On the influence of Si:Al ratio and hierarchical porosity of FAU zeolites in solid acid catalysed esterification pretreatment of bio-oil

A series of faujasite zeolites with various Si:Al ratio was acquired from Zeolyst International. The catalysts were calcined in static air at 550 °C with a heating rate of 5 °C min⁻¹ for 5 h. The hierarchical zeolite (HFAU30) was obtained by suspending the as-received FAU30 (3.3 g) in an aqueous solution (100 cm³) containing 0.1 M NaOH and 0.2 M tetrapropylammonium bromide (TPABr) at 65 °C for 30 min followed by three consecutive exchanges in aqueous 0.1 M NH₄NO₃ with a solid-to-liquid ratio of 10 g L⁻¹ at room temperature for 8 h. The resulting material was dried overnight at 65 °C and calcined in static air at 550 °C (heating rate = 5 °C min⁻¹) for 5 h. All catalysts were dried at 100 °C for 1 h prior to catalytic tests.

Batch esterification was performed using a Radleys Carousel Reactor Station at atmospheric pressure, unless for reactions with methanol which were performed in Ace pressure flasks equipped with a deep tube and valve, under autogenous pressure. 25 mmol alcohol (p-cresol, n-butanol, benzyl alcohol, anisyl alcohol and methanol, Sigma-Aldrich), 5 mmol of acetic acid (Fisher Scientific), 10 ml toluene (Fisher Scientific) as solvent and 0.5 mmol of dihexyl ether (Sigma-Aldrich) as an internal standard were added to the reactor under stirring at 100 °C. 100 mg of catalyst was also added to the reaction after 1 h drying at 100 °C. For reactions with acid:alcohol ratio = 1 or 5, the amount of acid and alcohol adjusted, respectively. Aliquots of the reaction mixture periodically withdrawn and filtered and analysed on a Varian 3900 GC equipped with a Phenomenex ZB-50 column. All catalytic profiles are an average of three injections per sample. Conversions reported are based upon change in the concentration of acetic acid, with initial rates calculated over the first hour of reaction, wherein the conversion profile was linear. Turnover frequencies (TOFs) were determined from the initial rate of acetic acid conversion (from the linear portion of the reaction profile) normalised to the Brønsted acid site loadings calculated from pyridine adsorption.

Bio-oil from thermal pyrolysis of oak woodchips was provided by the Centre for Research and Technology-Hellas (CERTH). The measurement of bio-oil acid content was carried out according to modified D664A acid number titration method [48]. Briefly, 1 g of bio-oil was added into 100 ml ethylene glycol/water solution (95:5 vol:vol), and then the mixture was titrated with 0.1 N KOH solution in 2-propanol, and the pH of the mixture was monitored by a pH meter (Jenway 3510) until pH 7 was obtained. For the esterification of bio-oil, the equivalent mass of 5 mmol acid-containing bio-oil was mixed with 150 mmol methanol, and 100 mg of catalyst and the reaction was carried out at 100 °C in a sealed Ace pressure flask. After 3 h, the reaction was stopped by cooling down the flask. Then, a sample was withdrawn, filtered and titrated with similar method as explained before (Fig. 1).

Bio-oil characterisation was obtained by GC×GC-ToFMS, after proper dilution in MeOH and without any other pretreatment. The GC×GC analytical system was an Agilent 7890A GC with injector Agilent7683B series (Agilent Technologies, PaloAlto, CA, USA) connected to a Pegasus 4D time-of-flight mass spectrometer from Leco Instruments (St. Joseph, MI, USA). The first chromatographic separation was performed with an apolar column BPX-5 (5% phenyl polysilphenylene-siloxane) 30 m, I.D. 0.25 mm, d.f. 0.25 μm. The second dimensional column was situated in a secondary internal oven and was a BPX-50 (50% phenyl polysilphenylene-siloxane) 1.5 m, I.D. 0.1 mm, d.f. 0.1 μm, both from SGE Analytical Science Pty Ltd. (Australia). Cryofocusing by liquid nitrogen and a quad jet dual stage modulator (Zoex, Houston, TX, USA) was applied. Instrument control, data acquisition and data processing were done by the ChromaToF (Leco) software. The ToFMS operated at an acquisition rate of 100 spectra and a mass range of m/z 45–400 amu. The modulation period was 5 s. The carrier gas (He grade 5) flow rate was 1 ml min⁻¹, injection volume of 0.5 μL at a split ratio of 1:20 and an injection temperature of 250 °C. Temperature programming was performed at an initial temperature of 35 °C of the primary GC oven and was kept stable for 10 min. Then, the temperature increased at a rate of 3 °C min⁻¹ up to 250 °C, and afterwards with a rate of 15 °C min⁻¹ up to the final temperature of 330 °C, where it was kept stable until the end of the program. Total run time was 102 min. The secondary oven was programmed 15 °C ahead of the primary GC oven gradient. Modulator temperature offset was 30 °C. The data acquisition and peak identification were based on the NIST05 library, using as minimum identification criteria similarity of 700 and S/N ratio of 50. The classification of the compounds was performed by borderline group type classification [49].

Recycle tests were performed by recovering the HFAU30 after 3 h esterification of thermal pyrolysis bio-oil via filtration. The spent catalyst was washed with 200 ml methanol and dried at 80 °C overnight. Half of the recovered catalyst was then calcined in a flow of O₂ at 550 °C for 2 h.

Figure 2 shows the N₂ adsorption-desorption isotherms for the series of FAU zeolites under investigation, which show a progressive development of the hysteresis loop and intraparticle mesoporosity as the Si:Al ratio of the zeolites increases. A further enhancement in mesoporosity is observed upon alkaline treatment of FAU30 to generate HFAU30, consistent with previous reports [50, 51]. XRD shown in Fig. S2 confirms that all FAU catalysts are highly crystalline, and that crystallinity of FAU30 sample was preserved after alkali treatment.

XPS analysis was carried out to analyse the surface composition and probe the changes in the surface properties upon variation in Si:Al ratio (Fig. 3). Table S1 reports the Si:Al surface and bulk ratios, as determined by XPS and ICP, respectively, the values for which are consistent, suggesting the aluminium is evenly distributed throughout the zeolite framework. Perturbation of the Si 2p region was observed with increasing Al content, previously attributed to the generation of Si–O–Al species (102 eV binding energy) [52], whose surface concentration was proportional to Al content (Fig. S3). The O 1s spectra also reveals the emergence of a new O state ~531 eV with increasing Al content and is likewise attributed to Si–O–Al species.

Zeolite acidity was probed by NH₃ TPD and pyridine adsorption (FT-IR) to quantify the number and type of acid sites, and the results of which are summarised in Table 1. NH₃ TPD revealed that the acid site density of FAU zeolites decreased significantly with increasing Si:Al ratio (Fig. 4), although there was no systematic variation in acid strength and pyridine adsorption showed minimal change in the corresponding Lewis:Brønsted ratio. Alkali etching to produce the mesoporous HFAU30 material did not significantly alter the acid site loading, but increased the Lewis:Brønsted ratio consistent with surface defects formation upon etching. [53]

IGC was employed to determine the surface polarity of zeolites, which is calculated from the dispersive and specific surface energies shown in Table 2. Surface polarity varies inversely with hydrophobicity and was found to decrease with increasing Si:Al ratio for the FAU samples, dropping considerably for the hierarchical HFAU30 zeolite sample. Considering the surface polarity of hydrophobic materials such as polyethylene (0.0) and PVC (0.05), and hydrophilic Pharmatose P450 (0.81) [54], polarity variations across the FAU family were relatively small, except for the HFAU30. The change in polarity of FAU zeolites is consistent with the expected increased density of ≡Si–O–Si≡ centres as the Si:Al ratio increases [55] and is in good agreement with studies of water adsorption on zeolites, where increased nonpolar ≡Si–O–Si≡ centres leads to increased hydrophobicity [56]. The specific free energy of adsorption of methanol (−ΔG_ads ^SP MeOH) also decreased with increasing Si:Al ratio indicative of a weaker interaction of polar molecules over surfaces with increased ≡Si–O–Si≡ groups. Surface polarity is determined by comparing the elution of (relatively small) polar and (relatively bulky) alkane probe molecules through each zeolite. Fast elution indicates a poor interaction with the surface and hence a dissimilar chemical nature between probe molecule and zeolite. Enhanced transport of alkanes purely through mesopore introduction would therefore result in an apparent increase in surface polarity, the opposite to the change observed in Table 2, wherein HFAU30 exhibits a decreased surface polarity relative to its microporous counterpart. The greater hydrophobicity of HFAU30 relative to FAU30 is also evident from the lower −ΔG_ads ^SP MeOH of the former and hence is not an artefact of superior mass transport due to mesopores. The lower surface polarity of HFAU30 may well reflect increased surface roughness due to the etching pretreatment. Increasing zeolite hydrophobicity may promote esterification through the displacement of reactively formed water from active sites which may otherwise cause reversible site blocking [57].

In situ, ATR-IR studies of acetic acid adsorption were further undertaken to assess the impact of Si:Al ratio on the adsorption behaviour of the parent FAU zeolites. Figures S4–S6 show the resulting vibrational spectra obtained during acetic acid adsorption from methanolic solution. Rapid surface saturation was observed within 10 min of acetic acid addition to a methanol flow as seen by the growth and plateauing of a 1710 cm⁻¹ band assigned to acetic acid hydrogen bonded via C=O to surface hydroxyls, a broad band at 1400 cm⁻¹ in the region for δ_asym of CH₃ and υ_sym of COO⁻ and a 1270 cm⁻¹ band attributed to the C–O band (γ_C-O) [58]. The concomitant loss of surface methoxy groups observed as a broad negative peak between 3100 and 2800 cm⁻¹ (Fig. S7) suggests that acetic acid displaces methanol from the zeolite surface. The simultaneous growth of acetic acid features and loss of methoxy species is indicative of competitive adsorption between methanol and acetic acid and consistent with a Langmuir–Hinshelwood surface reaction model for esterification.

Figure 5 shows the integrated peak area of the γ_C-O (1270 cm⁻¹) band during acetic acid adsorption/desorption over the FAU zeolites and reveals that the saturation coverage of acetic acid is inversely proportional to Si:Al ratio (Fig. 5 inset) and hence directly proportional to acid site density and surface polarity, consistent with vapour phase IR measurements [58] that suggest only zeolitic bridging hydroxyls Si–OH–Al (zeol) are able to reactively adsorb acetic acid via:

The utility of FAU zeolite family of materials for the esterification of acetic acid with various alcohols (Fig. 6) was subsequently explored in toluene as a simulated bio-oil matrix. Reaction profiles are shown in Fig. S8.

Figure 7 illustrates the performance of the FAU zeolites with different Si:Al ratios and the hierarchical version of FAU30, in esterification of acetic acid with various alcohols. The first striking observation is that methanol is the most reactive alcohol among the tested alcohols, followed by anisyl alcohol, benzyl alcohol, p-cresol and butanol. The higher reactivity of methanol is explicable by a Langmuir–Hinshelwood bimolecular surface reaction model in which coordination of the alcohol and the organic acid at adjacent Brønsted acid sites are necessary and likely reflect both electronic and steric effects. For alcohols, increased alkyl chain length (or addition of other organic groups such as aromatic rings) is accompanied by enhanced electron-donating properties which are expected to destabilise the protonated transition state, hence slowing esterification. Smaller alcohols are also expected to adsorb more weakly relative to longer chain alcohols (of higher pKa which experience greater dispersive interactions with any surface), with the latter favouring alkoxide formation and concomitant site blocking and hence lower reactivity. More sterically demanding alcohols may also experience poorer configurational access to protonated (or activated) carboxylates due to increased distance from neighbouring acids sites. Diffusion limitations may also restrict the accessibility of in-pore acid sites for bulky alcohols contributing to their lower TOFs [59]. Irrespective of the alcohol type or saturation acetic acid coverage, the per site activity increased with Si:Al ratio, correlating with increasing hydrophobicity (Fig. 8) which thus emerge as the most important factors controlling catalytic activity. Product selectivity to the acetic acid ester was >99% for methanol, n-butanol and p-cresol; however, significant side products were observed for benzyl and anisyl alcohols. For benzyl alcohol, dibenzyl ether (the product of benzyl alcohol etherification) and an alkylation product formed, whereas anisyl alcohol only yielded the ester and ether: anisyl alcohol was more susceptible to etherification than benzyl alcohol. Figure S9 shows that the Si:Al ratio had little impact on the ester selectivity in acetic acid esterification with benzyl or anisyl alcohols. HFAU30 was subsequently evaluated for acetic acid esterification with methanol and benzyl alcohol to assess the impact of mesoporosity on mass transport. This revealed that a 30–50% increase in TOF for both alcohols as a result of increased acid site accessibility (and hydrophobicity).

The effect of acid:alcohol ratio on activity was also investigated, with reaction profiles shown in Fig. S10. Figure 9 reveals that increasing this ratio from 1:5 → 5:1 progressively increased TOFs, consistent with a Langmuir–Hinshelwood bimolecular surface reaction, wherein strongly bound alcohol inhibits esterification [59].

The recyclability of FAU30 and HFAU30 was explored in acetic acid esterification with methanol in presence of toluene without any regeneration between the runs. No loss in catalytic activity was observed after three consecutive runs, indicating excellent recyclability of these catalysts under these reaction conditions.

The performance of FAU zeolites towards esterification pretreatment of a real bio-oil obtained by thermal pyrolysis of oak woodchips was subsequently explored in pressure flasks using excess methanol at 100 °C. Figure 10 shows the resulting carboxylic acid conversions determined by titration. Acid conversion increased with Si:Al ratio and upon the introduction of hierarchical porosity, mirroring the observations from acetic acid esterification with model alcohols, albeit with lower conversions than observed for the simulated bio-oil, likely as a result of the water content of the pyrolysis oil (which may reversibly poison the acid sites) or presence of bulky organic components (e.g. oligomers/polymers) which may induce pore blockage [60].

In accordance with the results obtained for acetic acid esterification, HFAU30 was the most active catalyst for bio-oil esterification, and the esterified product from this catalyst was therefore analysed by GCxGC-ToFMS to identify products. Figures 11 and 12 show chromatograms of the initial and esterified bio-oils, respectively. In both cases, the chromatographic space is divided into six discreet areas: acids and esters, aldehydes and ketones (including furanoics and cyclic carbonyls), hydrocarbons (saturated and unsaturated nonaromatic), aromatic hydrocarbons, phenolic compounds, and sugars. Compounds that were not identified by the library and/or did not meet the required identification criteria were classified as ‘unidentified’. Based on these criteria, a more detailed group classification of the compounds along with their relative chromatographic area is presented in Table 3.

Comparing the chromatograms of the initial (Fig. 11) and esterified bio-oil (Fig. 12), it is obvious that catalytic pretreatment altered the peak distribution in different chromatographic areas. In particular, there is a significant reduction in the number of phenolic, aldehyde and ketone species [61], as quantified in Table 3. A few aromatic hydrocarbons detected in the initial bio-oil were also absent from the esterified one. A pronounced decrease in acetic acid (in the acids area) and levoglucosan (in the sugars area) was apparent from the esterified bio-oil chromatogram. The main peak in Figure 12 is attributed to methyl acetate (representing 34.3% of the chromatographic area), verifying successful esterification of the principal acid present in the bio-oil. Smaller peaks corresponding to methyl propanoate and hexanoate were also detected. A peak attributed to glycolaldehyde dimenthyl acetal corresponds to the second highest peak area of the esterified bio-oil chromatogram. Previous studies have demonstrated that in alcohol media, levoglucosan can be transformed by acid catalysts to methyl levulinate with intermediate glycolaldehyde (GA) formation [62, 63]. However, under such conditions, GA may be partly deactivated via glycolaldehyde dimethyl acetal (GDA) formation. We propose that levoglucosan in the initial bio-oil was converted to GA (minor peak representing 0.2% of the chromatographic area, Fig. 12), which in turn formed GDA (major peak representing 32.2% of the chromatographic area, Fig. 12); methyl levulinate was not detected in the esterified bio-oil.

Deactivation of HFAU30 following esterification of thermal pyrolysis bio-oil was subsequently explored. Spent catalyst was divided into two portions: one washed with methanol and the other recalcined after a methanol wash with the goal of completely removing any carbon deposits. Temperature-programmed oxidation of an untreated spent catalyst revealed the presence of 11 wt% carbonaceous residues post-reaction. Reuse of the methanol-washed catalyst revealed around 65% of its initial activity was lost. In contrast, recalcination fully regenerated the activity of the fresh HFAU30 catalyst, evidencing site blocking by strongly adsorbed carbonaceous species as the principle deactivation pathway.