ALD) produced aluminum thin films, which are composed of amorphous alumina with predominantly four- and fivefold-coordinated Al (Al(IV): 56% and Al(V): 36% for PVD, and Al(IV): 54% and Al(V): 41% for AVD, respectively).^11,29 In the conventional precipitation method, pH influences the structure of the precursor (e.g., aluminum hydroxide or boehmite), which in turn affects the coordination distribution of Al(IV), Al(V), and Al(VI) in the resulting alumina. In the acid region, Al³⁺ (aq.) and Al(OH)²⁺ tend to form, while Al(OH)₃ becomes predominant with increasing pH. At pH > 8, Al(OH)₄⁻ is predominant.^30,31 Mashkovtsev et al. used the controlled double jet precipitation method to obtain an aluminum hydroxide precursor as a function of pH. At lower pH values, aluminum hydroxide xerogels incorporate a larger fraction of NO³⁻, derived from Al(NO₃)₃ reagent, directly coordinated to aluminum, which induces structural disorder and promotes the formation of metastable fivefold-coordinated Al species. As the precipitation pH increases, the amount of aluminum-bound nitrate decreases, leading to improved ordering of the oxide lattice, larger crystallite sizes, and broader pore size distributions after calcination. Notably, samples obtained at pH ≈ 6 yield well-sedimented spherical particles with narrow size distributions. Solid-state NMR analyses further confirm that the relative abundance of AlO₅ units is highest under acidic precipitation conditions, while higher pH favors a shift toward four- and six coordination environments.³²

Although fivefold-coordinated Al sites are known to be highly active, and amorphous alumina has frequently been reported as a support material, there are relatively few examples where pure amorphous alumina was employed as the catalyst itself. In this study, we aim to synthesize amorphous alumina with an increased concentration of Al(V) sites by controlling the pH during gelation, and its catalytic performance is investigated.

Volatile organic compounds (VOCs), emitted from industrial processes including chemical manufacturing, transportation, construction, printing, and automobiles,^33–36 are among the most abundant and hazardous air pollutants. VOCs exert extremely harmful effects on human health due to their toxic and carcinogenic nature. The control of VOC emissions has therefore become one of the most pressing global environmental issues. While many studies have reported the thermal decomposition of VOCs using noble-metal nanoparticles (Pt, Pd, and Rh)^6–8 or rare-^1,37 and rare-earth-metals^17,18 as catalysts, there is a strong demand for decomposition processes employing more abundant materials such as Al₂O₃. We here propose to exploit pH-controlled synthesis to enrich Al(V) in pure amorphous alumina and evaluate its intrinsic catalytic activity toward VOC decomposition, rather than its conventional role as a support.

Experimental

Sample preparation

An aluminum hydrogel was prepared using liquid metal galinstan (alloyed at 250 °C with Ga: 68.5 wt% (Kojundo Chemical Laboratory Co. Ltd), In: 21.5 wt%, and Sn: 10 wt%, respectively, refer to previous research³⁸). 0.33 g of Al particles (3N, 20 μm, Kojundo Chemical Laboratory Co. Ltd) were mixed with galinstan and 150 mL of NH₄NO₃ aqueous solution (pH 4.7) as a pH controlling buffer at 0.1, 1, 10, and 100 mM for 390 minutes in an Ar atmosphere. The obtained hydrogel was centrifuged at 15 000 rpm for 10 minutes, followed by drying in a vacuum at 80 °C for 18 hours. Dried samples were dehydrated at 150, 200, 250, 350, 550, and 650 °C at 1 °C min⁻¹ for 1 hour.

Characterization

The pH during hydroxylation was monitored using a pH meter (Horiba, Ltd) for 0–390 minutes. The crystal structure of amorphous alumina was confirmed using X-ray diffraction equipment (XRD: Ultima IV, Rigaku Corp., Japan) with a Cu Kα line at an operating current and a voltage of 40 mA and 40 kV, respectively. N₂ adsorption (BELSORP MAXII, MicrotracBEL Corp., Japan) measurement at 77 K was conducted to obtain the specific surface area (SSA) and particle size estimated using spherical assumption. The chemical structure of amorphous alumina was analyzed via a diffuse reflectance infrared Fourier transform spectroscopy (DRIFT: FT/IR-V6000, JASCO Corp., Japan) under heating up to 650 °C. Weight loss during dehydration was measured using thermal gravimetry (TG: Shimadzu Corp., Japan) at 25–550 °C at 1 °C min⁻¹. The coordination structure of amorphous alumina was measured using solid ²⁷Al-NMR (ECA600II, JEOL Ltd.).

NMR spectra were separated into three or four waves by using a Lorentzian function. The area of Al(V) ratio and area of Al(V) broad component ratio were defined as follows;

The area of Al(V) ratio is described as A_Al(V) ratio

The area of Al(V) broad component ratio is described as A_{Al(V) broad} ratio

VOC catalytic decomposition test

The catalytic decomposition test was carried out based on the experimental conditions as suggested in our previous papers.^39,40 50 mg of the sample powder was placed in a quartz tube. A mixed gas consisting of ethyl acetate (103 ppm, Taiyo Gas Co., Ltd., Japan) as a VOC source and pure air was then flowed through the tube at a flow rate of 125 mL min⁻¹ for each component through a mass flow controller, at reaction temperatures of 100, 200, 300, and 400 °C. After waiting for 20 minutes to reach equilibrium, the gas remaining after VOC decomposition was collected and analyzed by gas chromatography (GC: GC-2030, Shimadzu Corp., Japan) and detected using infrared absorption CO (UM-300: KITAGAWA KOMYO RIKAGAKU KOGYO, Japan) and CO₂ (RI-215D: RIKEN KEIKI Co., Ltd., Japan) monitors to determine the products and residual gas concentration. The VOC decomposition rate, C₂H₄ production rate, and CO/CO₂ selectivity were defined as follows;

VOC decomposition rate

where A_ΔVOC represents the peak area of ethyl acetate at room temperature minus the peak area at the target temperature, determined from GC. A_{VOC R.T.} represents the peak area of ethyl acetate at room temperature.

C₂H₄ production rate

where A_{C2H4 target temp.} indicates the peak area of C₂H₄ determined from GC.

CO/CO₂ selectivity

where CCO/CO₂ indicates the concentration of CO or CO₂. N_Carbon indicates the total number of carbon atoms in ethyl acetate (four). The initial VOC concentration was 51.5 ppm (103/2), and it was converted into % by multiplying 10⁴.

T ₅₀ is the temperature at 50% of VOC decomposed and is a common indicator to evaluate catalytic performance. T₅₀ in this study and the preceding comparative studies was determined by linear interpolation between two points that bracket the 50% VOC conversion rate within the 200–400 °C temperature range.

In situ FT-IR under an ethyl acetate flow at 30–400 °C was conducted using a DRIFT equipped with a self-designed in situ VOC flow chamber with a VOC and air flow rate of 10 mL min⁻¹ each; detailed construction has been reported elsewhere.

Results and discussion

Samples with different buffer concentrations here were simply denoted as a-Al₂O₃-Bx, where x represents the buffer concentration in mM. The pH during gelation was monitored with various NH₄NO₃ buffer concentrations (Fig. 1). The pH began at 4.7 and steeply increased in a few minutes to 6.5–8.0. Al reacted with water, resulting in the production of OH⁻ with hydrogen, leading to an increase of pH. a-Al₂O₃-B0.1 and a-Al₂O₃-B1 reached a maximum pH at 8.5 and 8.3 in 150 min, although the pH of a-Al₂O₃-B10 and a-Al₂O₃-B100 kept increasing until 390 min. Final pH values were 8.5, 8.3, 8.2, and 7.8 for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100, respectively.

Fig. 1 pH transition during hydrogelation of a-Al₂O₃-B0.1 (red), a-Al₂O₃-B1 (green), a-Al₂O₃-B10 (blue), and a-Al₂O₃-B100 (yellow).

XRD measurements were performed to obtain the crystal structure information. Fig. S1 shows the XRD patterns of each sample after heating at 80–550 °C. They exhibited mainly an amorphous phase in addition to In₃Sn (ref. 41) at 80–250 °C, and In₂O₃ (ref. 42) at 350–550 °C regardless of buffer concentration, and there were no peaks for precursors such as bayerite,⁴³ gibbsite,⁴⁴ and boehmite.⁴⁵ After heating at 650 °C, the γ-alumina⁴⁶ phase was observed in a-Al₂O₃-B1 and a-Al₂O₃-B10 while it was not observed in a-Al₂O₃-B0.1 and a-Al₂O₃-B100 (Fig. 2). A 100 mM buffer suppressed the pH increase and induced the shortage of local OH⁻ instead of abundant NO₃⁻, leading to the formation of Al–NO₃, which is more thermally stable than Al–OH. Then, the removal of NO₃ ions in the range of 280–500 °C leads to the formation of disordered structures.³² In contrast, under high-pH conditions (such as 0.1 mM), soluble complexes of Al(OH)₄⁻ dominate, inhibiting crystal nucleation. Consequently, the precipitate grew into an amorphous phase even after heating at 650 °C.³⁰ Thermogravimetry (TG) also showed a weight loss of 25 wt% at 0.1–10 mM and 550 °C, but it was approximately 40 wt% at 100 mM due to the removal of NO₃-derived gases (Fig. S2). Furthermore, in situ FT-IR experiments at 80–650 °C were conducted to investigate the influence of the buffer on a-Al₂O₃ structures (Fig. 3 and S3). In Fig. S3, broad spectra in the range of 3000–3600 cm⁻¹ might be attributed to gibbsite, bayerite, and boehmite OH stretching,^47,48 which decreased with increasing temperature. Al–OH stretching vibration emerged at 3600–3700 cm⁻¹ and also diminished with increasing temperature due to dehydration. On the other hand, the peaks at 1275–1480 cm⁻¹ are attributed to ν_s and ν_as of N–O derived vibration modes,³² and the peak at 2300–2400 cm⁻¹ was attributed to N–N stretching, which was produced by decomposition of NO₃ species in the Al network as well as the previous study in which Al(NO₃)₃ was used as a reagent.^49,50 Active Al(V) sites act as adsorption sites of N₂, evaluated by IR measurements (N₂ works as a probe for Al(V) sites).²⁰ In our results, a-Al₂O₃-B100 exhibited a high intensity peak at 2355 cm⁻¹ that still remained even at 650 °C because generated N₂ strongly adsorbed on the active Al(V) sites. In FT-IR focusing on the 1200–1900 cm⁻¹ region (Fig. 3), a-Al₂O₃-B1 only exhibited a distinct band at 1630–1650 cm⁻¹, which is assigned to the bending vibration of physically adsorbed water,⁵¹ consistent with the high density of accessible adsorption sites, as discussed below. In contrast, the bands observed around 1560 cm⁻¹ for all alumina samples are tentatively attributed to the bending mode (δ(H–O–H)) of confined molecular water within the alumina structure, as structurally restricted water molecules are known to exhibit red-shifted bending vibrations compared to free or weakly adsorbed water.⁵² a-Al₂O₃-B1, prepared at a medium buffer concentration, possesses a relatively ordered surface structure, consistent with XRD results, providing abundant adsorption sites that enable strong hydrogen-bond interactions with water molecules.

Fig. 2 XRD patterns of a-Al₂O₃-B0.1 (red), a-Al₂O₃-B1 (green), a-Al₂O₃-B10 (blue), and a-Al₂O₃-B100 (yellow) after heating at 650 °C, compared with reference data of γ-Al₂O₃, bayerite, gibbsite, boehmite, In₃Sn, and In₂O₃.

Fig. 3 In situ FT-IR measurements at 80–650 °C of (a) a-Al₂O₃-B0.1, (b) a-Al₂O₃-B1, (c) a-Al₂O₃-B10, and (d) a-Al₂O₃-B100.

N₂ adsorption measurement at 77 K was performed to obtain the specific surface area (SSA) and particle diameter under spherical assumption as shown in Fig. 4a. The observed micropores and mesopores were assumed to be interparticle spaces because there was no steep increase at a relatively low-pressure range (P/P₀ < 0.1) but there was small adsorption hysteresis at P/P₀ = 0.4–0.9. The SSAs of amorphous alumina samples were 164, 312, 283, and 112 m² g⁻¹ and diameters calculated from the SSA using spherical assumption were 9.6, 5.1, 5.6, and 14.1 nm for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100, respectively. Fig. 4b indicates SSA and particle diameter as a function of buffer concentration. a-Al₂O₃-B1 and a-Al₂O₃-B10 had large SSA and small particles although a-Al₂O₃-B0.1 and a-Al₂O₃-B100 had an opposite tendency. For a-Al₂O₃-B0.1 and a-Al₂O₃-B100, Al(OH)₄⁻ formation and NO₃ intrusion into Al networks might prevent nuclear growth of aluminum hydroxide and produced smaller particles, which are likely to aggregate into larger particles with low SSA in terms of the N₂ molecular probe. Although the final pH values were not significantly different, structural variations observed through XRD, ATR-FT/IR and N₂ adsorption measurements suggested that the fundamental structural framework was established during the initial stage of the reaction, where the pH difference was substantial.

Fig. 4 (a) N₂ adsorption isotherms at 77 K for a-Al₂O₃-B0.1 (red), a-Al₂O₃-B1 (green), a-Al₂O₃-B10 (blue), and a-Al₂O₃-B100 (yellow) after heating at 650 °C. Open and closed symbols represent adsorption and desorption branches, respectively. (b) SSA and particle diameter dependency on buffer concentration.

²⁷Al solid-state NMR measurements were conducted to determine the fraction of five-coordinate aluminum species (Fig. 5). Four-, five-, and six-coordinated Al components appear at 10–13, 35–45, and 67–74 ppm based on previous studies.^20,24,32 Al(V) can be created through the hydration of Al(IV) and/or dehydration of Al(VI).²⁴ At room temperature, Al(VI) sites were predominant, however with increasing dehydration temperature, a progressive increase in Al(V) species instead of a decrease in Al(VI) was observed in all samples. For quantitative analysis, peak deconvolution was carried out using Lorentzian functions as shown in Fig. S4–S7. Fig. 6 shows Al components as a function of temperature. The Al(IV) component ratio was almost consistent at 20–25% for all conditions, whereas Al(V) and Al(VI) ratios drastically changed. Al(V) was considered to be predominantly produced from the dehydration of Al(VI). Al(V) ratios were 19, 17, 21, and 14% at 80 °C and 42, 43, 47, and 53% at 650 °C for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100, respectively. These dehydration trends with increasing temperature are consistent with previous research, which indicated that the increase in Al(V) associated with dehydration occurred above 350 °C.^20,24 The fraction of Al(V) was slightly higher than that observed for thin films heated above 550 °C, and it corresponds more closely to the ratio typically found in nanoparticles rather than in thin films.¹⁵

Fig. 5 NMR spectra of (a) a-Al₂O₃-B0.1, (b) a-Al₂O₃-B1, (c) a-Al₂O₃-B10, and (d) a-Al₂O₃-B100 heated at 80 (dark blue)–650 °C (red).

Fig. 6 Al components: Al(IV) (green), Al(V) (sky blue), and Al(VI) (yellow) as a function of dehydration temperature for (a) 0.1, (b) 1, (c) 10, and (d) 100 mM samples.

Subsequently, VOC decomposition experiments were conducted by measuring the conversion of ethyl acetate in the temperature range of 100–400 °C to evaluate the catalytic performance using amorphous alumina after dehydration at 650 °C (Fig. 7a and S8). No decomposition was observed at 100 °C, whereas a pronounced increase in conversion was detected above 300 °C. At 400 °C, the conversions recorded for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100 were 97, 100, 89, and 94%, respectively. The decomposed ethyl acetate was converted predominantly into ethylene (Fig. S8), with ethylene production rates of 0.64, 0.68, 0.72, and 0.94 for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100, respectively (Fig. 6b). For CO/CO₂ conversion, 18–20% and 11–28% CO and CO₂ selectivities were recorded. Consequently, a-Al₂O₃-B1 showed 48% CO + CO₂ selectivity, indicating that 50% of ethyl acetate was perfectly oxidized into CO or CO₂. To the best of our knowledge, it was revealed that a-Al₂O₃-B1 as a pure amorphous alumina catalyst completely decomposed VOC for the first time. When compared with other materials such as perovskite oxide,¹ transition metal oxide,¹⁴ typical silica–alumina,¹⁶ rare- and rare-earth metal^17,18 catalysts combined with appropriate supports in previous research, the catalytic activity of pure amorphous alumina in this study exhibited that T₅₀ (temperature at 50% of VOC decomposed and an indicator for catalyst performance) was as low as that of rare-earth metals or perovskites and less than that of silica–alumina, while the gas hourly space velocity (GHSV) was significantly high as shown in Fig. 7d.

Fig. 7 (a) VOC decomposition rate as a function of temperature for a-Al₂O₃-B0.1 (red), a-Al₂O₃-B1 (green), a-Al₂O₃-B10 (blue), and a-Al₂O₃-B100 (yellow). (b) Ethylene production rate. (c) Selectivities of CO (green), CO₂ (blue), and CO + CO₂ (red). (d) Comparison of T₅₀ in this study (red) with perovskite oxide (black),¹ transition metal oxide (yellow),¹⁴ typical silica–alumina (pink),¹⁶ rare- (blue) and rare-earth metal (green)^17,18 catalysts combined with appropriate supports vs. GHSV.

Meanwhile, no clear correlation was observed between buffer concentration and the VOC decomposition rate (Fig. 8a). Therefore, we examined which structural parameters predominantly govern the catalytic activity and analyzed their correlation with the VOC decomposition rate. Since catalytic activity is generally known to improve with increasing specific surface area associated with decreasing particle size,⁵³ we first examined correlations with SSA and particle diameter, however, the resulting coefficient of determination (R²) values was essentially zero (Fig. 8b and c). As mentioned above, numerous reports have suggested that Al(V) species in amorphous alumina contribute to enhanced activity. Therefore, we next considered the effect of surface area by correlating the product of the Al(V) ratio and SSA with the VOC decomposition rate, but again the R² value was nearly zero (Fig. 8d). It should be noted that Al(V) can be further classified into two types: a symmetric trigonal bipyramidal structure and an asymmetric square-pyramidal structure (Fig. 9a). Recent studies have demonstrated that in ²⁷Al NMR spectra, the symmetric trigonal bipyramidal sites exhibit a small quadrupolar coupling constant, which represents the strength of the interaction between the nuclear quadrupole moment and the surrounding electric field gradient, and thus appear as sharp peaks in NMR, whereas the asymmetric pyramidal sites possess a larger quadrupolar coupling constant and consequently give rise to broad peaks.^54,55 The origin of Al(V) broad sites was considered to be the dehydration of OH groups and/or H₂O on the surface Al(VI), which is terminal OH/H₂O. By contrast, Al(V) sharp sites were derived from the dehydration of OH in the bulk of the Al–OH–Al network (doubly or triply-bonded).^20,24,56

Fig. 8 Correlation plots of VOC decomposition rate against (a) buffer concentration, (b) SSA, (c) particle diameter, (d) A_Al(V) ratio × SSA, (e) A_{Al(V) broad} ratio × SSA, and (f) CO + CO₂ selectivity against A_{Al(V) broad} ratio × SSA.

Fig. 9 (a) Structural images of trigonal bipyramid type and pyramid type five-coordinated Al. Re-deconvolution of NMR spectra in four components for (b) a-Al₂O₃-B0.1, (c) a-Al₂O₃-B1, (d) a-Al₂O₃-B10, and (e) a-Al₂O₃-B100.

Based on this insight, we re-deconvoluted the NMR spectra into four components—Al(IV), Al(V) sharp, Al(V) broad, and Al(VI) at 12.5, 33, 42, and 70 ppm, respectively—and obtained an excellent fit in Fig. 9b–e. Al(V) broad ratios were 0.19, 0.14, 0.09, and 0.26 for a-Al₂O₃-B0.1, a-Al₂O₃-B1, a-Al₂O₃-B10, and a-Al₂O₃-B100, respectively. Remarkably, when the peak area ratio of the Al(V) broad component was multiplied by SSA, a strong correlation with VOC decomposition rate was observed with R² = 0.74. Furthermore, CO + CO₂ selectivity also exhibited a significant correlation with R² = 0.78, indicating a significant contribution to the catalytic reaction. According to FT-IR results as mentioned above, pyramidal Al(V) could be partially covered by N₂. Therefore, a-Al₂O₃-B100 showed lower catalytic activity although a-Al₂O₃-B100 had a large Al(V) broad component.

To further elucidate the reaction pathway, in situ FT-IR measurements for a-Al₂O₃-B1 were conducted under an ethyl acetate flow at the temperature range of 30–400 °C, as shown in Fig. 10. A gradual decrease in the OH-related band around 3500–3700 cm⁻¹ was observed with increasing temperature, indicating progressive dehydration of surface hydroxyl groups coordinated to aluminum species. In contrast, no other distinct absorption bands attributable to adsorbed ethyl acetate species, such as C=O stretching vibrations, were observed throughout the measurement range, and no significant evolution of other organic-related peaks was observed. These observations support the reaction pathway proposed in Scheme 1, suggesting that the initial dehydration of surface OH groups is the primary thermally activated process and acts as the trigger for subsequent reactions. The dehydration of OH groups coordinated to surface Al species leads to the formation of pyramidal Al(V) sites. The resulting dehydrated sites retain localized electrons,⁵⁷ which can promote the formation of superoxide-like species derived from adsorbed oxygen in the air. Pre-existing and newly generated pyramidal Al(V) sites work as strong Lewis acid sites, enabling adsorption of the carbonyl group of ethyl acetate. The activated oxygen species subsequently attack the adsorbed molecule, leading to the formation of acetaldehyde and ethanol as primary intermediates.⁴⁰ By analogy with our previous research on hydroxyapatite,⁴⁰ ethanol then undergoes two competitive reactions depending on the neighboring surface structure. At the Lewis acidic pyramidal Al(V) sites, ethanol is dehydrated to ethylene, whereas at adjacent bridging oxygen sites acting as basic centers, β-H transfer promotes dehydrogenation to acetaldehyde.²⁵ The formed acetaldehyde is further oxidized by thermally activated oxygen species, eventually yielding CO and CO₂. For a-Al₂O₃-B100, the incorporation of nitrate species and their subsequent thermal removal led to local structural rearrangement, resulting in the formation of isolated pyramidal Al(V) Lewis acid sites and partial disruption of the Al–O–Al network. This structural isolation reduces the effectiveness of neighboring bridging oxygen base sites, thereby suppressing β-H transfer but retaining ethanol dehydration to ethylene.

Fig. 10 In situ FT-IR spectra of a-Al₂O₃-B1 under an ethyl acetate flow at 30–400 °C.

Scheme 1 Predicted catalytic reaction pathway of ethyl acetate on Al(V)-containing amorphous alumina.

Conclusion

Five-coordinated Al (Al(V)) sites are widely recognized as highly reactive motifs in alumina, and amorphous alumina typically contains a relatively large fraction of such Al(V) species. In general, amorphous alumina is synthesized through the dehydration of aluminum hydroxides. Because the structure of aluminum hydroxides evolves sequentially with pH, the control of pH strongly influences both the hydroxide precursor and the final alumina structure. The present study aimed to synthesize highly active amorphous alumina enriched in Al(V) species. To control the pH increase during hydroxide gelation, buffers of different concentrations (0.1, 1, 10, and 100 mM) were employed, and the structure and catalytic performance of the resulting amorphous alumina after dehydration at 80–650 °C was evaluated.

The influence of pH controlled by buffer addition on structure and catalytic activity can be rationalized as follows.

1) High pH (low buffer concentration): excess OH⁻ leads to the predominance of Al(OH)₄⁻, which suppresses the precipitation of Al(OH)₃. As a result, nucleation and growth of Al(OH)₃ are hindered, yielding very small nanoparticles. Even after heating to 650 °C, the amorphous phase remains predominant. Although the fraction of Al(V) broad components is high, the large surface energy of the small particles promotes aggregation, leading to a decrease in specific surface area and catalytic activity.

2) Low pH (high buffer concentration): although Al(OH)₃ precipitation occurs, local OH⁻ deficiency favors the formation of bonds between Al and buffer anions (e.g., Al–NO₃). These NO₃ groups are released between 280 and 500 °C, inducing structural disorder and amorphization. Consequently, aggregation occurs and the surface area decreases as well as in the case of high pH. Furthermore, N₂ derived from NO₃ decomposition can adsorb onto the pyramidal Al(V) active sites, leading to their blocking and lowering catalytic activity. In the exceptional case of a-Al₂O₃-B100, the amorphous fraction and Al(V) broad content were the highest, thus, despite partial site blocking, the activity remained higher than that of a-Al₂O₃-B10, resulting in the order a-Al₂O₃-B100 > a-Al₂O₃-B10.

From these considerations, it can be concluded that the optimal conditions for maximizing the Al(V) broad component in amorphous alumina are achieved by aluminum hydroxide gelation at around pH 7–8, where particle size can be controlled to avoid aggregation while generating a large fraction of Al(V). Experimentally, a-Al₂O₃-B1 was considered to be optimized, achieving 100% of VOC decomposition with pure amorphous alumina for the first time, which was comparable to other perovskite and rare-earth metal catalysts with suitable supports and superior to silica–alumina in terms of T₅₀.

We assumed the VOC decomposition mechanism based on a previous study. Thermally excited oxygen radicals decomposed ethyl acetate into acetaldehyde and ethanol, followed by further oxidation into ethylene at acid sites and complete oxidation by the assistance of oxygen radicals. These findings demonstrate that amorphous alumina enriched in five-coordinated Al sites can act as an efficient acid–base catalyst. Applications are expected not only in VOC decomposition but also in organic chemistry and petrochemical processes, such as esterification, alcohol conversion reactions, and aldol condensation.

Author contributions

T. S. was responsible for the overall design, direction, and supervision for the project. T. W. performed part of the experiments, characterized the results and wrote the manuscript. H. K. performed part of the experiments and data acquisition. Y. X. and Y. X. provided the interpretation of characterization. All authors approved the final version of the manuscript.

Conflicts of interest

The authors declare no competing interests.

Data availability

Data for this article, including results of XRD, ²⁷Al-NMR, NMR fitting, GC, N₂ adsorption, ATR-FT/IR, pH transition, TG-DTA, and monitored CO/CO₂ concentration are available at Science Data Bank at https://www.scidb.cn/en/anonymous/SVpaamV1 (the dataset is utilized for peer review currently and will be updated as public after publication). The supplementary information (SI) contains figures generated from the raw data available via the links provided in the data availability statement.

Supplementary information is available. See DOI: https://doi.org/10.1039/d5cy01418B.

References

1. G. Pecchi, P. Reyes, R. Zamora, L. E. Cadús and J. L. G. Fierro, J. Solid State Chem., 2008, 181, 905–912.
2. J. Robertson and B. Falabretti, J. Appl. Phys., 2006, 100, 014111.
3. C. Avis and J. Jang, J. Mater. Chem., 2011, 21, 10649–10652.
4. I. B. Singh, P. Gupta, A. Maheshwari and N. Agrawal, J. Sol-Gel Sci. Technol., 2015, 73, 127–132.
5. J. S. Daubert, G. T. Hill, H. N. Gotsch, A. P. Gremaud, J. S. Ovental, P. S. Williams, C. J. Oldham and G. N. Parsons, ACS Appl. Mater. Interfaces, 2017, 9, 4192–4201.
6. J. Chi-Sheng Wu and T.-Y. Chang, Catal. Today, 1998, 44, 111–118.
7. J. He, F. Zheng, Y. Zhou, X. Li, Y. Wang, J. Xiao, Y. Li, D. Chen and J. Lu, J. Colloid Interface Sci., 2022, 613, 155–167.
8. S. Pitkäaho, T. Nevanperä, L. Matejova, S. Ojala and R. L. Keiski, Appl. Catal., B, 2013, 138–139, 33–42.
9. S. Arora, R. Khan and S. Sivakumar, ChemCatChem, 2025, 17, e202401673.
10. V. Sarou-Kanian, A. N. Gleizes, P. Florian, D. Samélor, D. Massiot and C. Vahlas, J. Phys. Chem. C, 2013, 117, 21965–21971.
11. S. K. Lee, S. B. Lee, S. Y. Park, Y. S. Yi and C. W. Ahn, Phys. Rev. Lett., 2009, 103, 095501.
12. E. Lam, A. Comas-Vives and C. Copéret, J. Phys. Chem. C, 2017, 121, 19946–19957.
13. R. Lizárraga, E. Holmström, S. C. Parker and C. Arrouvel, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83, 094201.
14. Y. Zhou, H. Zhang and Y. Yan, Chem. Eng. Res. Des., 2020, 157, 13–24.
15. A. N. Gleizes, D. Samélor, C. Vahlas, V. Sarou-Kanian, P. Florian and D. Massiot, Adv. Sci. Technol., 2014, 91, 123–133.
16. T. K. Phung, A. A. Casazza, B. Aliakbarian, E. Finocchio, P. Perego and G. Busca, Chem. Eng. J., 2013, 215–216, 838–848.
17. R. Xiao, R. Qin, C. Zhang, S. Chen and J. Wang, J. Rare Earths, 2021, 39, 817–825.
18. S. A. Carabineiro, M. Konsolakis, G. E. Marnellos, M. F. Asad, O. S. Soares, P. B. Tavares, M. F. Pereira, J. J. Órfão and J. L. Figueiredo, Molecules, 2016, 21, 644.
19. C. V. Chandran, C. E. A. Kirschhock, S. Radhakrishnan, F. Taulelle, J. A. Martens and E. Breynaert, Chem. Soc. Rev., 2019, 48, 134–156.
20. K. Khivantsev, N. R. Jaegers, J.-H. Kwak, J. Szanyi and L. Kovarik, Angew. Chem., Int. Ed., 2021, 60, 17522–17530.
21. J. H. Kwak, J. Hu, D. Mei, C.-W. Yi, D. H. Kim, C. H. F. Peden, L. F. Allard and J. Szanyi, Science, 2009, 325, 1670–1673.
22. H. Duan, R. You, S. Xu, Z. Li, K. Qian, T. Cao, W. Huang and X. Bao, Angew. Chem., Int. Ed., 2019, 58, 12043–12048.
23. N. Tang, Y. Cong, Q. Shang, C. Wu, G. Xu and X. Wang, ACS Catal., 2017, 7, 5987–5991.
24. Z. Zhao, D. Xiao, K. Chen, R. Wang, L. Liang, Z. Liu, I. Hung, Z. Gan and G. Hou, ACS Cent. Sci., 2022, 8, 795–803.
25. J. F. DeWilde, C. J. Czopinski and A. Bhan, ACS Catal., 2014, 4, 4425–4433.
26. J. Lee, E. J. Jang and J. H. Kwak, Appl. Catal., A, 2019, 569, 8–19.
27. E. J. Jang, J. Lee, H. Y. Jeong and J. H. Kwak, Appl. Catal., A, 2019, 572, 1–8.
28. Y. Wang, C. Pei, X. Wang, G. Sun, Z.-J. Zhao and J. Gong, Fundam. Res., 2024, 4, 1480–1487.
29. S. K. Lee, S. Y. Park, Y. S. Yi and J. Moon, J. Phys. Chem. C, 2010, 114, 13890–13894.
30. P. A. Nikolaychuk and E. Kozeschnik, npj Mater. Degrad., 2024, 8, 94.
31. W. J. McHardy and A. P. Thomson, Mineral. Mag., 1971, 38, 358–368.
32. M. Mashkovtsev, N. Tarasova, E. Baksheev, V. Rychkov, N. Zhuravlev, P. Solodovnikova and M. Galiaskarova, Int. J. Mol. Sci., 2023, 24, 5151.
33. D. Baskaran, D. Dhamodharan, U. S. Behera and H.-S. Byun, Environ. Res., 2024, 251, 118472.
34. S. Li, Y. Lin, G. Liu and C. Shi, Environ. Sci.: Processes Impacts, 2023, 25, 727–740.
35. P. Yang, T. Zhang, Z. Ling, M. Liu and X. Zeng, Energies, 2025, 18, 3430.
36. Y. Huang, S. S. Ho, Y. Lu, R. Niu, L. Xu, J. Cao and S. Lee, Molecules, 2016, 21, 56.
37. F. Bi, J. Wei, Z. Zhou, Y. Zhang, B. Gao, N. Liu, J. Xu, B. Liu, Y. Huang and X. Zhang, JACS Au, 2025, 5, 363–380.
38. A. Zavabeti, B. Y. Zhang, I. A. de Castro, J. Z. Ou, B. J. Carey, M. Mohiuddin, R. Datta, C. Xu, A. P. Mouritz, C. F. McConville, A. P. O'Mullane, T. Daeneke and K. Kalantar-Zadeh, Adv. Funct. Mater., 2018, 28, 1804057.
39. Y. Xin and T. Shirai, Sci. Rep., 2021, 11, 7512.
40. Y. Xin, Y. Ando, S. Nakagawa, H. Nishikawa and T. Shirai, Catal. Sci. Technol., 2020, 10, 5453–5459.
41. J. Janczak, R. Kubiak, A. Zaleski and J. Olejniczak, J. Alloys Compd., 1994, 206, 215–216.
42. M. Marezio, Acta Crystallogr., 1966, 20, 723–728.
43. G. Yamaguchi and K. Sakamoto, Bull. Chem. Soc. Jpn., 1958, 31, 140–141.
44. H. Saalfeld and M. Wedde, Z. Kristallogr., 1974, 139, 129–135.
45. X. Bokhimi, J. A. Toledo-Antonio, M. L. Guzmán-Castillo and F. Hernández-Beltrán, J. Solid State Chem., 2001, 159, 32–40.
46. L. Smrcok, V. Langer and J. Krestan, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 2006, 62, i83–i84.
47. E. Balan, M. Blanchard, J.-F. Hochepied and M. Lazzeri, Phys. Chem. Miner., 2008, 35, 279–285.
48. C. Liu, K. Shih, Y. Gao, F. Li and L. Wei, J. Soils Sediments, 2012, 12, 724–733.
49. K. Atrak, A. Ramazani and S. Taghavi Fardood, J. Mater. Sci.: Mater. Electron., 2018, 29, 8347–8353.
50. P. A. Prashanth, R. S. Raveendra, R. Hari Krishna, S. Ananda, N. P. Bhagya, B. M. Nagabhushana, K. Lingaraju and H. Raja Naika, J. Asian Ceram. Soc., 2015, 3, 345–351.
51. H. A. Al-Abadleh and V. H. Grassian, Langmuir, 2003, 19, 341–347.
52. R. Kevorkyants, A. V. Rudakova, Y. V. Chizhov and K. M. Bulanin, Chem. Phys. Lett., 2016, 662, 97–101.
53. T. Watanabe and T. Ohba, ACS Sustainable Chem. Eng., 2021, 9, 3860–3873.
54. M. Taoufik, K. C. Szeto, N. Merle, I. D. Rosal, L. Maron, J. Trébosc, G. Tricot, R. M. Gauvin and L. Delevoye, Chem. – Eur. J., 2014, 20, 4038–4046.
55. R. Wischert, P. Florian, C. Copéret, D. Massiot and P. Sautet, J. Phys. Chem. C, 2014, 118, 15292–15299.
56. M. Digne, P. Sautet, P. Raybaud, P. Euzen and H. Toulhoat, J. Catal., 2004, 226, 54–68.
57. T. Ishizaka, S. Tero-Kubota, Y. Kurokawa and T. Ikoma, J. Phys. Chem. Solids, 2003, 64, 801–806.

---