Applied Catalysis A: General 227 (2002) 181–190

Selective etherification of glycerol to polyglycerols overimpregnated basic MCM-41 type mesoporous catalysts

J.-M. Clacens∗, Y. Pouilloux, J. BarraultLaboratoire de Catalyse en Chimie Organique, UMR 6503, CNRS ESIP,

40 Avenue du Recteur Pineau, 86022 Poitiers Cedex, France

Received 12 June 2001; received in revised form 19 October 2001; accepted 22 October 2001

Abstract

In the general context of the development of the use of agricultural products for non-food applications, and particularlyin the field of the glycerol valorisation (co-product of the triglycerides hydrolysis or methanolysis process), the selectiveetherification of glycerol was studied. The objective of this work is the direct and selective synthesis, from glycerol and withoutsolvent, of polyglycerols having a low polymerisation degree (di- and/or triglycerol), in the presence of solid mesoporouscatalysts. The main part of this study consists in the synthesis and the impregnation of mesoporous solids with different basicelements in order to make them active, selective and stable for the target reaction. The catalytic results obtained show thatthis impregnation method gives important activity, which must be correlated to an important active species incorporation.Concerning the selectivity of the modified mesoporous catalysts, the best value to (di-+ tri-) glycerol are obtained over solidsprepared by caesium impregnation. The re-use of these caesium impregnated catalysts does not affect the selectivity to the(di- + tri-) glycerol fraction. In the presence of lanthanum or magnesium containing catalysts, the glycerol dehydration toacrolein is very significant whereas this unwanted product is not formed when caesium is used as impregnation promoter.© 2002 Elsevier Science B.V. All rights reserved.

Keywords:Glycerol etherification; Mesoporous catalysts; Base catalysis; Polyglycerols

1. Introduction

Glycerol is mainly a natural product issued from themethanolysis of vegetable oils. In Europe, due to theincreasing use of methyl esters as fuel additives, onecan expect an increase of glycerol production, whichcould become a cheaper raw material from chem-istry [1]. For example, polyglycerols and speciallypolyglycerols-esters (PGEs) are gaining prominence

∗ Corresponding author. Present address: Institut de Recherchessur la Catalyse, CNRS, UPR 5401, 2 Av. A. Einstein, 69626Villeurbanne Cedex, France.E-mail address:clacens@catalyse.univ-lyon1.fr (J.-M. Clacens).

in new products for surfactants, lubricants, cosmetics,foods additives, etc. Indeed PGEs exhibit multifunc-tional properties and a wide range of formulatingoptions, if it is possible to control; (i) the length of thepolyglycerols chain; (ii) the degree of esterificationand (iii) the fatty acid molecular weight. These re-actions are quite interesting goals for shape-selectivecatalytic processes.

Previous works showed that the selectivity of thefirst step is not really controlled and that a mixtureof di- to hexaglycerol (linear or cyclic) is obtained.Then it is rather difficult to get a well-defined prod-uct and to predict the hydrophilic–lipophilic balance(HLB) after esterification. In our laboratory, it was

0926-860X/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0926-860X(01)00920-6

182 J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190

Fig. 1. Schematic representation of the etherification of glycerol to polyglycerols.

evidenced that the esterification of glycerol could beselective to�-monoglycerides over cationic resins [2].Nevertheless, polyglycerols and polyglycerols estersas well as acrolein were obtained as main by-products.Moreover, we observed that the modification of thepseudo-pore size of these materials improved the se-lectivity to PG but acrolein is always obtained overthese acid catalysts.

In order to obtain a selective formation of di-,tri-, tetra- or polyglycerols, we investigated firstthe selective etherification of glycerol (Fig. 1) oversolid bases and compared the results to those ob-tained with Na2CO3 catalysts. The main objectiveof this work is to prepare selectively diglycerol or amixture of (di-+ tri-) glycerol by direct etherifica-tion of glycerol without use of solvent and withoutformation of acrolein (Fig. 2) which is mainly ob-tained by double dehydration of glycerol over acidicsites. In order to achieve this goal, we have pre-pared and modified new mesoporous basic solids byimpregnation of different elements in the MCM-41type mesoporous materials as shown by Kloetstraet al. [3].

Previous study of this reaction using alkalineexchanged zeolites [4] also showed that caesium

Fig. 2. Acrolein formation by double dehydration of glycerol.

exchanged X zeolite was giving higher selectivity todiglycerol than Na2CO3 whereas mesoporous cata-lysts containing various elements (Mg, La, Al, Mn)in the framework did not show better selectivity.

2. Experimental

2.1. Reaction

Glycerol etherification is carried out at 533 K in abatch reactor at atmospheric pressure under N2 in thepresence of 2 wt.% of catalyst, water being eliminatedand collected using a Dean-Stark system. Reagentsand products are analysed with a GPC after silyla-tion [5]. Analysis conditions: GPC equipped with anon-column injector, a FID and a polar column (HT5)supplied by SGE (L = 25 m, i.d. = 0.22 mm, thick-ness of the film= 0.10�m). Some of the numerousisomers of diglycerol and triglycerol (Fig. 3) are sep-arated (cyclic and linear); among the linear isomers,it is possible to estimate the�,�-diglycerol content.Batch processes are generally used in lipochemistry,especially for the esterification and transesterificationreactions (except for the preparation of methyl esters).

J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190 183

Fig. 3. Typical polyglycerols analysis by GC.

Acrolein formation was evaluated from the excess ofwater produced during the reaction. Different poroussolids were used in these studies differing in their sur-face area, porosity as well as acido–basic properties.

2.2. Preparation and characterisation ofthe catalysts

Impregnated mesoporous materials are preparedas follows: a certain amount (see notation) of alkali,alkaline earth and/or rare earth salt is added to 5 gof pure silica or aluminosilica mesoporous material,prepared (with cetyltrimethylammonium bromide) aspreviously described [4] and 50 g of methanol. Themixture is agitated at ambient temperature during 2 h,the solvent is then rapidly evaporated under vacuum

and the solid is calcined under air at 723 K overnightat a heating rate of 1 K/min.

Notation: element(amount of element impregnated×10−4

mol/g) Al(Si/Al ratio), example: Cs25Al(20)Nature of the salts: Cs (acetate), Mg (nitrate), La

(nitrate), Li (nitrate) and Na (nitrate).The other catalysts used have been described previ-

ously [4]. The samples containing different elementsin their framework (named “incorporated mesoporousmaterials”, notation: Me(n), wheren = Si/Me ratio)were prepared according to a procedure developed inour laboratory: we add dropwise 0.2 mol of a solu-tion of sodium silicate (27% SiO2) to a solution con-taining 0.022 mol of NaOH, 23.31 mol of H2O, anamount of the metal nitrate salt and 0.043 mol of tem-plate (cetyltrimethylammonium bromide). The pH is

184 J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190

adjusted at 10.5 with diluted HCl; then the formed gelis placed in an autoclave at 373 K for 24 h. The result-ing solid is filtered, washed with water, dried at 373 Kovernight and calcined under air at 823 K overnight ata heating rate of 1 K/min. Some of the solids have beenrecovered after reaction by filtration using hot waterand/or ethanol to dissolve the polyglycerols formed,and re-used without reactivation. Characterisation ofthe solids was performed using X-ray diffraction, N2adsorption, transmission electronic microscopy andchemical analysis.

3. Results and discussion

In this section we will present results and inter-pretations obtained with catalysts prepared accordingto different procedures going from less selective andunstable materials to selective and stable catalysts.

3.1. Homogeneous or modified zeolite catalysts

As a resume of the previous study [4], the re-sults reported in the Table 1 shows that Na2CO3is a more active and the less selective catalyst; thedistribution of polyglycerols being rather large. Csexchanged zeolites are quite selective to diglyceroland triglycerol while exchanged ZSM-5 are lessactive and selective. This is the result of the differ-ent size of the channels of ZSM-5 solids whatevertheir Si/Al ratio. In that particular case, the reac-tion could proceed on the surface of zeolite particlesrather than inside the channels. All these experimentsalso demonstrated that the primary hydroxymethylgroups were the most reactive giving linear and cyclicpolyglycerols.

Table 1Activity and selectivity of homogeneous and modified zeolite catalysts

Catalyst (preparation)(Si/Me ratio)

Glycerol conversion(%) after 8 h or (24 h)

Selectivity (%) after 8 h or (24 h)

Diglycerol Triglycerol Tetraglycerol Others

Na2CO3 94 (–) 27 (–) 31 (–) 21 (–) 21 (–)K/Al 2O3 (5% K) 42 (62) 83 (65) 17 (31) 0 (3) 0 (1)CsX (exchanged) 51 (79) 83 (62) 17 (33) 0 (4) 0 (1)CsX (impregnated) 36 (65) 88 (73) 12 (25) 0 (1) 0 (0)CsZSM-5(28) 12 (17) 94 (92) 6 (8) 0 (0) 0 (0)CsZSM-5(1000) 13 (42) 100 (80) 0 (20) 0 (0) 0 (0)

3.2. Mesoporous catalysts: incorporation ofelements in their framework

Different solids were obtained in changingthe element incorporated in the framework of asilico-alumina mesoporous material [4]. These results(Table 2) show that the nature of the element incorpo-rated (Al, Mg, La) in the framework of mesoporousmaterials modified the activity. On the other hand, wehave not observed significant changes of selectivitybetween these catalysts and homogeneous catalystsused in the industry. In a second step concerningthese “direct synthesised materials”, we decided tointroduce some other elements in the framework ofmesoporous materials, such as lithium and caesium.Table 2 also shows that the incorporation of lithium orcaesium did not change the activity and the selectivityof the mesoporous materials.

3.3. Impregnation of basic elements in theframework of the mesoporous materials

To increase the activity and/or the selectivity, wehave impregnated different elements; Li, Cs, La, Naand Mg over mesoporous materials.

3.3.1. Comparison between incorporated andimpregnated materials

As shown in Table 3, the impregnated material ismore active and more selective to (di-+ triglycerol)than the incorporated material.

3.3.2. Influence of the nature of the impregnatedelement

We demonstrated here that other impregnated ele-ments than caesium could be useful in the synthesis

J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190 185

Table 2Etherification of glycerol with different modified mesoporous materials (incorporation method)

Element incorporated(Si/element ratio)

Conversion (%),8 h (24 h)

Selectivity (%) 8 h (24 h)

Di Tri Tetra Others

Al(20) 9 (21) 74 (91) 26 (9) – –Al(∞); purely siliceous 12 (31) 81 (86) 19 (14) – –Mg(20) 15 (65) 90 (63) 10 (27) 0 (9) –La(20) 33 (94) 82 (26) 18 (23) 0 (27) 0 (24)Li(20) 15 (39) 80 (78) 10 (13) 0 (4) –Cs(20) 10 (27) 83 (78) 5 (11) 2 (9) –

Table 3Etherification of glycerol in the presence of mesoporous catalysts impregnated and incorporated with Cs

Catalysts Conversion (%) 8 h (24 h) Selectivity (%) 8 h (24 h)

Di Tri Tetra Others

Incorporated Cs(20)Al(∞) 10 (27) 83 (78) 5 (11) 2 (9) 0 (2)Impregnated Cs6Al(20) 16 (43) 94 (83) 4 (14) 2 (3) 0 (8)

of polyglycerols. The results presented (Figs. 4 and5) show that a lanthanum impregnated catalyst is themost active but the less selective material. Magnesiumimpregnated material is not very active but more selec-tive. Nevertheless, these catalysts were not studied fur-ther because they favoured glycerol dehydration andthe formation of acrolein, an undesirable by-product.Finally, as the caesium impregnated material had bothquite good activity and the most important selectiv-ity for the (di-+ tri-) glycerol synthesis, it was finallyselected.

Fig. 4. Activity of mesoporous solids impregnated with different elements.

3.3.3. Influence of the amount of impregnated elementWe observe that the catalytic behaviour depends

on the amount of Cs impregnated (Figs. 6 and7). The activity increases with the amount of Csbut the selectivity decreases. Some explanationsof these results: the BET surface area of the cat-alysts were measured and Table 4 shows that theintroduction of Cs decreases the BET surface area.This is the result of a structure collapse due toacido–basic interactions between the caesium oxide,formed during the calcination, and the silanol groups

186 J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190

Fig. 5. Selectivity of mesoporous solids impregnated with different elements.

Fig. 6. Activity of mesoporous solids impregnated with a different amount of Cs.

Table 4Chemical and physico-chemical properties of mesoporous solids impregnated with a different amount of Cs

Catalysts Specific surface area (BET) (m2/g) Chemical analysis of Cs (×10−4 mol/g) X ray diffractionda (Å)

Beforeimpregnation

Afterimpregnation

Beforetest

Aftertest

Beforeimpregnation

Afterimpregnation

Cs6Al(20) 838 620 6.4 0.1 35.7 33.8Cs25Al(20) 908 66 18.6 3.2 37.3 No peakCs50Al(20) 410 12 28.2 8.9 36.2 No peakCs100Al(20) 410 5 40.7 12.0 36.2 No peak

a d = 1.155a0.

J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190 187

Fig. 7. Selectivity of mesoporous solids impregnated with different amount of Cs.

and/or Lewis aluminium centres at the mesoporousmaterial surface.

This collapsing of the structure could explain thedecrease of the selectivity of the reaction but alsothe more important activity. During this collapsing,the caesium oxides are dissolved in the homogeneousphase where an homogeneous and not selective pro-cess took place.

Nevertheless, there is one more problem left withthat type of catalysts because of the caesium leach-ing during the reaction. This instability of the caesiumis observed whatever the amount of caesium impreg-nated. One solution could be the synthesis of a morestable coating (CsLaO2) as proposed in the literature[3], but the presence of La induces acrolein formation;what we have to avoid.

As the structure of the mesoporous catalyst seemsto be destroyed after impregnation of the caesium,the selectivity of these catalysts should be lowered.Such a change of the selectivity was not observed inincreasing the amount of caesium. At that momentone of the most important point was to know if thestructure of the porous material was really destroyed ornot. In order to give an answer, some characterisationsof the catalysts by transmission electronic microscopy(TEM) coupled with elemental analysis (EDX) weredone.

Fig. 8 shows that the solid has a mesoporousstructure before impregnation. Indeed, we observe onthis picture the hexagonal array of the mesoporous

structure. The comparison of Fig. 8 with that of theFig. 9, which represents an impregnated mesoporoussolid, demonstrates that the characteristics of themesoporous arrangement are still observed (withoutamorphous domain). So we conclude that, even if theXRD structure is no more observed, due to a decreaseof the long range order of the porous system of thesolid, there is still a short range ordered solid whichcan explain the high selectivity obtained with theseimpregnated mesoporous solids.

Fig. 8. TEM picture of a mesoporous solid (Al(20)).

188 J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190

Fig. 9. TEM picture of a mesoporous solid impregnated with caesium (Cs25Al(20)).

Moreover, the EDX analysis shows that whereverthe analysis is done, the same amount of caesium isobserved. As caesium clusters are not observed overthe impregnated solid, it is concluded that the caesiumis highly dispersed all over the mesoporous support.

Fig. 10. Etherification of glycerol, activity of different Cs catalysts[homogeneous (CsOH), heterogeneous (Cs25Al(20)) and mixed(Al (20) + CsOH)].

3.3.4. Experiments with different types of caesiumcatalysts

In order to know the role of the leached caesium onthe selectivity of the reaction, we decided to comparethree different catalysts: an impregnated one, a pure

Fig. 11. Etherification of glycerol, selectivity of different Cs cat-alysts [homogeneous (CsOH), heterogeneous (Cs25Al(20)) andmixed (Al(20) + CsOH)]. Di: diglycerols (�, �, �), tri: triglyc-erols ( , , ), di + tri: diglycerols and triglycerols (�, �, �).

J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190 189

Fig. 12. Activity of the re-used catalyst.

homogeneous CsOH catalyst and a mixture of meso-porous aluminosilica and caesium hydroxide, all thesecatalysts containing the same amount of Cs.

Figs. 10 and 11 show that the homogeneous CsOHis the most active but the less selective catalyst. Whatis very important for industrial application, is the (di-+tri-) selectivity at high conversion (85%). We noticethat the impregnated catalyst is very selective (87%)compared to the homogeneous CsOH (63%). This re-sult would mean that the porous impregnated catalysthas a real influence on the selectivity and lead to a

Fig. 13. Selectivity of the re-used catalyst.

shape selectivity(selectivity due to the control of theporous structure, i.e. the pore size).

3.3.5. Re-use of the catalystsA catalyst giving a high selectivity to (di-+

triglycerol) has been re-use (Cs25Al(20)R) in orderto know if the leaching of caesium was affectingthe selectivity of the solid. Figs. 12 and 13 show thatthe two catalysts (Cs25Al(20) and Cs25Al(20)R) havethe same selectivity to di- and triglycerol. The activityof the re-use catalyst is lower than that of the initial

190 J.-M. Clacens et al. / Applied Catalysis A: General 227 (2002) 181–190

one, what can be explain by the different caesium con-tent of the two solids.

4. Conclusion

In this study, we have prepared new impregnatedmesoporous materials and observed that the bestcompromise between activity, selectivity and catalystleaching was obtained with caesium impregnated onpure mesoporous silica materials. At a conversion of80%, such catalysts have a very high selectivity to[di- + triglycerol] (90%) compared to that of a ho-mogeneous “industrial” processes (72%). This resultrepresents a strong improvement of both the selec-tivity and the yield to [di-+ triglycerol] even if theglycerol conversion is quite high (80–90%).

Acknowledgements

The authors gratefully acknowledged support fromthe “FAIR program” of the European Community.

References

[1] A.J. Kaufman, R.J. Ruebush, in: T.-H. Applewhite (Ed.),Proceedings of the World Conference on Oleochemicals into21st Century, American Oil Chemist Society, 1991, p. 10.

[2] S. Abro, Y. Pouilloux, J. Barrault, in: Proceedings of the 4thSymposium on Heterogeneous Catalysis and Fine Chemicals,Basel, 8–12 September 1996.

[3] K.R. Kloetstra, M. van Laren, H. van Bekkum, J. Chem. Soc.,Faraday Trans. 93 (6) (1997) 1211.

[4] J.-M. Clacens, Y. Pouilloux, J. Barrault, C. Linares, M.Goldwasser, Stud. Surf. Sci. Catal. 118 (1998) 895.

[5] M.R. Sahasrabudhe, JAOCS 44 (1996) 376.