J. CHEM. SOC. FARADAY TRANS., 1994, 90(4), 667-674 Al-Pillared Saponites Part 1.-IR Studies Sophie Chevalier, Raymonde Franck, Helene Suquet, Jean-Franqois Lambed* and Denise Barthomeuf Laboratoire de Reactivite de Surface, URA 1106 CNRS, Universite Pierre et Marie Curie, 4, Place Jussieu, 75252 Paris Cedex 05, France Al-Pillared saponites have been prepared from natural Na-exchanged Ballarat saponite. Acidic properties of these saponites have been studied by IR spectroscopy using pyridine as a probe molecule. Al-Pillared saponites exhibit both Brsnsted and Lewis acidity. Lewis sites are evidenced in the parent Na-exchanged clay as well as in the pillared clays. Brsnsted acid sites are not evidenced in the parent Na-exchanged saponite but Al-pillared saponites exhibit Brsnsted acidity as shown by a pyridinium band at 1549 cm-'. Brsnsted acid sites have been related to OH groups at 3740-3720 cm-' and 3597-3594 cm-'.The same wavenumbers are evidenced for v(0H) stretchings in acidified saponite and in Al-pillared saponites. It is assumed that an acidic attack of Si-0-AI linkages which may provide Brsnsted sites occurs on the clay sheets during the pillaring process. Many attempts have been made to improve thermal stability and pore size of pillared clays.'-5 Thermal stability and pore size are significant properties in the use of pillared clays as cracking catalysts. However, acidity is also relevant for this purpose. Al-Pillared saponites, stable up to 750"C, with a pore size greater than 8 8, have been ~repared.~.~ This study has been undertaken to check their acidic properties.Pyridine adsorp- tion by the pillared clays and subsequent desorption at increasing temperatures was followed by IR spectroscopy as a test method. Experimental Starting Materials Starting materials were a Ballarat saponite provided by the Source Clays Repository of the Clay Minerals Society (University of Missouri, USA) and an A1 polycation solution containing the 'Keggin &-isomer' polycation, designated All ,, and possibly other A1 polymers. A Chlorhydrol commercial solution from the Reheis Chemical Company was used as one of the sources of A],, . This All, polycation was also pre- pared from an AlCI, solution treated with NaOH.' Catalyst Preparation (i) The Ballarat saponite was previously purified by a sedi- mentation process.The fraction of particles less than 2 pm in size was collected and then exchanged, at least three times, with 1 mol NaCl 1-' solution. The resulting sample is desig- nated SNa and used for preparations at a concentration of 5 g 1-'. The unit cell corresponds to the formula: (Si7.262 A10.7 38)[Mg5.920Fe~.~04Ti0.006Mn0.0021Na0.600ca0.028K0.006 02,(OH), .A slightly different formula was previously proposed by Post for the Ballarat ~aponite.~ According to the chemical analyses he reported, the clay is assumed to have 0.16 half-cell octahedral positions filled with Al. In our sample, however, no octahedral A1 was detected by 27Al NMR spec- troscopy.6 (ii) The Chlorhydrol commercial solution was diluted to 0.1 mol 1-' and matured for 2 h at 60°C.This initial solu- tion, designated ACH, has a pH of 4.8. Adding a few drops of concentrated NaOH raises the ACH pH to 6. Adding a 2 mol 1-' ammonium acetate solution3 to the initial ACH solution with an NH4 :A1 ratio of 15 :1 leads to a pH of 7. These pillaring solutions of ACH were added to an SNa suspension to give an Al, in SNa of 5 mmol g- ' where Al, is the concen- tration of aluminium in solution. Samples A, B and C were prepared with ACH of pH 4.8, 6 and 7, respectively. Sample D was prepared with the solution of AlCl, titrated with NaOH and designated AHY (pH = 5.0 and Al, in SNa of 5 mmol g- '). After addition of ACH or AHY to SNa, the mix- tures have been kept at 80°C for 2 h.Further experimental details have been published el~ewhere.~. ' (iii) The preparation process used for sample A was repeat- ed replacing the ACH solution by 1 mol 1-' HCl. The corres- ponding sample is designated ASNa and called 'acidified saponite'. IR characterization has been achieved for SNa, ASNa, and samples A and C. In a companion paper," an NMR study of SNa, A, B, C and D samples is reported. Thermal Treatments SNa, ASNa and samples A, ByC and D were oven-dried at 60°C for 12 h. We refer to the All,-contacted samples after drying at 60 "C as 'Al-intercalated' samples. Amounts of SNa, ASNa and samples A, B, C and D were then heated at 500°C or 750°C for 4 h. The heating rate was 36°C h-' up to 500 "C and 90 "C h- ' between 500 and 750 "C.The heated samples were designated SNa60, SNa500, SNa750, A60, A750, B500 etc. according to the temperature of the thermal treatment. Only samples stabilized by heating at tem-peratures of 500 "C or more are called 'pillared clays'. Thermal Analyses Simultaneous thermogravimety (TG) and differential thermal analysis (DTA) have been performed on a Netzsch Simulta- neous Thermal Analyzer STA 409 in a covered alumina cru- cible. The sensivity of a 100% weight loss has been fixed at 25 mg and the DTA sensitivity at 50 pV. X-Ray Difbaction Spectra (XRD) XRD reflexion spectra were recorded on a SIEMENS D500 diffractometer using CU-Ka radiation. For the pillared sapon- ites, the d(001) spacings have been determined from the 001 line maximum of intensity.N,Adsorption (BET) Before adsorption, samples were previously degassed for 1 h at 150°C under N, flow. N, adsorption measurements were performed on a Quantasorb Junior apparatus and surface areas were determined from the BET treatment. The upper limit of partial pressure of N, was fixed at 0.1-0.12 to use the linear range of the adsorption isotherm. While the use of the BET model is questionable for pillared clays,12 we provide the BET surface areas to allow easy comparison with the lit- erature. IR Spectroscopy Al-Pillared saponites can be ground to fine powders. These powdered samples, stored at room temperature, were sieved and pelletized as self-supported wafers.SNa60, SNa500 or ASNa500 cannot be pulverized by dry grinding. It was thus necessary for these samples, to prepare films from the corre- sponding colloidal solutions. It has already been shown that hydration is a reversible process for SNa heated to 700 OC.13 Wet grinding allows formation of a colloidal solution of SNa500 and ASNa500 from which films may be obtained. The films or wafers were placed in a vacuum cell specially adapted for adsorption-desorption experiments. They were heated under oxygen flow, overnight, at 450°C and then kept for 12 h at this temperature, under a Torr vacuum. After cooling to room temperature, their IR spectra were checked. Pyridine adsorption was carried out at room tem- perature for 1 h.The pyridine desorption was followed by heating for several 6 h steps, under vacuum, at increasing temperatures. IR spectra were recorded at each stage on a Perkin Elmer 1700 FTIR spectrophotometer, equipped with a Perkin Elmer 3600 data station. Results Table 1 lists the results of the sample characterization of XRD and BET surface area measurements. The fixed A1 content varies with the preparation procedure indicating that the pillaring aluminium species may be different in charge and/or degree of polymerisation according to the pH and to the dilution." Pyridine adsorption-desorption experiments for IR mea- surements have been performed on SNa500, A500, A750 and C500 samples. IR Spectra in the v(0H) Stretching Vibration Range IR spectra in the v(0H) stretching vibration range are shown in Fig.1 and the corresponding wavenumbers are listed in Table 2. The dehydrated SNa5OO saponite [Fig. l(a)] shows three absorption bands referred to as vl, v2 and v3. Their assign- ment has been known for a long time (see e.g. ref. 14). The Table 1 XRD basal spacings and BET surface areas thermal fixed A1 sample treatment/ "C content"/ mmol g-' d(001)/A S(BET!/ m2 g- SNa 500 - 12.4 36 A 60 2.7 18.7 360 500 17.7 287 750 17.2 148 B 60 3.9 18.8 373 500 17.8 303 750 17.2 212 C 60 4.5 19.0 205 500 17.8 256 750 16.9 118 D 60 2.1 18.5 339 500 17.8 332 750 16.9 67 a From the solution. J. CHEM. SOC.FARADAY TRANS., 1994, VOL. 90 3500 3000 wavenumber/cm-' Fig. 1 IR spectra in the v(0H) range of (a) SNa500 (23.1 mg) film, or of (b) A500 (27.0 mg), (c) C500 (27.5 mg) and (d) A750 (25.8 mg) self-supported wafers strongest band (v, at 3674 cm- ') is assigned to the hydroxy- group stretching vibration of the Mg3(OH) units. The electric field associated with the interlayer cations, namely Na', Ca2+ and K+,toward which some OH groups are directed, raises the wavenumber of v1 = 3716 cm-'. The very weak v3 band at 3622 cm-' is assigned to the stretching vibration of hydroxy groups belonging to one trioctahedral Mg,R(OH) unit, i.e. to one unit where cations such as R = Mn or Fe replace Mg in the octahedral layer. In the dehydrated Ballarat saponite, the average unit cell contains four OH groups, 0.66 interlayer cations (Na' or equivalent Ca2 + and K +) and 0.12 divalent atoms replacing Mg in the octahedral sites.Assuming that the molar absorp- tion of the different hydroxy groups is identical, that no coupling between OH vibrations occurs and that no contri- bution from one type of OH group is present at the same wavenumber as another type of OH group, the relative inten- sity of the observed v(0H) bands would be consistent with the cation content in the dehydrated sample. Indeed, if one interlayer cation Na' or equivalent (K', Ca2+) perturbs one OH group, the expected intensity ratio Z(v1)/Z(v2) is found to be 0.17 and the experimental value is also 0.17. Assuming a random distribution of R substituents in the trioctahedral groups, the relative intensity of v3 versus v2 may Table 2 IR spectra in the v(0H) stretching vibration range (before pyridine adsorption) sample vl/cm- v2/cm-' vJ/cm-' vdcm-' SNa5OO 3716 3674 3622 A500 3722 3674 3622 3594 c500 3730 3674 3622 3597 A750 3736 3663 ? ? J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 be calculated as the ratio 3x2(1-x)/x3 if (1 -x) and x are the atomic proportions of substituent and Mg, respectively. For 0.12 R2+ replacing Mg in one unit cell, 1(v3)/1(v2)is cal- culated to be 0.06 and the experimental value is found to be 0.08. However, this band is weak and the experimental ratio I(v~)/I(v~)is probably not very accurate.The same assignment as for SNa500 has been proposed for a Llano Na-vermiculite" pretreated at 550 "C which exhibits two bands at 3675 cm-' for v(0H) of non-perturbed OH of Mg, units and at 3708 cm-' for v(0H) perturbed by Na' cations. The Av shift between these bands is 33 cm-' while in SNa it reaches 42 cm-'.This is the sign of a stronger inter- action of Na+ with the OH groups in SNa. Some changes occur in the spectra of the dehydrated pil- lared clays A500 [Fig. l(b)] and C500 [Fig. l(~)]as compared with SNa500: (i) The v2 band of the structural hydroxy group bonded to three Mg atoms does not shift but is strongly enhanced and broadened. The sampling may account for the enhancement. It is well known that film deposits are oriented and that the OH stretching interaction of the IR beam with hydroxy ions depends on the angle of incidence of the beam with respect to the 0-H bond axis.16 In the SNa sample prepared by depo- sition from aqueous solution, the film plane corresponds to the a-b plane of the layers and the direction of propagation of the infrared beam is roughly parallel to the 0-H bond axes resulting in a weak absorption band.The significant broadening of the v2 band may be due to the frequency shift of many OH oscillators owing to the change in their environment. This broadening is stronger in the lower wavenumber range as observed with the pillared hectorites.' (ii) In the vicinity of the v1 band assigned to OH groups perturbed by Na+ or K+ cations in SNa, a shoulder is still present at 3722 cm-' in A500 [Fig.l(b)]. This shoulder, although partially hidden by the broadening in C500, can still be guessed to be around 3730 cm-' [Fig. l(c)]. (iii) The main changes occur around 3600 cm-': two peaks at 3622 and 3594 cm-' in A500 and at 3622 and 3597 cm-' in C500 are sharply marked instead of the very weak band at 3622 cm-' in SNa500. It may be concluded that the new peak at 3594-3597 cm-' referred to as v4 corresponds to a new hydroxy type created by the pillaring process. A weak shoulder in the region 3625-3590 cm-' is assigned by Occelli and Finseth' to H-bonded hydroxy groups in pillared hecto- rite catalysts. For the A750 sample, the dehydroxylation of structural OH is probably almost achieved.The DTA and TGA curves in Fig. 2 show that structural hydroxy groups of the octa- hedral layer are condensed to give water at around 800°C lo( 30 when the sample is heated in air, at a 10°C min-' heating rate. A750 was kept at 750"C, under air flow, for 4 h. It may be assumed that this treatment is long enough to cause the corresponding dehydroxylation. Therefore, there may be expected a decrease of the v(0H) band intensities. In addi- tion, the thermal treatment at 750°C has another obvious effect on the pillared clay and the self-supported wafer of A750 shows an important diffusion in the 4000-3000 cm-' range as evidenced by the baseline. The poor transmittance of this sample results in a very noisy absorbance spectrum [Fig.l(d)]. The peaks in the 3600 cm-' range can no longer be evidenced and the one near 3720 cm-' band is shifted to 3736 cm-'. ASNa500 acidified saponite exhibits in the v(0H) range the spectrum shown in Fig. 3(b). The 3716 cm-' v1 band is shifted to 3738 cm-' and the new peak v4, already observed in A500 and C500, is also present and centred at 3595 cm-'. Pyridine AdsorptiowDesorption Eflect on the v(0H) Stretching Vibration Range Spectral changes in this range are shown in Fig. 4-7. The spectrum of SNa500 is hardly transformed by pyridine adsorption: after desorption at 150"C, no frequency shifts are observed and a less than 10% intensity decrease is seen on the v1 and v2 bands [Fig. 4(b)].The former intensity of these bands is recovered by desorption at 250°C [Fig.qc)] showing the very weak acidic character of the corresponding hydroxy groups of the SNa. As shown by the d(001) value and by the surface area (Table l),the interlayer space is col-lapsed for SNa500 and pyridine does not enter this space. A500 and C500 pillared clays both behave in the same way with respect to pyridine adsorption-desorption : (i) As for SNa500, a slight decrease of the 3674 cm-' v2 band is observed after adsorption of pyridine and desorption 3674 Q) 3716C es: -Jn 3716 3594 II I I 0 200 400 600 800 1000- 40 1 3500 3000 temperaturePC waven u mber/cm -' Fig. 2 (a) TGA and (b) DTA curves of oven-dried sample A (in air, Fig. 3 IR spectra in the v(0H) range of dehydrated films of (a) heating rate 10"Cmin-;sample weight 68.80 mg) SNaSOO and (b) ASNaSOO J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 3622 P) C e % 2P I I 4c I 3500 30( wavenum ber/cm -' Fig. 4 IR spectra in the v(0H) range of SNa500 (a)before pyridine adsorption and after desorption at (b)150, (c)250 and (6) 350 "C at 150°C [Fig. 5(b) and qb)] and the former intensity is recovered by desorption at 250°C [Fig. 5(c) and qc)]. No frequency shift is observed for this band. (ii) The v1 shoulder near 3722 cm-in A500 and near 3730 cm-' in C500 disappears on pyridine adsorption and is seen again after desorption at 450°C [Fig. 5(e) and qe)]. After desorption at 520°C [Fig. 5(f) and qf), the sharpening of the v2 band allows to see that the v1 peak is flattened and broad and its maximum may be centred between 3738 and 3722 cm-', that is, significantly higher than the correspond- ing peak at 3716 cm-' in SNa500.It seems likely that OH groups responsible for this peak in the SNa are different to those in A500 and the C500 samples. (iii) The intensity of the 3622 cm-' vj band decreases slightly on pyridine adsorption and desorption at 150 "C [Fig. 5(b) and qb)] and v3 is seen again after desorption at 250°C [Fig. 5(c) and qc)]. This behaviour is the same as for the v2 band, but after desorption at 450"C, the v3 band dis- appears again because of dehydroxylation. It is known" that hydroxy groups bonded to Fe2+, for example, are the first to undergo decomposition by hydrogen loss with concomitant oxidation of octahedral Fe2+.(iv) The 3594 cm-' v4 band in A500 disappears completely on adsorption of pyridine and subsequent partial desorption at 150°C [Fig. 5(b)]. It is seen again after desorption at higher temperatures with its strongest intensity after desorp- tion at 350°C. However, it does not recover its former inten- sity and then decreases with increasing temperatures of desorption. It is still evidenced at 520°C [Fig. 5(f)]. The cor- responding OH group is obviously more acidic than the others. For C500, a similar evolution of the 3597 cm-' v4 band is seen but at lower temperatures of pyridine desorption: the strongest intensity of v4 is recovered at 250°C [Fig. qc)] and the band is completely flattened at 520 "C [Fig.qf)]. 4000 3500 3000 waven um ber/cm -' Fig. 5 IR spectra in the v(0H) range of A500 (a) before pyridine adsorption and after desorption at (b)150, (c)250, (6)350, (e) 450 and (f) 520 "C A750 behaves in an unexpected way with respect to pyri- dine adsorption (Fig. 7). A net broadening of the main peak at 3663 cm-' is observed from 150 to 250°C just as if H bonds were created. This broadening disappears at 350 "C, as is expected of H bonds. Egect in the 1700-1300 crr-' Range Fig. 8-11 present the IR spectra of SNa500, A500, C500 and A750 after pyridine adsorption and subsequent pyridine desorption at increasing temperatures, in the 1700-1 300 cm-' range. The numerical values are listed in Table 3.SNa500 after pyridine adsorption and subsequent desorp- tion at 150°C [Fig. 8(a)] shows only vibration bands assign- ed to pyridine coordinately bonded on Lewis acid sites at 1574 and 1442 cm-' and to H-bonded pyridine at 1594 and 1608 cm-'. No Brernsted acid sites (or at least very few) are evidenced. Water molecules have been totally evacuated at 500°C [Fig. l(a)] and the interlayer space is collapsed (Table 1). It is not surprising that pyridine is not sorbed much on SNa500 and that it is bound only by H bonding with surface oxygens and by nitrogen coordination to Lewis sites on layer edges. Partial desorption of pyridine at higher temperatures [Fig. 8(b) and (c)] causes a splitting of the 1442 cm-' band as fre- quently observed on oxides'gi20 smectites2' or zeolite samples.22 It has been s~ggested'~*~' that the presence of at least two groups of Lewis sites on the surface may account J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 67 1 a) C e s 9, 4000 3500 3000 wavenumber/crn -' Fig. 6 IR spectra in the v(0H) range of C500 (a) before pyridine adsorption and after desorption at (b) 150, (c) 250, (d) 350, (e) 450 and (f) 520°C for this splitting. These groups would differ in acid strength and behave differently with respect to the temperature. Simi- larly, it has been reported by Ward23 that the exchangeable cations in zeolites can also act as Lewis acid centres and give rise to similar bands as do low-coordinated aluminium ions. However, it has been recently shown in a fl zeolite that a surface change cannot account for the splitting of the band near 1450 cm-' 24 and that the decomposition of pyridine with increasing temperatures may be considered.Thermal Q) C e s 9, 1 4000 3500 300 wavenurnber/crn-' Fig. 7 IR spectra in the v(OH) range of A750 (a) before pyridine adsorption and after desorption at (b) 150, (c) 250, (d) 350 and (e) 450 "C analysis of pyridine-treated sepiolite and palyg~rskite~~ leads to the same conclusion. A500 after pyridine desorption at 150°C [Fig. 9(a)] shows bands which may be assigned not only to pyridine coordi- nately bonded on Lewis acid sites (at 1450, 1575 and 1622 cm-') and to H-bonded pyridine (at 1613 cm-') but also to pyridinium ion (at 1549 and 1640 cm-I).The corresponding Brmsted acidity is thus correlated to the presence of a new species introduced by the pillaring process. This acidity is no longer evidenced at 450"C [Fig. 9(d)]. Thus, it may be related Table 3 IR spectra in the 1700-1300 cm-' vibration range after pyridine desorption at different temperatures sample temperaturefc BPY LPY H4, LPY BPY BPy + LPy + HPy LPY SNa5OO 1 50 1608 1594 1574 1490 1442 350 1608 1448 1594 1575 1490 1436 A500 150 1640 1622 1613 1575 1549 1492 1456 1450 520 1622 1492 1456 c500 150 1640 1622 1613 1575 1549 1492 1456 1450 520 1633? 1622 1492 1456 A750 150 1640 1622 1610 1575 1549 1492 1448 450 1610 1492 1448 BPy = pyridine adsorbed on Brmsted acid site.LPy = pyridine adsorbed on Lewis acid site. HPy = pyridine adsorbed by H bonding. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1450T I 1700 1500 1300 wavenumber/cm-' Fig. 8 IR spectra in the 1700-1300 cm-' range of SNa500 after pyridine adsorption and subsequent desorption at (a)150, (b)250 and (c) 350°C to the hydroxy group which gives a new band at 3594 cm-' in the v(0H) vibration range [Fig. 5(b)-(f)]. This new hydroxy group is acidic and behaves differently from the other OH groups. The band assigned to pyridine on Lewis acid sites is already split, at 150"C, into two components: 1450 and 1456 cm-'. At higher temperatures the 1456 cm-' component decreases more slowly than the 1450 cm-' one [Fig.9(b)-(41.C500 after pyridine desorption at 150 "C [Fig. lqa)] shows a spectrum very similar to that of A500 with vibration bands assigned to coordinatively bonded pyridine on Lewis sites at 1450, 1575 and 1622 cm-', to H-bonded pyridine at 1613 cm-' and to pyridine bonded on Brernsted sites at 1549 and 1640 cm-'. The 1450 cm-' band assigned to pyridine on Q, t e 8P Q, C -ESIf? I 1'; 0 1500 1300 wavenumber/cm-' Fig. 10 IR spectra in the 1700-1300 cm-' range of C500 after pyri- dine adsorption and subsequent desorption at (a) 150, (b)250, (c) 350, (6)450 and (e)520°C Lewis sites is not so sharply split but shifts to 1456 cm-' with increasing temperatures of desorption [Fig.lqb)-(e)]. The Lewis :Brernsted site ratio which may be drawn from 1450 : 1549 cm-' band intensity ratios is greater for C500 than for A500 and may be related to the fixed A1 content (Table 1) of the corresponding samples. A750 shows an enhanced 1610 cm-' band and rather weak bands due to pyridine on Brernsted and Lewis sites [Fig. ll(a)]. Even though the 1610 cm-' band cannot be assigned only to H-bonded pyridine, it is consistent with observations in the v(0H) vibration range. The peak assigned to pyridine coordinated on Lewis sites is unique and centred at 1448 cm-' as one of the components in SNa500. It does not shift with increasing temperatures of desorption. 1700 1500 1300 wavenumber/crn -' Fig. 9 IR spectra in the 1700-1300 cm-' range of A500 after pyri- dine adsorption and subsequent desorption at (a)150, (b)250, (c) 350, (d)450 and (e) 520 "C J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 As already mentioned, the splitting of the band assigned to pyridine coordinately bonded to Lewis acid sites is frequent. Assuming that each component of this band corresponds to a different acid site, it would be attractive to assign the 1448- 1450 cm-' component to sites on the clay layer and the second component to sites influenced by the interlayer: 1436 cm-' in SNa500 where Na' is the major cation and 1456 cm-' in C500 and A500 where the pillar is in the interlayer. However, decomposition of pyridine with increasing tem- perature cannot be discarded and further experiments would be necessary to support one of these explanations.Discussion Both Lewis and Brsnsted acid sites have usually been evi- denced in most studies on pillared clays. In the various systems investigated, it is generally agreed to impute the main source of the Lewis acidity to the pillar^.^^-^' Indeed, if Lewis acidity is evidenced in the parent clay, the pillaring obviously enhances this acidity at least by facilitating access to the interlayer space. Generally, Lewis acidity is directly related to the kind and to the amount of interlayer pillars. On the contrary, controversial conclusions are reported on the Brsnsted acidity which seems to depend on the parent clay and on parameters such as pretreatment temperature or .~~outgassing temperature.Plee et ~1 have assigned Brsnsted acidity of Al-pillared beidellite to OH groups formed by protons captured by Si-0-A1 linkages. Occelli and Tindwa3' reported a protonic acidity due to OH groups associated with ACH pillars and He et aL3' asserted that the Brsnsted acidity is provided mainly by structural OH groups. Some discrepancies may be noticed too on the Lewis :Brsnsted ratios and on the acid strengths reported. In the IR spectra, Brsnsted acidity has been related to an OH group belonging to Si-OH-A1; its stretching frequency is centred at 3440 cm- ' in Al-pillared montmorillonite and beidellite.26 This band is coupled with a pyridinium band at 1540 cm-'. Acidic OH groups have been evidenced in Al-pillared bentonite3' by a band near 3700 cm-' which disap- pears upon pyridine adsorption.According to our results on saponite, Brsnsted acidity is not evidenced on SNa while it is shown on the Al-pillared saponites: in SNa500, vl, v2 and v3 are almost insensitive to pyridine sorption ; in Al-pillared saponites, v1 (near 3740- 3720 cm-') and v4 (at 3594-3597 cm-') bands are sensitive to the pyridine sorption and may be considered as due to acidic groups. So, with pillared saponites, we found OH groups involved in pyridinium ion formation both at higher wavenumbers than structural OH groups such as in benton- ite and at lower wavenumbers than structural OH groups such as in beidellite. Hydroxy groups of the v4 band slowly disappear from 350°C [Fig.5(c) and qc)] while those of the v1 band are still evidenced at 520°C [Fig. 5(f) and 6(f)].The spectra of synthetic saponite samples expanded by SiO, * TiO, colloidal particles after air drying3, exhibit bands at 3738, 3668 and 3590 cm-' which are very close to vl, v2 and v4 in our samples. These observations suggest that the Brsnsted acidity of the OH groups responsible for v1 and v4 absorptions does not depend on the nature of pillars. More- over, if we consider the results for acidified saponite, ASNa500, prepared in the same way as A500 by replacing the ACH solution by 1 mol 1-' HCI, and sampled indentically to SNa500 [Fig. 3(b)], the presence of a band at 3594 cm-' (wavenumber of v4 in A500) and the shift of v1 from 3716 to 3738 cm-' may be immediately noticed on the spectrum.Thus, the two bands which may be considered as due to acidic OH groups, namely v1 and v4 in Al-pillared saponites, are also evidenced exactly at the same wavenumbers in the 673 acidified saponite. When Brsnsted acidity is found in a parent clay, it may be related to an acidic treatment, cf. the case of acid-washed hectorite' as compared with ACH- pillared hectorite. By treating Na-beidellite with a 0.05 mol 1-' HCI solution, Schutz et observed that Na' is exchanged by H,Of. A strong band at 3440 cm-' in the IR spectrum is attributed to the corresponding bridging OH (Al-OH-Si). This band is also present in the spectrum of pillared beidellite26.29.33*34 at the same wavenumber. On the contrary, the parent Mg-hectorite studied by Occelli3' was not acid washed and does not sorb significant amounts of pyridine, similar to SNa500, and the band of pyridinium ions at 1546 cm-' is not present while it is seen on pillared hecto- rite samples.From our experiments, we may conclude that Brsnsted acidity results from H+ attack on the clay sheets. As it is evidenced on the non-pillared sample, the attack of Al-0-Si linkages of the layer seems to be involved, as already pr~posed.~~,~~ NMR experiments which will be developed further are also consistent with this hypothesis.' ' Moreover, the spectrum in Fig. 3(b) shows that the 3738 cm-' band has obviously a different origin from that at 3716 cm-' assigned to the Na' perturbation of OH.Na' exchange by H + is not completely achieved as seen by a very weak residual absorption and by the weaker intensity of the 3594 cm-' band as compared with that of A500. in the saponite expanded by SiO, TiO, colloidal particle^,^' a band at 3738 cm-' is also observed; it is assigned by the authors to the OH of silanol groups on the pillars. The inten- sity of this band is obviously higher with SiO, pillars than with ACH pillars. Thus, attention may be paid to the assign- ment of this band in our saponite samples to OH from silanol groups of amorphous silica as frequently observed on leached clays or in many zeolite samples. The behaviour of A750 in the v(0H) range may be interpreted by the increasing of the silica amount in this sample, probably because of the beginning of layer disorganization with simultaneous water formation.However, in Occelli's samples, the 3738 cm-band does not decrease drastically on pyridine adsorption as seen for A500 and A750 (Fig. 5 and 7). In the saponite sample treated by supercritical drying,32 it is more intense than in the air-dried sample and again does not decrease on pyridine adsorption. Usually, silanols of amorphous silica are not sen- sitive to pyridine sorption. Moreover, amorphous silica has not been evidenced by 29Si NMR spectroscopy in our samples. The assumption that a similar perturbation on structural OH groups as that due to Na+ cations is created by H30+ cations with a higher shift towards high wavenum- bers is not consistent with the experimental temperature range.H30+ cations can no longer be observed at 520°C and are not evidenced in the near 1700 cm-' range. Accord- ing to the observed wavenumber, the corresponding OH groups are rather 'free' as compared with the structural ones and we may assume that they are surface groups or groups located in the sheet edges which generally represent 10% of the total OH amount. Conclusion Both Lewis and Brernsted acidity have been evidenced in the Al-pillared saponites. Lewis sites are present both in the pil- lared and in the parent clay. According to the temperature of pyridine desorption, it may be concluded that Al-pillared saponites exhibit a very strong Lewis acidity. Brsnsted acidity of Al-pillared saponites is also strong.It is linked with the appearance of two OH-stretching vibration bands near 3735 and 3695 cm-'. Pillars do not bear Brransted acidic OH 674 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 groups directly but they do facilitate the access of protons to clay sheets. Our results are in good agreement with the assumption of a protonic attack of Si-0-A1 linkages to give this acidity. It has to be taken into account that protons may be yielded 14 15 16 V. C. Farmer, in The Znfrared Spectra of Minerals, ed. V. C. Farmer, Mineralogical Society, London, 1974, p. 331. J. M. Serratosa and J. A. Rausell-Colom, Mineral Petrogr. Acta, Part A, 1985,29,409. J. M. Serratosa and W. F. Bradley, J. Phys. Chem., 1958, 62, 1164.either by hydroxonium ions in the interlayer space provided by Na exchange with HCl or by reactions during the pillaring process, especially proton release upon pillar calcination with oxide formation. Saponite is a very convenient material to support this hypothesis: it easily dehydrates and does not show Brarnsted acidity in the Na form. Al-0-Si linkages of the layers undergo acid attack by giving acidic OH groups which may be seen at the same wavenumber (3594-3597 cm-') in the spectra of acidified saponite and of pillared saponites. A new assignment of the band near 3740-3720 an-' to free hydroxy groups located on the edges may be proposed. 17 18 19 20 21 22 23 24 M. L. Occelli and D. H. Finseth, J. Catal., 1986,99, 316. J. M. Serratosa and J. A.Rausell-Colom, Mineral Petrogr. Acta, Part A, 1985,29,399. E. P. Parry, J. Catal., 1963,2, 371. D. Ballivet, D. Barthomeuf and D. Pichat, J. Chem. SOC., Faraday Trans. I, 1972,68, 1712. J. P. Rupert, W. T. Grandquist and T. J. Pinnavaia, in Chemistry of Clays and Clay Minerals, ed. A. C. D. Newman, Mineralogical Society, Monograph no. 6, Longman, London, 1987, p. 282. R. Beaumont, P. Pichat, D. Barthomeuf and Y. Trambouze, Catalysis, 1973, 1, 343. J. Ward, in Zeolite Chemistry and Catalysis, ed. J. A. Rabo, ACS Monograph 171, Washington DC, 1976, p. 225. C-J. Jia, P. Massiani and D. Barthomeuf, J. Chem. Soc., Faraday Trans., 1993,89, 3659. References 25 U. Shuali, S. Yariv, M. Steinberg, M. Muller-Vonmoos, G. Kahr and A. Rub, Clay Miner., 1991,26,497. 1 2 3 4 5 6 7 8 9 10 11 12 J.Shabtai, F. E. Massoth, M. Tokarz, G. M. Tsai, J. McCauley, Proc. 8th Znt. Congr. Catal., Dechema, Verlag Chemie, Wein- heim, Berlin, 1984, p. 735. Jingjie Guan, Enze Min, and Zhiqing Yu, Proc. 9th Znt. Congr. Catal., ed. M. J. Phillips and M. Ternan, The Chemical Institute of Canada, Calgary, 1988, p. 104. F. Figueras, Catal. Rev.-Sci. Eng., 1988,30,457. D. Tichit, F. Fajula, F. Figueras, C. Gueguen and J. Bousquet, in Advances in Fluid Catalytic Cracking: Role in Modern Rejin- ing, ed. M. L. Occelli, ACS Symposium Series 375, American Chemical Society, Washington DC, 1988, p. 237. M. L. Occelli, in Keynotes in Energy-related Catalysis, ed. S. Kaliaguine, Stud. Su$ Sci. Catal., Elsevier, Amsterdam, 1988, S. Chevalier, R. Franck, J-F.Lambert, D. Barthomeuf and H. Suquet, Proc. 7th Euroclay Conference, ed. M. Storr, K-H. Henning and P. Adolphi, European Clay Groups Association, Dresden, 1991, p. 207. S. Chevalier, Thesis, Paris, 1992. S. Schonherr, H. Gorz, D. Muller and W. Gessner, Z. Anorg. Allg. Chem., 1981, 476, 188. J. L. Post, Clays Clay Miner., 1984,32, 147. S. Chevalier, R. Franck, H. Suquet, C. Marcilly and D. Bartho- meuf, in Expanded Clays and Other Microporous Solids, ed. M. L. Occelli and H. Robson, Synthesis of Microporous Materials, Van Nostrand, Reinhold, New York, 1992, p. 32. J-F. Lambert, S. Chevalier, R. Franck, H. Suquet, and D. Bar- thomeuf, J. Chem. Soc., Faraday Trans., 1994,90,675. F. Bergaya, L. Gatineau and H. Van Damme, in Multifunctional Mesoporous Inorganic Solids, ed.C. A. C. Sequeira and M. J. vol. 35, p. 101. 26 27 28 29 30 31 32 33 34 35 D. Plee, A. Schutz, G. Poncelet and J. J. Fripiat, in Catalysis by Acids and Bases, ed. B. Imelik, C. Naccache, G. Goudurier, Y. Bentaarit and J. C. Vedrine, Elsevier, Amsterdam, 1985, p. 343. Liu Zhonghui and Sun Guida, in Zeolites, ed. B. Drzaj, S. Hocevar and S. Pejovnik, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 1985, vol. 24, p. 493. D. Tichit, F. Fajula, F. Figueras, J. Bousquet and C. Gueguen, in Catalysis by Acids and Bases, ed. B. Imelik, C. Naccache, G. Goudurier, Y. Bentaarit and J. C. Vedrine, Elsevier, Amsterdam, 1985, p. 351. G. Poncelet and A. Schutz, in Chemical Reactions in Organic and Inorganic Constrained Systems, ed. R. Setton, Reidel, Dordrecht, 1986, p. 165. M-Y. He, Z. Liu and E. Min, Catal. Today, 1988,2, 321. M. L. Occelli and R. M. Tindwa, Clays Clay Miner., 1983, 31, 22. M. L. Occelli, K. Takahama, M. Yokoyama and S. Hirao, in Expanded Clays and Other Microporous Solids, ed. M. L. Occelli and H. Robson, Synthesis of Microporous Materials, Van Nos- trand Reinhold, New York, 1992, p. 57. A. Schutz, D. Plee, F. Borg, P. Jacobs, G. Poncelet and J. J. Fripiat, Proc. Znt. Clay ConJ, Denuer, 1985, ed. L. G. Schutz, H. van Olphen and F. A. Mumpton, The Clay Minerals Society, Bloomington, Indiana, 1987, p. 305. A. Schutz, W. E. E. Stone, G. Poncelet and J. J. Fripiat, Clays Clay Miner., 1987,35, 251. M. L. Occelli, Ffth Znternational Symposium on the Scientific Bases for the Preparation of Heterogeneous Catalysts, ed. G. Poncelet, P. A. Jacobs, P. Grange and B. Delmon, Elsevier, Amsterdan, 1991, p. 287 Hudson, NATO AS1 Series, vol. 400, Kluwer, Dordrecht, 1993, 13 p. 19. M. Kawano and K. Tomita, Clays Clay Miner., 1991,39, 174. Paper 3/04608G; Received 2nd August, 1993