574
Langmuir 1995,11,574-577
Water Sorption on Mesoporous Aluminosilicate MCM-41 P. L. Llewellyn,*3t F. Schuth,#Y. Grillet,? F. Rouquero1,s J. Rouquero1,t and K. K. UngerS CTM du CNRS, 26 rue du 141 R.I.A., F-13331 Marseille cedex 3, France, Institut fur Anorg. Ch,emie & Analyt. Chemie, Johannes Gutenberg Universitat, 0-55099 Mainz, Germany, and Universitk de Provence, Place Victor Hugo, F-13331 Marseille cedex 3, France Received May 4, 1994. In Final Form: November 8, 1994@ Characterization of t h e interaction of water with t h e highly ordered mesoporous solid MCM-41 (pore diameter -2.5 nm) is undertaken with the aid of several techniques (adsorption gravimetry, X-ray diffraction, Fourier transform infrared spectroscopy, a n d controlled rate-evolved gas analysis). The relatively complex water-MCM-41 interactions are characterized by a type V isotherm indicating an initial repulsive character followed by a capillary condensation step of the adsorbate. This highlights both hydrophobic a n d hydrophilic properties of this potential model mesoporous adsorbent.
Introduction Since its relatively recent introduction in 1992,1-4the MCM-41 class of materials have created interest in many fields of chemistry. The ability t o synthesize material with a pore diameter between 2 and 20 nm within a n a r r o w size distribution c r e a t e s its potential applications for use as a physisorption standard5s6and as a h o s t s t r u c t u r e in occlusion chemistry’ and in c a t a l y ~ i s . ~ ~ ~ In all of the above cases, the interaction (of the structure and surface) with water is of interest. Although water adsorption is a relatively complicated process,1o its measurement allows examination of surface chemistry (as it i n t e r a c t s strongly with hydroxyl species) and of the solid texture (mesopore filling). Thus, o n e c a n predict both the stability of the structure in wet air as well as the interaction of the surface with o t h e r specific molecules. This study aims t o provide information both o n the water-surface interaction (adsorption gravimetry, controlled rate-evolved gas analysis and Fourier t r a n s f o r m infrared spectroscopy) and o n the s t r u c t u r a l stability of MCM-41 (X-ray diffraction).
Experimental Section A. Sample. MCM-41 comprises a family of mesoporous aluminosilicates which are claimed to be able to be synthesized with a pore diameter between 2 and 20 nm.1-4 The pores are of CTM du CNRS. Johannes Gutenberg Universitat. Universite de Provence. Abstract publishedinAdvanceACSAbstracts,January 1,1995. (1)Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.;Vartuli, J. C. U.S. Patent No. 5,098,689, 1992. (2) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.;Vartuli, J. C. US. Patent No. 5.102.643. 1992. (3) Kresge; C . T.; Leonowicz, M. E.; Roth, W. J.;Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710. (4) Beck, J. S.; Vartuli, J. C.; Roth, W. J.;Leonowicz, M. E.; Kresge, C. T.; Schmitt, K. D.; Chu, C. T-W.; Olson, D. H.; Sheppard, E. W.; McCullen, S. B.; Higgins, J. B.; Schlenker,J. L. J . Am. Chem. SOC.1992, 114, 10834. (5) Franke, 0.;Schulz-Ekloff,G.; Rathousky, J.;Starek J.;Zukal, A. J. Chem. Soc., Chem. Commun. 1993,724. (6) Branton, P. J.; Hall, P. G.; Sing, K. S. W. J.Chem. SOC.,Chem. Commun. 1993,1257. (7) Llewellyn, P. L.; Ciesla, U.; Decher, H.; Stadler, R.; Schuth, F.; Unger, K. K. Zeolites and related Microporous Materials: State of the Art 1994; Weitkamp, J . , Karge, H. G., Pfeifer, H., Holderich, W., Eds.; Elsevier: Amsterdam, 1994; p 2013. ( 8 )Le, Q.N.; Thomson, R. T.; Yokomizo, G. H. U.S. Patent No. 5,134,242, 1992. (9) Bhore, N. A.; Le, Q.H.; Yokomizo, G. H. US. Patent No. 5,134,243, 1992. (10) Gregg, S. J.;Sing, K. S. W.Adsorption Surface Area & Porosity, 2nd ed.; Academic Press: London, 1982. 0 8
@
~~
cylindrical nature aligned in a hexagonal, honeycomb-like structure without intersections. The pore wall thickness is about 0.6 nm. The MCM-41 sample was obtained using a synthesis procedure based on that presented in the patents1r2 and in ref 4. An aqueous solution was made containing pyrogenic silica, waterglass, aluminum oxide, tetramethylammonium bromide, and finally dodecyltrimethylammonium bromide as the micelle builder. This solution was placed into an autoclave which was placed into an oven at a crystallization temperature of 393 K for 72 h. The resultant mixture was then filtered and washed with water. The solid material obtained was calcined to eliminate the residual water and organic material. The synthesized sample was heated from room temperature by 1 Kemin-l to a final temperature of 823 K which was maintained for 4 h. The final solid has a silica to aluminium ratio of greater than 1000. The adsorption isotherm of argon at 77 K(for which no specific interaction is expected) showed a reversible type lV isotherm leading to a BET surface area of 700 m2g-’ (assuming the area of a n adsorbed argon molecule of 0.138 nm2). A pore diameter of 2.5 nm is calculated taking the position of the first peak in the X-ray diffraction pattern (cf. Figure 6) assuming the hexagonal structure (Le. {(2/43) x d-spacing} - pore wall thickness). B. Gravimetric Adsorption. The gravimetric adsorption isotherms were obtained at a temperature of 297 K using an apparatus built in-house (CTM, Marseille1’ ). The sample was outgassed at 423 K for 16 h before each experiment. Water used in these experiments was doubly distilled to give a purity, measured using electrical conductivity, of greater than 5 MQ. The water was introduced to the sample in a slow and continuous manner chosen to provide quasi-equilibrium conditions; a continuous recording of the isotherm is thus obtained. The adsorption branch of the isotherm is completed in 1 week. As a test of the adsorbate-adsorbent equilibrium, at various times during the experiment, the water introduction was stopped and both the pressure and weight signals were monitored to ensure that sorbate-sorbent equilibrium was preserved before restarting the water introduction. C. Controlled Rate-Evolved Gas Analysis (CR-EGA). CR-EGA12 is in some respects the “reverse” or the ‘image” of conventional evolved gas analysis since, instead of controlling the heating rate, one controls the rate of gas evolution. In the present study, the rate of gas evolution is kept constant; it is measured from the pressure drop through a n orifice located in the pumping line and is controlled by appropriate heating of the sample. The preoutgassed (423 K for 16 h) samples were exposed to water vapor for 96 hat 291 K before thermal analysis was started a t 243 K. Once the sample was cooled to 243 K, it was connected to a vacuum line. Aresidual pressure below that selected for the experiment was obtained and the heating of the sample was automatically started. The heating continued such that the (11)Rouquerol, J.; Davy, L. Thermochim. Acta 1978,24,391. (12) Rouquerol, J. Thermochim. Acta 1989, 144, 209.
0743-746319512411-0574$09.00/0 0 1995 American Chemical Society
Langmuir, Vol. 11, No. 2, 1995 575
Water Sorption on Aluminosilicate 25
1
70 I
I
I
I 20
'
0.00
0.0
0.2
0.4
0.6
0.8
1 0.50
1.0
residual pressure (and hence the pressure drop through the orifice) was maintained constant at the predetermined value. This was maintained until a temperature of 823 K was reached. The results can be transformed to a curve of the degree of reaction (a)against temperature, and from this one can obtain information about the reaction process. D. Fourier Transform Infrared Spectroscopy (FTIR). The FTIR measurements were carried out on a Nicolet 5SXB instrument equipped with a microscope attachment (SpectraTech). The samples were placed in a special thermal cell under the microscope which is attached to a vacuum system and adsorptionapparatus. The sample (afterpreadsorptionofwater for 96 h at 296 K)was then placed under a vacuum and heated to various levels for 20 min before taking the spectra at these points (373,473,573,and 623 K).Readsorption ofwater is then carried out in situ at 296 K for 16 h. The FTIR spectra were obtained at various points during this procedure. E. X-ray Diffraction The powder XRD patterns were obtained on a Seifert3000l" instrument with an automatic divergence slit using copper Ka radiation (0.154 nm). The prevailing humidity within the apparatusis around 30%. Spectra starting with 28 values of around 0.5" were obtained.
o).
Results and Discussion The water adsorption isotherm shown in Figure 1is of type V in the IUPAC re~0mmendations.l~ This is indicative of relatively hydrophobiccharacter in the low-pressure region of the adsorption isotherm. The large step a t around relative pressure of 0.55 is probably due to a cooperative condensation mechanism. Due to the continuous method of adsorbate introduction allowing a virtually unlimited number of points, a slight curvature of the isotherm can be seen at low relative pressures (plp" < 0.1). The point at which this curvature stops, giving way to a linear uptake, can be estimated at around 1.67 mmo1.g-l. This corresponds to a total surface coverage of 105 m2*g-l (considering the surface area of a water molecule to be 0.105 nm2). This is much lower than the BET surface obtained from a n argon adsorption isotherm obtained on the same sample a t 77 K (712 mag-l). At the foot of the condensation step (0.5 plp"),the quantity of water adsorbed translates to a covered area of 320 mZ*g-'. This would suggest that the pore filling begins before completion of a compact monolayer. At the top of the step in the isotherm, a volume uptake of 0.450 cm3.g-l is obtained. Assuming the sample to have 2.5 nm diameter pores (see later XRD results) and the argon BET surface area to correspond to the internal surface of the pores, a theoretical pore volume of 0.445 cm3g-l is obtained. It would thus seem that condensation of the water occurs within the pores leading to total pore
1.50
2.00
2.50
na I mg.g-l
P I PO Figure 1. Water adsorption isotherm on MCM-41, obtained at 297.15 K.
1.00
Figure 2. Isosteric heats of adsorption calculated from the application of the Clausius-Clapeyron equation to two water adsorption curves on MCM-41 obtained at 287.35 and 297.15 K.
-
Y
a
L
a
c.
E a
; a
c.
0.0
0.2
0.4
0.8
reaction co-ordinate
0.8
1.o
(a)
Figure 3. CR-EGA curve obtained for MCM-41 pre-equilibrated with water.
filling. It was not possible, for all of the desorption, to obtain a n equilibrium within a reasonable time (less than 1week under our experimental conditions) which is why we do not present in the figure the significant hysteresis loop obtained. The quantity of water vapor retained by the sample after pumping to a pressure of 1.33 Pa for 16 h is 0.020 mgg-l a t the experimental temperature. The isosteric heats of adsorption (Figure 2) increase with uptake to a value of around 58 kJ*mol-l. A similar form of heat curve is observed with silica treated with a silane15and supports the idea that the MCM-41 surface is relatively hydrophobic. The maximum value of the isosteric heat is somewhat higher than the enthalpy of liquefaction (-44 kJ-mol-l). This would suggest that, initially, a hydrophobicwater-MCM-41 interaction occurs followed by a more important water-water interaction. The CR-EGA curve (Figure 3) shows two distinct regions of water loss (I and 11). As most of the adsorbate is lost when the sample is placed under vacuum, the remaining water corresponds to that directly bound to the surface. The first region (I)to a temperature of around 500 Kcorresponding to 0.012 mgg-l ofwater is due to the loss of physisorbed water from the surface. The second region (11)above 500 K and corresponding to 0.008 mgg-l of water is probably due, in analogy with the pyrogenic silica used in the synthesis,16 to the dehydroxylation of the surface. From X-ray diffraction studies it can be seen that structure breakdown occurs a t temperatures starting
(13)Sing,K.S.W.;Everett,D.H.;Haul,R.A.W.;Moscou,L.;Pierotti, R. A.; Rouquerol, J.; Siemieniewska, T.Pure Appl. Chem. 1985,57,603. (15)Sing, K. S. W.; Ramakrishna, V. R. Thermochimie; C.N.R.S. (14)Brunauer, 5. The Adsorption of Gases and Vapors; Princeton Ed.: Paris, 1972;p 435. University Press: Princeton, NJ, 1945.
Llewellyn et al.
576 Langmuir, Vol. 11, No. 2, 1995
4000
3500
3000
2500
2000
1500
wavenumbers / cm-’ Figure 4. FTIR spectrum of MCM-41 saturated with water. a t 923 K. The total weight loss from 500 to 1273 K is 0.012 mgg-l. Assuming that this weight loss between 500 and 1273 K is due t o total surface dehydroxylation, an estimation of the quantity of surface hydroxyl sites can be obtained from this water loss is close to 1.2OHnm-2. This is much less than that of a pyrogenic silica (3-5 OH.nm-2 l7 ) such as that used for the synthesis of MCM41. From the first region of water loss a fitting of the curve was carried out using a combination of the k r h e n i u s law and several laws describing different reaction mechanisms.18 It was found that the best fit came from a threedimensional diffusion law (the Jander equation, g ( a )= [ l - { 1 - a)1/312 l9 which suggests a free diffusion of water through the relatively large pores (with respect to the water molecule) and from the external surface. The activation energy for this process obtained from Figure 3 was found to be 25 kJ*mol-l. The second region (11) of the CR-EGA (Figure 3) is fitted by a second three-dimensional diffusion law (the Ginstling-Brounshtein equation, g ( a ) = [{ 1 - 2 d 3 ) - (1- a)u3120 1which leads to an activation energy of 62 kJ-mol-’. From these results it would seem that a small amount of water is initially adsorbed on specific sites present on the sample surface (probably -OH sites). However with the CR-EGA results, it would seem that half of this initially adsorbed water is chemisorbed with a desorption activation energy of 62 kJ*mol-l. The other half is physisorbed with the surprisingly low desorption activation energy of 25 kJ-mol-’. This low value is nevertheless consistent with the initial increase of the isosteric heats a t moderately low uptake. The initial values of the isosteric heats are not able to be obtainedvia the ClausiusClapeyron equation due to the large error accumulated in the measure of the low equilibrium pressures. Finally we suggest that cluster formation occurs around the initial adsorption sites. The confined geometry of the MCM-41 mesopores may explain the fact that the isosteric heats obtained are well above the enthalpy of water liquefaction (44 kJ-mol-l). The FTIR measurements are shown in Figures 4 and 5. It can be seen that on saturation with water vapor under atmospheric conditions, a spectrum characterized by broad peak due to liquid water a t 3100 cm-l is obtained (Figure 4). On putting the sample under a vacuum of (16) Legrand, A. P.;Hommel, H.; Tuel, A.; Vidal, A,; Balard, H.; Papirer, E.; Levitz, P.; Czernichowski, M.; Erre, R.; Van Damme, H.; Gallas, J. P.; Hemidy, J. F.;Lavalley, J. C.; Barres, 0.; Bumeau, A.; Grillet, Y. Adu. Colloid Interface Sci. 1990,23,91. (17) Iler, R.K.The Chemistry ofSilica; J.Wiley & Sons: New York, 1979. (18) Sharp, J. H.; Brindley, G. W.; Narahari Achar, B. N. J. Am. Ceram. SOC.1966,49, 379. (19) Jander, W.2. Anorg. Allg. Chem. 1927,163,1. (20) Ginstling, A. M.;Brounshtein, B. I. 2.Prikl. Khim. 1950,23, 1327.
3800
3600
3400
3200
wavenumben I cm-1 Figure 5. FTIR spectrum of MCM-41 saturated with water and then outgassed under a vacuum of 1.33Pa and heated up to (a) 296 K, (b) 373 K, (c) 473 K, (d) 573 K, and (e) 673 K.
1.33 Pa, this large peak decomposes to two peaks of much lower intensity a t 3747 and 3542 cm-’, respectively (Figure 5a). These peaks are probably due to surface hydroxyls (marked “I” a t 3747 cm-’) and to physisorbed and chemisorbed surface water (marked “11”a t 3542 cm-’1. After the sample is heated to various temperatures under pressure of 1.33 Pa, it can be seen that a small decrease in intensity of the peak a t 3542 cm-l occurs up to a temperature of 473 K (Figure 5b,c) which becomes more noticeable a t 573 (Figure 5d) and 673 K (Figure 5e), respectively. The peak a t 3747 cm-’ does not change in intensity until 673 K where a small relative increase can be observed. These phenomena can be explained by the initial loss of the small amount of remaining physisorbed water up to a temperature of 473 K (corresponding to region I of the CR-EGA curve, Figure 3). This is then followed by a loss of some of the chemisorbed water a t 573 K, and a t 673 K, further loss of chemisorbed water occurs which may lead to the formation of siloxane bridges. The XRD results in Figure 6 show the progression from the initial calcined sample (spectrum a)to the same sample after water contact for 96 h with water vapor (spectrum b) and finally the sample after thermal treatment to 823 K (spectrum c). After calcination, the sample exhibits a pattern typical of MCM-41 samples.21 The low d-spacing of the first peak (2.86 nm) in spectrum A confirms the pore size of around 2.5 nm. The second and third peaks can also be clearly seen. After contact with water vapor, the second and third smaller peaks become merged into the first peak (d-spacing = 2.92 nm) which is probably due to interference with the water adsorbed as has been seen elsewhere for an aluminophosphate of similar pore structure.22 Finally, on thermal treatment to 823 K as (21) Feuston, B. P.;Higgins, J. B. J. Phys. Chem. 1994,98, 4459. (22) Coulomb, J.P.; Martin, C.; Grillet, Y.; Tosi-Pellenq, N. Zeolites and related Microporous Materials: State of the Art 1994; Weitkamp, J.,Karge, H. G., Pfeifer, H., Holderich,W., Eds.;Elsevier: Amsterdam, 1994; p 445.
Water Sorption on Aluminosilicate
.-cm>r c
Q)
c C
.-
Langmuir, Vol. 11, No.2, 1995 577
I:1 2
4
degree 29 Figure 6. XRD spectra of MCM-41: (a) aRer calcination to 623 K, (b) after equilibrium with water vapor for 1 week; (c) aRer a second thermal treatment of the previous sample t o 823
Conclusion The mesopores of MCMdl,a potential standard adsorption material, have a relatively hydrophobic character with a small amount of surface -OH sites. However, it would seem that water initially adsorbs on these sites with the formation of clusters a t these sites. At a certain pressure, capillary condensation occurs leading to a total filling of the pore volume and thus to a type-V isotherm. Under vacuum conditions, most of the physisorbed water is removed. Heating the sample under vacuum conditions reveals the initial loss of the last of the physisorbed water up to 500 K. Above 500 K loss of chemisorbed water and dehydroxylation occurs.
K.
the corresponding spectrum shows the re-emergence of the second and third peaks (d-spacing of the first peak = 2.85 nm).
Acknowledgment. P.L.L. thanks the EEC “Human Capital & Mobility Program” for financial support. LA940371R