Au/CeO₂ Photocatalyst for the Selective Oxidation of Aromatic Alcohols in Water under UV, Visible and Solar Irradiation

Abstract

Au nanoparticles supported on CeO₂ have been prepared and investigated as photocatalysts for the photocatalytic selective oxidation of benzyl alcohol and 4-methoxybenzyl alcohol to the correspondent benzaldehydes, in aqueous suspensions and room conditions under UV, visible and natural solar light irradiation. Au nanoparticles have been supported by impregnation (1 and 3 wt.%) on two types of CeO₂ (i.e., a commercial one and a home prepared oxide obtained in the presence of NaOH as precipitation agent). The Au impregnated samples showed strong visible radiation absorption at 565–570 nm associated to localized surface plasmon resonance (LSPR). The bare CeO₂ samples are activated by UV light and resulted virtually inactive under visible irradiation, whereas the presence of Au improved both the conversion of the alcohols and the selectivity of the reaction towards the aldehyde, giving rise to good results, particularly under visible and natural solar light irradiation. The activity of the materials increased by increasing the Au content.

1. Introduction

The development of green and sustainable technologies is a big challenge of the current scientific research in applied sciences. The use of solar energy as clean and renewable energy source is a priority for the technologies aiming to develop photocatalytic transformations where the energy of photons can be used to drive many useful chemical reactions, including solar production of fuels, water purification and selective reactions.

In a photocatalytic process when a semiconductor is irradiated by suitable radiation, electrons are promoted from the valence band to the conduction band. The generation of electron/hole couples is responsible of oxidation and reduction processes. The photogenerated electrons reduce the adsorbed oxygen (O₂), if the process is carried out in air, to form superoxide radicals (O₂^−•) and the photogenerated holes react with H₂O to form hydroxyl radicals (OH^•), generally responsible for the oxidation of substrates resulting in their mineralization and/or partial oxidation [1].

The partial oxidation of alcohols to obtain aldehydes and/or ketones is one of the most important reactions in organic synthesis both, at industrial scale and in the laboratory. Photocatalysis is a promising technology for selective oxidation, alternative to traditional chemical methods based on reactions that are usually carried out at high temperatures and pressure with hazardous and harmful reagents. Photocatalytic reactions are carried out under mild conditions (i.e., room temperature and atmospheric pressure with molecular oxygen or air as oxidant). The use of water for these reactions is highly desirable even if the selectivity can be lower with respect to the use of organic solvents. Photocatalysts responsive to visible light are particularly interesting, mostly those active under natural sunlight as the energy source.

The efficiency of the photocatalytic surface will depend on the light absorption ability that governs the number of active electrons/holes photoproduced. The greatest used photocatalysts are transition metal oxides, being TiO₂ the most popular. They require ultraviolet (UV) irradiation to induce photocatalytic reactions due to their relatively large band gap (ca. 3.2 eV for TiO₂). The total solar energy accounts only for ca. 5% of UV irradiation so they can be used with a moderate success also under sunlight. Consequently, the development of photocatalysts prone to be activated under visible light irradiation is paramountcy important from the practical point of view because visible light accounts for ~50% of the total solar energy. In this regard, metal nanoparticles such as gold (Au), silver (Ag) or copper (Cu) have been extensively studied as co-catalysts due to their unique properties associated with their strong photoabsorption in the visible light region, which is due to localized surface plasmon resonance (LSPR) [2].

As one of the rare earth oxides, with a bandgap of ca. 3.2 eV, CeO₂ has attracted intense interest because it plays an important role in environmental and energy related applications. The selective photocatalytic oxidation of benzyl alcohols into the corresponding aldehydes has been poorly investigated with CeO₂-based photocatalysts. Li et al. report that under different irradiation conditions CeO₂ showed some partial oxidation of benzyl alcohol to benzaldehyde in water. The benzyl alcohol conversion achieved 27% in the presence of CeO₂ and 33% when Pt/CeO₂ was used; whereas the selectivity to benzaldehyde was 12% or 37% in the presence of CeO₂ or Pt-CeO₂, respectively, after 5 h of reaction under irradiation with a 300 W Xe lamp [3].

Recently, a great deal of research has been conducted in order to understand the catalytic activity of CeO₂-supported noble metals such as Au, Pt and Ag. An enhanced photodegradation of organic compounds has been largely reported after adding Au or Ag nanoparticles to the surface of semiconductor photocatalysts, including CeO₂ [4]. These metal nanoparticles (MNPs) have been used in photocatalysis for the inhibition of the recombination of the photogenerated electron/hole couples because they are excellent scavengers of electrons [5]. An alternative benefit of using metal NPs is their localized surface plasmon resonance (LSPR, commonly called as SPR), which results from the confinement of a surface plasmon in a nanoparticle (NP) smaller than the wavelength of light exciting the plasmon (i.e., when a NP is irradiated, the oscillating electric field causes the coherent oscillation of conduction electrons) [2]. Irradiating metal nanoparticles with light at their plasmon frequency generates intense electric fields at the surface of the nanoparticles. The frequency of this resonance can be tuned by varying the nanoparticle size, shape, material, and proximity to other nanoparticles [6].

The use of a semiconductor supporting metal NPs gives rise to a plasmonic photocatalysts where the mechanism of the electron/hole photogeneration is improved by the visible activated plasmon resonance [7]. Absorption of light is a characteristic property of metal NPs, particularly for Au, Ag and Cu NPs which are visible to the unaided eye, indeed, the peaks of photoabsorption due to SPR of Au, Ag and Cu nanoparticles are generally observed at around 550, 450 and 600 nm, respectively. In contrast to reports on supported Au nanoparticles, less research has been published on chemical reactions induced by the SPR of Ag and Cu nanoparticles, probably due to their instability under working conditions.

The plasmonic effect of metal NPs has found applications in medical therapy, surface-enhanced Raman spectroscopy, etc. [8]. Kozuka et al. carried out pioneer work introducing Au and Ag on the TiO₂ layer in photoelectrochemical (PEC) evaluation [9]. Great advances have been achieved in this field, and plasmonic photocatalysts have become a promising strategy for developing efficient visible-light-responsive composite materials.

The photocatalytic activity of these nanocomposites was enhanced due to the following two physical effects: (i) The SPR-induced light for enhancing the light absorption and (ii) the electron transfer from the metal to the oxide in the Au/semiconductor structure [10,11,12]. Many reports have been devoted to the use of Au/CeO₂ photocatalysts for dye bleaching but more interestingly, recent literature explores their application as robust photothermal catalysts active under solar irradiation for CO₂ reduction [13]. The selective oxidation of alcohols to their corresponding aldehydes in water, the greenest solvent for photocatalytic reactions, is an important organic transformation, because aldehydes are widely used in food, pharmaceutical and chemical industries [14].

Li et al. studied Au/CeO₂ hybrid nanofibers as photocatalysts for selective oxidation of benzyl alcohol to benzaldehyde in acetonitrile using a Xe lamp and a filter to get just visible irradiation (λ > 420 nm). They obtained a complete conversion of BA to BAL by using increasing loading amounts of Au NPs from 0.25 to 2.5 wt.% showing SPR absorption centered at ca. 605 nm [15]. The reaction rate obtained was in the range 500 to 2250 μmol h⁻¹ g⁻¹ by using the Xe lamp and from 50 to 400 μmol h⁻¹ g⁻¹ when the system was irradiated with λ > 420 nm. Kominami et al. prepared Au/CeO₂ nanocomposites exhibiting SPR absorption at ca. 550 nm, so they used a green LED as irradiation source [16]. The intensity of the photoabsorption due to the SPR of Au was higher for Au NPs with bigger size. The photocatalytic oxidation of benzyl alcohol in aqueous suspensions of Au/CeO₂ samples gave rise to a 100% conversion of the substrate to benzaldehyde in ca. 15–20 h. The rates of benzaldehyde formation were in the range 1.9 to 3.0 μmol h⁻¹.

According to the literature, the most important factors influencing the activity and selectivity of Au/CeO₂ as thermal catalysts for the partial oxidation of benzyl alcohols are the size of the supported gold nanoparticles [17] and the concentration of lattice defects in the ceria support [18]. Wolsky et al. report that the use of NaOH as precipitation (or co-precipitation) agent instead of urea or hexamethylenetetramine in the CeO₂ preparation has a significant impact on the structure, texture and catalytic properties of Au/CeO₂ catalysts resulting beneficial in the formation of small and uniform gold nanoparticles possessing enhanced catalytic activity s in low-temperature oxidation of benzyl alcohol [19].

Based on the so far reported literature, we have considered it of interest to investigate the properties and photocatalytic activity of two sets of Au catalysts, with loading 1 and 3 wt.%, deposited on ceria. One set of samples is based on commercial CeO₂, and the other one is based on a home prepared (hp) CeO₂ oxide using NaOH, as precipitating agent. The prepared samples were used as photocatalysts for the selective oxidation of both benzyl alcohol and 4-methoxy benzyl alcohol to the corresponding aldehydes in water and aerobic environment under UV, visible and solar irradiations. In particular, this study investigates for the first time how two types of ceria (one commercial and one home prepared) both pure and loaded with gold are able to exploit either UV or Vis radiation for partial photocatalytic oxidation reactions of alcohol. In fact, UV radiation is able to make ceria work as a photocatalyst and visible radiation to promote the plasmonic effect on Au nanoparticles. Furthermore, by making the reactions take place using natural solar radiation, the possible synergistic effect between the two effects (photocatalytic and plasmonic) was studied.

2. Results and Discussion

2.1. Characterisation of the Au/CeO₂ Photocatalysts

The XRD patterns of the Au catalysts supported on commercial and hp ceria are shown in Figure 1A,B, respectively. In all cases, the diffraction peaks of the fluorite structure of ceria were detected, with a more crystalline structure in the case of the Aldrich sample. In

Table 1. Textural properties of ceria oxides and Au supported catalysts.

Sample	SSA [m2 g−1]	Mean Mesopore Diameter (BJH) nm	Mesopore Volume [cm3 g−1] 1	Micropore Volume [cm3 g−1] 2	Crystallite Size [nm] 3
CeO2 A	79.0	12.3	0.22	-	22.0
1%Au/CeO2 A	77.0	11.7	0.20	-	18.0 (nd)
3%Au/CeO2 A	75.0	10.8	0.19	-	18.0 (3.5)
CeO2 hp	101.0	6.0	0.69	0.0090	10.0
1%Au/CeO2 hp	99.0	6.0	0.68	0.0086	12.0 (nd)
3%Au/CeO2 hp	95.0	6.0	0.64	0.0077	12.0 (2.5)

¹ By BJH method; ² By t-Plot; ³ CeO₂ crystallite size and Au particle size (in parenthesis) determined by Debye-Scherrer equation.

the support crystallite sizes and the gold particle sizes, as determined from the line broadening of the CeO₂ and metal Au (111) XRD peaks, are listed. No gold signals are visible in the XRD patterns of both Au1% catalysts, while slightly detectable features were observed in the case of the Au3% catalysts, corresponding to an average size of ca. 3.5 and ca. 2.5 nm, in the case of the CeO₂ Aldrich and the hp supported samples, respectively.

In Figure 2A,B the nitrogen adsorption–desorption isotherms and pore size distribution curves are displayed for CeO₂ A, CeO₂ hp and for the corresponding Au catalysts. The isotherms are classical type IV, as defined by IUPAC, with hysteresis typical of mesoporous materials, H1-H2 types. The adsorption–desorption curves, for CeO₂ hp and the corresponding Au catalysts, result at low p/p₀ values in a relatively larger adsorption than for CeO₂ A and the related catalysts, suggesting that the former samples contain also some micropores (see Table 1).

In Table 1 the specific surface area (SSA), mesopore diameter and mesopores volume values are listed. The micropore volume calculated by t-Plot for CeO₂ hp and the corresponding gold supported catalysts is also reported. Both ceria oxides are characterized by relatively high SSA, ~80 and 100 m²/g, respectively. After Au deposition and calcination, for both catalysts the surface area values along with the pore volume slightly decreased. For the Au catalysts over CeO₂ A the mean pore diameter also decreased, suggesting that some metallic Au nanoparticles may partially fill the pores, while no modifications were observed in the pore size distribution of gold deposited over CeO₂ hp, according to the stabilization of gold mainly as Au⁺¹ species that partially diffuse into the bulk of the support. Furthermore, the Au/CeO₂ hp catalysts kept some microporosity based on the micropore volume values listed in Table 1.

The SEM micrographs shown in Figure 3 and Figure 4 are useful to study the morphology of the investigated samples. Figure 3 reports the micrographs of the samples based on the home prepared CeO₂ material. The CeO₂ hp sample (prepared in the presence of NaOH) consists of aggregates of nanoparticles whose size ranged between ca. 24 and ca. 40 nm. The empty spaces between these particles have roughly the same size of the particles and thus this sample appears to be essentially mesoporous. As far as the commercial CeO₂ A (Aldrich) is concerned, as shown in Figure 4, it is constituted of aggregates of nanoparticles as well, even if in this case their size is generally smaller (14–25 nm) than those of the CeO₂ hp sample. However, the biggest differences between CeO₂ A and CeO₂ hp sample are observed in the nanoparticles aggregation manner. Indeed, in the CeO₂ A catalyst, in addition to the presence of mesopores, existing in the CeO₂ hp, are also observed macropores. The morphology of the CeO₂ A catalyst looks like that of certain volcanic rocks. For all Au catalysts, the presence of 1% or 3% Au does not modify significantly the morphology of the corresponding ceria supports, in agreement with the so far discussed SSA and porosity values (Table 1).

As far as the Au content detected by EDX investigation is concerned, it can be noted that the two samples with a nominal content of 1% seem a little richer in the metal. Indeed, for the 1% Au-CeO₂ hp and 1% Au-CeO₂ A the Au content measured was 1.7% and 1.4%, respectively. On the contrary, in the case of the two catalysts with the higher Au content, the EDX result is practically coincident with the nominal one, 3.2 and 3.1 for the 3% Au-CeO₂ hp and 3% Au-CeO₂ A samples, respectively. The higher amount of gold detected by EDX for both 1% Au catalysts is in agreement with the higher metal dispersion with respect to the 3% Au samples.

Scanning/transmission electron microscopy was used to get information on the morphology of the samples. The cubic morphology of the bare CeO₂ Aldrich and hp is apparent in the micrographs (a) and (d), respectively, of Figure 5. While the Aldrich sample appears more homogeneous, some elongated particles can be noted in the hp sample. The presence of gold has been highlighted through EDXS mapping as shown in micrographs (b2) and (e2) for the samples containing 1% gold, and through high-angle annular dark-field (HAADF) detector coupled with EDXS analysis of selected areas for the samples containing 3% gold. Gold is homogeneously distributed throughout the samples and it is present mainly as spherical nanoparticles ranging between 2 and 3 nanometers with some bigger aggregates up to 6 nanometers.

To get deep insight into the chemical composition of the catalysts, the samples were further characterized using XPS. Cerium region is very complex and the determination of the relative percentage of Ce(III) and Ce(IV) is usually done according to the initial classification by Burrough [20] with ten components (six for the Ce3d5/2–Ce3d3/2 of Ce(IV) and four for Ce3d5/2–Ce3d3/2 of Ce(III)). Anyway, the correct evaluation of the relative percentage of the two different oxidation states, suffer of the complexity of the region. For comparison reasons, it is possible to use the simplification proposed by Henderson [21]. In agreement with this method, the Ce(IV)% is estimated by calculating the attenuation of the u‴ component at 917 eV with respect to the total area of the Ce3d peak. According to their study in a pure CeO₂ oxide the u‴ component would be 14% of the total Ce3d area. The Ce(III) content in % is then calculated by using Equation (1):

Ce(III)% = 100 − Ce(IV)% = 100 − (100 * (u‴/14))

(1)

where u‴ is the area fraction of the peak at 917 eV [22,23].

Ce (III)–Ce(IV) relative amount was calculated applying this method (see Figure 6A). Moreover, in order to minimize cerium reduction by the beam, for all samples, Ce region was recorded with a fast modality (3 min each scan). The results obtained are compiled in

Table 2. XPS data of the two series of Au/Ceria catalysts in terms of Ce(III)/Ce(IV) percentages, Au 4f5/2 binding energy and Au/Ce atomic ratios. The values calculated by EDX and the nominal Au/Ce ratios are reported for comparison.

Sample	Ce(III) [%]	Ce(IV) [%]	Au 4f5/2 [eV] a	Au/Ce (XPS)	Au/Ce (EDX)	Au/Ce Nominal Value
1%Au/CeO2 A	14	86	84.3 (1.7)	0.05	0.014	0.010
3%Au/CeO2 A	13	87	84.8 (1.7)	0.18	0.032	0.030
1%Au/CeO2 hp	18	82	85.2 (2.3)	0.06	0.017	0.010
3%Au/CeO2 hp	15	85	85.6 (2.6)	0.11	0.033	0.030

^a In parenthesis the FWHM of Au 4f5/2 component.

. The samples supported on CeO₂ hp show a higher percentage of Ce(III) with respect to the samples prepared on CeO₂ A. Gold nature is sensitive to the support type. For CeO₂ A both samples show an Au 4f7/2 peak centered at 84.5 ± 0.3 eV typical of metallic gold, while the CeO₂ hp supported series exhibit for both composition the Au 4f7/2 peak centered at 85.4 ± 0.3 eV typical of Au⁺¹ species [24]. These results coupled with the increase of Ce(III) in CeO₂ hp supports point out a withdraw of electron from gold to CeO₂ in the hp series. Moreover, for the samples CeO₂ hp the FWHM of gold peak is quite high indicating the presence of several slightly different chemical environments for gold. For both series, the increase of the amount of gold caused a decrease of cerium reduction probably due to the formation of bigger particles, as seen by XRD, which lessen the influence of gold on ceria.

Table 2 reports the XPS-derived Au/Ce atomic ratio along with the above commented EDX atomic ratio. As expected by the surface nature of XPS technique, the XPS-derived ratio are higher than the EDX derived ratio by a factor of ca. 3 for all samples except than 3%Au/CeO₂ A where the Au/Ce by XPS is 5.6 higher than the one by EDX. This fact is probably due to the smaller pore diameter of CeO₂ A. which causes a minor diffusion of gold precursors inside the pores with a more pronounced surface deposition emphasized by XPS analysis. On the other hand, it is likely that Au⁺¹ species tend to diffuse into the bulk of CeO₂ hp oxides being less detectable on the surface.

Diffuse reflectance spectroscopy (DRS) has been employed to study the optical properties of the samples. Figure 7 shows the DRS of the CeO₂ and Au/CeO₂, along with their absorption of UV light. The DRS of the CeO₂ and Au/CeO₂ samples, depicted in Figure 7A evidenced that both CeO₂ showed only one absorption edge around 400 nm corresponding to the band gap energy of the semiconductor, whereas for the Au supported materials, the visible light absorbance was evident, as revealed in Figure 7B corresponding to the Kubelka-Munk function F(R_∞) of the diffuse reflectance spectra, which can be assumed as the absorption of the samples. The plasmonic absorption band of the Au is observed at 575 and 560 nm for the Au/CeO₂ A and Au/CeO₂ hp samples, respectively. The SPR absorption becomes more intensive as the Au loading increases from 1% to 3 wt.%. The increased visible light absorption from Au SPR should help enhance the harvesting of solar energy and promote related photocatalytic processes.

Ke et al. supported Au nanoparticles on different oxides, including CeO₂ and TiO₂, among others, to test these photocatalysts under visible light for the reduction of ketones to alcohols. They reported the DRS spectra of the powders and comparing Au/CeO₂ with Au/TiO₂, the LSPR band of Au/CeO₂ resulted red shifted and a more intense peak was recorded for the one of Au/TiO₂. They suggested a stronger LSPR effect of Au/CeO₂, explaining the stronger absorption light of Au/CeO₂ by a possible strong interface action between Au nanoparticles CeO₂ that those existing in the Au/TiO₂ powder [25]. The red-shift of the LSPR band for the present samples, with respect to those observed by Ke (ca. 520 nm), could indicate a strong interaction with the CeO₂ support, particularly in the case of the CeO₂ A samples.

To determine the optical band gap energy of the materials, the Kubelka-Munk function F(R_∞) of the diffuse reflectance spectra has been used and the band gap value was estimated by extrapolating a linear fitting in the Tauc plot [26] (i.e., the plot of (F(R_∞)·hν)^1/2 vs. the incident light energy in eV) by considering the CeO₂ as an indirect semiconductor, as shown in Figure 7C,D. The presence of Au on the surface of CeO₂ does not change the E_gap nevertheless it affords absorbance in the visible region, so that both UV and visible radiation can contribute to the photoactivity of the powders [27]. In addition, the photoactivity can increase via the delay of the recombination rate of photogenerated electron and hole pairs [28].

2.2. Photocatalytic Activity under UV or under Visible Light Irradiation

All the Au/CeO₂ powders showed to be photoactive under UV and, as well, under Vis light irradiation for the photocatalytic partial oxidation of both alcohols. The results obtained for the selective oxidation of benzyl alcohol (BA) and 4-methoxy benzyl alcohol (4-MBA) in water, in terms of alcohol conversion and selectivity to the corresponding aldehyde after 4 h of reaction, are reported in

Table 3. Conversion percentage (X) of BA and 4-MBA and selectivity percentage (S) towards the corresponding aldehyde after 4 h of UV or visible (VIS) irradiation.

Catalyst	BA				4-MBA
	UV		VIS		UV		VIS
	X	S	X	S	X	S	X	S
CeO2 A	-	-	-	-	11	25	8	7
1%Au/CeO2 A	2	100	5	100	20	72	22	98
3%Au/CeO2 A	4	100	6	100	35	78	44	95
CeO2 hp	-	-	-	-	5	18	-	-
1%Au/CeO2 hp	2	100	4	100	6	19	7	40
3%Au/CeO2 hp	2	100	5	100	13	75	15	100

.

The perusal of Table 3 shows that by irradiating the system containing a benzyl alcohol suspension with both UV or visible light, the CeO₂ pristine powders resulted completely inactive, whereas a modest conversion of BA was observed in the presence of Au-loaded CeO₂ samples showing a selectivity to benzaldehyde of 100%. As a general consideration, the conversion values were, for any photocatalyst, nearly the same by irradiating with UV light or under visible irradiation, albeit for the latter the presence of Au seemed to be slightly beneficial.

The conversion of 4-MBA was much higher than that obtained for BA and the selectivity to 4-MBAL was remarkable only in the case of the samples loaded with Au. Under UV irradiation, both CeO₂ pristine powders gave rise to a certain conversion, slightly higher for the commercial powder, as reported in Table 3. Interestingly, only the CeO₂ A sample gave rise also conversion of 4-MBA under visible light irradiation. The Au/CeO₂ samples lead to a higher conversion of the alcohol than the pristine CeO₂, and the conversion increased by increasing the Au amount. The Au supported samples resulted particularly active by using the CeO₂ A as support. These samples showed higher conversion of 4-MBA under visible irradiation than under UV light. The most active sample in terms of conversion resulted to be the 3%Au/CeO₂ A, particularly under visible light. Selectivity to 4-MBAL under UV irradiation resulted 25% and 18% for the pristine CeO₂ A and hp, respectively. The presence of Au on both of the supports increased not only the conversion but also the selectivity to the aldehyde and in general the activity increased by increasing the Au content. The most active sample in terms of conversion and selectivity resulted the 3%Au/CeO₂ A, followed by the 3%Au/CeO₂ hp. These insights evidence that the photocatalysts based on the CeO₂ hp are in general less oxidant than the analogous based on the commercial CeO₂ A. This behavior is the same than that observed for the BA partial oxidation, albeit BA resulted by far more difficult to be oxidized than 4-MBA. The Au-loaded CeO₂ materials can be used under visible light irradiation for 4-MBA partial oxidation quite successfully. Furthermore, these materials seem more effective under visible light irradiation with respect to UV light.

The mechanism of the photo-oxidation in the presence of Au nanoparticles is distinctly different from that of the semiconductor photocatalysts. Indeed, in a photocatalytic process when a semiconductor is irradiated by suitable radiation, electrons are promoted from the valence band to the conduction band. Consequently, the generated electron/hole couples are responsible of oxidation and reduction processes. The photogenerated electrons reduce the adsorbed oxygen (O₂), if the process is carried out in air, to form superoxide radicals (O₂^−•) and the photogenerated holes react with H₂O to form hydroxyl radicals (OH^•), generally responsible for the oxidation of substrates resulting in their mineralization and/or partial oxidation. On the contrary, irradiated Au-NPs are able to absorb visible light due to the surface plasmon resonance effect. This effect is due to the joint oscillation of conduction electrons in the gold NPs which, resonating with the electromagnetic field of the incident light, give rise to a significant enhancement of the local electromagnetic fields near the surfaces of the Au-NPs. Consequently, the conducting electrons on the NP surfaces increase their energy content and they can interact with the O₂ molecules adsorbed on the surface of Au-NP or even on the support surface to form superoxide radicals (O₂^−•). Furthermore, the excited electrons may relax back to their equilibrium states and release heat to the Au-NP lattice, resulting in a rapid and localized heating of the Au-NPs [25]. Such two effects can induce together chemical reactions of adsorbed molecules both on the Au-NPs and on the support. Furthermore, the localized heating of the Au-NPs can favorite the desorption of the reaction intermediate. In the case of the reactions reported in this study, the desorption of the formed aldehydes is of paramountcy importance to avoid their successive oxidation to benzoic acids that are responsible for the blockage and consequently the inactivation of the catalytic sites [29].

To explain the higher conversion of 4-MBA with respect to BA by using the same photocatalyst and experimental conditions, it is necessary to remind that the aromatic alcohols with electron donating substituent groups (EDG), as -O-CH₃ in different positions, show different photocatalytic activity and selectivity, in the presence of TiO₂ [30], but also by using C₃N₄ based photocatalysts [31,32]. Indeed, it is already known that the conversion of aromatic alcohols depends on the type and position of the substituent in the aromatic ring. In particular, the methoxy group in para position of the benzyl alcohol, increased both conversion and selectivity to the aldehyde. This because, the methoxy group, as Electron Donating Group (EDG), is an ortho-para orienting group, and its presence in the para position induces the attack by oxidant species to the benzyl group, thus favoring the alcohol-to-aldehyde transformation. The result obtained by using TiO₂ and C₃N₄ as photocatalysts under UV was also observed during the present investigation in the presence of CeO₂ and Au/CeO₂ photocatalysts both under UV and visible irradiation. The adsorption of 4-MBA by means of the alcoholic group on the surface of the photocatalyst and the inductive and delocalization effects caused by the –OCH₃ group on the aromatic ring, hinder the strong oxidizing attacks that can cause the mineralization of the molecule [31].

The photocatalytic alcohol degradation rate obtained by using all the samples are reported in

Table 4. Photoactivity results in terms of degradation rate of benzyl alcohol (r_BA) and 4-methoxybenzyl alcohol (r_4-MBA) expressed in [mM·min⁻¹], both under UV or visible light irradiation.

	UV	Vis	UV	Vis
Photocatalyst	rBA × 105	rBA × 105	r4-MBA × 105	r4-MBA × 105
CeO2 A	-	-	23	15
1%Au/CeO2 A	8	11	37	42
3%Au/CeO2 A	11	20	69	84
CeO2 hp	-	-	9	-
1%Au/CeO2 hp	8	9	12	15
3%Au/CeO2 hp	8	11	26	28

. The results are in agreement with the conversion data reported in Table 3.

Bare CeO₂ does not produce any oxidation of BA, whereas 4-MBA was partially oxidized, faster in the presence of CeO₂ A than with CeO₂ hp. Notably, despite the band gap values of the CeO₂ samples, ca. 2.9 and 3.0 eV (see Figure 7) for the home prepared and Aldrich, respectively, the presence of the Au allowed the reaction to proceed with higher rates under visible irradiation than under UV one. This result is attributed to the surface plasmon resonance effect. The faster reaction was observed in the presence of 3%Au/CeO₂ A.

The reaction proceeds under both UV and visible light irradiation, by the two distinct mechanisms already discussed above. For that reason, the best photocatalysts (i.e., those containing 3% of Au) were tested for the 4-MBA partial oxidation also under natural solar irradiation supplying contemporaneously both UV and visible light.

2.3. Photocatalytic Activity under Natural Solar Light Irradiation

The evolution of 4-MBA concentration along with that of 4-MBAL and 4-methoxy benzoic acid (4-MBAcid) during photocatalytic experiments carried out under solar irradiation in the presence of the two 3% Au-loaded CeO₂ samples are reported in Figure 8. Both experiments were carried out contemporaneously, so the two reactors used received the same number of photons. During the first 30 min of the experiments, conducted in dark conditions, both catalysts adsorbed ca. 20% of the initial 4-MBA. This adsorption was slightly more evident (ca. 24%) in the case of 3%Au/CeO₂ A. In Figure 8 it is observed that, for both materials, during the first 30 min of irradiation the amount of aldehyde formed was greater than the amount of alcohol disappeared. This fact indicates that during the first steps of the reaction the amount of adsorbed alcohol reacting to give the aldehyde is not completely replaced by the alcohol present in the bulk of the suspension. This suggests that the amount of alcohol consistently adsorbed on the catalyst surface under irradiation is lower than that adsorbed in dark conditions. Furthermore, it cannot be excluded that part of the aldehyde formed remained adsorbed on the catalyst surface to be furtherly oxidized to 4-methoxybenzoic acid.

From the results reported in Figure 8, it was possible to calculate the conversion of 4-MBA and the selectivity versus the formation of 4-MBAL and 4-MBAcid during the evolution of the reaction. These results are reported in Figure 9.

As reported in Figure 9 for both 3% Au-loaded CeO₂ samples, the selectivity towards 4-MBAL was always in the range 95–98% (the highest values were observed for 3%Au/CeO₂ A) and it seems independent to the percentage of 4-MBA converted. The selectivity towards the formation of 4-methoxybenzoic acid was close to 5–2% (the highest values were observed for 3%Au/CeO₂ hp), indicating that no other species were formed during the 4-MBA partial oxidation.

In order to compare the tests carried out under natural sunlight irradiation with those conducted under UV or Vis light, the evolution of 4-MBA, 4-MBAL and 4-MBAcid concentration were also reported versus the cumulative energy entering in the reacting system per unit of volume of suspension. The results obtained under solar light irradiation are reported in Figure 10.

Figure 8 and Figure 10 are representative of the same runs, consequently after 4 h of irradiation (240 min) the cumulative energy entering the reactor per unit of volume of suspension was ca. 340 kJ·L⁻¹. It is important to consider that ca. 6% of the measured solar irradiation (20 kJ·L⁻¹) was in the UV (315–400 nm) and the remaining part (320 kJ·L⁻¹) in the visible range (450–560 nm). As far as the runs carried out under UV irradiation are concerned, the cumulative energy entering the reactor per unit of volume of suspension after 4 h was ca. 5 kJ·L⁻¹, then it was much lower if compared with the amount entering under solar light irradiation. On the contrary, in the case of the tests conducted under visible light irradiation, the cumulative energy entering the reactor per unit of volume of suspension after 4 h was ca. 540 kJ·L⁻¹ (i.e., ca. 220 kJ·L⁻¹) more with respect to that entering during the solar tests. In

Table 5. Conversion percentage of 4-MBA and selectivity percentage towards 4-MBAL after 4 h of UV, visible or natural solar irradiation for runs carried out in the presence of 3%Au/CeO₂ A and 3%Au/CeO₂ hp as photocatalysts.

Irradiation Source	Cumulative Energy per Unit of Volume [kJ·L−1]	Photocatalyst	Conversion of 4-MBA [%]	Selectivity to 4-MBAL [%]
UV light	5 a	3%Au/CeO2 A	35	78
		3%Au/CeO2 hp	13	75
Visible light	540 b	3%Au/CeO2 A	44	95
		3%Au/CeO2 hp	15	100
Natural solar light	20 a 320 b	3%Au/CeO2 A	37	98
		3%Au/CeO2 hp	32	96

^a Value measured in the range 315–400 nm; ^b value measured in the range 450–560 nm.

are reported the results obtained in the presence of the two photocatalysts loaded with 3% of gold after the runs carried out by irradiating the photoreactor with UV, visible or natural solar light.

From a perusal of Table 5 it can be concluded that the 3%Au/CeO₂ A photocatalyst was always the most active in terms of alcohol conversion. In particular, under both UV and visible irradiation the activity of the 3%Au/CeO₂ A sample was substantially higher than that obtained with the 3%Au/CeO₂ hp photocatalyst. On the contrary, when the reaction was carried out under natural solar light the activity of the two catalysts was similar even if the sample based on CeO₂ A showed to be always the most active. On the basis of the irradiation source reaching the system in the three cases, it is possible to observe that the two photocatalysts are active both under UV and visible light but the 3%Au/CeO₂ A sample is able to better exploit both UV and visible light irradiation when used separately. On the other hand, the 3%Au/CeO₂ hp photocatalyst is not very active when irradiated with UV or visible light alone. Conversely, when the latter catalyst is irradiated with sunlight containing both UV and visible, a considerable increase of activity occurs. This synergistic effect was not observed in the case of the other catalyst and consequently the activity of the two samples appeared similar.

To explain the different behavior of the two catalysts under the various types of irradiation, it is necessary to consider various factors. In particular, the porosity and the presence of metallic gold on the surface of the materials seem to be the most important ones, while the specific surface area does not seem to play an important role.

Let us now understand what happens when the catalysts are irradiated with the different sources of radiation. From the results shown in Table 5 it can be seen that the 3% Au/CeO₂ hp catalyst is less active under UV irradiation than the other one. This fact can be justified by considering that the 3% Au/CeO₂ hp material possess a greater amount of mesopores with respect the 3% Au/CeO₂ A sample. Mesopores that results also smaller in comparison with those showed by the latter catalyst. Consequently, the 4-MBAL formed during the early stages of irradiation could in part be retained in the pores to be further oxidized to 4-MBAcid which blocking [29] the catalytic sites strongly reduce the activity of the catalyst. This effect is observed, although to a lower extent, also in the case of the 3% Au/CeO₂ A catalyst; in fact, even in this case the selectivity towards 4-MBAL does not reach 100%, indicating the formation of 4-MBAcid which remains adsorbed on the catalyst. Interestingly, under UV irradiation no 4-MBAcid was observed in solution.

Under visible irradiation, the plasmonic effect, causing strong local heating, favors the 4-MBAL desorption, avoiding its subsequent oxidation to 4-MBAcid. This fact is evidenced by the high selectivity towards 4-MBAL (ca. 100%) observed under visible light irradiation with both catalysts. The fact that the 3% Au/CeO₂ hp catalyst was still not very active also under visible irradiation can be justified considering that the gold deposited in this catalyst is essentially in ionic form (see XPS study) and consequently the plasmonic effect responsible for the photoactivity was less marked.

Furthermore, the lower gold-support interaction in the case of the 3% Au/CeO₂ hp catalyst, as evidenced by diffuse reflectance investigation, can contribute to the low activity of this material. Under natural solar light irradiation it was observed that the activity of the 3% Au/CeO₂ hp catalyst increased. This fact can be related to the UV energy density entering the system that was 20 kJ·L⁻¹ (i.e., four times higher than that measured during the test carried out under UV lamp irradiation (5 kJ·L⁻¹)). Consequently, it is possible to assume that the 3% Au/CeO₂ hp catalyst is essentially activated by UV; indeed, under solar irradiation the amount of visible light reaching the reactor (320 kJ·L⁻¹) was lower with respect to the test conducted by irradiating the system by using visible LED (540 kJ·L⁻¹). However, the presence of visible light was still beneficial because the plasmonic effect help the reaction by favoring the desorption of the 4-MBAL; indeed, for these runs the selectivity towards the formation of aldehyde was always close to 100% (see Table 5). Resuming for the 3% Au/CeO₂ hp catalyst, the enhanced photocatalytic activities can be attributed to the synergistic effect between the Au nanocrystal acting as the plasmonic component for efficiently harvesting the light and the CeO₂ support providing catalytically active sites for the oxidation reaction. On the other hand, in the case of the 3% Au/CeO₂ A catalyst, which was almost active even under irradiation with visible light, the additional UV contribution of solar radiation did not significantly increase the activity of the catalyst.

It is interesting to compare the conversion, selectivity and reaction rate obtained with the photocatalysts in this study with respect to those reported in literature for the same reaction.

Li et al. study the partial oxidation of benzyl alcohol with a 100% selectivity to benzaldehyde in the presence of 0.25–0.5 wt.% Au/CeO₂ samples using acetonitrile as the solvent and the irradiation of λ > 420 nm [15]. They reported a maximum conversion rate of ca. 2250 and ca. 390 μmol·h⁻¹·g⁻¹ under UV–Vis or only visible irradiation, respectively. These results are better than those obtained in our work (ca. 36 μmol·h⁻¹·g⁻¹ under visible light) but they are not comparable because we have used water as solvent. Moreover, the UV and visible photon flux emitted by the Xe lamp used by Li et al. were much higher with respect to our light source. Kominami et al. studied the photocatalytic oxidation of benzyl alcohol in aqueous suspensions of Au/CeO₂ samples under irradiation by green light from an LED. They prepared Au/CeO₂ nanocomposites by two photodeposition methods, both exhibiting LSPR absorption at around 550 nm matching with the wavelength of the green LED light used in their research [16]. They conclude that the size of the nanoparticles affected the intensity of the photoabsorption and hence the photocatalytic activity as previously observed in plasmonic Au/TiO₂/photocatalysts [33]. The authors have shown that the benzyl alcohol was completely consumed after 15 to 20 h depending upon de material and experimental conditions. Benzaldehyde was formed from benzyl alcohol partial photo-oxidation with a >99% selectivity. The rates of benzaldehyde formation were determined to be in the range 1.9 to 3.0 μmol h⁻¹. These results were very close to those obtained in this work using the visible LED (in Table 4, 20·10⁻⁵ mM min⁻¹ correspond to 1.8 μmol h⁻¹). Cui et al. studied the selective photocatalytic conversion of alcohols (benzyl alcohol, 4-methoxybenzyl alcohol) using Au/CeO₂ composites with different Au loadings under visible light irradiation (λ > 400 nm, light intensity of 0.7 W cm⁻² by using benzotrifluoride as solvent [34]. The conversion achieved a 100% in the presence of Au 4 wt.%/CeO₂ after 4 h of reaction and the selectivity to benzaldehyde was ca. 95%. Wolski et al. prepared three CeO₂ samples using different co-precipitation agents (urea, hexamethylenetetramine or NaOH) and Au-loaded photocatalysts. The catalytic study was carried out in dark condition at 40 °C (thermal catalysis) and evidenced that the most active catalysts (highest conversion of benzyl alcohol) were those containing gold and in particular the one prepared by using NaOH as co-precipitation agent [19]. The results obtained by Wolski et al. are not comparable with those reported in this work but indicate that, in order to avoid the over oxidation of benzyl alcohol to benzoic acid, it is better to carry out the experiments at room temperature, as in our case. By using the best catalyst they obtained benzoic acid. With a conversion after 40 min of 47% with a selectivity to benzoic acid of 91%, the benzaldehyde was obtained with a 9% of selectivity. Furthermore, in our opinion, the use of a basic solution for the experiments could significantly influence the results.

3. Materials and Methods

3.1. Preparation of Au/CeO₂ Photocatalysts

All the used reagents (Sigma-Aldrich) were of analytical grade with purity ≥99.5%. The home-prepared (hp) cerium oxide was synthesized via hydrothermal method. In details, for a typical preparation, 6 g of Ce(NO₃)₃·6H₂O were dissolved in 30 mL of H₂O then, 40 mL of NaOH solution (2.5 mol·L⁻¹) were added under vigorous stirring. The resulting slurry was rapidly transferred into a 100 mL Teflon autoclave and heated in oven at 120 °C for 23 h. The aged precipitate was collected by filtration, washed with abundant deionized water until neutrality and finally with ethanol. The powder was dried overnight at 60 °C and calcined at 350 °C for 3 h (heating rate 2 °C·min⁻¹). This sample was denoted CeO₂ hp. For comparison purpose, a commercial cerium oxide was purchased from Sigma-Aldrich and herein is labelled as CeO₂ A.

Supported gold catalysts were prepared by deposition-precipitation method. The support (CeO₂ A or CeO₂ hp) was preliminary suspended in water, then, gold was deposited by adding a solution of HAuCl₄·3H₂O (0.01 mol·L⁻¹) corresponding to 1 or 3 wt.% of Au in the catalyst. After stirring at room temperature for 1 h in order to favor a preliminary interaction between the Au precursor and the hydroxyl groups of the support, the pH of the solution was set to ~7.5–8.0 by adding K₂CO₃ (0.05 mol·L⁻¹) and the temperature was increased to 65 °C. The suspension was stirred overnight at this temperature. After filtering and careful washing with hot deionized water, the catalyst was dried overnight at 100 °C, and then calcined at 350 °C for 2 h (heating rate 2 °C·min⁻¹). The resulting catalysts were labelled as 1% and 3%Au/CeO₂ A or 1% and 3%Au/CeO₂ hp.

3.2. Characterization of the Materials

The crystalline structure of the samples was determined by powder X-ray diffraction patterns (XRD), performed on a Bruker D5000 diffractometer equipped with a Kristalloflex 760 X-ray generator (Bruker AXS GmbH, Karlsruhe, Germany) and with a curved graphite monochromator using Cu Kα radiation (40 kV/30 mA). The data were recorded in a 2θ range of 20°–80° with a step size of 0.05° and time per step of 20 s. The crystalline phases of samples were analyzed according to ICSD (Inorganic Crystal Structure Database) files. The mean crystallite size was calculated by the Debye–Scherrer equation: D = 0.9λ/Bcosθ, where D represents the average crystalline size, 0.9 is the Scherrer parameter, λ is the wavelength of the X-ray radiation (0.15406 nm), B denotes the full width at half maximum of the peak (FWHM), and θ is the angular position of the main peak, the (111) for CeO₂ and Au phases, as well.

The Specific surface area (SSA), pore volume and mean pore diameter of the materials were measured by N₂ adsorption−desorption isotherms using a Micromeritics ASAP 2020 apparatus. Before analysis, the samples were degassed in vacuum at 250 °C for 2 h, and then the measurements were performed at liquid nitrogen temperature (−196 °C). The Brunauer−Emmett−Teller (BET) method was used to calculate the SSA. The Barrett−Joyner−Halenda (BJH) method was applied to the desorption branch to estimate the mesoporous pore volume and the average pore diameter (in the range 2–50 nm). The total pore volume was measured as single desorption point, at p/p₀ = 0.95. The micropores volume was calculated by t-Plot.

Scanning electron microscopy (SEM) was performed by using a FEI Quanta 200 ESEM microscope, operating at 20 kV on specimens upon which a thin layer of gold had been evaporated. On the other hand, an electron microprobe used in an energy dispersive mode (EDX) was employed to obtain information on the actual Au and Ce content present in the samples.

TEM (transmission electron microscopy) analysis was performed with an S/TEM ThermoFisher Talos F200S operating at 200 kV. The microscope is equipped with an integrated EDS (Energy Dispersive X-rays Spectroscopy) system with two windowless silicon drift detectors (SDD). The samples were observed in both TEM and STEM mode by using bright field (BF) and high-angle annular dark-field (HAADF) detectors. Moreover, the STEM mode allows evaluating the actual elemental distribution in the sample collecting EDS maps. Samples were dispersed in ethanol and a 50 µL drop of the dispersion were deposited on TEM copper support grids covered by an amorphous carbon film. The X-ray photoelectron spectroscopy (XPS) analyses of the samples were performed with a VG Microtech ESCA 3000 Multilab (VG Scientific, Sussex, UK), using Al Kα source (1486.6 eV) run at 14 kV and 15 mA, and CAE analyzer mode. For the individual peak energy regions, a pass energy of 20 eV set across the hemispheres was used. The constant charging of the samples was removed by referencing all the energies to the C 1s peak energy set at 285.1 eV, arising from adventitious carbon. Analyses of the peaks were performed using the CASA XPS software (version 2.3.17, Casa Software Ltd. Wilmslow, Cheshire, UK, 2009). For the peak shape a Gaussian (70%)-Lorentzian (30%) line shape, defined in Casa XPS as GL(30) profiles were used for each component of the main peaks after a Shirley type baseline subtraction. The binding energy values were quoted with a precision of ±0.15 eV, and the atomic percentage with a precision of ±10%.

UV–visible diffused reflectance spectra (DRS) of the samples were obtained for the dry-pressed film samples using a UV–visible spectrophotometer (UV-2550, Shimadzu, Japan). BaSO₄ was used as a reflectance standard in a UV–visible diffuse reflectance experiment.

3.3. Photocatalytic Activity Tests

The photocatalytic experiments were carried out by using three different set-ups. In the first one, a Pyrex cylindrical photoreactor (internal diameter: 32 mm, height: 188 mm) containing 150 mL of aqueous suspension was used. The photoreactor was irradiated externally by three Actinic BL TL MINI 15 W/10 Philips fluorescent lamps located at 3 cm distance from the reactor axis. The lamps emitted in the 340–420 wavelength range with the main emission peak at 365 nm. The impinging radiation energy in the range 315–400 nm was measured by a radiometer Delta Ohm DO9721 with an UVA probe and its average value was 2.7 W m⁻². In the second set-up, a Pyrex batch photoreactor of cylindrical shape containing 150 mL of aqueous suspension was used. 120 W visible LED was positioned externally in a co-axial position surrounding the photoreactor. The average radiation energy impinging the reactor was ca. 300 W·m⁻² in the 450–560 nm range. The emission in 315–400 nm range was null. In both the set-ups all the photons emitted by the lamps completely reached the suspension.

During the experiments, both reactors were open, and the equilibrium between dissolved O₂ in the aqueous suspension and in the atmosphere was achieved. The reaction was carried out at ca. 25 °C as the reactors were provided by a thimble where water was allowed to circulate. Benzyl alcohol (BA) and 4-hydroxy benzyl alcohol (4-MBA) were used as the substrates, and their initial concentration in water was 0.5 mM at natural pH.

Before starting the irradiation, 50 mg of catalyst (0.33 g L⁻¹) were added to the solution and the obtained suspension was kept in an ultrasonic bath for 10 min and stirred under dark for 30 min, to attain the adsorption equilibrium. Throughout the reaction, samples of the irradiated suspension were withdrawn every 30 min and filtered through 0.25 μm membranes (HA, Millipore) to separate the photocatalyst particles before the HPLC analyses. All the aromatic molecules (i.e., BA and 4-MBA) but also the corresponding aldehydes (i.e., benzaldehyde (BAL) and 4-methoxybenzaldehyde (4-MBAL)), along with the corresponding carboxylic acids, were analyzed by a Beckman coulter HPLC apparatus equipped with a Diode Array detector and a Phenomenex KINETEK 5 μm C18 instead. The eluent (0.8 mL min⁻¹) consisted of a mixture of acetonitrile and 13 mM trifluoroacetic acid (20:80 v:v). Standards purchased from Sigma-Aldrich with a purity >99% were used to identify the products and to obtain the calibration curves.

Further experiments, by using the natural solar light irradiation, were carried out on clear sunny days in Palermo (Italy) from 9:30 to 13:30. Typically, 75 mL of 0.5 mM 4MBA solution and 25 mg of photocatalyst were introduced inside a round-shaped Pyrex batch reactor having a total volume of 125 mL and a diameter of 10 cm. The reactor was closed and no gases were fed during the tests as preliminary experiments indicated that O₂ deriving from air and present in the system was enough to induce the oxidation reaction. The suspension was continuously magnetically stirred and approximately 2.5 mL were withdrawn every 30 min and analyzed by using the previously described analytical procedure. In order to assess the cumulative photon energy entering in the photoreactor, the photon flux reaching the reacting suspension in the range 315–560 nm was measured every 10 min throughout the photocatalytic tests by using the same radiometer above described.

4. Conclusions

In this work, two type of CeO₂ (i.e., a commercial one and a home prepared one) were used as support of Au nanoparticles at two different percentages of metal (1 and 3 wt. %). All materials were investigated as photocatalysts for the photocatalytic selective oxidation of benzyl alcohol and 4-methoxybenzyl alcohol to the correspondent benzaldehydes, in aqueous suspensions and room conditions under UV, visible and natural solar light irradiation. Bare CeO₂ samples resulted active under UV light and virtually inactive under visible light irradiation. The presence of Au improved both the conversion of the alcohols and the selectivity of the reaction towards the aldehyde, showing good activity, particularly under visible and natural solar light irradiation. This fact was attributed to the presence of the Au nanoparticles that caused a strong visible radiation absorption at 565–570 nm associated to the localized surface plasmon resonance (LSPR) of this metal. Moreover, the photo-induced activity of the materials increased by increasing the Au content.

The activity of Au/CeO₂ plasmonic photocatalysts prepared in this work was similar to those reported in previous studies, but here it was shown the possibility to successfully carry out the partial oxidation of benzyl alcohols in water solution under natural sunlight, reaching high conversion after only 4 h of irradiation (ca. 37% in the case of 3%Au/CeO₂ A sample), and in particular high selectivity versus the aldehyde production (ca. 98%). This fact was explained by considering the ability of these materials to exploiting the synergistic effect between the Au nanoparticles acting as plasmonic component for efficiently harvesting the visible light and the CeO₂, that provides photo-active sites for the oxidation reaction under UV light irradiation.