*3.1. Catalyst Characterization*

The catalysts were analyzed by PXRD in order to characterize their crystalline structure (Figure 2).

**Figure 2.** X-ray diffractograms of the prepared catalysts.

The original, as-synthesized HT catalyst has the typical hydrotalcite structure, that is, a layered double hydroxide structure of Mg–Al mixed hydroxides organized in brucite-like layers containing water and carbonate anions in the interlayer [32]: sharp (003), (006), (009), (110), and (113) diffraction lines, and two broad (015) and (018) ones. The crystalline structure of hydrotalcites have a high morphological anisotropy because of the layered structure. Hence, it is common to report the crystallite size for every <hkl> direction (Table 1).


**Table 1.** Results of the ICP-AES analysis of the prepared catalysts.

The calculated lattice parameter *c* corresponds to the basal spacing of the layered double hydroxide; dividing the crystallite size in the 003 direction by *c* gives the number of layers in one crystallite, corresponding to about four in this case. The calcination of HT causes the collapse of the layered double hydroxide structure. As a consequence, the MMO X-ray diffractogram does not include any of the aforementioned reflections, but instead presents two broad (200) and (220) diffraction lines, corresponding to the MgO cubic phase (periclase) [33]. Immersion of MMO in aqueous media enables the double hydroxide structure layer to be reconstructed, thanks to the memory properties of these oxides which, for this reason, are also called layered double oxides. XRD analysis of MX does indeed confirm that the rehydration has induced the reconstruction of the original layered double hydroxide structure. As the rehydration was carried out in an inert atmosphere, the calcination–rehydration cycle has the overall effect of replacing the initial carbonate anions with hydroxyls, which enhance the Brønsted basicity of the catalyst. Even so, this anion exchange does not seem to have a notable effect on the diffractograms. Nevertheless, the lattice parameter *c* slightly changes as a result of the different interlayer composition. Moreover, the layered double hydroxide obtained after calcination and rehydration appears to have a bigger crystallite size in all directions, indicating agglomeration of the layers: the number of layers per crystallite is now about eight, approximately double the value obtained with HT (Table 1).

Next, the catalysts were analyzed by ICP-AES to obtain the atomic composition of the prepared materials. The information of interest was the final Mg: Al ratio in the solids and the Na content, which may be due to incomplete HT washing after the co-precipitation. In particular, this parameter is known to have a remarkable influence on the catalytic performance of the solids [34]. Table 2 summarizes the results of the ICP-AES analyses.


**Table 2.** Results of the ICP-AES analysis of the prepared catalysts.

The atomic ratio of Mg and Al matches the theoretical one (2) in all cases. The Na content of the original HT sample is below the detection limit (Table 2, entry 1). The Na, Mg, and Al concentrations increase with calcination as a result of the loss of water, hydroxyls, and carbonates, although their percentage is still very low (Table 2, entry 2). On rehydration, the introduction of water and hydroxyls in the solid pushes the Na content back to below the detection limits (Table 2, entry 3). It can therefore be argued that Na has little or no effect on the aldol condensation results, which will be discussed in the sections below, since the sodium content in HT is below the level of that which is considered to have an impact on the catalytic activity (0.04%).

TGA-DSC was used to study the non-metal composition of the catalysts, and the degradation behavior of the solids. The results of the analysis are reported in Figure 3.

**Figure 3.** TGA (colored lines) and DSC (black lines, exo up) curves of the prepared catalysts.

The thermogravimetric curve obtained for HT fits the previously reported ones well [25]. The curve points to two distinct mass loss events: one of these is at a DSC negative peak at about 200 ◦C, which has been attributed to the loss of physisorbed and inter-layer water; the other is at about 400 ◦C, and should correspond to decarboxylation and dehydroxylation events. The MMO was analyzed as a freshly calcined material, so little or no mass loss was expected. As anticipated, there was no distinct mass loss, and the sample mass drifted slowly to 90% of its original value. As for MX, the profile very much resembled HT's. Yet there were some differences: for example, the different mass loss of water, and the different shape of the second mass loss step, which is probably due to the absence (or at least a reduced quantity) of carbonates in the structure. These observations are in agreemen<sup>t</sup> with previously reported data [25].

N2 physisorption was used to obtain the nitrogen adsorption–desorption isotherms, Brunauer–Emmett–Teller (BET) surface area, and pore volume of the catalysts. The results obtained are summarized in Figure 4 and Table 3.

**Figure 4.** (**a**) Isotherms of N2 adsorption–desorption, and (**b**) Barrett–Joyner–Halenda (BJH) pore size distributions of the prepared catalysts.


**Table 3.** Results of the N2 physisorption analysis of the prepared catalysts.

Generally, the results of N2 physisorption analysis of HT and related materials very much depend on the procedure followed for the preparation and the sample degassing [25,34–36]. The shape of the HT isotherm resembles the shape of type III (Figure 4a), which commonly corresponds to macroporous materials (even though HT has pores in the mesopore range (Table 3, entry 1)) with weak adsorbent–adsorbate interactions [37]. The same considerations apply to the isotherms of MMO and MX (Table 3, entries 2 and 3, respectively). As for adsorption hysteresis, the loops of HT and MMO seem to be type H3, which is associated with clays made of non-rigid aggregates of plate-like particles [37]. On the other hand, the hysteresis loop of MX appears to be more pronounced and seems to correspond to type H2b, which indicates pore blocking. The BET area of the prepared HT is quite high (Table 3, entry 1), and calcination of the hydrotalcite is supposed to increase this value [25,34,36], which is indeed the case for our MMO (Table 3, entry 2). The surface area of MX depends on the preparation conditions [36], and in our case, it was lower than the area of both its parent catalysts (Table 3, entry 3). Moreover, the Barrett–Joyner–Halenda (BJH) results indicated that while HT and MMO are overall quite porous, the porosity of MX is considerably lower.

#### *3.2. Evaluation of the Catalytic Performance of HT at Di*ff*erent Concentrations*

The as-synthesized HT was tested as a catalyst in the microwave-assisted aldol condensation of acetone and furfural. In all the GC-FID chromatograms obtained after analysis, only three products were present in detectable amounts: the mono-aldol product, C8-OH, the mono-condensation product, C8, and the double-condensation product, C13. The only other product detected was diacetone alcohol, the product of the self-aldolization of acetone, which was produced in small amounts. Since, in our system, acetone was present in large excess, the formation of diacetone alcohol did not compromise the selectivity of the reaction. However, the presence of this side product in the final reaction mixtures means that a purification may be required at the end of a potential industrial process.

The concentration of furfural is the first parameter that was investigated. We expected that the different acetone:furfural ratio would have an effect on the C8:C13 selectivity. The chosen reaction temperature was 100 ◦C. This temperature was reached in about 10 min by irradiating the water bath, and it was then kept constant for 20 min. The central value of concentration was 0.1 g of furfural in 5 mL of acetone (0.02 g/mL, 0.21 M), with a furfural:HT weight ratio of 2:1. Two other concentrations were explored, namely 0.11 and 0.42 M, corresponding to half and double the central value, respectively. The volume of acetone and the furfural:HT ratio was kept constant for all experiments. Figure 5 reports the results of the concentration optimization.

**Figure 5.** Influence of furfural concentration (at 100 ◦C, 30 min, furfural:HT 2:1).

The best concentration in terms of activity appears to be 0.21 M. The reaction is slightly slower at lower concentration, perhaps as a result of a slower dehydration, whereas the lower conversion obtained at 0.42 M may be the result of a lower irradiation power per mass of catalyst, which may influence the local temperature of the solid, so fewer or less intense hotspots are formed in the catalyst. Hence, the central concentration value was chosen for further optimization. As expected, the C8:C13 selectivity seems to decrease when the concentration is increased from 54 to 22 to 14. This means that although it is possible to obtain almost complete selectivity to C8 with a somewhat lower concentration, it is not feasible in an HT-catalyzed neat reaction to shift the selectivity to C13, because that would require an excess of furfural and, therefore, to use a solvent.

#### *3.3. Influence of the Catalyst Quantity*

Next, the effect of decreasing the furfural:HT ratio on the reaction outcome was studied (Figure 6).

**Figure 6.** Influence of the furfural:HT ratio (at 100 ◦C, 30 min, concentration 0.21 M).

The reaction rate decreases dramatically when the furfural:HT ratio is increased to 5:1, and only traces of the condensation products are detected. However, with an intermediate ratio of 3:1, the activity is still fairly high, although a consistent part of the converted furfural is still present as the intermediate C8-OH.

#### *3.4. Influence of the Reaction Time*

Then, the formation of the various products over time was studied (Figure 7).

**Figure 7.** Influence of the reaction time (at 100 ◦C, furfural:HT 2:1, 0.21 M).

The conversion increases consistently with the reaction time. After 10 min, the major product is C8-OH, which disappears after 30 min to make way for C8. Aldol condensations are reversible reactions, so even after complete conversion has been reached, it could still be possible to obtain C13 from C8, even if formally furfural is absent, especially when one product is more stable than the other. Nevertheless, no such transformation is seen, and the composition of the mixture remains unaltered for 30 min after complete conversion. In addition, this test shows how the condensation products are substantially stable, and do not deteriorate in the reaction conditions.

#### *3.5. Influence of the Temperature*

The reaction temperature is the next parameter that was studied (Figure 8).

**Figure 8.** Influence of the reaction temperature (30 min, furfural:HT 2:1, 0.21 M).

The reaction at 60 ◦C is considerably slower than the reaction at 100 ◦C. Only traces of condensation products are observed. At 80 ◦C, an intermediate level of conversion is obtained, and the dehydration also appears to be favored by the higher temperature. However, the reaction at this temperature is still much slower than at 100 ◦C, and even when the reaction time is increased to 2 h, the conversion does not match the conversion in the experiment at 100 ◦C for 30 min. This might also be associated with the lower irradiation power. The C8:C13 ratio does not appear to be affected by the lower temperature.

#### *3.6. Choice of the Catalyst*

A blank test was performed to confirm the role of the catalyst in the transformation. Indeed, no conversion and formation of product was seen in our reaction conditions. Other HT-derived catalysts were tested: HT was calcined to MMO, which was tested as a catalyst (Figure 9).

**Figure 9.** Influence of the catalyst choice (at 100 ◦C, 10 min, 0.21 M).

MMO proved to be a more active catalyst than HT, in terms of both furfural conversion and dehydration of C8-OH. As seen for HT, the MMO activity decreases if a 5:1 furfural:catalyst ratio is used. MMO was rehydrated in the liquid phase, and the solid collected after this modification, MX, was used as a catalyst for the condensation. MX was the most active of the catalysts tested, with conversion reaching >99% by the end of the heating step of only 10 min, and C8-OH dehydration being almost complete. Remarkably, the performance of this catalyst at a furfural:catalyst ratio of 5:1 was greater than when HT was used at the same ratio, and even at a 2:1 ratio, in the same conditions. Interestingly, the C8:C13 selectivity obtained with MMO as a catalyst was lower than when the parent catalyst was used (13 vs. 22). The selectivity obtained with MX was not significantly different (26). The solid in the mixture of an MX-catalyzed reaction was recovered by filtration, washed with abundant acetone, and then used as a catalyst in a subsequent reaction. This catalyst (MXrec) proved to be basically inactive; differently from the parent catalyst, which is a white solid, MXrec is a brown powder. Apparently, after reaction, organic matter is deposited on the catalyst surface and deactivates the catalyst.

MMO and MX were also tested at lower temperatures (Figure 10).

**Figure 10.** Performances of the various catalysts at lower temperatures (0.21 M).

The conversion obtained with MMO at 80 ◦C is comparable to the conversion obtained at 100 ◦C, although the reaction time is much longer (30 min vs. 10). The same applies to the C8:C13 selectivity (13 in both cases). Decreasing the reaction temperature to 60 ◦C leads to a somewhat lower conversion, even with a reaction time as long as 1 h. Also, in this case, the activity decreases dramatically when the furfural:MMO ratio is decreased to 5:1, as was observed for the reaction at 100 ◦C. When using MX at 80 ◦C for 30 min the activity decreases, but with a 5:1 ratio, about half of the furfural can still be converted. When the same reaction is performed at 60 ◦C for 1 h, the conversion increases slightly. Using a 2:1 furfural:MX ratio leads to complete conversions and C8-OH dehydrations at these lower temperatures.

Differently from what has previously been reported with conventional heating (almost complete conversion at 100 ◦C in 2 h, excluding the pre-heating, with a catalyst analogous to MX at a 3.25:1 furfural:MX ratio [27]), in our MW-assisted process it was possible to obtain a 70% conversion with a 5:1 furfural:MX ratio with 10 min irradiation in the heating step to the same temperature. Likely, it is the interaction of microwaves with the solid catalysts that is the source of this activity boost. The reaction is also much faster than the reactions catalyzed by the most common heterogeneous catalysts (e.g., Mg–Al and Mg–Zr) in the aqueous phase [18].
