Aqueous-phase oxidation of 5-hydroxymethylfurfural over Pt/ZrO₂ catalysts: exploiting the alkalinity of the reaction medium and catalyst basicity

2.1 Base and metal catalyst preparation

Aqueous solutions of NaOH (Vetec Química Fina Ltda, Rio de Janeiro, Brazil) and Na₂CO₃ (Vetec) were used as homogeneous base along with the Pt/ZrO₂ metal catalyst. Due to their very low solubility in water, Mg(OH)₂ and Ca(OH)₂ were taken as solid base catalysts to compose a dual metal/base heterogeneous catalytic system along with Pt/ZrO₂, keeping a constant mass ratio of 1.4. Many commercial or synthesized ZrO₂-based materials were taken as support for the Pt bifunctional catalysts. Two commercially available ZrO₂ polymorphs (tetragonal and monoclinic: t-ZrO₂ and m-ZrO₂) and aTi-doped ZrO₂ (Ti-ZrO₂) (Saint-Gobain NorPro, Stow, OH, USA) were used without any further treatment.

Different Mg-doped ZrO₂ solids were also synthesized by coprecipitation from aqueous solutions of both precursor salts and the precipitating agent. Basically, a solution containing Mg(NO₃)₂ and ZrO(NO₃)₂ in the desired Mg:Zr ratio was gradually added to 100 ml of distilled water in a Kettle reactor at room temperature. The pH was kept constant at 10 throughout the precipitation step with the aid of a NaOH solution at 2 mol l^-1. A white gel was immediately formed and then aged for 48 h at room temperature. The precipitate was filtered and extensively washed with water until constant pH. Thereafter, the powder was dried overnight at 100°C and finally calcined in a muffle at 500°C for 12 h following a 10°C min^-1 linear temperature increase. The samples were named according to the nominal loading of Mg; Mg10ZrO₂ stands for ZrO₂ doped with 10 wt.% of MgO. Pure MgO and ZrO₂ samples were also prepared following the precipitation procedure.

Two additional doped ZrO₂ samples, containing calcium and yttrium (Ca50ZrO₂ and Y5ZrO₂), were also synthesized following the exact same procedure described previously. In these cases, Ca(NO₃)₂ and Y(NO₃)₃ were used as precursor salts.

Pt-based catalysts were prepared by incipient wetness impregnation using an H₂PtCl₆ aqueous solution so as to obtain a platinum content of 5 wt.%. All supported samples were calcined in a stream (50 ml min^-1) of synthetic air, a mixture containing 20 vol.% O₂/N₂, for 4 h at 500°C. After calcination, the catalysts were activated by reduction at 350°C under a pure hydrogen flow (50 ml min^-1) for 1 h.

2.2 Characterization of fresh catalysts

Chemical analyses were carried out in a Bruker S8 Tiger X-ray fluorescence spectrometer without any pretreatment. The BET surface area measurements were performed at -196°C in a Micromeritics ASAP 2020 equipment. The samples were pretreated under vacuum at 300°C prior to analysis. Powder X-ray diffraction (XRD) patterns were collected on a D8 Advance Bruker, using CuKα radiation and by increasing 2θ from 20° to 90° with increments of 0.02° and counting time of 0.5 s per step.

The reaction rate of cyclohexane dehydrogenation was used to estimate the apparent platinum dispersion of all catalysts. The experiments were performed following a detailed protocol described elsewhere [23]. Prior to reaction, the catalysts were reduced at 350°C (10°C min^-1) for 1 h under pure hydrogen. A mixture of cyclohexane in hydrogen was then admitted into the reactor at a H₂/C₆H₁₂ ratio of 13.2. The reaction rates were calculated under differential conditions at 260°C, and conversion was monitored by using an on-line gas chromatograph (Agilent 6890N) equipped with a flame ionization detector. Dispersion was estimated by correlating the reaction rates with dispersion values of Pt/ZrO₂ samples previously obtained by hydrogen chemisorption.

Temperature-programmed reduction (TPR) analysis of the calcined catalysts was carried out in an AutoChem Micromeritics equipment. The samples were heated from 30°C to 600°C under a 10 vol.% H₂ in N₂ flow at a heating rate of 10°C min^-1. Before analysis, the catalysts were submitted to a thermal treatment at 500°C for 60 min under synthetic airflow. Afterward, the samples were cooled down to room temperature and then the gas was switched on to TPR reducing gas mixture.

2.3 Catalytic runs

Oxidation of HMF (99.9%, Aldrich) was carried out in the liquid phase in a semibatch 300-ml steel reactor (Parr Instruments). The reactions were performed at constant pressure (1 MPa) of synthetic air (20 vol.% O₂/N₂) and at 80°C. HMF was used without further purification from commercial products. The experiments were conducted with stirring at 600 rpm, which proved to favor gas (synthetic air) diffusion and ensure kinetic control. Liquid samples were taken periodically to be analyzed by high-performance liquid chromatography. A metallic frit was used in order to obtain clear liquid samples. In a typical run, the reduced catalyst and 80 ml of an aqueous solution of HMF at 0.1 mol l^-1 (HMF/Pt molar ratio=39) was charged into the reactor vessel, which was then repeatedly flushed with N₂. After heating to the reaction temperature, the reaction was started by switching the gas to synthetic air (20 vol.% O₂/N₂) and the pressure adjusted to 1 MPa. When solid base catalysts were used to compose a dual metal/base heterogeneous catalytic system in a mixture with Pt/ZrO₂, the amount of solid base was equivalent to the highest amount of the corresponding alkali-earth metal in bifunctional catalysts supported on solid solution (Pt/ZrO₂:solid base mass ratio=1:0.725).

The reaction runs were analyzed in terms of HMF conversion, selectivity, and yield of the products calculated on a molar basis by the following equations:

Conversion of HMF(XHMF)=[HMF]i-[HMF]f[HMF]i×100

Selectivity(S)=[Product][HMF]i-[HMF]f×100.

Yield(Y)=[product][HMF]i×100.

2.4 Analysis of the liquid phase

All liquid aliquots collected from the reactor were filtered with a polyethylene 0.22-μm filter and then diluted with distilled water (1:10). Samples were then analyzed in a Waters Alliance equipment coupled to a photodiode array detector (PDA) and a refractive index detector. A Biorad Aminex HPX-87H ion exchange column operating at 65°C was used to separate all products; the analyses were performed in isocratic elution mode (0.7 ml min^-1), using an H₂SO₄ aqueous solution at 0.005 mol l^-1 as mobile phase. Analyses in PDA detector were performed at 280 nm.

Atomic absorption spectrometry (AAS) was used to analyze the concentration of eventual dopant element in the aqueous solutions after the reaction. Analyses were performed in a Varian AA 280 FS spectrometer.

Total organic carbon (TOC) was determined in a Shimadzu TOC-VCPH analyzer in order to estimate the overall carbon balance after reaction.

2.5 Analysis of spent catalysts

After each catalytic run, the catalysts were filtered and dried to be analyzed regarding their crystalline structure. Similar to the fresh samples, XRD analyses were performed on a D8 Advance Bruker, using CuKα radiation and by increasing 2θ from 20° to 90° with increments of 0.02° and counting time of 0.5 s per step.

3.1 Characterization of base and metal catalysts

The commercially available solids – ZrO₂ polymorphs and Ti-ZrO₂ – and all synthesized supports – ZrO₂, Mg10ZrO₂, Mg50ZrO₂, MgO, Ca50ZrO₂, and Y5ZrO₂ – were analyzed regarding their more primary chemical and physical features, i.e. chemical composition and surface area and crystalline structure.

The data collected in Table 1 show that the chemical composition of the synthesized samples satisfactorily meets the nominal ones. Moreover, all commercial and synthesized oxides present moderate to high surface areas (S_BET, Table 1).

Powder X-ray diffractograms of commercial materials showed that pure zirconias are indeed distinct polymorphs with all typical diffractions related to either a tetragonal (t-ZrO₂) or a monoclinic (m-ZrO₂) crystalline structure (Figure 1A), as compared with PDF 01-071-1282 and PDF 01-070-2491, respectively. On the other hand, even though the synthesized ZrO₂ revealed to present mostly a monoclinic structure, tiny peaks associated with tetragonal lattice system (30.6°, 81.5°) could also be identified in the XRD pattern.

Figure 1:

XRD patterns of the commercial and synthesized supports (A) and a detailed range within 25–35° (B).

Apart from Ti-ZrO₂, zirconia samples doped with oxides of Mg, Ca, and Y exhibited the same diffraction peaks as those from t-ZrO₂, showing that the tetragonal lattice system is preferentially formed upon doping. Nevertheless, t-ZrO₂ peaks were seen to shift to higher or lower Bragg angles, indicating a contraction or expansion of the zirconia lattice. Such a trend may be ascribed to the formation of solid solutions at which Zr⁴⁺ (0.84 A) is substituted by smaller Mg²⁺ (0.72 A) or larger Ca²⁺ (0.99 A) and Y³⁺ (0.90 A) cations in the crystal lattice. Figure 1B shows, as expected, that the shift is more pronounced with Ca50ZrO₂ and Mg50ZrO₂ due to the higher amount of a foreign cation, leading to a more significant lattice distortion. It is worth mentioning that no reflections associated with any pure dopant crystalline phase were identified on Ca50ZrO₂ and Y5ZrO₂, i.e. CaO or Y₂O₃. On the other hand, a peak related to MgO (2θ=43°) was clearly seen in the Mg50ZrO₂ XRD profile (Figure 1), indicating the occurrence of an isolated crystalline phase in this solid.

As for the commercial Ti-ZrO₂, at least three different crystalline phases were identified related to both t-ZrO₂ and m-ZrO₂ as well as TiO₂ with anatase crystal structure. Lastly, MgO presented all peaks associated with a periclase phase with a cubic structure.

This group of ZrO₂-based supports can thus allow evaluating catalysts retaining a similar crystalline structure (a tetragonal ZrO₂ structure) while presenting distinct base properties. Their different basicity stems from their chemical composition, particularly the nature of the doping cation as revealed by their position in the periodic table (electropositivity). More important, it has to be outlined that doping cations are entrapped into the ZrO₂ crystal lattice as shown by XRD. However, one should bear in mind that the occurrence of amorphous or small isolated and highly dispersed crystalline phases cannot be completely ruled out for all samples.

Standard characterization of Pt catalysts prepared on the aforementioned supports was also performed. Metal content and dispersion as well as the catalyst reducibility were determined, and the results are summarized in Table 2.

Platinum loadings were within the expected range considering the nominal catalyst formulation. Dispersion of the supported catalysts was estimated by determining the rate of cyclohexane dehydrogenation, which is reckoned as a structure-insensitive reaction and has been extensively used to assess the apparent dispersion of supported metal catalysts [23], [24]. This approach is of particular interest when ordinary techniques, such as hydrogen chemisorption, may provide misleading results due to the spillover phenomenon, as may be the case for MgO-supported metal catalysts [25]. In general terms, it is considered that the higher the metal dispersion (metallic surface area), the higher the reaction rate of the cyclohexane dehydrogenation. In this contribution, dispersion was assessed from an experimental correlation between the cyclohexane dehydrogenation rates and the dispersions obtained from hydrogen chemisorption measurements carried out with some reference Pt-based samples [21].

The rates of cyclohexane dehydrogenation at 260°C as normalized by catalyst or metal weight are collected in Table 2 along with their corresponding metal dispersion values. It can be seen that, apart from Pt/MgO, dispersion of supported catalysts are high, usually above ~50%.

The reducibility of the catalysts was assessed by conventional H₂-TPR analyses, and the results are shown in Figure 2. All samples presented similar profiles, featuring essentially one main reduction peak related to the reduction of PtO₂ to metallic Pt. However, such event is observed at different temperatures according to the oxide used as support. It is usually associated with the chemical nature of the support itself. Pt reduction is seen between 100°C and 150°C when dispersed on ZrO₂ irrespective of its polymorph. Introduction of Mg into its lattice gradually shifts the reduction peak to higher temperatures, moving from 120°C to around 210°C for catalysts supported on bare ZrO₂ and MgO, respectively. This trend is indeed in close agreement with the literature regarding similar catalysts [26], [27]. No relevant influence in PtO₂ reduction is seen when using Y to dope ZrO₂, as it is also registered at around 130°C. On the other hand, Pt supported on Ti-ZrO₂ mixed oxide seems to be more easily reduced, presenting the lowest reduction temperature (~75°C). It seems thus apparent that the mixed oxide surface chemical features are determining the metal-support interaction.

Figure 2:

TPR profiles of mixed oxide-supported Pt catalysts.

3.2 Liquid-phase oxidation of HMF

Oxidation of HMF may occur through either oxidative dehydrogenation of the hydroxymethyl group in the furanic ring side chain leading to 2,5-diformylfuran (DFF), or carbonyl oxidation to HMFCA. Either of these intermediates can be further oxidized/dehydrogenated to 5-formyl-2-furancarboxylic acid (FFCA), which lastly leads to FDCA. The required high pH to accomplish quantitative yields of FDCA in water can be tricky, as HMF may degrade [9], [11], [28], triggering a significant loss in carbon yield. Alternatively, an FDCA intramolecular rearrangement may be favored and, as a consequence, 2,5-bis-hydroxymethylfuran could be formed [29]. A simplified reaction scheme is depicted in Figure 3.

Figure 3:

Simplified reaction scheme of HMF oxidation in aqueous phase.

The alkalinity of the reaction medium in the oxidation of HMF was firstly examined by using homogeneous alkali solutions (NaOH and Na₂CO₃) along with the synthesized Pt/ZrO₂ catalyst (HMF:Pt=39). These runs can provide a benchmark with the widespread literature focused on this oxidation reaction. It is relevant, as it has been shown that the behavior of a specific association of a catalyst (metal and support) and the chosen alkali (cation and anion) may significantly vary. Using a homogeneous alkali would also allow contrasting the performance of ZrO₂-supported catalyst in distinct alkaline medium, particularly those promoted by heterogeneous solid systems.

The first experiments were carried out with NaOH (NaOH/HMF=4), as it has been claimed to lead to quantitative FDCA yields by some authors [13], [14]. However, fast austere degradation of HMF was observed under this condition; the aqueous solution turned from transparent to dark brown and finally black within a few seconds right after adding the alkali solution. An alike behavior was indeed recently reported by Rass et al. [12] when testing a Pt/C catalyst under similar reaction conditions; the authors could only control degradation by performing the reaction at a moderate basic environment. Following a similar strategy, NaOH was then switched to Na₂CO₃ keeping the same ratio (Na₂CO₃/HMF=4). The time-dependent profiles of HMF conversion and product yields are displayed in Figure 4 along with the results from a control experiment carried out without adding any alkali solution in the reactor (Figure 4A). The reaction data are listed in Table 3, entries 1 and 2.

It can be seen that the performance of the Pt/ZrO₂ sample is rather different according to the reaction medium, both from the kinetics and the product distribution point of view. Total conversion of HMF is only achieved after 6–7 h when the reaction is carried out without any pH control, i.e. at the natural pH of the HMF aqueous solution (pH 4) (Figure 4A). By adding the Na₂CO₃ solution, and as a consequence raising the pH of the reaction medium to 11.5, the catalyst global activity is significantly increased and complete conversion is accomplished after 3 h (Figure 4B). HMFCA was the only partially oxidized intermediate detected, suggesting that oxidation of the carbonyl group in the HMF side chain preferably occurs on this catalyst in the initial reaction stage, in close agreement with some previous reports [7], [30]. It could indeed be expected as an aldehyde carbonyl is more promptly oxidized in the presence of a hydroxymethyl group [31]. Conversely, some authors have reported the preferential formation of DFF as the main oxidized intermediate [12], [32]. Nonetheless, such apparent ambiguous results are likely ascribed to the alkalinity of the reaction medium, as DFF is usually identified under moderate basic pH.

Figure 4:

Time-dependent profiles for HMF oxidation on Pt/ZrO₂ (A) without any alkali solution and (B) with Na₂CO₃ (HMF:Na₂CO₃=4).

Reaction conditions: 0.1 mol l^-1 HMF, 80°C, 1 MPa, HMF:Pt=39 (molar ratio). (□) Conversion, (▲) S_HMFCA, (•) S_FDCA, () pH.

The dynamic of HMFCA conversion is also rather distinctive; while HMFCA selectivity remains roughly the same along many hours under acidic conditions with an unimportant production of FDCA (Figure 4A), it is rapidly oxidized to the desired dicarboxylic acid at high selectivity at basic pH (Figure 4B). The interdependence between HMFCA and FDCA is clear under alkaline conditions, indicating that the formation of FDCA is not a simple, straightforward oxidation reaction but indeed involves successive steps. It should be mentioned that an FFCA intermediate was never observed, implying that its oxidation is very fast, not even allowing its detection. This proposal is in accordance with widespread literature on this matter [5], [6], [7], [13], [17], [18].

This reaction pattern with a well-defined improvement in activity and high yield to FDCA has indeed been broadly reported for this reaction over a wide variety of supported Pt catalysts in association with different alkali solutions [12], [16], [33], [34]. These findings set a positive benchmark for the Pt/ZrO₂ catalyst studied herein and provide a reliable reference for further exploiting the basicity of potential dual or bifunctional heterogeneous catalysts.

Alkaline-earth metal hydroxides were used as base solid catalysts due to their fairly low solubility in water [Mg(OH)₂, K_sp=1.5×10^-11; Ca(OH)₂, K_sp=5.5×10^-6]. These solid bases were then taken along with Pt/ZrO₂ to compose a dual metal/base heterogeneous catalytic system. The standard reaction conditions used previously with Na₂CO₃ solution, including HMF:Pt=39, was used in these runs as well; the results are also summarized in Table 3 and the time-resolved curves are depicted in Figure 5.

Figure 5:

Time-dependent profiles for HMF oxidation on Pt/ZrO₂ with (A) Mg(OH)₂ and (B) Ca(OH)₂ as solid base.

Reaction conditions: 0.1 mol l^-1 HMF, 80°C, 1 MPa, HMF:Pt=39 (molar ratio). (□) Conversion, (▲) S_HMFCA, (•) S_FDCA, (■) S_FA, () pH.

It is curious to observe that the global activity, as examined by HMF conversion, is only slightly improved by introducing solid Mg(OH)₂ to constitute a dual catalytic system with Pt/ZrO₂ (Figure 5A compared to Figure 4A). It is even clearer by comparing the turnover frequencies (TOF) of Pt/ZrO₂ and Pt/ZrO₂+Mg(OH)₂ in Table 3 (entries 1 and 3). On the other hand, the product distribution is noticeably different (compare Figures 4A and 5A). By using Mg(OH)₂ to provide reaction medium basicity, an HMFCA intermediate is registered at a very low selectivity (<5%), and the formation of FDCA is found right from the first reaction minutes (Figure 5A). It suggests that all oxidation steps from HMF (aldehyde carbonyl oxidation followed by oxidative dehydrogenation of the hydroxymethyl group with further oxidation) are kinetically favored over this dual catalytic system.

Comparing now Pt/ZrO₂ performance under basic conditions – provided either by homogeneous Na₂CO₃ solution or poorly soluble Mg(OH)₂ – one can see that FDCA selectivities (Figures 4B and 5A) and yields (Table 3, entries 2 and 3) are rather equivalent. Nevertheless, no HMFCA formation is seen over the dual catalytic system [Pt/ZrO₂+Mg(OH)₂], suggesting that oxidation of this intermediate is even faster in this case. These results evidence the efficient use of a solid base catalyst [hardly soluble Mg(OH)₂]. The high selective performance of dual metal/base heterogeneous catalytic systems has indeed been reported earlier for some oxidative processes [35], [36].

As for the use of Ca(OH)₂ in the dual catalytic system [Pt/ZrO₂+Ca(OH)₂], HMF conversion is considerably higher, reaching nearly 100% after only 1 h (Figure 5B). However, the formation of furanic compounds is oddly limited when compared to the Pt/ZrO₂+Mg(OH)₂ system (Figure 5A), taking into account that the same alkaline-earth hydroxide:HMF:Pt molar ratio was used, and they both provided the same basic pH (pH 9). It is worth mentioning that in this reaction, the carbon balance is low compared to the others; as a matter of fact, it is among the lowest balances measured in this study (Table 3). This tendency is probably due to the high rate of resinification of furanic compounds in the presence of Ca(OH)₂, as a dense viscous brown material was obtained afterward. An analogous effect has been reported by Lux and Siebenhofer when studying aqueous-phase conversion of dihydroxyacetone, which also holds a hydroxyl and a carbonyl group, over Ca(OH)₂ [35]; the dramatic increase in activity is claimed to be related to its ionic strength [35], [37]. The authors also observed instantaneous dimerization of the chemical feedstock, and provided evidences for the formation of oligo- and polymerization products as well [35], [38].

The performance of all catalysts supported on ZrO₂-doped oxides in the aqueous-phase oxidation of HMF was evaluated under base-free conditions. An overview of catalyst behaviors is provided in Table 4. Catalytic activities were compared by calculating TOFs from the initial rates (at 1 h), conditions at which HMF conversion was commonly <40%. Selectivity was compared at matching conversions (isoconversion conditions), which was chosen at high levels (~75%) as it is a sequential multistep oxidation reaction as previously discussed. Additionally, the selectivity at total conversion was also compared to follow the reaction process. Product yields were calculated after a 9-h reaction.

The behavior of zirconia polymorphs was firstly examined in order to infer on any contribution brought about by their different structures and to compare and contrast with the performance of the synthesized Pt/ZrO₂ sample, which was shown to present unbalanced amounts of both crystalline structures (Figure 1). Diverse performance has been reported for different polymorphic phases of ZrO₂ on a wide variety of gas-phase reactions [21], [39], [40], [41], enlightening that the coordination environment of zirconium and oxygen atoms in the oxide can indeed determine catalyst activity due to a distinctive metal-oxide interface [41].

As it can be seen in Table 4, all three ZrO₂-supported catalysts (entries 1–3) show roughly the same behavior; FDCA is produced at very low yields, and HMFCA is the only relevant partially oxidized intermediate. This finding is consistent with other reports found in the literature regarding the performance of similar Pt/ZrO₂ catalysts [16]. Carbon balance as measured by TOC analyses (Table 4) revealed that all organic carbon was still in the aqueous media, ruling out the possibility of an overoxidation of the organic compounds to gaseous CO₂. Low yields to FDCA and HMFCA may be rationalized by degradation or, most likely, condensation of HMF in unidentified/undetected products (high molecular weight products and humins). These findings suggest that the differences in any eventual metal-oxide interface due to zirconia polymorphism are not pertinent in this liquid-phase transformation.

Gradual introduction of magnesium into zirconia lattice completely change the reaction path, as shown in Figure 6 (sets A and B) and Table 4 (entries 4 and 5). While HMF conversion is only slightly improved by increasing magnesium content, the product distribution is more significantly affected and FDCA formation is clearly favored at higher amounts of the alkaline metal in the catalyst. It should be firstly emphasized that the production of FDCA is favored in the presence of even small amounts of MgO, which is clear by comparing Pt/ZrO₂ and Pt/Mg10ZrO₂ catalysts (Table 4, entries 3 and 4). Nonetheless, quantitative yields of FDCA is accomplished on Pt/Mg50ZrO₂ (Table 4, entry 5) but it does not exceed ~30% by using Pt/Mg10ZrO₂ sample (Table 4, entry 4). Either way, the alcohol-acid intermediate HMFCA is the only other relevant organic compound identified.

Figure 6:

Time-dependent profiles for HMF oxidation on (A) Pt/Mg10ZrO₂, (B) Pt/Mg50ZrO₂, and (C) Pt/MgO.

Reaction conditions: 0.1 mol l^-1 HMF, 80°C, 1 MPa, HMF:Pt=39 (molar ratio). (□) Conversion, (▲) S_HMFCA, (•) S_FDCA, () pH.

The Pt/MgO catalyst taken as a reference for those Mg-doped samples also presented an improved performance (Table 4, entry 6; Figure 6, set C); however, a lower FDCA yield (~60%) was accomplished.

Surprisingly, none of the other doped catalysts were selective to FDCA under the same standard conditions (Table 4, entries 7–9), leading to yields <5%. It is likely associated with the lower pH these catalysts imposed on the reaction medium; it only ranged from 4 to 2, while it was around 9 for Mg-containing catalysts.

Improved performance by using Mg as a catalyst promoter stands out once again in close agreement with some previous reports in the literature [17], [18]. The stability of these systems was then investigated by analyzing both the post-reaction water solution and the recovered spent powder catalyst.

3.3 Post-reaction analyses

The aqueous reaction medium and the used catalyst were separated by filtration right after the reaction. The AAS results from the aqueous solutions after the reaction disclosed that Mg²⁺ ions were leached from the catalysts (Table 5), even though they were initially embedded into the ZrO₂ lattice. Such Mg²⁺ removal from the solid catalyst would also afford the alkaline conditions required for FDCA production, as previous claimed by other authors for some Mg-containing catalysts [5], [18], [20]. As a matter of fact, the FDCA yield was shown to virtually linearly increase with the concentration of Mg²⁺ in solution, as depicted in Figure 7. It should be recalled that those samples presented a solid solution at which magnesium was effectively settled into the ZrO₂ crystalline lattice. This interdependence between FDCA yield and Mg²⁺ ions in solution (Figure 7) suggests that cation extraction from solid solution might occur by its complexation with the dicarboxylic acid formed. Moreover, it reveals that hosting Mg into the ZrO₂ crystalline structure does not totally avoid its leaching upon reaction in aqueous medium, making the design of a heterogeneous bifunctional catalyst rather a challenge.

Figure 7:

Correlation between FDCA yield and concentration of Mg²⁺ ions leached into the reaction solution.

In the experiments carried out in this study, the amount of solid base was equivalent to the highest amount of the corresponding alkali-earth metal in the solid solution. That is to say that the amount of Mg(OH)₂ used was equal to the total amount of Mg²⁺ in the catalyst with the highest amount of magnesium (Pt/Mg50ZrO₂). In this sense, a straightforward comparison between those two systems is feasible. Firstly, it could be noted that either way, the pH of the reaction medium was rather similar [9 for Mg(OH)₂ and 8.5 for Pt/Mg50ZrO₂]; however, anyhow, some differences arose. Pt/Mg50ZrO₂ was revealed to be only slightly more active (Figures 5A and 6B) but accomplished significantly higher FDCA yield (Tables 3 and 4). It suggests that HMF transformation to FDCA occurs more effectively when Mg²⁺ complexation from solid solution (leaching) is taking place. The formation of an Mg²⁺ chelate might indeed be an important issue in this conversion, and one could think about its role in stabilizing a transition species leading to FDCA. A specific mechanistic study would be required to infer more deeply on that.

No traces of any other dopant (Ca, Y, or Ti) were detected in solution, which might be related to the rather low formation of FDCA (Table 4, entries 7–9), and consequently no metal chelation would occur.

Due to the evidence of magnesium leaching, spent Mg-containing catalysts were recovered by filtration, dried, and examined concerning their crystalline structures. The XRD patterns are displayed in Figure 8. It is seen that the structure of the MgO-supported catalyst completely changed upon HMF aqueous-phase oxidation. As expected, the XRD pattern indicates that the MgO phase is hydrated to the Mg(OH)₂ brucite phase, as all typical diffraction peaks can be easily identified (Figure 8A). As for the catalysts supported on MgO-ZrO₂, some important changes were also observed. Although no reflection related to MgO could be detected anymore, the expected formation of the hydrated brucite phase was not significant as only tiny peaks could be identified (37.9° and 50.7°). Furthermore, the peaks initially associated with a distorted t-ZrO₂ phase were seen to shift to lower Bragg angles, matching now the expected diffraction for a pure tetragonal crystalline structure (Figure 8B). These findings suggest that Mg²⁺ cations were indeed extracted from the ZrO₂ crystal lattice, which are in line with previous results indicating an expressive Mg leaching from this sample (Table 5).

Figure 8:

XRD patterns of the spent Mg-containing catalysts (A) and a detailed range within 25–35° (B).

The results presented herein indicated that small amounts of Mg in the catalyst composition may have a positive impact on the selectivity and yield to FDCA despite leaching vulnerability (Pt/Mg10ZrO₂). Nevertheless, hosting high amounts of Mg²⁺cations into the ZrO₂ crystalline lattice could not prevent MgO from dissolution into the reaction medium due to the chelating capacity of the formed dicarboxylic acid.