Synthesis of Cu–M_xO_y/Al₂O₃ (M = Fe, Zn, W or Sb) catalysts for the conversion of glycerol to acetol: effect of texture and acidity of the supports

Abstract

The purpose of this work is to study the relationships between the copper oxide structure, specific surface area, acidity of M_xO_y/Al₂O₃ (M = Fe, Zn, W and Sb) supports and subsequently the catalytic proprieties of the solids. The samples were characterized by XRD, N₂ adsorption/desorption isotherms and microcalorimetry of NH₃ adsorption. The XRD results (after and before reaction) of the Fe and Zn-containing solids present the formation of a copper-modified alumina structure (high dispersion of Cu species with a strong interaction with the support); however, it is not observed a copper-modified phase for the catalysts composed of W and Sb, since it was identified the isolated CuO phase (a lower dispersion of Cu with a poor interaction with the support). N₂ adsorption/desorption isotherms analysis shows that the copper-modified alumina samples are mesoporous materials and have a high surface area compared to the other catalysts, which did not presented a copper-modified phase. The catalytic tests ascribed that the presence of a copper-modified structure improves the activity, selectivity and mainly the stability for conversion of glycerol to acetol. W and Sb-based samples present low stability in glycerol transformation to acetol probably due to its acidity leading to formation of coke, blocking the active sites as a result of the low number of sites per area (low surface area) as well as the sintering of the copper phase. Thus, the results explain that the choice of metal elements in the solids composition affects its structural, textural, acidity proprieties and consequently the catalytic performance.

---

Introduction

Copper-based catalysts are well known as very active heterogeneous catalysts for various reactions, such as steam reforming of methanol,¹ CO oxidation² and dehydrogenation reactions.³ Particularly, copper-containing catalysts are being studied for the hydrogenolysis, dehydrogenation and dehydration of glycerol.^4–6 Glycerol is a major byproduct in the biodiesel manufacturing process. The declining price of glycerol has sparked much interest in transforming crude glycerol through catalytic processes.^7,8 Specifically, the catalytic synthesis of acetol from glycerol has received great attention, since it is the product of great commercial value, which is used as reduced dyes, food industry (give aroma to foods), flavor compounds in heated milk, skin tanning agent; it is also applied to produce chemicals such as propylene glycol, propionaldehyde, acetone and furan derivatives (quite valuable chemical intermediate).^9–11 In addition, acetol is an interesting molecule from an astrophysical perspective.^12,13

There are many cases using the acidic catalysts in the chemical industries or for the production of fine chemicals.^14–16 Alcohols dehydration is an important catalytic test to identify acid in heterogeneous catalysts.¹⁷ Thus, a number of solid acid catalysts for the dehydration of glycerol (polyol) have been reported including metal sulphates and phosphates, zeolites, metal oxides (Al₂O₃, TiO₂, ZrO₂) and heteropoly acids.^18,19 Furthermore, Sato et al. showed that copper-based catalysts combined with an acidic support may promote the dehydration of glycerol to form acetol with interesting catalytic properties.⁷

In spite of this, these catalysts need to overcome various problems including harsh reactions conditions, rapid catalyst deactivation due to coke deposition and the formation of unrequired by-products as well as the thermal catalytic sintering.^20,21 Therefore, the development of a catalyst resistant to deactivation during the conversion of glycerol mainly in gas phase is one of the most interesting research challenges.

The strong metal-support interactions (SMSI) between Cu phases and surface oxygen defects are supposed to have a positive effect on catalyst proprieties.²² CuAl₂O₄ and copper-modified alumina is formed by solid–solid interaction between CuO and Al₂O₃ and the addition of small amounts of a certain foreign oxide such as ZnO and CeO₂ to the system has found to positively influence in the solid–solid interaction between CuO and Al₂O₃. These effects may also affect in the catalytic performance.^23,24 Recently, Cu–ZnO-based catalysts were found to be extremely efficient for catalytic hydrogenolysis of glycerol. Additionally, high catalyst stability during the transformation of glycerol to propylene glycol was obtained over Cu–ZnO–Al₂O₃ based samples.^25,26 It is important to emphasize that the copper-modified alumina due to the interaction between CuO and Al₂O₃ and the effect of the addition a certain foreign oxide such as Fe₂O₃, ZnO, WO₃ and Sb₂O₅ on the copper oxide structure and on the textural properties of the supports and consequently on the gas-phase conversion is rarely reported in the literature, requiring further investigation.

Our objective is to investigate the effects of copper oxide structure as well as the texture and acidity of M_xO_y/Al₂O₃ (M = Fe, Zn, W and Sb) supports on the gas-phase conversion of glycerol to acetol.

Experimental section

Catalyst preparation

Polymeric precursor method. For 10FeAl catalyst, aluminum nitrate nonahydrate {Al(NO₃)₃·9H₂O}, iron nitrate nonahydrate {Fe(NO₃)₃·9H₂O}, citric acid monohydrate (CA) {C₆H₈O₇·H₂O}, and ethylene glycol (EG) {C₂H₆O₂} were used as starting chemicals. A CA/metal ratio of 2 : 1 (mol) was used for all the samples. The metal amount is the sum of M (M = Fe, Zn, W or Sb) and Al. The mass ratio of CA/EG was kept at 2 : 3. The sample was labeled XMAl, with X denoting the M/Al weight ratio (10FeAl, 10ZnAl, 10SbAl and 10WAl).

For the synthesis process of the 10FeAl sample, where 0.07 is the Fe/Al weight ratio, 0.0075 and 0.11 mol of Fe and Al salts, respectively, were dissolved in distilled water at room temperature. CA (0.2269 mol) was dissolved in ethanol at 50 °C. Later the aqueous solution was added to the CA–ethanol solution and stirred during 60 min at 50 °C. Subsequently, the EG (0.3403 mol) was added, and the mixture was stirred for 4 h at 100 °C until it became a viscous resin. The resin was treated at 250 °C for 1 h under air. The resulting precursor composite was ground and treated at 500 °C under air flow for 1 h. The same conditions were used to prepare 10ZnAl, 10SbAl and 10WAl catalysts.

Wet impregnation method. Copper-based catalysts were prepared by wet impregnation from 10FeAl, 10ZnAl, 10SbAl and 10WAl samples synthesized by polymeric precursor method (as described above) with aqueous solutions using appropriate amounts of copper nitrate trihydrate {Cu(NO₃)₂·3H₂O} in a rotary evaporator. After impregnation, the samples were calcined at 500 °C in muffle furnace for 1 h (5 °C min⁻¹). The sample was denominated XCuMAl, which X representing the percentage by weight of copper (5CuFeAl, 5CuZnAl, 5CuSbAl and 5CuWAl).

Catalyst characterization. The crystal structure of the metal oxides was characterized by X-ray diffraction analysis (XRD), with CuKα irradiation source (λ = 1.540 Å, 40 kV and 40 mA). Specific surface area (BET) and pore volume of the catalysts were determined by N₂ adsorption/desorption isotherms, at liquid nitrogen temperature. The catalyst samples (80 mg) were vacuum treated at 200 °C for 1 h.

The acid properties were measured by NH₃ adsorption at 80 °C, using a TianCalvet calorimeter coupled with a volumetric equipment. The solids (0.1 g) were first vacuum treated at 400 °C for 2 h. After the pre-treatment, the cell was then introduced in the calorimeter at 80 °C and the experiment was performed. The oxide samples were then contacted with small doses of NH₃ up to equilibrium and the differential enthalpy of adsorption was recorded together with the amount of adsorbed ammonia.

Thermogravimetric analysis (TG) was carried out to estimate the amount of carbon deposited, using a 5 °C min⁻¹ heating rate, under an air flow of 40 mL min⁻¹ and 30 mg of sample.

Catalytic activity. Catalytic tests were carried out in a fixed bed microreactor, using 200 mg of sample. All fresh catalysts were firstly reduced in H₂ (25 mL min⁻¹) at 350 °C for 1 h. After the temperature of the catalyst bed had been maintained at 350 °C in hydrogen flow for 1 h, an aqueous solution of glycerol at 10 wt% was fed through the reactor top by a micro-pump at the feed rate of 0.03 cm³ min⁻¹, which corresponded to 0.001 mol of glycerol per minute, in H₂ flow at 25 mL min⁻¹. The catalysts were tested at atmospheric pressure. The liquid products were collected in an ice-water-salt trap (−6 °C) in order to be analyzed every hour on a gas chromatograph with a mass spectrometer and gfha gas chromatograph connected to a hydrogen flame ionization detector (FID-GC), using a capillary column purchased from Varian, CP 9048 (length: 30 m, diameter: 0.32 mm, thickness of the stationary phase: 0.1 mm). Carbon balances were between 80 and 96% in the catalytic experiments. Glycerol conversion and acetol selectivity were calculated in the following way:

Results and discussion

X-ray diffraction

The X-ray diffraction patterns are presented in Fig. 1. The crystalline phases were identified in comparison with ICDD files. The 10FeAl, 10ZnAl, 10SbAl and 10WAl samples, without copper, are poorly crystalline material and present X-ray patterns of an amorphous solid, Fig. 1a. In spite of this, 10WAl catalyst show a broader peak related to WO₃ phase formation (JCPDS 04-013-0859). The XRD-amorphous diffractograms probably indicate that the metal oxides structures are highly dispersed in the Al₂O₃ matrix, which are interesting features for catalytic supports.

Fig. 1 X-ray diffraction profile of the samples, after the calcination. (a) Samples without copper (b) Cu-based catalysts.

On the other hand, copper-containing solids prepared by wet impregnation method presented peaks related to the formation of crystalline phases, Fig. 1b. 5CuSbAl and 5CuWAl samples points to the formation of a copper oxide phase (CuO, JCPDS 04-045-0937), besides the presence of a broader peak, concerning the formation of tungsten oxide (WO₃, JCPDS 04-013-0859) for 10WAl catalyst.

Contrarily, the 5CuFeAl and 5CuZnAl samples also synthesized by wet impregnation route present broader peaks, which suggest the formation of a copper-modified alumina phase (CuO-γ-Al₂O₃, JCPDS 00-034-0425-Fe and JCPDS 00-001-1153-Zn), indicating that copper oxide is well dispersed in the support and probably interacting with its structure.

From these results one can notice that the choice of metal elements affect the structural proprieties of the samples, whereas the Zn and Fe-based catalysts stimulate the formation of a copper-modified phase (highly dispersed copper oxide) compared to the Sb and W-containing solids, which showed the formation of single CuO (copper oxide with low dispersion and less interaction with the support).

As mentioned above, the CuO-γ-Al₂O₃ (high dispersed copper oxide phase) is probably formed as a result of the thermal solid–solid reaction according to the following reaction: CuO + Al₂O₃ → CuO-γ-Al₂O₃, which indicates that CuO diffused onto the surface of defective spinel-type-γ-Al₂O₃. This indicate that CuO–A1₂O₃ interaction is promoted in the samples composed of Fe and Zn, however, it is not possible in the catalyst containing Sb and W. It is known that the addition of small amounts of other cations, such as Zn²⁺ may influence the chemical interaction between the CuO and Al₂O₃,^27–30 which may justify the results observed in Fig. 1. The close relationship of charge/radius ratio for Zn and Fe compared to Cu or Al may justify this interaction for the Zn and Fe-based catalysts. However, the relationship of charge/radius ratio for W and Sb retated to Cu or Al is not very close, which may explain the difficulty of interaction between species. Previous papers demonstrate that the copper oxide crystalline structure directly affects on the catalytic performance during the conversion of glycerol.^31,32 Therefore, in this paper, the transition phase was designated Cu-modified γ-Al₂O₃, which the copper is highly dispersed on γ-Al₂O₃, whereas it is not observed defined XRD peaks concerning the crystalline phase of copper oxide.

The average crystallite diameter was estimated from the diffraction data using the Scherrer formula, and the results are shown in Table 1. The average crystallites diameters of the copper-modified crystallites were found between 3.9 and 4.3 nm. Furthermore, the average crystallite size for CuO phase was found between 4.8 and 5.7 nm. The reflections are quite broad for the copper-modified peaks, indicating the formation of fine nanostructure.

Table 1 Average crystallite sizes of the Cu-containing solids after calcination process

Samples	Crystallite sizes (nm)
	CuO	CuO-γ-Al2O3
5CuFeAl	—	3.9
5CuSbAl	29.8	—
5CuWAl	26.3	—
5CuZnAl	—	4.3

---

The X-ray diffraction may be used to indirectly measure (qualitatively) the metal oxides dispersed throughout the base matrix from the diffractograms profile. Despite no quantitative characterization of the copper dispersion/surface area has been performed, XRD results is a good indication that the 5CuFeAl and 5CuZnAl samples synthesized by Pechini method is a successful method for preparing nanometer-sized containing copper-modified alumina structure with high metal dispersion and strong interaction copper support compared to 5CuWAl and 5CuSbAl solids, which probably will have very interesting catalytic properties, since the catalytic activity usually has a direct relationship with the dispersion of active sites.

Specific surface area measurements

The nitrogen adsorption/desorption isotherms of the calcined sample are shown in Fig. 2 and subsequently the textural properties obtained are presented in Table 2.

Fig. 2 N₂ adsorption/desorption isotherms of the catalysts; (a) samples without copper (b) Cu-based solids.

Table 2 Textural properties of the samples

Samples	SBET (m^2 g^−1)	Vp (cm^3 g^−1)	Dp (nm)
10FeAl	330	0.30	3.0
10ZnAl	244	0.20	4.1
10SbAl	245	0.16	4.2
10WAl	294	0.16	2.6
5CuFeAl	270	0.30	4.1
5CuZnAl	409	0.45	3.9
5CuSbAl	18	0.03	5.3
5CuWAl	32	0.04	4.1

---

Only the catalyst 10WAl exhibits a classic type I isotherm characteristic of microporous materials according to the IUPAC classification. Contrarily, the other solids (10FeAl, 10ZnAl, 10SbAl, 5CuFeAl, 5CuZnAl, 5CuSbAl and 5CuWAl) present a typical type IV curve with H2 hysteresis loops, also characteristic of mesoporous materials. These observations are confirmed by the pore diameter distribution.

Table 2 lists the surface area (S_BET), total pore volume (V_p) and average pore diameter (D_p) of solids. The 10FeAl, 10ZnAl, 10SbAl and 10WAl samples, without copper, presented a range of specific surface area between 330 and 244 m² g⁻¹. Therefore, one can note that the choice of metal elements (Fe, Zn, Sb or W) change the textural proprieties values of the supports.

However, after the addition of copper by wet impregnation one can observed changes in the surface area, total pore volume and average pore diameter data. For example, the specific surface area for the 5CuZnAl materials compared to the 10ZnAl catalyst increases from 244 to 409 m² g⁻¹ and the specific pore volume of the solid also increases from 0.20 to 0.45 cm³ g⁻¹. On the other hand, the S_BET for the 5CuWAl catalyst compared to the 10WAl sample decrease from 294 to 32 m² g⁻¹ and the V_p of the solid also decrease from 0.16 to 0.04 cm³ g⁻¹.

These textural effects in the porosity of the solids may be related to the information evidenced in the XRD patterns (Fig. 1). That is probably manifested by phase segregation, chemical interactions and polymorphous transformations of the alumina support interacting with the copper oxide. Interactions of the alumina support and copper oxide and formation of copper-modified structure, may explain the significant changes in the textural properties; this phenomenon was observed in the Zn and Fe-based catalysts, however, it is not showed in the Sb and W-containing samples. Thus, the choice of metal elements affects clearly its structure and texture, as observed in Fig. 1 and 2 as well as the Tables 1 and 2 The structural change accompanied by the significant increase of surface area after impregnation of Cu²⁺ and recalcination using this synthetic route was previously explained for the catalysts containing pure copper oxide and aluminum oxide due to the redissolution and recrystallization of transition alumina.

Microcalorimetry of NH₃ adsorption

Calorimetric measurements are generally considered as reliable techniques to measure the acidic strength of solid acids. Thus, the catalysts without and with copper were also analyzed by calorimetry of NH₃ adsorption in order to determine the total acidity and the acid strength distribution, Fig. 3.

Fig. 3 Differential heat of NH₃ adsorption as a function of ammonia coverage. (a and c): copper free samples; (b and d): copper-based solids. (a and b): normalized by the mass of the catalyst; (c and d): normalized by specific surface area from N₂ physisorption results.

Initially, the sample 10ZnAl, 10FeAl, 10SbAl and 10WAl, without copper, presented heat of ammonia adsorption of 193, 161, 126 and 52 kJ mol⁻¹, respectively, Fig. 3a. Therefore, the first values confirm the following order of acid strength: 10ZnAl > 10FeAl > 10SbAl > 10WAl for the copper-free solids. However, with greater coverage of ammonia, increasing the number of moles of NH₃, it is noticed a progressive decrease in the heat of ammonia adsorption, reaching a value of 36 and 8 kJ mol⁻¹ for the samples 10ZnAl and 10WAl, respectively. This profile characterizes the heterogeneity of the present acid sites, and that the 10ZnAl catalyst, despite higher acidity sites compared to other solids, these sites are present in lower amount. Nevertheless, the sample 10WAl shows a profile, which indicates a solid with a weak acidity and low amount of sites. Contrary, the supports 10FeAl and 10SbAl did not show a rapid decrease of the heat of adsorption with the addition of the molecule probe doses, indicating that these samples have higher amount of acid sites, with superior homogeneity.

Fig. 3c, relative to the samples without copper, which the number of moles of ammonia adsorbed were normalized by the specific surface area, obtained from the N₂ adsorption/desorption isotherms (Table 2). It is noted a profile similar to the Fig. 3a probably due to the fact that the surface area of these solids are very close.

On the other hand, after the addition of copper by wet impregnation, it was observed differences in the microcalorimetry data, Fig. 3b. In the first dose, the heat of adsorption for the samples 5CuFeAl, 5CuWAl, 5CuSbAl and 5CuZnAl were 192, 140, 139 and 114 kJ mol⁻¹, respectively, indicating the following acid strength: 5CuFeAl > 5CuWAl ≈ 5CuSbAl > 5CuZnAl. However, increasing doses of ammonia, it was observed an abrupt drop of heat of NH₃ adsorption with values of 27 and 25 kJ mol⁻¹, concerning the 5CuWAl and 5CuSbAl catalysts, respectively. Therefore, it may be related to the elevated heterogeneity of sites and the low surface area for these two solids (Table 2), providing lower number of acid sites per gram of material. However, the 5CuFeAl and 5CuZnAl materials, otherwise shows only a slight decrease in acidic sites with increased content of adsorbed NH₃ and consequently a better homogeneity (better distribution of the acids sites), as well as a higher amount of adsorbed sites per gram of material. The high surface areas of these catalysts provide greater number of acid sites.

Thus, the fact that 5CuWAl and 5CuSbAl samples indicate a rapid decrease in the adsorption heat with the increase of the amount adsorbed may be correlated with the obtained textural properties by N₂ adsorption/desorption isotherms, Fig. 2 and Table 2. Among the synthesized catalysts, Sb and W-containing solids presented a low total pore volume (Table 2), hence, the ammonia accessibility towards the acid sites may be limited for these two materials. This fact is confirmed in Fig. 3d, in which the number of moles of NH₃ was normalized by surface area values shown in Table 2. It is noticed that the sample 5CuSbAl, 5CuFeAl and 5CuZnAl presented a similar amount and strength of acid sites. The catalyst 5CuWAl confirms a minor amount of acidic sites. Thus, if the reaction which the catalysts is dependent on acid sites (as the glycerol conversion), the surface area influence the number of adsorption sites and will certainly influence on the catalytic performance.

Glycerol dehydration is a classic acid catalyzed reaction. The amount, strength and type of acid sites (Brønsted or Lewis) strongly affect the catalytic performance of solid acid catalysts.^19,33 It is well known that the product selectivity from glycerol dehydration depends on which hydroxyl group is firstly removed, terminal or secondary hydroxyl. Acetol is the product when the terminal hydroxyl is initially removed from glycerol. On the other hand, 3-hydroxypropanal is formed as an intermediate, when the secondary hydroxyl group is abstracted first, and posteriorly is converted to acrolein during the second dehydration.^33,34

Furthermore, the type of acid sites (Bronsted and/or Lewis) is well known to affect the glycerol dehydration mechanism. The Bronsted acid sites favor the removal of middle hydroxyl groups to form acrolein, and Lewis acid sites tend to remove the terminal hydroxyl groups toward acetol.^33,34 Thus, it is expected that the synthesized catalysts have higher selectivity to acetol, since unhydrated metal oxides (M–O) solids preferably have Lewis sites.

Catalytic activity

The conversion of glycerol and the acetol selectivity as a function of time for the four different catalysts (10FeAl, 10ZnAl, 10SbAl and 10WAl), without copper, is shown in Fig. 4.

Fig. 4 Catalytic performance of the samples without copper; (a) conversion of glycerol, (b) product selectivity.

The results explain that the samples 10ZnAl, 10SbAl and 10WAl are almost inactive after 300 min of reaction, indicating that only the acid properties probably play a secondary role during the transformation of glycerol, since the pure supports showed no interesting activities in reaction. On the other hand, 10FeAl catalyst presents higher activity and stability compared to the others three solids. Additionally, Fig. 4b presents that the solid 10ZnAl, 10FeAl and 10SbAl area higher selectivity for the acetol formation. Contrarily, 10WAl sample, especially during the first 60 minutes, presented a large amount of acrolein.

It is reported in literature that the promotional effect of H₂ is related to the formation of H^δ+(WO₃)_n^δ− Brønsted acid sites through the local reduction of polytungstate domains, without the removal of lattice oxygen atoms. These sites catalyze dehydration reactions much more effectively than acidic OH groups in W–O–W species under a reducing environment.^33,35,36 The presence of Bronsted sites for the W-based solids may justify the observed selectivity to acrolein for the catalysts 10WAl and 5CuWAl. The acrolein was observed only for these two samples.

On the other hand, all the four different Cu-containing catalysts (5CuFeAl, 5CuZnAl, 5CuSbAl and 5CuWAl) are very active during the first sixty minutes of reaction, Fig. 5a. These results propose that copper species combined with an acid site provides an active species for the conversion of glycerol to acetol formation. However, the differences in catalytic stability can be observed after 300 min of time on stream. 5CuSbAl and 5CuWAl samples present a rapid deactivation, while the 5CuFeAl and 5CuZnAl solids show an excellent performance for the conversion of glycerol to acetol (high stability).

Fig. 5 Catalytic tests of the Cu-containing catalysts: (a) conversion of glycerol, (b) acetol selectivity.

The profiles of acetol selectivity are presented in Fig. 5b. 5CuFeAl and 5CuZnAl catalysts display similar acetol selectivity, near 90%, after 4 h of time on stream. Conversely, 5CuWAl sample demonstrate lower acetol selectivity probably due to the tungsten ability to form bronsted acid sites, hence, leading to the formation of unrequired products such as acrolein.

Sato et al. studied the vapour phase glycerol dehydration/dehydrogenation to acetol, on copper based catalysts, which the dehydrogenation was catalyzed by Cu through the formation of Cu alkoxide species. Glycerol may be dehydrated through the dehydrogenation–dehydration–hydrogenation sequence of 1,3-propanediol. In analogy to the dehydration of 1,3-propanediol, glycerol may be converted to 2-hydroxypropanal, which is readily tautomerized to acetol. It was mentioned that the presence of an acid support may promote the dehydration via elimination of OH anion, optimizing the conversion of glycerol to acetol.^7,34 Thus, copper-based catalysts combined with an acid support are very selective to acetol according to the mechanism proposed in the literature, in agreement with the results presented in Fig. 5 and 6.

Fig. 6 Catalyst reusability for the sample 5CuFeAl.

The other products selectivity (1,2-propanediol, acrolein, acetaldehyde, allyl alcohol, acetone and condensation products) is presented in Table 3. For example, the sample 5CuWAl presented the 1,2-propanediol and others products selectivity of 31.2, and 34.6%, respectively.

Table 3 Catalytic selectivity observed for others products and amount of coke for the spent catalysts

Samples	Selectivity^a (%)
	1,2-Propanediol	Others products^b	C (wt%)
a Values for reaction period of 60 min. C: carbon content in catalysts after reaction from TGA analysis.b Other products: acrolein, acetaldehyde, allyl alcohol, acetone and condensation products.
5CuFeAl	8.9	4.6	3.4
5CuSbAl	10.0	13.8	2.0
5CuWAl	31.2	34.6	1.1
5CuZnAl	4.6	10.1	7.9

---

The gradual catalyst deactivation for the 5CuSbAl and 5CuWAl solids, observed in Fig. 5a, can be attributed to high sintering of copper (low metal support interaction) due the formation of single CuO phase (low metallic dispersion), Fig. 1 and 7. It is very well known that copper species are quite favorable to suffer sintering during the catalytic process due to the drastic reaction conditions for the transformation of glycerol in gas phase. The acidity of the WO₃/Al₂O₃ and Sb₂O₅/Al₂O₃ supports (Fig. 3b and d), and consequently, the formation of undesired by-products causing the carbon deposit may also favor the catalytic deactivation (covering of the sites and blockage of pore). The low dispersion of copper species (well-defined peaks observed in the XRD patterns) and the low number of sites per gram (low surface area, Table 2) for the 5CuSbAl and 5CuWAl samples favor the rapid deactivation by carbon deposit (blocking active sites), justifying the sharp drop in glycerol conversion after 60 min.

Fig. 7 X-ray diffraction diffractograms of the solids after the catalytic tests.

On the contrary, the best stability for 5CuFeAl and 5CuZnAl catalysts can be ascribed the presence of a copper-modified phase (high metallic dispersion, strong metal support interaction and higher surface area), which presents greater resistance to suffer sintering (high structural stability), as observed in XRD results, Fig. 1 and 7. The carbon deposit due to its acidity (Fig. 3b and d) is also occurring for these two catalysts, however, the greater number of sites per area (Table 2) and the largest dispersion of the copper sites (Fig. 1) slows the total coverage of active sites and blockage of pore by carbon and the catalyst deactivation, favoring its high conversion and stability (Fig. 5 and 6).

From these results presented one can note that the structural, textural and acidity proprieties and accordingly the catalytic performance, depend on the choice of metal elements employed (Fe, Zn, Sb and W) due to the fact that the Fe and Zn-based samples presented the copper-modified structure, greater number of acid sites per area and higher surface area, subsequently, better conversion and high stability during the glycerol to acetol related to W and Sb-based solids.

It is important to highlight that experiments were performed in the presence of hydrogen in order to promote the hydrogenation of acetol to 1,2-propanediol, however, under the applied reaction conditions the reaction rate in the hydrogenation is much lower compared to the reaction rate of the dehydration of glycerol, justifying the higher selectivity to acetol related to the 1,2-propanediol (Table 3 and Fig. 4b). The low selectivity to 1,2-propanediol compared to acetol is due to the fact that under atmospheric pressure and elevated temperature its production is disadvantaged, since the hydrogenation of acetol is an exothermic reaction.^34,37

The remarkable activity and stability for the 5CuFeAl catalyst make the study of recycling particularly interesting. The four catalytic tests are presented in Fig. 6. For recycling experiments, the sample used in a first run was reduced again in H₂ (25 mL min⁻¹) at 350 °C for 1 h after each run and reused in the conversion of glycerol to acetol at 250 °C for 6 h, as described for the fresh catalysts.

These results clearly indicated that for the 5CuFeAl sample no significant decrease in the catalytic activity and acetol selectivity were observed after the second recycling test. However, the conversion of glycerol achieved over 63% after the fourth recycling experiment, but the acetol selectivity is similar to the first run. Thus, the results ascribed that the catalytic activity and the acetol selectivity did not decrease dramatically during the recycling experiments, considering that no special treatment (reoxidation of deposited coke) was employed after every reuse. Thus, the recycling experiments also confirm the high resistance to deactivation for the 5CuFeAl solid, as previously mentioned, is due to high metallic dispersion, strong metal support interaction and high surface area, according to the XRD and N₂ physisorption results.

In order to obtain information concerning the reasons of the decrease in the conversion of glycerol, principally for the samples 5CuSbAl and 5CuWAl, the TG analysis was carried out for all the Cu-based catalysts, after the catalytic test. The TG results for all the samples present a carbon elimination (burning) in the temperature range of 300 and 600 °C; it shows that the carbon deposit is more pronounced for the 5CuZnAl sample, as it is illustrated in Table 3. Furthermore, a relationship between the amount of acid sites from calorimetric measurements (Fig. 4b) and the carbon deposited is observed, indicating that a higher acidity leads to a superior coke deposition.

Additionally, the spent Cu-based solids were also analyzed by XRD in order to obtain information relating to the thermal sintering (structural stability) and the metal-support interaction. The diffractograms are presented in Fig. 7 and the average crystallite sizes were estimated using the Scherrer's formula, Table 4.

Table 4 Average crystallite sizes of the Cu-based solids for the spent catalysts

Samples	Crystallite sizes (nm)
	Cu2O	Cu	CuO-γ-Al2O3
5CuFeAl	—	—	4.9
5CuSbAl	10.0	41.8	—
5CuWAl	18.2	28.0	—
5CuZnAl	—	—	4.5

---

The diffraction pattern presented in Fig. 7 for the 5CuSbAl and 5CuWAl samples suggest the Cu²⁺ reduction during the reaction process. The results of the Sb and W-based catalysts point to the cuprous oxide formation (Cu₂O, JCPDS 04-003-6433) and metallic copper phase (Cu, JCPDS 01-071-4611), indicating that the oxidation states of copper are 0 and +1 after reaction.

Great attention was given to the transformation of the catalytic phase during reaction, and the oxidation state of copper during the conversion of glycerol.^6,7 It was already mentioned that the presence of metallic copper may favor the conversion of glycerol to acetol.³⁸

On the other hand, the XRD patterns for the 5CuFeAl and 5CuZnAl catalysts ascribed that the copper-modified alumina structure (JCPDS 00-034-0425-Fe and JCPDS 00-001-1153-Zn) was maintained after transformation of glycerol, whereas that the Fe and Zn-containing diffractograms after the reaction (Fig. 7) are very similar to the fresh catalyst (Fig. 1), which may justify the best catalytic stability, Fig. 5a. Thus, the copper-support interaction for the 5CuFeAl and 5CuZnAl samples was stronger compared 5CuSbAl and 5CuWAl to solid, justifying its better stability.

The average crystallite diameter using the Scherrer formula for the spent catalysts is illustrated in Table 4. The average crystallites diameters of the metallic Cu particles were found between 28.0 and 42.0 nm for the 5CuSbAl and 5CuWAl samples. Contrarily, the copper-modified phase for the 5CuFeAl and 5CuZnAl catalysts show high sintering-resistance ability (strong metal-support interaction) due to the particle sizes after and before reaction are very similar, as can observed in Tables 1 and 4

Conclusions

The XRD, N₂ adsorption isotherms and microcalorimetry of NH₃ adsorption results showed that the choice of metal elements affect its structure, texture, acidity and subsequently the catalytic performance. Interactions of the copper oxide and aluminum oxide and formation of copper-modified alumina phase occurred in the Fe and Zn-containing solids, however, it is not observed in the Sb and W-based samples. The solids composed of Cu/Zn/Al and Cu/Fe/Al showed the best performance for conversion of glycerol to acetol. The catalysts with a copper-modified structure (Fe and Zn-based solids) achieved higher catalytic stability compared to the samples containing only isolated copper oxide (W and Sb-containing solids). It was found that sintering of Cu was more pronounced in the Sb and W-based samples related to Fe and Zn-containing solids.

Acknowledgements

The authors acknowledge the “Universidade Federal do Ceará” (UFC) and Institut de recherches sur la catalyse et l'environnement (IRCE) in Lyon-France.

Notes and references

1. Y. Men, H. Gnaser, R. Zapf, V. Hessel and C. Ziegler, Catal. Commun., 2004, 5, 671.
2. G. Sun, X. Mu, Y. Zhang, Y. Cui, G. Xia and Z. Chen, Catal. Commun., 2011, 12, 349.
3. A. Romero, A. Santosa, D. Escrig and E. Simona, Appl. Catal., A, 2011, 392, 19.
4. E. S. Vasiliadou and A. A. Lemonidou, Appl. Catal., A, 2011, 396, 177.
5. A. Bienholz, H. Hofmann and P. Claus, Appl. Catal., A, 2011, 391, 153.
6. M. Velasquez, A. Santamaria and C. Batiot-Dupeyrat, Appl. Catal., B, 2014, 160, 606.
7. S. Sato, M. Akiyama, R. Takahashi, T. Hara, K. Inui and M. Yokota, Appl. Catal., A, 2008, 347, 186.
8. A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107, 2411.
9. C. W. Chiu, M. A. Dasari, G. J. Suppes and W. R. Sutterlin, AIChE J., 2006, 52, 3543.
10. K. Lacasse and W. Baumann, Textile Chemicals Environmental Data and Facts, Springer, Berlin, 1st edn, 2004, p. 1180, ISBN: 3540408150.
11. M. H. Mohamad, R. Awang and W. M. Z. W. Yunus, Am. J. Appl. Sci., 2011, 8, 1135.
12. A. F. Jalbout, F. F. Contreras and L. Adamowicz, Interface Sci., 2008, 108, 279–288.
13. A. J. Apponi, J. J. Hoy, D. T. Halfen and L. M. Ziurys, Astrophys. J., 2006, 652, 1797.
14. H. Matsuhashi, T. Tanaka and K. Arata, J. Phys. Chem. B, 2001, 105, 969.
15. K. Arata, H. Matsuhashi, M. Hino and H. Nakamura, Catal. Today, 2003, 81, 250.
16. W. Chu, J. Hu, Z. Xie and Q. Chen, Catal. Today, 2004, 90, 349.
17. M. E. Manrıquez, T. López, R. Gómez and J. Navarrete, J. Mol. Catal. A: Chem., 2004, 220, 229.
18. B. Katryniok, S. Paul, M. Capron and F. Dumeignil, ChemSusChem, 2009, 2, 719.
19. A. Alhanash, E. F. Kozhevnikova and I. V. Kozhevnikov, Appl. Catal., A, 2010, 378, 11.
20. W. Suprun, M. Lutecki, T. Haber and H. Papp, J. Mol. Catal. A: Chem., 2009, 309, 71.
21. E. S. Vasiliadou, T. M. Eggenhuisen, P. Munnik, P. E. de Jongh, K. P. de Jong and A. A. Lemonido, Appl. Catal., A, 2014, 145, 108.
22. P. Djinovic, J. Levec and A. Pintar, Catal. Today, 2008, 138, 222.
23. W. M. Shaheen, Thermochim. Acta, 2002, 385, 105.
24. H. Mai, L. Mengfei and F. Ping, J. Rare Earths, 2006, 24, 188.
25. Z. Zhiming, L. Xun, Z. Tianying, H. Wenbin, C. Zhenmin and Y. Weikang, Chin. J. Chem. Eng., 2010, 18, 384.
26. S. Panyad, S. Jongpatiwut and S. Osuwan, Catal. Today, 2011, 174, 59.
27. W. M. Shaheen, Thermochim. Acta, 2002, 385, 105.
28. J. Antona, J. Nebel, H. Song, C. Froese, P. Weide, H. Ruland, M. Muhler and S. Kaluza, Appl. Catal., A, 2015, 505, 326.
29. C. Jeong and Y. Suh, Catal. Today, 2015 , in press.
30. G. Vijayaprasath, R. Murugan, T. Mahalingam, Y. Hayakawa and G. Ravi, J. Alloys Compd., 2015, 649, 275.
31. Z. Xiao, S. Jin, X. Wang, W. Li, J. Wang and C. Liang, J. Mater. Chem., 2012, 22, 16598.
32. Z. Xiao, X. Wang, J. Xiu, Y. Wang, C. T. Williams and C. Liang, Catal. Today, 2014, 234, 200.
33. L. Liu, B. Wang, Y. Du and A. Borgna, Appl. Catal., A, 2015, 489, 32.
34. J. T. Dam and U. Hanefeld, ChemSusChem, 2011, 4, 1017.
35. C. D. Baertsch, K. T. Komala, Y. Chua and E. Iglesia, J. Catal., 2002, 205, 44.
36. M. Valigi, D. Gazzoli, I. Pettiti, G. Mattei, S. Colonna, S. de Rossi and G. Ferraris, Appl. Catal., A, 2002, 231, 159.
37. A. Bienholz, H. Hofmann and P. Claus, Appl. Catal., A, 2011, 391, 153.
38. R. B. Mane, A. Yamaguchi, A. Malawadkar, M. Shirai and C. V. Rode, RSC Adv., 2013, 3, 16499.

---