Esteriﬁcation of Levulinic Acid to Methyl Levulinate over Zr-MOFs Catalysts

: At present, the trend towards partial replacement of petroleum-derived fuels by those from the revaluation of biomass has become of great importance. An effective strategy for processing complex biomass feedstocks involves prior conversion to simpler compounds (platform molecules) that are more easily transformed in subsequent reactions. This study analyzes the metal–organic frameworks (MOFs) that contain Zr metal clusters formed by ligands of terephthalic acid (UiO-66) and aminoterephthalic acid (UiO-66-NH 2 ), as active and stable catalysts for the esteriﬁcation of levulinic acid with methanol. An alternative synthesis is presented by means of ultrasonic stirring at room temperature and 60 ◦ C, in order to improve the structural properties of the catalysts. They were analyzed by X-ray diffraction, scanning electron microscopy, infrared spectroscopy, X-ray photoelectron spectroscopy, microwave plasma atomic emission spectroscopy, acidity measurement, and N 2 adsorption. The catalytic reaction was carried out in a batch system and under pressure in an autoclave. Its progress was followed by gas chromatography and mass spectrometry. Parameters such as temperature, catalyst mass, and molar ratio of reactants were optimized to improve the catalytic performance. The MOF that presented the highest activity and selectivity to the desired product was obtained by synthesis with ultrasound and 60 ◦ C with aminoterephthalic acid. The methyl levulinate yield was 67.77% in batch at 5 h and 85.89% in an autoclave at 1 h. An analysis of the kinetic parameters of the reaction is presented. The spent material can be activated by ethanol washing allowing the catalytic activity to be maintained in the recycles.


Introduction
In recent decades, due to the finite reserves and environmental pollution problems of fossil energy, governments have put forward the "energy strategy", stimulating the utilization of renewable energy and resources [1]. The development of new technologies for producing energy and chemicals from them has prompted biomass valorization to become an important area of research. Biomass provides an ideal alternative to fossil resources; indeed, biomass is the only sustainable source of organic compounds and has been proposed as the ideal equivalent to petroleum for the production of fuels and chemicals [2]. In general, biomass is defined as all organic matter, including crops, food, plants, and

Synthesis of MOF UiO-66 and UiO-66-NH 2
MOFs UiO-66 and UiO-66-NH 2 were obtained by solvothermal method [21,31,32]. Two different binding agents of terephthalic acid (BDC) or aminoterephthalic acid (NH 2 -BDC) were used. The synthesis was carried out by magnetic and ultrasonic stirring, the latter at room temperature or 60 • C. The samples were named as M X , M X -U, or M X -UT, where "X" can be "A" related to synthesis with BDC or "B" related to synthesis with NH 2 -BDC. The letter U indicates the synthesis on ultrasound at room temperature. The letters U-T indicate the synthesis on ultrasound at 60 • C of temperature.
The synthesis consisted of dissolving 1.042 g in 50 mL g of ZrCl 4 in 50 mL of the solvent (DMF); the solution was magnetic or ultrasonic stirred for 5 min. In the case of ultrasound synthesis, the container is placed inside ultrasound equipment, which works at room temperature or at 60 • C. Then, 0.708 g of terephthalic acid or 0.788 g of aminoterephthalic acid was incorporated. The mixture was kept under stirring (magnetic or ultrasonic) for 30 min. The gel was transferred to a Teflon-lined stainless-steel autoclave and kept in an oven at 120 • C for 24 h. The material obtained was immersed in DMF and then washed with acetone with a lower boiling point and easily removed. The precipitate was separated by centrifugation. Finally, the solid was dried at 90 • C for 24 h, obtaining~1.404 g and 1.735 g of a white (A) or yellow (B) solid powder, respectively.
Infrared analyses were carried out on a spectrophotometer Smartomi-Transmission Nicolet iS10 Thermo Scientific (Thermo Scientific, Waltham, MA, USA) in a range of 4000-400 cm −1 .
The micrographs of the mixed oxides were obtained by SEM instrument model JSM-6380 LV (JEOL, Japan) equipped with a Supra 40 (Carl Zeiss, Oberkochen, Germany), the samples were metalized with chromium.
The specific surface area (SSA), analysis was carried out in an ASAP 2020 instrument (Micromeritics, Norcross, GA, USA) and was calculated by the Brunauer-Emmett−Teller (BET) method.
X-ray photoelectron spectra (XPS) were recorded with a PHI VersaProbe II Scanning XPS Microprobe (Physical Electronics, Chanhassen, MN, USA) with scanning monochromatic X-ray Al Kα radiation as the excitation source (100 µm area analyzed, 52.8 W, 15 kV, 1486.6 eV), and a charge neutralizer. The pressure in the analysis chamber was maintained below 2.0 × 10 −6 Pa. High-resolution spectra were recorded at a given take-off angle of 45 by a multi-channel hemispherical electron analyzer operating in the constant pass energy mode at 29.35 eV. Spectra were charge referenced with the C 1s of adventitious carbon at 284.8 eV. The energy scale was calibrated using Cu 2p 3/2 , Ag 3d 5/2 , and Au 4f 7/2 photoelectron lines at 932.7, 368.2, and 83.95 eV, respectively. The Multipack software version 9.6.0.15 was employed to analyze in detail the recorded spectra. The obtained spectra were fitted using Gaussian-Lorentzian curves to more accurately extract the binding energies of the different element core levels.
In order to carry out the elemental analysis, samples were acid-digested in a closedvessel microwave oven (SCP Science, Baie-D'Urfe, QC, Canada). The chemical quantification was performed by microwave plasma atomic emission spectroscopy (MP-AES), using an Agilent 4200 MP AES instrument (Agilent, Santa Clara, CA, USA).
To study surface acidity, the adsorbed CO molecules were measured by FTIR analysis at two different temperatures (50 • C and 100 • C) in a Nicolet iS10 instrument. Initially, samples were treated under vacuum and then were put in contact with CO molecules during 2 h at room temperature.

Catalytic Esterification Reaction
The synthesized MOFs were used for the catalytic esterification of levulinic acid and methanol (Figure 1), with a 1:15 molar ratio and 0.05 g of the solid [25,[33][34][35][36]. Two reaction systems were used. On the one hand, the reaction was carried out in a glass batch reactor with magnetic stirring at 65 • C, coupled to a reflux condenser. The total time was 5 h and 0.15 mL of samples were collected from the reaction medium at 0, 1, 2, 3, 4, and 5 h. The time t = 0 h was the time at which the temperature of the reaction medium reached 65 • C. Samples were collected using a microsyringe equipped with a filter (Whatman paper filter n • 5) to remove catalyst particles. On the other hand, a pressure system was used constituted of an autoclave reactor. The reactor was placed inside an oven at 130 • C with magnetic stirring at 400 rpm. It was previously purged with N 2 and then pressurized to 30 bar. The reaction products were analyzed by gas chromatography and mass spectrometry. It was used on an Agilent Technologies 7820A instrument equipped with a HP-20M column and a Perkin Elmer Clarus 560 instrument. We calculated the LA conversion (%C), selectivity (%S), and yield (%Y) of products and identified n (LA)I and n (LA)F as the initial and final concentrations of levulinic acid, and n (ML) as the concentration of methyl levulinate.

Physicochemical Characterization
3.1.1. X-ray Diffraction Figure 2 shows the X-ray diffraction patterns of the MOFs. Two peaks located close to 2θ of 7.4, 8.5, and 25.6 • are associated with the diffraction by (1 1 1), (2 0 0), and (600) planes characteristic of the MOF UiO-66-NH 2 . [22,[37][38][39]. The M A -U and M A -UT samples show an exchange in the intensity of the main peaks (111) and (002) and the appearance of a new peak at 2θ angle of~17.6 • related to a phase change, is the same as that reported in the literature [22]. This behavior could be promoted by temperature and ultrasound applied in the synthesis. On the other hand, the XRD patterns in series B show peaks of similar intensity with good crystalline order, regardless of the applied treatment [40].

FTIR Spectroscopy
The chemical states and functional groups of the different samples were investigated by FT-IR ( Figure 3). The peaks at 1586 and 1395 cm −1 are attributed to the asymmetric and symmetric stretching vibrations of the COO − groups in the terephthalic acid, and the peak at 1506 cm −1 is the typical peaks of the C=C peak of the vibration of the aromatic ring. The two lowest frequencies at 488 cm −1 belong to the stretching vibrations of Zr-O in the Zr 6 cluster, and the band at 551 cm −1 is assigned to the asymmetric stretching vibration of Zr-(OC). For the case of materials with the BDC-NH 2 ligand, the signals at 3449 and 3348 cm −1 corresponded with the amine symmetric and asymmetric stretching bands, while the signals at 1258 and 1386 cm −1 were attributed to C-N binding absorption [41][42][43].

SEM-EDS Analysis
The morphology was characterized by a scanning electron microscope (SEM). Figure 4 shows the synthesized UiO-66 of regular octahedral structure, smooth surface, and good dispersity.
The UiO-66 samples (series A) exhibit a narrow crystal size distribution ranging between 285.7 nm and 488.5 nm, while the UiO-66-NH 2 samples (series B) exhibit smaller particles with a mean crystal size between 209.7 and 406.0 nm ( Figure S1).
The Zr, C, O, and N elemental mapping images ( Figure S2), show dispersion and homogeneous distribution of these elements on the catalyst surface [44,45].

Microwave Plasma Atomic Emission Spectrometry (MP-AES) and the Surface Area by BET
Through elemental analysis, the percentage of Zr present in the bulk of the catalyst was determined (Table 1). It is observed that the samples that had a temperature in the synthesis managed to anchor a greater amount of the Zr. Furthermore, comparing the binding agents, those with BDC have a higher % Zr than the others.
To investigate the textural characteristics, N 2 adsorption−desorption isotherms for UiO-66 and UiO-66-NH 2 were measured and these are shown in Figure 5. All materials exhibit type Ib isotherms, where the pore size distributions are in a range including micropores wider than ∼1 nm and narrow mesopores (of width < ∼2.5 nm) [46]. The N 2 uptake at the low-pressure range (P/P 0 < 0.3) suggests the existence of micropores, while the uptake at the middle-pressure range (P/P 0 > 0.5) with hysteresis between adsorption and desorption branches is characteristic of mesopores. For the sample M A, which was synthesized without ultrasonic and temperature treatment, no hysteresis loop was observed on the isotherm, indicating a typical microporous structure. The rest of the samples show H1 type hysteresis loops [46], which at low and high relative pressure belong to disparate types of isotherms characteristic of solids with microporous windows and partial mesoporous cages [27,47]. Ultrasonic and temperature treatment applied, have decreased the pore size from 2.64 nm to 2.30 nm for series A and from 2.36 nm to 1.07 nm for series B. These comparative results suggest that the ultrasonic and temperature treatments have a significant influence on the pore structures, promoting the presence of a narrow mesopore and micropore distribution. Pore size and surface area in the series B samples have been largely reduced compared with that of series A. This might be because of the introduction of amino groups (-NH 2 ) in the BDC-NH 2 blocking the pores [48]. The surface area reported by the literature indicates higher values than reported in this work, along with lower pore volume values. Wang et al. [49], Su et al. [50], and Hu et al. [48], were reported a surface area of 1110 m 2 g −1 , 967 m 2 g −1 , and 1525 m 2 g −1 , respectively, while the pore volume reported was 0.60 cm 3 g −1 , 0.57 cm 3 g −1 , and 0.66 cm 3 g −1 corresponding to micropore size distribution. The lower values of pore volume reported in the literature could be attributed to the different time and temperature autoclave steps, as well as the different ways and times stirring was applied in this work.

XPS-Spectroscopy
The surface composition of the studied catalysts was studied by XPS. Table 2 shows this composition in atomic concentration %. As expected, all samples exhibit a relatively high concentration of C, typical of MOFs materials. The % of C slightly decreases upon using ultrasound and after thermal treatment. Samples of series A show a small % of N due to the use of DMF in the synthesis, without being part of the structure. The % of Zr increases upon using ultrasound. In the case of the catalysts of series B, as expected, the % of N increases due to the presence of the amino group. This percentage increases when using ultrasound. The XPS spectra are shown in the Supplementary Material ( Figure S3), and in Table 3, the binding energy values in eV for the C 1s, Zr 3d 5/2 , O 1s, and N 1s signals are included. The catalysts of series A show C 1s core-level spectra composed of two contributions at 284.8 and 288.8 eV. The former and more intense is assigned to -C-C-and -C=C-bonds of adventitious carbon and mainly to the aromatic ring, and the latter to the presence of carboxylic groups. The Zr 3d 5/2 maxima appear at about 182.8 eV, typical of Zr(IV). The O 1s core-level spectra can be decomposed into two contributions at about 530.1 eV and 531.8 eV. The latter and more intense is assigned to carboxylic oxygen, and the former to lattice oxygen bonded to Zr. The observed higher relative intensity of the contribution at 531.8 eV for samples synthesized with ultrasound is concomitant with the increase in the %Zr (see Table 2). The catalysts of series B show N 1s binding energy values typical of amino groups. The C 1s core-level spectra of these catalysts with amino functional groups show a new contribution at 286.0 eV due to C-N bonds [51].

Acid Analysis by FTIR-CO
The acid sites were studied by FTIR CO adsorbed at different temperatures (50 and 100 • C). The Lewis and Brønsted acid sites were measured by the relative area percentage ( Table 4). The sites measured at 50 • C are defined as weak, while those measured at 100 • C are defined as medium [52,53]. The graphs are attached in the Supplementary Material ( Figure S4). Making an analysis of the acid sites present, it can be observed that when using ultrasound in the synthesis, an increase in the amount of Lewis acid sites is produced, which could be related to the better catalytic results obtained with said materials. As the acid-base duality is also needed, those with BDC-NH 2 are the ones that produce the best conversion [54].

Catalytic Activity: Esterification for Methyl Levulinate Production
Esterification products were identified by gas chromatography with FID and mass spectrometry. The main esterification product was methyl levulinate (ML) and the byproduct was β-lactone from the dehydration of levulinic acid [55][56][57][58][59]. A control experiment was carried out in the absence of catalyst, which gave a negligible methyl levulinate yield of only 1.56%, attributed to the auto-catalysis reaction due to the acidity of levulinic acid [60].
The M B -UT presented the best yield to the ML (Table 5). These esterification reactions with the presence of the amino group, occur through a dual activation mechanism, related to an acid-base site, in which the Zr sites (acid site) interact with the adsorbed LA while the amino group (basic site), near the Zr, forms an adduct linked to hydrogen in the alcohol molecule, which increases the nucleophilic character of the O atom, favoring the first reaction step [30,[33][34][35]. Due to this behavior, series A without the basic sites (-NH 2 ), showed a lower yield than series B. By XPS, M B -U showed a higher Zr (5.63%) content indicating Lewis acid sites, while M B -UT contains a higher content of N (5.09%) by -NH 2 groups, which would indicate more number of basic sites. Better catalytic results are shown in M B -UT, probably due to the excess of Lewis acid sites as shown in the relation L/B in Table 4 and related with the Zr at the surface as shown in the XPS result. That would ensure that each basic site (-NH 2 ) is close to a Zr, generating a synergistic effect by acid-base duality in the catalyst.
To improve the catalytic performance, some reaction parameters were modified. Prior to it, the catalyst was washed with acetone using ultrasound, centrifuged, and dried at 90 • C, to take out the solvent present in the pores. This allowed the conversion and yield rate to be increased to 70.87% and 67.77%, respectively.

Effect of Reaction Parameters on the LA Esterification with Methanol Catalyzed by M B -UT
In order to determine the optimal conditions in the catalytic activity of M B -UT for esterification from LA to ML under reflux conditions, the roles of temperature and reaction time, catalyst loading, and molar ratio of LA to methanol were studied.

Effect of Catalyst Loading
The catalyst mass was varied based on 0.05 g. The catalytic reaction was tested with half, double, and triple of said mass ( Table 6). As the catalyst mass increases, the conversion also increases, but this slight increase is not economically justified. Once all the active surface area of the catalyst is occupied by the reactants, the addition of a greater quantity of mass blocks part of its active sites. For this reason, a catalyst mass of 0.05 g was taken as the optimal value.

Effect of Temperature
The original reaction was performed at a temperature close to the boiling point of methanol, so that it does not evaporate while it occurs. Then, the reaction temperature was modified to 55 • C, 60 • C, and 70 • C ( Table 7). Although the reaction at 70 • C has the highest conversion, the temperature increase of 5 • C was not justified to increase only 3% of conversion; therefore, the optimal working temperature is 65 • C, a temperature with which a good conversion to the desired product is ensured and also the evaporation of one of the reagents is avoided, which could change the molar ratio between them.

Effect of Molar Ratio
Regarding the molar ratio of the reagents, we initially worked with a 1:15 ratio of LA: methanol. Variations were performed at a more concentrated point (1:10) and a more dilute point (1:20) (Table 8). In this case, the optimal value corresponds to a ratio of 1:15.

Kinetic Model and Estimation of Kinetic Parameters
The esterification reaction of levulinic acid (LA) in the presence of methanol (M) to produce methyl levulinate (ML) and water (W) is given in the reaction below: The adjustment was made with the experimental data obtained in the batch reaction catalyzed by the M B-UT material. To determine the reaction kinetics, an irreversible reaction and a pseudo-homogeneous model were considered [25,[61][62][63]. Since the alcohol (methanol) is in excess, a zero-order reaction was assumed for it. According to this, the reaction rate is expressed in Equation (5) Furthermore, with the variation of the reaction temperature and by using the Arrhenius Equation (7), it was possible to determine the activation energy and the heat of the reaction [61].
where Ea, A, and Rg are the activation energy (J/mol), constant "pre-exponential or frequency factor" and ideal gas constant (8.3143 J/mol K), respectively. Both the frequency factor A and the activation energy Ea, are obtained by non-linear regression. The plot of ln k vs. 1/T is plotted by a straight line ( Figure S6), the activation energy (Ea) is the slope of the line, and the pre-exponential factor (A) is the ordinate at the origin. The experimental data were fitted, with pseudo-first-order kinetics, through Equation (6). With the adjustment, a constant kinetic value was obtained k = 3.57 × 10 −3 min −1 with an R 2 of 0.98. As for the activation energy (Ea), adjusted with Equation (8), it yielded a value of 48.99 kJ/mol with an R 2 of 0.98. Additional data have been included as electronic Supplementary Material (Tables S1 and S2, Figures S5 and S6).

Esterification of Levulinic Acid to Methyl Levulinate in Pressure System, Study of the Stability
For the pressure system, the values obtained are shown in Table 9. It is observed that the conversion and yield results at 1 h of reaction were better for all materials, compared to the batch system at 5 h. The presence of pressure and high temperature in the autoclave system promotes the diffusion of the substrates through the pores of the material and their coordination with the active acid-basic sites, favoring the catalytic performance [41,[64][65][66][67]. To study the stability of the catalyst, three recycles of the reaction were carried out with sample M B-UT (M B -UT R1, M B -UT R2, M B -UT R3); in each recycle, the catalyst was separated from the reaction mixture and dried at 90 • C. It was observed that the yield decreases while the selectivity is maintained in the first two recycles and falls in the third (Table 9).
By MP-AES, the concentration of Zr in the recycles supernatant was measured, finding values between 5 ppm and 24 ppm, that values represent~(0.5-2.9)% Zr p/p leached from the solid in each recycle. The activity of the reaction in a homogeneous phase in the presence of Cl 4 Zr was also measured, obtaining 98.24% yield to ML. This indicates that the presence of Zr leached in the supernatant did not generate a homogeneous phase reaction or increase the conversion in the recycles as shown in Table 9.
The decrease in catalytic activity can be mainly attributed to the partial blocking of Zr +4 sites by electrostatic attraction with ML molecules which cannot be removed by methanol in successive recycles. To support this explanation, after the third recycle catalyst, two reactions with ethanol (M B -UT R4, M B -UT R5) were carried out, obtaining an increase in the conversion (58.66-76.03%) with 100% selectivity (Figure 6). This behavior is attributed to the removal of methyl levulinate molecules on the surface catalyst due to the greater affinity and solubility of ethanol molecules, releasing the Zr +4 acid sites.
On the other hand, the M B -UT R3 catalyst was washed with ethanol and ultrasonic stirring, then dried at 90 • C and analyzed by FTIR. Figure 7 Figure S7. The intensity and the crystalline order of the MOFs phase were maintained after each recycle and mainly after the third one.
Finally, a decrease in catalytic activity with the successive reuse of the catalyst could be mainly related to the blockade of Zr +4 acid sites by ML and to a lesser extent by the leachate of this metal.

Conclusions
The synthesized materials presented outstanding crystallinity and porosity properties. The implementation of ultrasound and temperature in the synthesis promoted the formation of a micro/mesoporous structure, with crystal sizes between~209.7 and~488.5 nm and a homogeneous distribution of Zr, C, O, and N on the surface, which can be corroborated by BET and SEM analyses. Regarding the catalytic evaluation in a batch system, the materials of each series synthesized with ultrasound at 60 • C exhibited the highest conversions, where the M B -UT sample is the one that presented the best activity, with a yield of 67.77% for the methyl levulinate. The optimal working conditions for this were 65 • C, 0.05 g of the catalyst and a mol ratio of reactants of 1:15 (LA:M). These results are associated with the greater dispersion of Lewis acid sites available to catalyze the reaction and also with the acid-base duality due to the -NH 2 group. This dual mechanism involves LA interacting with Zr (Lewis acid site) while the alcohol forms a hydrogen bond with the amino group (basic site). This simultaneous acid-base interaction favors the first reaction step.
A pseudo-homogeneous model was adjusted to first-order kinetics with a value of the reaction constant k = 3.57 × 10 −3 min −1 and activation energy of 48.99 kJ/mol, with an R 2 of 0.98 for both parameters.
Improvements in the catalytic results were achieved, using more severe conditions, through the use of pressure reactors with greater selectivity to the desired product and reduction in the reaction time from 5 to 1 h. The deactivated catalyst observed in the successive recycles is mainly related to the blocking of the Zr +4 acid sites by electrostatic attraction of the ML molecules and, to a lesser extent, to the leaching of this metal. The spent material can be activated by ethanol washing allowing the catalytic activity to be maintained in the recycles.
Finally, it can be said that the synthesis of levulinate compounds from levulinic acid derived from biomass using heterogeneous catalysts such as UiO-66 and UiO-66-NH 2 is a highly viable and efficient alternative, which allows eco-compatible reactions to be carried out with the medium, with easy separation of the products of interest.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/chemengineering6020026/s1, Additional information about the XPS graphics, the particle size, the EDS images, the XRD pattern for the used catalyst, the CO-FTIR absorption spectra, the kinetics of the reaction, and the adjustments made have been included in the file supplied as Supplementary Materials.