B-TUD-1 : a versatile mesoporous catalyst †

Novel amorphous mesoporous borosilicate, B-TUD-1, was prepared to test its performance for different sustainable reactions. The structure of the material, the effective incorporation of boron into the framework as well as the nature of incorporated boron were verified by N2-sorption, XRD, ICP-OES, TEM, NH3-desorption, MAS NMR and FTIR. The potential of these materials as catalysts was tested in two reactions under different conditions. They showed an apparently good activity and recycling potential for HMF (5-hydroxymethyl-furfural) synthesis. However leaching of boron occurred during this aqueous reaction. We could demonstrate that the leached boron was actually the catalyst of the reaction. In a second reaction, this time in organic solvent, the B-TUD-1 materials were tested for the Prins cyclisation of citronellal. A very promising activity was obtained. The catalyst could be recycled and no boron leached, as demonstrated by hot filtration experiments.


Introduction
Heterogeneous catalysts were widely studied during the last few years, in particular for the development of sustainable processes.Indeed, while representing powerful catalysts, they have the additional advantage that they are easily recyclable and thus lead to signicant reductions of waste.In this context, TUD-1 materials were developed.
The special properties of mesoporous TUD-1, such as its three dimensional pore structure with a good accessibility for substrates and products, tuneable pore properties, 1,2 and a costeffective one-pot, surfactant-free synthesis by the sol-gel method, have made it a versatile heterogeneous catalyst for many interesting synthesis applications. 3TUD-1 has been successfully functionalized with numerous incorporated metals (Al, Ti, Co, Fe, V, Cr, Mo, Hf, Ga, Ni, Mn, Zr and Cu) aiming at a variety of catalytic properties.  Surisingly, the incorporation of B, in the same group as Al, was never investigated.
Boron derivatives have proven to be efficient catalysts for a large range of reactions (reduction, Diels-Alder, aldol reactions). 30,31][36] Both TUD-1 materials and boric acid have been demonstrated as excellent catalysts for sustainable reactions, conversion of renewable materials and synthesis of platform chemicals (for example, boric acid catalysed synthesis of HMF 37 or Hf-TUD-1 catalysed conversion of glycerol to solketal). 14eviously other boron-incorporated zeolites and mesoporous materials (Beta-and ZSM-5 zeolites, SBA-15, MCM-41) [38][39][40][41][42][43][44][45][46][47][48][49] have been reported.Key examples are the B-MCM-41-catalysed Beckmann rearrangement of cyclohexanone oxime 44 permitting the efficient synthesis of caprolactam, and, the B-MCM-41-catalysed three component Strecker reaction also leading to the formation of a-aminonitriles, 50 intermediates for the synthesis of a-aminoacids or N-containing heterocycles.Boron is herewith a promising, cheap and environmentally benign element that imparts activity to mesoporous materials.In this paper a novel approach towards the synthesis of heterogeneous mesoporous B-catalyst is presented, B-TUD-1.Combining the advantageous properties of the TUD-1 materials (notably their 3D structure suppressing diffusion limitation of the substrates) and excellent catalysis efficiency of the boron derivatives, the obtained B-TUD-1 catalysts should present new versatile catalytic properties.We present here not only the preparation of the catalyst and its full characterization, and additionally performance tests for two important, sustainable reactions, the HMF synthesis, and, the Prins cyclisation of citronellal.

Preparation of the material
The M-TUD-1 materials (Al and B) were prepared by the sol-gel method using triethanolamine as a complexing agent following the protocols described earlier. 9The M was included in the form of the isopropoxide M(iPrO) 3 .
The use of triethanolamine permits to avoid the use of any surfactants or liquid crystal templates.Complexing with the metal triethanolamine leads to the formation of the mesopores (Fig. 1).Initially the active metal is complexed to form an atrane. 51,52Then these atranes complex together with free triethanolamine creating the framework of the material.During gelation and also aer calcination and consequent removal of organic compounds, the metal is incorporated into the framework of the material as oxide species. 3,4This is also known as the atrane route. 53The formation of the boratranes has been demonstrated as to be a set of complex equilibria between the different coordination forms.5][56] Another interesting point is that by the use of a similar method introducing surfactants, ordered mesoporous materials could be obtained (for example SBA-15 57 ) in contrast to the amorphous, mesoporous TUD-1 materials.
Different ratios of M were introduced, Si/Al 25 (Al-TUD-1 25), and Si/B 50, 25 and 10 (B-TUD-1 50, 25 and 10 resp.).A hydrothermal treatment of 6 h at 180 C was applied that has dened the size of the pores and the mesoporous structure. 1Subsequent calcination at 600 C for 10 hours led to the nal material.

Characterization: a classical mesoporous catalyst
The ratio of incorporated boron was veried by elemental analysis by inductively coupled plasma atomic emission spectroscopy (ICP-OES).The mesoporous character of the materials was conrmed by X-ray diffraction (XRD), N 2 sorption and highresolution transmission electron micrographs (TEM).The acid sites were characterised by temperature-programmed desorption of NH 3 .Finally, the incorporation of boron was veried by XRD at small angles and by MAS NMR.The latter permitted also to characterise the nature of the boron species incorporated into the mesoporous framework.
Results of the elemental analysis (ICP-OES) and the porosity measurements are given in Table 1.The elemental analysis reveals Si/B ratios of the calcined catalysts very close to the actual added amount.$95% of the added boron compounds during the preparation of the gel was incorporated.This demonstrates the good predictability of the preparation method, already observed with other M-TUD-1 materials. 8,58ccording to Jansen et al., 1,2 the pore size of TUD-1 material is tuneable and can be controlled by the time of the hydrothermal treatment.Aer a treatment of 6 h, the expected average pore size was between 6 and 8 nm.The measured pore diameters correspond to the theoretical values and are slightly larger than those for classical M-TUD-1 (average around 4.0 nm) 3 and close to other boron containing mesoporous silicates (for example, B-MCM-41, average around 4.5 nm, B-SBA-15, average around 6.5 nm). 43,46,47The surface area of B-TUD-1 increases upon increasing of the metal loading.At the same time, the average pore diameter decreases in line with the results for other M-TUD-1 materials 3 and the boron substituted mesoporous SBA-15. 43he N 2 adsorption and desorption isotherms of the B-TUD-1 50 and 25 (Fig. 2) reveal a typical "type IV" hysteresis loop, with a large uptake of nitrogen at relative pressures between 0.5 and 0.9 due to capillary condensation in the mesopores.A plateau at relative pressures of above 0.9 p/p 0 indicates the absence of large mesopores or macropores.The B-TUD-1 10 isotherm (Fig. 2) differs slightly by the fact that it shows a lower uptake in the rst phase, possibly related to a lower mesoporosity.The material has the lowest pore diameter and micropore volume (Table 1).This might result from the intensied interaction of the B atoms with the silica structure as it has been observed for the Zr-TUD-1 material. 58he X-ray diffractograms of the materials (Fig. 3) show a broad intense peak at low angles around 0.5-1 2q in line with the mesostructured character of these materials.No correlation could be established between the peak intensity and the Si/B ratio.
The complete XRD spectra of B-TUD-1 10, 25 and 50, as well as that of Al-TUD-1 25 for comparison, were recorded.No crystalline structure was observed in any case.A spectrum of boric acid was measured as a reference (multiple peaks, characteristic of this crystalline product).None of the characteristic signals of boric acid could be detected in the B-TUD-1 spectra, not even at a Si/B ratio of 10.This is quite remarkable compared to other M-TUD-1 materials.It indicates a complete framework incorporation even at the highest metal loading, where most of   the M-TUD-1's contain a certain percentage of metal oxides visible on TEM pictures. 3,8Complete boron incorporation at high loading has also been reported for other boron-containing zeolites 38,45 or mesoporous silicates. 43,44he TEM pictures (Fig. 4) show "worm/sponge-like" structures typical for amorphous material and TUD-1 samples. 1,3This conrms the mesoporous structure of the material.Furthermore, no crystalline boron particles could be observed even at higher loading conrming the incorporation of the boron species.Therefore, the material appeared completely amorphous, as no ordered pores or channels were visible, and as no boron (or aluminium) oxide could be detected (Fig. 3 and 4).
A temperature-programmed desorption (TPD) prole (from 100 to 500 C) of ammonia for different B-and Al-TUD-1 samples (Fig. 5) was measured.For all the boron-incorporated materials two peaks were observed, an intense peak at 150 C corresponding to weak acids, and a broad peak with low intensity at around 400-500 C, corresponding to strong acid sites (Table 2).The medium acid sites (temperature around 250-300 C) are hardly present.The intensity of the peaks (especially for the weak acid sites) increases with the B loading (i.e.decreasing Si/B ratio) as well as the total acidity.In the case of Al-TUD-1 25, we could observe one broad band from 100 to 350 C corresponding to weak and medium acids, more classically observed with M-TUD-1 catalysts. 11he solid-state 11 B MAS NMR (Fig. 6) showed three main peaks for the B-TUD-1 samples, corresponding to the earlier results for the boron containing mesoporous silicates as MCM-41 (ref.42) or SBA-15 (ref.43) (and to the results obtained for many zeolites 39,40,49 ).The rst peak at around À22.9 ppm, sharp and narrow, corresponds to the 4-coordinated tetrahedral boron (mainly, 2-coordinated also possible, 43 symmetric form).The two other broader bands from À11.5 ppm to À2.5 ppm correspond to different trigonal forms of boron species with different degrees of coordination to silica (in line with literature 39,40,42,49 ), namely 3-coordinated (11.5 ppm), and trigonal B with a lower coordination number (structures in Fig. 7).With a decrease in boron loading, i.e. higher Si/B ratio, relatively more tetrahedral boron is observed.It is however important to notice the dynamic nature of these incorporated boron species.Diverse equilibria are established between the different species, especially between tetrahedral and trigonal forms, as function of the temperature and media (hydration, presence of cations), 45,49 explaining the indistinguishable broad peaks observed in the NMR spectra.For this reason, the nature of the acid sites is also dynamic.Due to this dynamic equilibrium that is also inuenced by water, it is not meaningful to determine the ratio of Lewis to Brønsted acid sites as this is dynamic too.Furthermore, the 27 Al MAS NMR spectrum of the Al-TUD-1 material was also measured (cf.ESI; Fig. S1 †).[61] The presence of tetrahedral boron species in B-TUD-1 materials was conrmed by FTIR as a shoulder at 920-930 cm À1 (mainly visible for B-TUD-1 10, Fig. 8).Another peak characteristic for the presence of trigonal boron can be identied at 1395 cm À1 . 43,47he intensity of these two peaks, related to the presence of B, decreases in line with the B loading.For comparison, these peaks are absent in the spectra of Al-TUD-1.The absorption bands at 460, 810 and 1070 with a shoulder at 1200 cm À1 , present in all the B-and Al-TUD-1, can be assigned to the symmetric and asymmetric stretching of Si-O-Si vibrations of tetrahedral SiO 2 units (Fig. 8).Based on these results, we can postulate that B-TUD-1 materials with fully framework incorporated boron were prepared.They contain trigonal boron representing Lewis acidity as well as tetrahedral boron representing Brønsted acidity (Fig. 6  and 7) in a dynamic equilibrium.
Performance test in aqueous or biphasic media: HMF synthesis HMF, 5-hydroxymethylfurfural, is a biobased platform chemical, obtained from sugars, especially fructose, by acid-catalysed dehydration (Scheme 1). 62However, despite the many studies that have been reported, no real efficient, sustainable and scalable route has been proposed yet. 62any Lewis acids, 63,64 zeolites and mesoporous materials 15,22,[65][66][67][68][69][70] were used as catalysts for this reaction.Because of the numerous advantages (recyclability, easy separation of the product, no corrosion of equipment), heterogeneous catalysts in general have been the main focus.
As Al-TUD-1 60 and boric acid 37,71 were already successfully tested for HMF synthesis in biphasic systems or ionic liquids, B-TUD-1 materials appear to be promising heterogeneous, thus potentially recyclable, catalysts for HMF production.Indeed, highly stable, TUD-1 materials have been shown to be easily recovered and recycled aer reaction by a calcination similar to the one used during the preparation. 3he synthesis of HMF in aqueous media is a challenging reaction.The B-TUD-1 catalysts showed very good activity for this reaction, especially under biphasic conditions with toluene.
In Fig. 9, experimental results are presented.With the B-TUD-1 catalysts under purely aqueous conditions, the conversion and yield appeared limited (same results at 150 C for 90 min or 4 h, # 20% yield); whereas in biphasic media, at 150 C for 4 h, 65 to 70% conversion of fructose was observed with a yield between 40 and 45%.Interestingly, these results were similar for all Si/B ratios and very close to what can be obtained under similar conditions with Al-TUD-1. 60ecycling experiments were successfully carried out in biphasic media (Fig. 10).At 190 C up to 90% conversion, a yield of 60% can be reached aer 40 minutes.This can be repeated in three consecutive cycles.
In the same way, consecutive recycling experiments were carried out in a purely aqueous phase (Fig. 11A).However, the liquid 11 B NMR analysis of the reaction media revealed the presence of a broad peak at 0 ppm, corresponding to the presence of boric acid.Furthermore, the elemental analysis of the  recycled catalysts aer one, two or three reactions revealed an increasing Si/B ratio (from 25 to 85 aer 3 reactions, Fig. 11B), thus a decreasing B content on the mesoporous material.The B is probably slowly released during the reaction.Furthermore, a porosity analysis of the used catalysts aer calcination showed a clear change in their mesoporous structure (Table 3).Notably a net decrease of the mesoporous surface and pore volume could be observed.The structure is probably affected by the release of B and the hydrolysis of the Si-O-B bonds.Finally, a 11 B MAS NMR of the used catalysts (cf.ESI; Fig. S2 †) showed a net decrease of the peaks corresponding to tetrahedral 4-coordinated B (sharp peak at À22.9 ppm), but also to the trigonal 3-coordinated B (broad peak at À11.5 ppm).We observed thus under acidic conditions, in line with literature descriptions in the case of different boron containing zeolites, 45,49 both the conversion of tetrahedral to trigonal boron, as well as the progressive cleavage of the Si-O-B bridges, leading to a partial deboronation.This observed leaching, questions the character of the catalysis, heterogeneous ensured by the solid catalyst, or    homogeneous ensured by the released boric acid.To clarify this point, a recycling experiment of the aqueous phase of the reaction was carried out (Fig. 12).The catalyst was ltered off aer one hour and the reaction was continued with the remaining aqueous phase for an extra hour.In parallel, a reference sample was running, where the catalyst was not removed.
The reaction was shown to continue despite the removal of B-TUD-1.Both conversion and yield continued to increase.Moreover, very similar results were obtained with the solid catalyst or aer removal of the catalyst (Fig. 12).Our conclusion is that the released boric acid catalyses the HMF formation.Finally, a last experiment (conditions biphasic, 190 C, 40 min), conducted with a small concentration of boric acid (4 wt%) led to similar results (89% conversion, 59% HMF yield) as the ones observed with the solid catalysts (91% conversion, 60% HMF yield), thus conrming this assertion.
This result with B-TUD-1 can be compared to the stability of Al-TUD-1 under similar conditions.Lima et al. 60 showed a good recyclability of Al-TUD-1 under similar reaction conditions.Moreover, our measurement of the 27 Al MAS NMR aer reaction showed identical spectra to the original calcined sample.This difference in stability can be explained by a smaller size of the boron and its greater dynamic reactivity in the silica framework.

Performance test in organic media: Prins cyclisation
The catalyst performance was also tested in the Prins reaction 72-74 (Scheme 2), an important C-C bond forming reaction catalysed by acids. 75,76his intramolecular cyclisation of citronellal, a renewable product, is an essential step in the Takasago synthesis of menthol. 77Industrially, the reaction is catalysed by stoichiometric amounts of solid ZnBr 2 in benzene. 780][81][82][83][84][85][86][87][88][89] The Prins reaction was carried out with the different B-TUD-1 (50, 25 and 10) catalysts as well as with Al-TUD-1 25 as a reference (Scheme 3).The obtained activity of the B-TUD-1 catalysts was quite high (Fig. 13) and was increasing with the amount of incorporated boron (thus with a decreasing Si/B ratio).It should be noticed that the experiments were carried out with industrial grade citronellal containing around 5% isopulegol.As it has previously been shown, the presence of isopulegol partially inhibits the conversion of citronellal.Higher activities can be expected starting from optically pure citronellal. 58e could observe full conversion aer one hour for the B-TUD-1 10.This result is very close to the conversion that we obtained via Al-TUD-1 (full conversion aer approximately 50 minutes).Full conversion was reached aer four hours for B-TUD-1 25, 80% conversion was obtained aer 6 h with B-TUD-1 50.Despite an apparent lower activity of the latter two catalysts, the comparison of their turn over frequencies (TOF, Fig. 13B) showed their activity per active site to be actually higher than for B-TUD-1 10.
These results are similar to the results with Zr-TUD-1 100.We can also compare this performance to other examples of zeolite catalysts for this reaction.Similar performance was observed under the exact same conditions with Al-MSU-50 (ref.83) (88.6% conversion aer 30 minutes, and 96.7% aer 60 minutes), whereas Zr-incorporated MOF 89 only presented a full conversion aer a minimum of 4 h at 100 C for a citronellal/Zr ratio of 10 (in our conditions, with B-TUD-1 10, citronellal/B ratio of 9, with B-TUD-1 25, citronellal/B ratio of 21).Our catalyst thus presents an interesting activity for the Prins cyclisation of citronellal.The product selectivity to isopulegol (all diastereoisomers) for all the catalysts appeared excellent ($95%, Table 4).The observed diastereoselectivity to isopulegol (vs.its other diastereoisomers) is around 63/37 for all B-containing TUD-1 catalysts (Table 4).This result is close to the measured diastereoselectivity of Al-TUD-1 25 (66/34) and to the values described in literature for Zr-TUD-1 (average 65/35). 58The reduced pore diameter of B-TUD-1 10 does not inuence this ratio. 58 liquid 11 B NMR of the reaction medium aer ltration of the catalyst did not show the presence of boron compounds, indicating the absence of leaching.Furthermore a porosity measurement of the catalysts aer reaction and calcination showed very similar results (Table 3) to the initials catalysts (Table 1).Finally, a hot ltration test was carried out.Aer 20 min, the reaction medium is sampled, immediately ltered, and le again at 80 C.This operation is performed on a hot sample to prevent any possible precipitation of diluted compounds. 90Aer the removal of the solid catalyst, the reaction clearly stopped (Fig. 14), proving the heterogeneous nature of the catalytic activity.
Recycling steps were realised with the catalysts B-TUD-1 10 and 25, with or without calcination before the recycling reaction (Fig. 15).Very similar results were obtained with the calcined recycled catalysts, whereas we could observe a net decrease of the activity of the non-calcined samples, probably partially inhibited by the product remaining on the catalyst as described earlier. 58

Conclusions
We successfully obtained and characterized a new, easy to prepare, three-dimensional and mesoporous acidic B-TUD-1 catalyst.We veried the mesoporous and amorphous character of the material as well as the incorporation of the boron into the    framework even at low Si/B ratio.We also characterized the mixed acidity of the material (Brønsted and Lewis acid sites) as well as the different possible structures of the incorporated boron (trigonal and tetrahedral) and their dynamic equilibrium.Catalytic tests showed the efficiency of this material for the catalysis of different sustainable reactions.However, in the case of HMF synthesis, the acidic aqueous reaction conditions caused boron leakage and a change of the structure of the material was observed.On the contrary, B-TUD-1 demonstrated high reactivity as well as reasonable recyclability for the Prins cyclisation of citronellal in toluene.B-TUD-1 is there with a promising heterogeneous acid that catalyses reactions under non-aqueous conditions.

Materials and methods
All chemicals were purchased from Aldrich, Janssen or Acros.
For the catalysis experiments, the anhydrous solvents and solids were used as received, all other liquids were dried and distilled prior use.M-TUD-1 catalysts were activated in the presence of air at up to 600 C at a ramp of 1 C min À1 and heated at 600 C for 10 h.The experiments were performed in dried glassware and under N 2 atmosphere.

Catalyst preparation
General protocol.Dry isopropanol and ethanol were added to a plastic ask (washed with acetone and placed 30 min in an oven at 60 C) and the ask was closed.M(iPrO) 3 (aluminium or boron isopropoxide) was ground (in case of Al; B is liquid) before addition to the isopropanol/ethanol mix.Then TEOS (tetraethyl orthosilicate) was added.This mixture was stirred 20 min, resulting in a milky solution.TEA (triethanolamine) was weighed and diluted with water.The TEA/water mix was added dropwise over a period of 25 min, whilst stirring, then the mixture was stirred an additional 20 min.This resulted in a milky/glassy white gel.TEAOH (tetraethylammonium hydroxide solution 35%) was then added dropwise over a period of 30 min.This yielded a gel with a molar ratio of SiO 2 /xM(iPrO) 3 /(0.3-0.5)TEAOH/(0.7-1.1)TEA/(10-20)H 2 O. Aer 30-45 min extra stirring, the gel formed was transferred to a porcelain bowl and was allowed to dry at room temperature overnight.The porcelain bowl was placed in an oven at 98 C for 24 h (drying step).It turned into a yellow, transparent solid.This solid was ground, then transferred in Teon lined steel autoclaves, and placed in an oven at 180 C for 6 h (hydrothermal treatment).Aer this hydrothermal treatment, the material was transferred again into the porcelain bowl and placed in the calcination oven.The following program was applied (calcination step): heating at 1 C min À1 until 600 C, hold at this temperature for 10 h, then cooled down to 50 C at a rate of 1 C min À1 .The M-TUD-1 was recovered as a white solid.

Catalyst characterization
The volumetric nitrogen adsorption was realised on a Quantachrome Autosorb-6B at 77 K. Prior to the physisorption experiments, samples were degassed at 250 C for 16 h.
Powder XRD patterns were obtained on a Philips PW 1840 diffractometer equipped with a graphite monochromator using CuKa radiation.The powder XRD patterns at small angles were obtained with a Bruker D8 Discover X-ray diffractometer equipped with the 2-dimensional Hi-Star Area Detector and Cross Coupled Göbel Mirrors.The measurements were performed in transmission mode at room temperature using monochromatic CuKa1 radiation and a sample to detector face of 30 cm.
The elemental analysis of the materials was carried out using an ICP-OES 4300DV plasma emission spectrometer (Perkin Elmer, United States), enabling simultaneous recording of the full emission spectrum of the sample in the range from 166.25 to 847 nm with the help of a Charge Injection Device (CID).The spectrometer was equipped with a cyclonic spray chamber with concentric nebulizer.A double system of observation of plasma was used (axial and radial).Approximately 10-40 mg sample was weighed, and transported into a thoroughly rinsed Teon vessel.All samples are dissolved in an aqua regia acid mixture containing 1.2 mL concentrated hydrochloric acid (36% v/v), 0.4 mL concentrated nitric acid (69% v/v) plus 0.4 mL concentrated hydrouoric acid (40% v/v).
Temperature-programmed desorption (TPD) of ammonia was carried out on a Micromeritics TPR/TPD 2920 equipped with a thermal conductivity detector (TCD).The sample (30 mg) was pre-treated at 600 C to remove volatile components.Prior to the TPD measurements the samples were degassed at 250 C (aer ramping from room temperature at a ramp rate of 10 C min À1 ) for 1 h then saturated with ammonia gas at 100 C for 30 min at a ow of 20 cm 3 min À1 .Aer a purge of ammonia by helium for 30 min at a ow of 10 cm 3 min À1 at 100 C, desorption of NH 3 was monitored in the range between 100 and 600 C at a ramp rate of 10 C min À1 .
The FT-IR spectra were measured on a Perkin Elmer Spectrum ONE instrument.A KBr wafer was prepared containing a particular amount of the calcined sample that allowed for a transmission of minimally 50% (<1 mg).The spectrum was taken over a range of 450-4000 cm À1 with a resolution of 1 cm À1 .
Transmission electron microscopy was performed by using a Philips CM30T electron microscope with a LaB6 lament as source of electrons operated at 300 kV.
The MAS NMR was recorded on a Bruker Avance-400 Spectrometer with a 4 mm zirconium rotor sample holder spinning at 10 kHz.The 11 B/ 27 Al MAS NMR was measured at a frequency of 128.31/104.20 MHz with a spectral width of 25/100 kHz, acquisition time of 0.05 s/0.08 s, and acquisition delay 1 s, the total number of scans was 2000/4000.Liquid 11 B NMR spectra were recorded with a Bruker Avance 400 spectrometer with chemical shis (d) reported in ppm downeld with B(OH) 3 0.1 M in D 2 O.

HMF synthesis
General protocol.600 mg of fructose and 200 mg of M-TUD-1 were weighed and lled in an AceÒ pressure 15 mL-glass tube.The solution was completed to 2 mL by adding demi-water, then 6 mL of toluene was added in the case of the biphasic experiments.The solution was placed in an oil bath at the desired temperature.Aer the dened time of reaction, the tube was removed from the oil bath and allowed to cool down to room temperature (in 5 min approx.).In the case of biphasic reactions, the two phases were allowed to separate for a few minutes.100 mL of the organic phase was then sampled, derivatised and analyzed by GC (replicated).The aqueous phase was diluted 5 times by adding 8 mL 0.005 M H 2 SO 4 .2 mL of this diluted solution was transferred into an Eppendorf tube, centrifuged and the supernatant analyzed by HPLC.
Recycling step (recycled catalyst).Just aer reaction and cooling (cf.protocol above), the AceÒ tubes were centrifuged 10 min at 1800 rpm.The recovered pellets (humins and catalyst) were placed in an oven and calcined at 600 C for 6 h, then used in a recycling step (same protocol followed).
Recycling step (recycled solution).Just aer reaction and cooling (cf.protocol above), the AceÒ tubes were centrifuged 10 min at 1800 rpm.The supernatant was placed in the same conditions as a classical reaction.
HMF synthesis: HPLC analysis (aqueous phase).Samples were diluted by a factor 10 or 20 with an aqueous solution of H 2 SO 4 at 0.005 M. 10 mL of the diluted samples were measured by HPLC.The used eluent was an aqueous solution of H 2 SO 4 at 0.005 M. The method lasted 35 min with a ow rate of 0.8 mL min À1 .Both UV (for HMF and furaldehyde, Fal) and RI (for HMF, fructose, levulinic acid, LA, formic acid, FA, and Fal) detections were used.Retention times: fructose: 7.3 min, FA: 10.3 min, LA: 11.5 min, HMF: 21.7 min, Fal: 32.6 min.
HMF synthesis: derivatisation and GC analysis (organic phase).100 mL of the toluene phase was sampled.50 mL of the IS solution (hydroquinone, 20 mg in 10 mL, 2 g L À1 ) was added.

Prins reaction
General protocol (following method reported earlier. 11,83) In a Schlenk ask of 50 mL (placed in an oven at 70 C o/n), 50 mg of catalyst (activated in an oven overnight at 100 C) were introduced.Aer ushing the atmosphere with N 2 , 5 g of dry toluene were introduced, 0.1 mL of triisopropylbenzene (IS) were added and nally 4 mmol of AEcitronellal (industrial grade, containing approx.5% isopulegol, 0.725 mL).The mixture was then introduced in an oil bath at 80 C and stirred under N 2 .Regularly, samples were withdrawn using a capillary (introduced in the reaction medium via a needle), ltered over a cotton plug (in a Pasteur pipette), diluted and analysed by GC.
Recycling step.The catalysts were recycled aer ltration and calcination (600 C, 6 h) or activation (100 C overnight) and used following the same protocol.
Hot ltration test.The reaction is started as described above.Aer 20 minutes the reaction medium is sampled via a dried syringed and immediately ltered through a PTFE syringe lter previously washed with anhydrous toluene at 80 C. The liquid medium is introduced directly into a Schlenk ask of 50 mL, previously placed overnight in an oven at 70 C and ushed with N 2 , and let at 80 C. The monitoring of the reaction was realized before and aer ltration following the same protocol as describe before.

Fig. 6
Fig. 6 11 B MAS NMR of B-TUD-1 catalysts (25, 10) referenced to 0.1 M aq.B(OH) 3 (most of the cited articles use BF 3 (OEt) 2 as reference leading to a shift of the signals of 19.8 ppm).

Fig. 7
Fig. 7 Possible tetrahedral and trigonal structures of boron incorporated in TUD-1 framework.

Fig. 13 (
Fig. 13 (A) Evolution of the yield of isopulegol formed during the Prins cyclisation catalysed by various B-and Al-TUD-1 catalyst; (B) evolution of the TOF of the catalysts during the reaction (in mol converted product per mol B or Al per min).

Fig. 15
Fig. 15 Results of the recycling experiments with or without intermediate calcination of the B-TUD-1 25 and 10 catalysts, yield of isopulegol (%).

Table 1
Physico-chemical characterization of M-TUD-1 with different Al or B loadings a From ICP-OES measurements.b Not determined because not measurable.

Table 4
Conversion, selectivity and diastereoselectivity for M-TUD-1 catalysts in the Prins cyclisation of citronellal