Identifying the Role of Brønsted and Lewis Acid Sites in the Diels‐Alder Cycloaddition of 2,5‐DMF and Ethylene

The role of Lewis and Brønsted acid sites in the Diels‐Alder cycloaddition (DAC) of ethylene to 2,5‐dimethylfuran (2,5‐DMF) to p‐xylene was investigated. Amorphous silica catalysts containing Al3+ (ASA), Ga3+ (ASG), and In3+ (ASI) were prepared via homogeneous deposition‐precipitation. Silica modified with Zr4+ (ASZ) was prepared by impregnation. Their acidic properties were characterized by various IR and NMR spectroscopic techniques. Measurements using pyridine as a probe molecule highlighted the presence of mostly Lewis acid sites (LAS) in all materials. Using CO as a probe, in contrast, demonstrated the existence of Brønsted acid sites (BAS) in ASA and ASG, which were nearly absent in ASI and ASZ. Differences in basic strength can explain the contrast in results observed between the two probe molecules. The highest p‐xylene yield (~20 %) in the DAC reaction, could be achieved with ASA and ASG. The lack of BAS in ASI and ASZ resulted in inferior performance in the DAC, with p‐xylene yields below 5 %. These results indicate the importance of BAS for the DAC reaction. Several other heterogeneous and homogeneous catalysts were explored for the DAC reaction to show the generality of our conclusion that BAS play a critical role in obtaining p‐xylene from 2,5‐DMF and ethylene.


Introduction
[3][4][5][6][7] Therefore, it is highly compatible with the demands of the current terephthalic acid (TA) processes, which require high isomeric purities. [8]This renders 2,5-DMF a suitable drop-in molecule for the production of TA, which is a key polyethylene terephthalate (PET) monomer.PET is one of the most recycled plastics in the world (~80 % of PET bottles) [9] , suggesting that only a (relatively) small fraction of virgin polymer is needed.This may provide an incentive to replace fossil-based precursors with biobased alternatives for PET production.[3][4][5][6] It is widely recognized that the HOMO-LUMO gap between diene and dienophile determines the reaction rate, which Lewis or Brønsted acids can enhance. [10,11]On the other hand, recent studies have shown that the Pauli repulsion between the πorbitals of the diene and dienophile influences the DAC rates. [12,13]However, most density functional theory (DFT) calculations on DAC reactions to obtain p-xylene predate these new insights.[16][17][18][19] Generally, BAS are considered to be only effective in the dehydration of the DAC intermediate, while providing no apparent enhancement in the actual cycloaddition step. [18,19][16][17] However, no systematic trends were observed by the DFT calculations.It has been mentioned that strong Lewis acids, such as Li + and K + -loaded zeolites, can facilitate the dehydration of the cycloadduct. [16]Several studies demonstrate high p-xylene selectivities with Lewis acidic catalysts, such as metal-modified zeolites, [29][30][31][32] phosphated zeolites, [33] and homogeneous metal triflates. [34]However, most of these works did not investigate the acidity in detail, some of these catalysts may contain small amounts of BAS.[37][38] The oxanorbornene intermediates derived from these activated reagents were stable enough to be isolated in high purity due to the stabilizing effects of the functional groups.The final aromatic product can be obtained in a separate dehydration step using Brønsted acids.Such a two-step process could be envisioned for p-xylene production as well.Although many experimental and theoretical studies investigate the role of the active sites, a comparative investigation between LAS and BAS has, to the best of our knowledge, not been reported yet.
Herein, we investigate silica-based catalysts modified to impart them with distinct Lewis or Brønsted acidic properties for the DAC reaction.Previously, homogeneous depositionprecipitation (HDP) has been employed to synthesize well-defined amorphous silica-alumina (ASA) by grafting different amounts of Al on silica. [28,40]ASA are generally weaker Brønsted acids than zeolites, resulting from a comparatively small amount of zeolite-like strong acid sites next to a large number of weaker BAS due to the interaction of strong Lewis acidic Al 3 + species with silanol groups (Figure 1). [41]Replacing Al 3 + with other Lewis acid metal cations like Ga 3 + and In 3 + can weaken the M 3 + -HOSi interactions and, therefore, weaken Brønsted acidity or generate Lewis acid sites.A zirconium-modified silica was also prepared, as these materials have been used as efficient Lewis acid catalysts for the DAC reaction. [29,30]The acidic properties of these materials were investigated by IR spectroscopy using CO and pyridine as probe molecules and magic angle spinning (MAS) NMR.Recently, NMR measurements with 15 N-labelled pyridine demonstrated its potential in characterizing zeolite acid sites. [42,43]Here, we employ this approach to investigate the acid sites of the amorphous silica-based materials.The catalytic performance in the DAC reaction of 2,5-DMF with ethylene was investigated in a batch reactor at a temperature of 200 °C and a pressure of 50 bar.

Results and Discussion
Amorphous silica-alumina (ASA) materials can be readily synthesized via homogeneous deposition-precipitation (HDP). [28,44]Herein, the HDP method is extended to graft Ga 3 + and In 3 + , next to Al 3 + , from their corresponding nitrate salts on fumed silica.Table 1 shows that the resulting ASA analogs, denoted here as ASG and ASI, containing other trivalent cations than Al 3 + can be obtained in this manner.The pH evolution during the HDP of Al 3 + , Ga 3 + , and In 3 + is shown in Figure S1.The synthesis of ASG proceeds at a comparable rate as comparably to ASA, while the grafting stage with ASI is considerably faster.For all materials, the amount of metal corresponds reasonably well with the targeted Si/M ratio of 10.The calcined metal-loaded samples have lower surface areas than the parent silica (Table 1), and the surface area decreases with decreasing atom size in the order ASZ > ASI > ASG > ASA.The total pore volume (V p ) increases slightly upon grafting Al 3 + , Ga 3 + and In 3 + , whereas a lower V p is obtained with the larger Zr 4 + .The pore size distributions derived from the NLDFT method (Figure S3) emphasize the mesoporous nature of the parent silica and the metal-modified samples.
The XRD patterns are dominated by the broad reflection of amorphous silica without substantial crystalline phases other than minor contributions of metal oxides (Figure S4).These findings show that the grafted metals are well dispersed over the silica support.The small contributions of crystalline metal oxides are most prominent for ASI, showing the presence of cubic In 2 O 3 (2θ 31 and 51°), [45] while the XRD pattern of ASZ contains features of cubic ZrO 2 (2θ of 30 and 50°). [46]A very weak Ga 2 O 3 contribution feature may be observed in the XRD pattern of ASG.On the other hand, no evidence of (semiÀ )crystalline Al 2 O 3 phases is observed in ASA.Nevertheless, the 27 Al MAS NMR spectrum shows the contributions of Al(IV) and Al(VI), which may suggest the presence of Al 2 O 3 domains (Figure S5).The formation of small patches of alumina at a relatively high Al content is expected for ASA prepared with HDP. [28] 1  MAS NMR experiments were carried out on dehydrated samples to investigate the proton speciation and potential presence of Brønsted acid sites.As shown in Figure 2, all spectra are dominated by the typical silanol contribution at ~1.7 ppm.The broadness of the silanol signal indicates the presence of a range of SiÀ OH species. [47]Some additional features in the tail of the silanol peak can be observed in the NMR spectrum of ASA.The contribution at 2.5 ppm is due to OH groups Table 1.Physico-chemical properties of the fumed silica starting material (AS) and the metal-modified samples.

Sample M -
Si/M [a] -M [a] wt.%  connected to Al sites (e. g., aluminols, AlÀ OH) [48] , while bridged hydroxyl groups (AlÀ OHÀ Si) are observed at 3.7 ppm. [48,49]Such bridged hydroxyl groups are also present in the 1H spectrum of ASG, albeit very faint, whereas no such features are observed with ASI.The NMR spectrum of ASZ includes features related to the silanols (1.8 ppm), non-acidic zirconyls (1.0 ppm), [50] bridged OH species (3.7 ppm), and a broad shoulder of acidic zirconyl hydroxides at 5.0 ppm. [51,52]he acidity of the samples was further characterized by IR spectroscopy using pyridine as a probe molecule.The spectra obtained after the evacuation of pyridine at 150 °C reflect the overall acidity (Figure 3).The spectra of ASA and ASG contain the typical bands at 1638 and 1545 cm À 1 of pyridine interacting with Brønsted acid sites (BAS), while those at 1622, 1491, and 1455 cm À 1 are due to the interaction with Lewis acid sites (LAS).The areas of the corresponding IR bands were used to determine the total amount of BAS and LAS (Table 2).Among these samples, ASA has a higher BAS concentration (33 μmol g À 1 ) than ASG (9 μmol g À 1 ), while the amount of LAS in these samples is comparable (~98 μmol g À 1 ).In contrast, ASI has a low LAS concentration of 29 μmol g À 1 .The lower wavenumber of these bands in ASI compared to ASA and ASG may point to LAS of weaker strength. [53]The IR spectrum of ASZ contains features of LAS at 1491 and 1448 cm À 1 and BAS at 1639 and 1545 cm À 1 with concentrations (80 and 30 μmol g À 1 ) almost similar to ASA.Additional pyridine IR spectra were recorded after evacuation at 400 °C, which can be used to determine the concentration of strong acid sites.The corresponding spectra of ASA and ASG mainly contain features of LAS, implying the absence of strong BAS (Figure 3).No proper IR spectrum of the ASI sample could be recorded after the high-temperature treatment, likely due to the reduction of In 3 + to metallic In.Like ASA and ASG, the ASZ sample contains mainly LAS bands after evacuation at 400 °C.As shown in Table 2, the high-temperature evacuation data shows that most LAS are relatively strong, with ASA having the highest amount (62 μmol g À 1 ) followed by ASG (55 μmol g À 1 ) and ASZ (15 μmol g À 1 ).Based on the pyridine IR results, we might conclude that ASA and ASG have the highest density of predominantly strong LAS.ASZ contains slightly less LAS, and their strength is weaker than the sites in ASA and ASG.The ASI exhibits the lowest amount of LAS compared to the other samples.The total BAS concentration is as follows: ASA � ASZ > ASG > ASI.The absence of a substantial amount of BAS in ASG might be related to the weak interaction between Ga 3 + and silanol groups.Still, it can also be due to a lower dispersion of the Ga-oxide phase on the silica surface.In contrast, the high BAS content of ASZ is associated with the presence of both bridged ZrÀ OHÀ Si species and acidic zirconyl groups.Pyridine IR reveals that most BAS are relatively weak, typical for acid catalysts obtained by modification of amorphous silica. [41]urther insight into the acidity was obtained using 1 H- 15 N CP MAS NMR spectroscopy of adsorbed 15 N-labeled pyridine. [42,43,54,55]Typically, dehydrated samples are exposed under an inert atmosphere to 15 N-pyridine, followed by evacuation at 150 °C and air-free transferred to the NMR apparatus. 1H MAS NMR spectra were first measured to verify the adsorption of labeled pyridine (Figure 4a).][58][59][60] From the relative intensities in the 1 H NMR spectra, we might conclude that ASA and ASG contain more pyridine than ASZ, while even less base is absorbed on ASI.The corresponding 1 H- 15 N CP MAS NMR spectra have the typical bands in the 200-300 ppm range.These bands are broader than those typically observed with zeolites (Figure 4b). [42,55]The 15 N chemical shift depends on the interaction Figure 3. IR spectra of pyridine adsorbed on the amorphous silica-derived samples after pyridine adsorption at 150 °C followed by evacuation for 1 h at 150 °C (solid) and 400 °C (dashed).
Table 2.The concentration of BAS and LAS obtained with Pyridine and CO IR.

Sample
Pyridine CO [a] Concentration of the total Brønsted (1545 cm À 1 ) and Lewis (1455 cm À 1 ) was obtained after the evacuation of pyridine at 150 °C, while the concentration of strong Brønsted was determined after evacuation at 400 °C.
[b] Amount of Brønsted acid sites obtained after peak fitting of the CO saturated sample.[c] Infrared band position of COÀ OH complex.
strength of pyridine with LAS or BAS, which can be used as an indicator of the acid strength. [42,55]The spectra do not contain the typical signature band of physisorbed pyridine at 300 ppm, which shows the evacuation procedure was effective.The peak around 200 ppm is assigned to the pyridinium ion, which results from the protonation of pyridine by BAS.Peaks in the 240-280 ppm range correspond to pyridine interacting with LAS.In some cases, however, these bands are related to BAS of weaker strength, for instance, for Ga-Beta or B-Beta. [42]This makes an unequivocal assignment of the bands to BAS or LAS challenging.The ASA spectrum contains three contributions at 205, 244, and 263 ppm (Figure 4b).The band at 205 ppm is related to pyridine interacting with BAS, while the latter two bands can be assigned to LAS of varying strength.The 15  sites depends significantly on the selected probe molecules. [41]t should be mentioned that the interaction between probe molecules and acid sites in modified amorphous materials is often discussed similarly for crystalline solid acids like zeolites (Scheme 1).However, BAS in ASA mainly rely on the van der Waals interactions between the Lewis acidic Al 3 + atoms and surface hydroxyl groups, although it is suggested that ASA can contain a very small number of BAS of zeolitic strength [40] .Previous studies on silica-grafted zirconia sulfates showed that using pyridine as a probe molecule results in substantial overestimation of the LAS content. [61,62]Morterra et al. found that competitive ligand exchange occurs between pyridine and surface sulfate groups, which causes the probe molecule to bind with the metal center (i.e., LAS) instead of the BAS.When using a weaker base, like CO, higher BAS concentrations were quantified. [61,62]Similar events might occur in amorphous silicaalumina and related materials, which can lead to an overestimation of the amount of LAS (Scheme 1).In principle, CO can bind to the Lewis acidic metal atoms as well, although CO is much less basic than pyridine.However, Poduval et al. used CO IR spectroscopy to distinguish between BAS and LAS in ASA. [41]Figure 5a shows CO IR spectra for the various metalmodified silica samples upon the initial exposure of CO to dehydrated samples at À 183 °C.The following bands are present: 2190 (LAS), 2170 (BAS), 2158 (silanols), and 2137 cm À 1 (physisorbed CO). [28,41]The IR spectra of ASA and ASG appear to be qualitatively similar.However, the BAS-related band in ASG is much broader and located at a lower wavenumber (2167 cm À 1 ) than the corresponding feature in ASA (2170 cm À 1 ).
In contrast, the ASI and ASZ spectra are dominated by the band related to the interaction of CO with LAS at 2186 and 2178 cm À 1 , respectively.The broadness of the 2178 cm À 1 feature in the ASI spectrum indicates a large distribution in the strength of the LAS.The BAS were quantified after deconvolution of the CO-saturated IR spectra (Figure 5b).As listed in Table 2, ASG has a slightly higher BAS concentration (64 μmol g À 1 ) than ASA (58 μmol g À 1 ).Deconvolution confirms the predominant Lewis acidic nature of ASI and ASZ, although a minor feature at around 2171 cm À 1 shows the presence of acidic zirconyl groups (8 μmol g À 1 ) in the latter sample.Based on the CO IR experiments, ASA and ASG have similar BAS concentrations, whereas the pyridine IR measurements show that the former sample has a significantly higher BAS content.This can be rationalized by the larger orbital size of Ga 3 + compared to Al 3 + , which results in a weakening of the M 3 + -HOSi interaction and makes the metal ion more susceptible to pyridine bonding.The wavenumber of the CO IR bands due to BAS suggests that the BAS in ASG are weaker than in ASA, confirming the lesser activation of the silanols with Ga 3 + than by Al 3 + .For ASZ, the opposite is observed when comparing the two probe molecules, and the observed concentration of acidic zirconium species might be related to the different strengths of the bases.The pyridine experiments (i.e., IR and NMR) indicate that ASA, ASG, and ASZ contain more LAS than BAS.CO IR spectroscopy, on the contrary, suggests that ASA and ASG have only BAS, whereas ASZ is mainly Lewis acidic.Interestingly, for ASI, both probe molecules show that this sample only contains LAS.Despite these contradictory results, these data can still provide insight into the acidity situation in these amorphous silica samples.In case ASA and ASG would contain strong LAS, as pyridine shows, these sites would also be observed when using a weaker base like CO.However, the absence of bands related to LAS in Figure 5 suggests that the observation of these sites depends on acid-base interactions, as Morterra et al. proposed. [61,62]herefore, ASA and ASG are most likely solid Brønsted acids, although the complete absence of Lewis acid sites cannot be confirmed.When comparing the total amount of BAS and LAS quantified with both probe molecules, lower amounts are observed with CO than with pyridine.CO IR revealed the acidic zirconyl sites next to LAS in ASZ, which were also observed with 1 H MAS NMR.Based on acidity characterization, we might propose that the samples can be arranged based on their Brønsted acidity with ASA � ASG > ASZ > ASI, with the main distinction between Al 3 + and Ga 3 + is the relative acid strength difference.
The effect of BAS and LAS on the DAC reaction between 2,5-DMF and ethylene was investigated in an autoclave at 200 °C at 50 bar (Figure 6).After 6 h, 2,5-DMF conversions of 35 and 28 % were achieved by ASA and ASG. with p-xylene yields of 21 and 18 %, respectively.These two catalysts also achieve comparable amounts of 2,5-hexanedione (2 and 1 %) and cyclopentenones (~3 %).Previous studies have shown that pxylene formation is less affected by BAS strength [19] , explaining the comparable yields of ASA and ASG.In contrast, stronger acid sites might promote side-product formation (i. e., unidentified oligomers), leading to higher conversion levels.The absence of significant amounts of BAS in ASZ and ASI results in a considerable drop in the performance, with conversions of 10 and 5 % and p-xylene yields of 5 and 2 %, respectively.The higher activity of ASZ compared to ASI might be attributed to the small amount of BAS in the zirconium sample.In order to further demonstrate the ineffectiveness of LAS in the DAC between 2,5-DMF and ethylene, an additional set of Lewis acidic materials was prepared (Figure S6).However, both Sc/SiO 2 and Sn/SiO 2 displayed, like the other Lewis acidic samples, no activity that is comparable to ASA or ASG.Previous studies have shown that Zr-Beta zeolites are active in the DAC of 2,5-DMF and ethylene. [29,30]hen comparing the activity of H-Beta with a partial Zrexchanged zeolite, a reduced catalytic performance is observed with the latter material.The decreased activity of Zr-Beta can be related to a reduction of the BAS concentration due to the partial ion exchange.Thus far, the Brønsted acidic ASA and ASG demonstrate superior performance in the production of pxylene from 2,5-DMF and ethylene.16][17] To study the formation of the cycloadduct, ex situ highpressure 1 H NMR experiments were conducted (Figure S7).Small sample aliquots were taken from the reactor under operating conditions, while a pressure of 10 bar was maintained in the sampling tube throughout the NMR measurement.It was envisioned that the oxanorbornene intermediate could be partially preserved by retaining a pressurized system.However, the experiments revealed that the intermediate is absent in all cases (i.e., blank, ASA, and ASI), and p-xylene was only detected with the ASA catalyst (Figure S7).Unfortunately, maintaining higher pressures was impossible with the current system, so the formation of the cycloadduct remains elusive.We also explored the effect of combining Brønsted and Lewis acid catalysts.However, additional experiments involving a physical mixture (1 : 1) of ASA and ASI demonstrated a nearly two-fold drop in the activity compared to ASA (Figure S8).This decrease in catalytic performance can be related to reducing the BAS concentration.
Thus far, solid Brønsted acids were the only active materials forming p-xylene.Homogeneous metal catalysts have also been reported to be active in the DAC reaction. [34,63]When verified in the DAC of 2,5-DMF and ethylene, the metal triflates perform better than their chloride counterparts (Figure 7).The high activities observed are in the range of benchmark zeolites (i.e., H-Beta) with p-xylene yields of 35-39 %.However, literature has shown that small amounts of water in the organic media can yield strong Brønsted acidity (i.e., triflic acid) from metal triflates.Whereas the chloride derivatives produce weak acids under similar conditions due to the incomplete ionization of the metal salts. [64,65]The effect of water on the DAC reaction between 2,5-DMF and acrylic acid has been corroborated before, and no activity was observed upon adding a molecular sieve to the reaction mixture. [34]When the DAC experiment was reproduced with Sc(OTf) 3 and additional molecular sieve 3 Å, a 2,5-DMF conversion of 10 % and a p-xylene yield of 3 % were achieved, similar to the performance of the metal chlorides.These findings can be reasonably explained by triflic acid being the active catalyst, its formation requiring small amounts of water.This result furthermore reaffirms that Brønsted acids are key for producing p-xylene, albeit potentially only relevant for the dehydration step, as shown by previous DFT calculations. [18,19]The inability of LAS to produce p-xylene is furthermore shown by the homogenous catalysts and the physical experiments.Furthermore, combining Brønsted and Lewis acid catalysts did not result in an improvement in the performance.Further efforts must be made to resolve the mechanistic aspects, for which detection of the oxanorbornene intermediate is likely the critical challenge.We emphasize again that recent theoretical studies have shown that Lewis acids can promote the DAC reaction by lowering the Pauli repulsion via activation of the side group (e. g., carbonyl functionality) of the  dienophiles. [12,13]The absence of such functional groups in ethylene could explain the lack of any catalytic effect of Lewis acids in this specific reaction.

Conclusions
Herein, we investigated the catalytic effects of Brønsted and Lewis acidity in the DAC of 2,5-DMF and ethylene.Amorphous catalysts were synthesized with distinct acid characters by varying the metal grafted on the silica support.Experiments with pyridine as probe molecules revealed the mainly Lewis acidic nature of the samples.CO IR spectroscopy, in contrast, highlighted the Brønsted acidic nature of the aluminum and gallium-based materials.Whereas the indium and zirconium demonstrated mainly Lewis acidity.Catalytic testing revealed that high activity in p-xylene formation could be achieved with the Al 3 + and Ga 3 + -based materials.On the other hand, the In 3 + and Zr 4 + samples showed poor performance in the DAC reaction.The marginally higher p-xylene yield achieved with the zirconium catalyst can be attributed to the presence of some BAS.Extending the material scope to other homogeneous and heterogeneous acids revealed similar trends.The results presented in this work make a compelling case for the effectiveness of solid Brønsted acids in the DAC of 2,5-DMF and ethylene to p-xylene.

Experimental Section Materials
Homogeneous deposition-precipitation (HDP) was carried out as previously reported. [28,44]In a typical procedure, a 30 g L À 1 suspension containing fumed silica (Sigma Aldrich), 0.76 M urea (Sigma Aldrich, 99.5 %), and the desired metal nitrate salt was prepared.The metal precursors were Al(NO 3 ) 3 • 9H 2 O (Merck, purity 99 %), Ga(NO 3 ) 3 • 9H 2 O (Alfa Aesar, purity 99.9 %), and In(NO 3 ) 3 • 9H 2 O (Alfa Aesar, purity 99.99 %).The suspension temperature was raised to 90 °C and kept there for a certain period while recording the pH.The mixture was cooled in an ice bath once a pH above 6 was reached.The as-prepared solids were retrieved by filtration, washed with water, and dried overnight at 110 °C.The samples are denoted ASA (Al 3 + ), ASG (Ga 3 + ), and ASI (In 3 + ).Zirconium was loaded onto fumed silica via impregnation with ZrCl 4 (Sigma Aldrich, purity 98 %).The silica was suspended in water, and the desired amount of the metal precursor was added.The mixture was heated to 80 °C until the water was evaporated.The obtained powder was dried overnight at 110 °C and the sample is denoted as ASZ.All materials were calcined in air for 5 h at 500 °C (0.5 °C min À 1 ).Before the measurement, samples were degassed at 300 °C for 8 h under N 2 flow.Pore size distributions were obtained using the NLDFT method, using the slit-pore model for N 2 adsorption on carbon at À 196 °C. [66] spectroscopy.The Lewis and Brønsted acid sites were quantified by IR spectroscopy using pyridine on a Bruker Vertex 70v spectrometer.All solids were dehydrated under an O 2 flow (33 vol.% in He) at 400 °C for 1 h before recording the initial background spectra at 150 °C under vacuum.After that, pyridine was dosed until the samples were thoroughly saturated with pyridine, followed by outgassing for 1 h to remove excess pyridine.After evacuation at 150 °C, a spectrum was recorded to quantify the total amount of acid sites.The concentration of strong Brønsted and Lewis acid sites was determined at 150 °C after evacuation of the material at 400 °C for 1 h.The BAS and LAS bands at 1545 and 1455 cm À 1 were quantified using previously reported extinction coefficients (ɛ BAS, 1545 cm À 1 = 1.67 cm • μmol À 1 and ɛ LAS, 1455 cm À 1 = 2.22 cm • μmol À 1 ). [67]CO adsorption measurements were performed on a Bruker Vertex 70v spectrometer at À 183 °C using liquid nitrogen.Before recording the background, sample wafers were dehydrated at 400 °C under an O 2 flow (33 vol.% in He) for 1 hour.CO was dosed to the cell using a six-way valve and sample loop (50 μl).The amount of adsorbed CO was obtained after subtraction of the background spectrum.An extinction coefficient of 2.6 cm • μmol À 1 was used to quantify the amount of BAS. [41]R spectroscopy.Al speciation was investigated by 27 Al magicangle spinning nuclear magnetic resonance (MAS NMR) spectroscopy.Measurements were performed using an 11.7 T Bruker NEO500 NMR spectrometer with a 2.5 mm MAS probe head spinning at 25 kHz.Spectra were recorded with a single pulse sequence with an 18°pulse, duration of 1 μs, and an interscan delay of 0.5 s.The samples were hydrated in a desiccator before the experiments. 1H NMR measurements were carried out using a 4 mm MAS probe head with a sample rotation rate of 10 kHz.The spectra were obtained with a Hahn-echo pulse sequence p1-τ1-p2-τ2-aq with a 90°pulse p1 = 3.150 μs and a 180°p2 = 2.5 μs.The interscan delay of 120 s was chosen for quantitative spectra.Samples were dehydrated in a diluted O 2 atmosphere (33 vol.% in He) for 2 h at 400 °C.After removal of the oxygen, the rotors were prepared in a glovebox before the measurements.Cross-polarization magic angle spinning (CP MAS) 1 H- 15 N NMR spectra were measured on samples loaded with pyridine-15 N (98 %, Sigma).Samples were calcined in a diluted O 2 atmosphere (33 vol.% in He) for 2 h at 400 °C and transferred to a glovebox after removing oxygen.Small aliquots of pyridine-15 N (98 %, Sigma) were added to the pretreated sample.Before the measurements, physisorbed pyridine was removed after evacuation at 150 °C for 1 h at 7 mbar.Ex-situ high-pressure NMR was carried out using a 10 mm heavy wall precision NMR tube from SP Wilmad-LabGlasss.The 1 H NMR spectra were recorded using a Varian Mercury Vx 400 MHz.Proton chemical shifts are reported in ppm downfield from trimethylsilane (TMS) using the deuterated CDCl 3 resonance frequency as an internal standard.
Thermogravimetric analysis.The amount of carbonaceous deposits on the catalyst used in the DAC reaction was determined with a Mettler Toledo TGA/DSC 1 instrument.In a typical analysis, 20 mg of material was heated to 800 °C at a rate of 5 °C min À 1 in a diluted O 2 atmosphere (33 vol.% in He).

Diels-Alder cycloaddition
The catalytic activity in the DAC reaction of 2,5-DMF and ethylene was investigated in a 100 mL TOP Industrie autoclave equipped with a mechanical stirrer and a pressure control system.First, 0.1 g of sieved catalyst (250-500 μm) was added to 30 mL of 1.0 M 2,5-DMF and 0.03 M n-dodecane (as internal standard) in 1,4-dioxane.
The autoclave was purged three times with helium.Next, the system was pressurized to 25 bar with ethylene, and the reactor was heated to 200 °C (final pressure was 50 bar).The reaction was ended by disconnecting the heat source and depressurizing the vessel.Then, the reaction mixture was separated from the catalyst by filtration and analyzed using GC-FID (Shimadzu GC-FID GC-17A equipped with a Rxi-5 MS column).The 2,5-DMF conversion and product yields were calculated below.

Figure 1 .
Figure 1.Silica-based catalyst with distinct Brønsted or Lewis acidic characteristics by changing the atom size and valency.
N spectrum of ASG contains two bands at 202 ppm and 241 ppm, which can be attributed to BAS and LAS, respectively.The absence of a band around 200 ppm for ASI suggests that this sample does not contain BAS, while a broad signal at 264 ppm reveals the LAS in this sample.The low field shift of these bands might indicate the lower strength of the LAS in ASI than those in ASA and ASG.The spectrum of the ASZ sample also contains two bands at 201 and 277 ppm due to BAS and LAS.Based on the chemical shift, the strength of the LAS appears to decrease from ASA~ASG > ASI > ASZ.Both pyridine IR and NMR experiments, thus far, have demonstrated that considerable amounts of LAS are present in ASA and ASG.However, previous work on ASA by Poduval et al. has shown that the quantification of acid

Figure 5 .
Figure 5. a) IR spectra upon sequential dosing of CO to pretreated samples at À 183 °C.b) Peak deconvolution of the IR spectra upon CO saturation (green: LAS, red: BAS, blue: silanols, and yellow: physisorbed CO).