97
J. Phys. Chem. 1982, 86, 97-102
Investigation of Structural Charge Transfer in Zeolites by Ultraviolet Spectroscopy E. D. Oarbowskit and C. Mirodator” Instnut de Recherctms sw le Cetal~se,69826 Vllkwbanne &lex,
France (Recehred: January 6, 1981; In Flnal Form: Awust 31, 1981)
Charge-transfer processes in various zeolites (faujasite, mordenite, ZSMB,erionite, and offretite) are evidenced by means of UV spectroscopy. UV bands at 240 and 320 nm are assigned to two different A1-0 units. The band at 240 nm, present whatever the zeolite and whatever the chemical or thermal treatment, is related to framework Al-0 units which are “inert”toward catalysis and easily removed by dealumination or dehydroxylation. The band at 320 nm, more stable toward dealumination and dehydroxylation, is specifically detected or significantly enhanced for catalytically active samples (methanol conversion,hydrocarbon cracking, and disproportionation). This band could be related to oxoaluminum structures inside the zeolite matrix.
Introduction Much work has already been devoted to determine the nature and location of active centers in zeolite catalysts which combine very high catalytic performance with rather well-defined structural characteristics. Despite general relationships between protonic density, acid strength, and carbonium ion catalysis,112the origin of proton lability, the specific role played by Lewis acidity, and more generally collective framework interactions which occur in zeolites between Si, Al, 0, protonic and cationic species are still controversial. Thus, Hopkinsa and Lunsford4 ascribe the very high strength of protonic acidity in zeolites to electronic inductive influence of Lewis centers. More recently, it has been suggested that, in zeolite frameworks considered as electrolytic media! “superacidity”would arise from interactions between oxoaluminum deposits in the channels, as widely described by Breck and Skeels in steamed or ultrastabilized faujasites and hydroxyl groups linked to the l a t t i ~ e . ~ Besides classical schemes proposed by several author^^^^ which involve trigonal silicon and explain Bronsted and Lewis acidity in zeolite, the specific role played by aluminum deficiencies in the lattice or deposits in cationic positions has attracted increasing attention, especially in highly siliceous zeolites as mordenite, ZSM, or silicalite. Despite recent studies comparing zeolites with very different Si/Al ratios and structures,lOJ1attempts at correlating basic zeolite properties as for instance aluminum electronic and crystallographicstates remain rather scarce. Collucia et a1.12 and Stone and Zecchinals used UV spectra to obtain some very detailed information about surface states of alkaline-earth oxides and magnesia. They assumed that active centers are associated with low coordinated surface ions. These results prompted us to investigate zeolite surfaces by UV reflectance spectroscopy in order to get a better understanding of zeolite structures, hence of zeolite active centers. Experimental Section Diffuse reflectance spectra were recorded on a CF4DR Optica Milano spectrometer, equipped with an integrating sphere covered with magnesia. The reference was also MgO. The spectra were recorded at room temperature in a quartz cell described e1~ewhere.l~ Materials. Table I lists the chemical composition of catalysts used in that work. X,A, Y, mordenite, and erionite zeolites were prepared from their commercial form. Samples of synthetic offretite and ZSMB were provided Laboratoire de Chimie Indwtrielle, Universite Claude Bernard Lyon I. 0022-3654/82/2~86-~097$0 7.2510
TABLE I: Catalyst Compositions
--
unit-cell composition catalysts mordenite Na-2 H-2 Y faujasite Na-Y H-Y dealuminated Y Na-X Na-A offretite K-Of H-Of erionite K-E H-E ZSM5 silicalite
Si/AI
5.0 5.0 2.3 2.3 8.6
1.2 1.0
Na 8
8 8
0.2 54
56
7
56
2.3
20
86
86
12
12
0.07 0.01
3.1
4.0
3.2 3.2
2.5 0.1
4.7
4.0
14 1420
Al
K
0.6
3.6 3.6
0.1
7.7 7.6 6.4
0.1
0.135
2.6
respectively by B a r r e P and Vedrine.I6 Ammoniated samples were prepared by NH4Cl exchange. In order to get protonated and Lewis forms, the solids were calcined under flowing oxygen and then desorbed at temperatures that are reported in the Results section. Additional treatments snch as steaming or dealumination were performed on some mordenite and faujasite samples according to procedures described elsewhere.”J8 (1) Hirschler, A. E. J. Catal. 1963, 2, 428. (2) Rabo, J. A.; Poutsma, M. L. Adv. Chem. Ser. 1971,102,201. (3) Hopkins, P. D. J. Catal. 1968, 12, 325. (4) Lunsford, J. H. J. Phys. Chem. 1968, 72, 4163. (5) Rabo, J. A. Stud. Surf. Sei. Catal. 1980,5, 341. (6) Breck, D. W.; Skeels, G. W. Roc. Int. Congr.Catal., 6th, 1976 1977, 2,645; ACS Symp. Ser. 1977, No. 40,271; R o c . Int. Conf. Zeolites, 5th 1980,335. (7) Mirodatos, C.; Barthomeuf, D. J. Chem. SOC.,Chem. Conmun. 1981, 39. (8) Uytterhoeven, J. B.; Christner, L. G.; Hall, W. K. J. Phys. Chem. 1966,69, 2117. (9) Ward, J. W. J. Catal. 1967,9, 225. (10) Barthomeuf, D. J. Chem. Soc., Chem. Commun. 1977,743. (11) Flanigen. E. M. Proc. Int. Conf.Zeolites. 5th 1980. 760. (12) Colu&ia,’ S.; Tench, A. J.; Segdl, R. L. J: Chem. Soc., Faraday Trans. 1 1979. 75. 1769. (13) Zecchina, A.; Stone, F. S. J. Chem. Soc., Faraday Trans. 1 1978, 74, 2278. (14) Praliaud, H.; Forissier, M. React. Kinet. Catal. Lett. 1978,8,461. (15) Barrer, R. M.; Harding, D. A. Sep. Sci. 1974,9, 195. (16) Derouane, E. G.; Nagy, J. B.; Dejaifve, P.; Van Hoof, J. H. C.; Spekma, B. P.; Naccache, C.; Vedrine, J. C. J. Catal. 1978,53,40. (17) Marcilly, C. French Pattents 75/33.601,76/31.620, and 77/01.265. (18) Mirodatos, C.; Ha, B. H.; Otsuka, K.; Barthomeuf, D. Proc. Int. Conf. Zeolites 5th 1980, 382. (19) Nakamoto, K. ‘Infrared Spectra of Inorganic and Coordination Compounds“; Wiley: New York, 1963, p 159.
0 1982 American Chemical Society
98
The Journal of Physical Chemlstty, Vol. 86, No. 1, 1982
Garbowski and Mlrodatos
ib8orbancda.u)
a
500
400
300
220
Figwe 2. UV spectra of NH,-Z samples: (a) initial s p e c " , (b) after treatment under vacuum at 450 "C,(c) 500 "C,(d) 550 "C,(e) 600 OC, (f) 650 "C. 5 00
400
300
2 20
Figure 1. UV spectra of Na-2 samples; (a) after heat treatment at 700 "C for 15 h, (b) after steaming at 400 "C for 1 h, (c) after heat treatment under vacuum at 700 "C for 15 h.
Catalytic Experiments. Most of our catalysta have been contacted with methanol a t room temperature and then progressively heated to 500 "C, UV spectra being recorded at each step. Evidence of coke formation, hence of catalytic reaction, is provided by the growth of absorption bands at ca. 300 "C, characteristic of highly dehydrogenated hydrocarbons, i.e., coke precursors. Direct evidence of catalytic activity has also been provided by n-octane conversion, performed at 450 "C under flowing hydrogen in a microcatalytic reactor connected to an automatized gas-chromatographic analysis system (Pnca,/PH,= 0.06, flow rate = 2.4 L/h, ma= 100 mg).Mordenite samples were also tested for toluene disproportionation.l8ps
Results Mordenite. Na-Z mordenite exhibits two strong, very stable absorption bands at 240 nm (5.15 eV) and 320 nm (3.85 eV) whether the sample is dehydrated at 700 "C overnight or not (Figure la). These bands will be later (20) Mirodatos, C.; Barthomeuf, D. J. Catal. 1979,57, 136. (21) Mataga, N.; Kubota, T. 'Molecular Interactions and Electronic Spectra"; Marcel Dekker: New York, 1970; Chapter 6. (22) Urbain, H., personnal communication. (23) Olsson, R. W.; Rollmann, L. D. Inorg. Chem. 1977,16, 651. (24) Ione, K. G.;Stepanov, V. G.; Maatik Hin, V.; Paukshtis, E. A. h o c . Znt. Conf. Zeolites, 5th 1980, 223. (25) Knozinger, H.; Rataanamy, P. Catal. Reu. Sci. Eng. 1978,17,31. (26) Mirodatos, C.; Barthomeuf, D. to be submitted for publication. (27) Gnep, N. S.; Martin de Armando, M. L.; Marcilly, C.; Guisnet, M. Stud. Surf. Sa. Catal. 1980, 6, 79.
referred to as transitions T1 and T2, respectively. No significant changes were observed when activated samples were contacted with water at room temperature (p N 20 torr); but steaming at 400 "C for 1 h caused small decreases of both bands (Figure lb). Further dehydration at 700 "C for a second time cause the two bands to increase slightly in intensity (Figure IC). Accordingly, vacuum treatments or heating under water pressure have almost no effect on cationated mordenite. In the case of ammoniated samples, NH,-Z, two bands still appear, although the overall spectrum changes significantly. As depicted in Figure 2, the T1 band is a little broadened in comparison with Na-Z (curve a). When this sample is heated and desorbed successively from 450 to 700 "C, drastic changes occur. Transition T1 decreases in intensity as long as the desorption temperature increasea to 550 "C, while the transition T2 remains nearly stable. At higher temperature, Le., 600 "C, T2 starts to decrease in the same way as T1. These results clearly show that, for decationted samples, T2 is much more stable than T1 with respect to dehydration. This statement prompted us to check how surface OH groups behaved during the same treatment. Near-infrared spectra were record4 on the same sample from 100oO to 4350 cm-l. Only two bands appear, at 7120 and 4690 cm-' (Figure 3). Both the intensities and the frequencies of these IR bands suggest that they are an overtone and a combination of OH vibrations, 2 X 3600 and 1100 + 3600 cm-', respe~tive1y.l~As shown in Figure 4, the intensities of the two IR bands (curves c and d) decrease at the same time and to the same extent; that is, in a way different from the T1 and T2 bands (curves a and b). Therefore, no direct correlation was found between UV transitions and IR OH bands.
The Journal of Physical Chemistry, Vol. 86, No. 1, 1982 09
Structural Charge Transfer in Zeolites
Absorbance ( a & )
Absorbance( a.u )
Flgwe 3. Near-IR spectra of NH4-Z samples recorded after treatment under vacuum at (a) 400, (b) 450, (c) 500, (d) 550, (e) 600, and (f) 650
I
1
1
1
500
I
I
I
l
I
I
I
I
I
I
221
300
400
Figure 5. (a) UV spectra of NH4-Z: (b) after steaming at 500 OC for
15 h, (c) after HCI deaiuminatlon.
I
Absorbance (a.u)
I
500
600
Figure 4. Intensity of the (a) 240- and (b) 320-nm UV bands and of the (c) 7120- and (d) 4690-cm-' I R bands as a function of heattreatment temperature for NH4-Z samples.
In order to check whether there were any water effects on Lewis mordenite, a sample that had been dehydrated at 700 "C was contacted with H 2 0 at room temperature. As for Na-Z, the intensity decreased only slightly even if the watered sample was heated for 1h at 400 "C. In no case could the original spectrum be restored. The effect of mordenite dealumination was studied on two samples. One was chemically dealuminated by HC1, i.e., with an actual loss of aluminic material, the other was partially dealuminated by steaming, according to ref 18, i.e., with a partial transfer of lattice aluminum into channel deposits. UV spectra reported in Figure 5 for nondealuminated and dealuminated samples indicate that (i) chemical dealumination significantly decreases the two transitions to nearly the same degree, (Figure 5c) and (ii) steaming specifically affeds transition T1 in contrast with T2, whose intensity is even slightly reinforced after the treatment (Figure 5b). ZSM5 and Silicalite. These zeolites have very high Si/A1 ratios, up to ca. 1000. They were activated at 400, 550, or 700 OC. For this type of structure, the same pair
1
50 0
1
1
1
1
1
1
400
1
1
I
l
l
I
220
300
Flgure 6. UV spectra of Y zeolite samples: (a) lnitiil Na-Y sample, (b) protonated form H-Y, (c) dehydroxylated form after treatment under vacuum at 600 OC for 15 h.
TABLE 11: Values of the T1 Band Intensity As a Function of Si/M Ratio catalysts NH,-Z H-ZSM5 Si/A1
T1 band intensity, au
5 2.4
14 0.20
24 0.08
27 0.07
of transitions is observed but with lowered intensities. Table I1 clearly shows that the transition intensityespecially for the 320-nm band-parallels the aluminum content. For silicalite, only a small absorption without a maximum is observed. Faujasite. Figure 6 reports the main trend in UV absorption for faujasite-type zeolites. For Na-Y material
100
The Journal of Physical Chemistry, Vol. 86, No. 1, 1982
Garbowski and Mirodatos
Absorbance ( a u )
14brorb.n~. (0.u)
/
a
C
I -
Flgure 7. UV spectra of offretke (a, b) and erionite (c, d) samples:
(a, c) cationated form (exchanged with K'), (b, d) protonated form.
TABLE 111: Catalytic Activity (au) in n-Octane Conversion for T1 and T2 Y faujasite
mordenite
offretite
erionite
Na-Y H-Y Na-Z H-Z K-Of H-Of K-E H-E convrn, %
T1 T2 a E
I
ea
4
6
12
2
0.60 0.47 2.40 1.70 0.25 0.02 0.13 0.30 0.40 less than 1%.
22
1
10
0.22 0.10 0.09 0.07
0.04
(curve a), only a rather sharp maximum at 245 nm (Tl) is observed. In the case of decationated samples H-Y (curve b), a second band at 320 nm (T2) appears whereas the first one is decreased. When the ammoniated zeolite is thoroughly dehydroxylated at 600 " C for 15 h, i.e., in its Lewis form as checked by IFt spectroscopy and pyridine adsorption, only a broad maximum is observed at 270 nm. For dealuminated H-Y sample, only very small (curve c) absorption occurs in that frequency range. Offretite and Erionite. UV spectra recorded on variously exchanged potassium offretite and erionite are reported in Figure 7. Highly cationated samples display only transition T1 at 245 nm. When potassium ions are removed and exchanged by protons, a rather broad shoulder appears at 320 nm, which we assign to the transition T2. For these exchanged samples, the influence of temperature and band stability is the same as for the other zeolites. Alumina and Silica. Only one weak and very broad band at around 260 nm is detected. It is stable with respect to calcination temperature. No absorption occurs on Aerosil-type silica. Catalytic Results. Table I11 compares the catalytic activities and the intensities of transitions T1 and T2 for catalysts used in this work. Both parameters were determined for identically pretreated samples. Figure 8 reports changes in UV spectra which occur when a ZSM5 sample is contacted with methanol at increasing temperatures. These data indicate the following: (i) The catalytic activity of samples which display specifically the transition T2 is significant. Thus,highly cationated offretite, erionite, and faujasite on which transition T1 is essentially detected are not coked when contacted with methanol and convert n-octane only poorly. In contrast, the decationated form of those catalysts and to a lesser extent the sodium mordenite sample which display also the transition T2 are coked when contacted with methanol at 300 "C and cat-
Flgure 8. Changes in UV spectra of ZSM5 samples upon methanol adsorption and conversion: (a) initial material, (b) after methanol adsorption at room temperature, (c) heated at 150 OC. (d) heated at 200 OC, (e) heated at 250 "C.
alyze n-octane cracking. Similar results are observed for other acidic reactions such as the isooctane conversionexcept for erionite catalysts due to a shape-selectivity effect=-and and toluene disporportionation.18 (ii) Before coke bands appear in UV spectra of active catalyst contacted with methanol, i.e., from room temperature to ca. 150 "C, transition T2 is selectively lowered and then disappears while the T1 one remains intact, as depicted in Figure 8. Accordingly, the occurrence of T2 transition seems to correlate with the catalytic ability to convert methanol and more generally to favor carbonium ion reactions. On the other hand, transition T1 seems to be unrelated to catalytic activity.
Discussion Origin of the Bands. The bands at 240 and 320 nm which are observed in the present study appear in the UV range which is related to charge-transfer processes.21 Species which can participate in such transfers (at least that corresponding to the transition T1) exist in all of the zeolites, whatever the structure, and can be eliminated by chemical dealumination. They may be impurities, special entities embedded in the zeolite matrix, or intrinsic parts of the zeolite framework. All zeolites contain impurities, principally iron which substitutes for A1 or Si as FeOd tetrahedra. Recent Mossbauer experiments performed on our mordenite samples have shown that most iron impurities are in the +2 oxidation state and some in the +3 one. Moreover, magnetic measurements on the same samples ruled out any possibility of metallic iron embedded in the zeolite matrix. Inferentially, iron impurities are in zeolites as in a mixed oxide. Iron(I1) (3ds is tetrahedral coordination) should exhibit an absorption band in the near-infrared but not in the visible or UV range. The reflectance spectrum of
The Journal of Physical Chemistty, Vol. 86, No. 1, 1982 101
Structural Charge Transfer in Zeolites
TABLE IV: Values of the T1 band Intensitv As a Function of Iron Content catalysts Fe content, wt % T1 band intensity, au
Na-Z 0.14 2.4
offretite
commercial Na-Y
“pure” Na-Y
erionite
Na-X
ZSM5
0.13 0.25
0.12 0.60
0.05 0.45
0.03 0.1
0.02 0.1
ca. 0.01 0.2
Fe203,which contains both octahedral and tetrahedral Fe3+ ions, shows strong bands in the visible and near-UV. Moreover, as shown in Table IV,the iron content does not correlate at all with the intensities of the two UV bands. For instance, offretite contains as much iron as mordenite, whereas the UV absorption behaves in a very different way. A sample of “pure” Y zeolite containing less than 500 ppm of iron has almost the same spectrum as the usual commercial material with ca. 1200 ppm of iron. Thus, iron impurities cannot explain the observed UV bands. Impurities also might be introduced in the crystallization or cation-exchange processes. Nitrate salts are often used for these purposes, and some NO, ions may remain inside the zeolite framework even after thorough washing and heating treatments. The spectrum of NO< ions typically displays a strong and sharp band at 300 nm and a very large absorption band below 250 nm, which can hardly be confused with the Tl-T2 set observed in that study. Furthermore, NO, ion cannot resist high-temperature and vacuum treatments; it would have been degraded under our experimental conditions. Thus, NO< impurities can be rejected as the sources of these bands. As shown for the mordenite zeolites, no direct correlation was found between the UV transitions and OH groups. Moreover, this set of UV bands occurs in carefully dehydrated Na-Y zeolite whose IR spectrum in the hydroxyl region remains perfectly flat. Thus, the observed UV phenomena cannot be directly related to surface hydroxyl groups. In a similar way, the occurrence of isolated aluminum or silicon atoms inside the framework cannot explain the spectrum. In the case of AlO, atomic absorption predicts bands at 237 (medium), 310 (strong), and 395 nm (strong); the latter is never observed. For Sio,bands at 222 and 252 nm should be observed.22Furthermore, dispersed Si or A1 atoms are very unlikely to be stabilized in a highly oxidizing medium like zeolite. Since the T1 and T2 bands are detected in a charge-transfer region only for aluminated material, we assume that such an absorption process originates from a transfer occurring between aluminum atoms and their surroundings, i.e., oxygen atoms in the case of zeolites. From previous UV studies on alkaline-earth oxides,12J3 it has been inferred that some surface cations are in a coordination state low enough to allow electron transfer from oxygen atoms to cations under appropriate energetic radiation. A similar process might occur in zeolites. Aluminum atoms at specific locations (e.g., edges, corners, surfaces, or clusters) which distort their coordinative oxygen surroundings could undergo a kind of internal redox process which can be tentatively written as A13+-02-
& A12f-O-
If the observed UV phenomena are rationalized in terms of electron transfer in such A1-0 species, the frequencies and intensities of the T1 and T2 transitions may be related to various parameters such as aluminum and oxygen charge density, A1-0 distance, crystallographic position of Al atoms in the framework, and origin and nature of the surroundings (proton, cation). Experimental results have clearly shown that the two transitions T1 and T2 behaved in a very different way according to cation exchange, heat treatment, or dealumination. It is then inferred that they correspond re-
4 1 SIIAI
1
2
3
4
I
5
Flguro Q. Intensity of the T1 band for (1) A, (2) X, (3) Y, (4) erionite, (5) offretlte, and (6) mordenite as a function of the SMAi content.
spectively to two distinguishable Al-0 species. Examining the specificity of each band (energy, frequency) should allow one to determine the nature and the function of the corresponding species. Nature and Location of AI-0 Units. The difference of energy between the two transitions is ca. 1.3 eV. This implies that the charge distributions in the corresponding A1-0 units are not identical. One can reasonably expect that the more the Al-O unit is polarized, the more favored charge transfer from oxygen to aluminum atoms will be. Thus, the energetically favored T2 transition (320 nm) is assigned to Al-O units characterized by a highly charged oxygen atom and a highly electron-deficient aluminum atom by comparison with the A1-0 units of the T1 transition (240 nm). Different bond lengths can also be expected for the different Al-0 units since, in an electrostatic model, the T-0 (T = Si or Al) distances in zeolites are related to the charge imbalance of the oxygen atoms.2s Transition Tl. The band intensities and frequencies vary for the different zeolite samples. For a given zeolite (such as ZSM5 in Table 11) the UV band corresponding to the transition T1 decreases in intensity upon dealumination. By analogy with the extensively studied variations of structural IR bands upon dealumination,29s30 this trend can be easily understood since lowering the A1 concentration in the lattice is expected to reduce the probability of A1-0 charge transfer. The same observation (significant and selective decrease of the T1 band intensity) is noted during dehydroxylation of the protonated samples (Figure 2). In this case, the concentration of A1-0 units allowing charge transfer T1 would be lowered by oxygen abstraction from the lattice, according to the usual scheme of zeolite dehydroxylation. For cationated samples, i.e., in the absence of surface OH groups, all dehydroxylation or steaming treatments leave the zeolite nearly unchanged and thereby no significant changes in the A1-0 transfer intensity are expected, as is actually observed. (28)Smith, J. V. ACS monogr. 1976, No. 171, 1. (29)Pichat, P.;Beaumont, R.; Barthomeuf, D. J. Chem. Soc., Faraday Trans. 1 , 1974, 70, 1402. (30)Flanigen, E.M.ACS Monogr. 1976, No. 171, 80.
102
The Journal of Physlcal Chemistty, Vol. 86,No. 1, 1982
TABLE V: Values of the T1 Band Frequency for Variouslv Treated Mordenites catalysts
T1 band
Na-Z NH,-Z steamed Z 240
245
265
dealuminated Z (HCI) 270
freq, n m
These inferences about the T1 band intensity and its variation upon dealumination and dehydroxylation derive from the implicit assumption that the related A-0 species belong to the zeolite lattice, i.e., to the structural tetrahedra of the framework. In Figure 9, the T1 band intensity is plotted as a function of the Si/Al ratio for various zeolites (the data represented in this figure are from cationic zeolites, which minimize any effect of dehydroxylation or mild dealumination induced by pretreatments, as shown previously). Firstly, note that there is a discontinuity between the curves drawn through the points from the cubic zeolites and from the orthorhombic zeolites. This supports the previous contention that the T1 transition is correlated with the different type of structural Al-0 units. A second point is that, in a given crystallographic system, the T1 band surprisingly increases in intensity when the A1 density decreases. This would indicate that the extent of charge transfer occuring in an Al-0 unit is reinforced when the density of other A1 species is decreased (a kind of long-distance effect of the structural surrounding atoms). Summing up, the T1 band intensity tends to (i) increase as a function of the aluminum concentration, hence of Al-0 unit concentration in a given zeolite, and (ii) decrease as a function of the aluminum concentration when comparing zeolites in a given crystallographic system. Table V shows that the frequency of the transition T1 for mordenite samples significantly increases when the material is progressively dealuminated (recall that a steaming treatment produces a milder dealumination compared with the chemical treatment by HCll'). Such a change implies that the charge transfers which occur on the remaining T1 units are energetically more favored. One possible explanation is that the observed T1 band is the sum of several overlapping bands, with slightly different frequency, as a function of the precise configuration of the related structural aluminum atoms. This is plausible because all zeolite frameworks display several distinct types of aluminum from a crystallographic point of view. For instance, in mordenite, four types of A1 locations related to various Al-0 distances are postulated.23 Accordingly, the observed frequency shift upon dealumination comes from a selective elimination of the less-polarized A1-0 units, the average of the charge transfer of the remaining A1-0 units being displaced toward lower energy (higher frequency). This is quite consistent with the conclusions of Olsson and R01lmann~~ that, in mordenite, framework aluminum atoms are selectivity extracted from the lattice according to their crystallographic location. Alternatively, the increase in frequency could be due to a global alteration of the electrostatic crystal field inside the zeolite induced by a modification of the Si/A1 ratio. Transition T2. The transition T2 has been shown to occur specifically for catalysts which are active in acidic reactions. The question arises as to whether the A1-0 species involved in this transition are framework units like
Garbowski and Mirodatos
the T1 ones. In a previous study on mordenite zeolites,l8 it has been assumed that, during steaming treatment, A1 atoms are removed from the lattice and form aluminum hydroxy compounds embedded in the zeolite matrix in agreement with Kuhl's conclusions.31 At the same time, a type of superacidity has been evidenced by NH3 thermodesorption measurements.' In that work, using the same mordenite catalysts, the transition T2 is shown to be favored by the steaming treatment while the concentration of T1 structural units is decreased. In numerous papers, Breck and co-workers6have proposed that extra lattice aluminum in the form of cationic oxo or hydroxy complexes could be the origin of Lewis and Bronsted acidity involved in catalysis. Ione et al.24have assumed hexacoordinated aluminum to be responsible for catalytic activity. Jacobs and Beyer have also recently postulated the existence of (A10)+ units which were attributed to true Lewis sites in faujasite-type zeolites.32 The charge transfer that is observed at 320 nm could be explained on the same basis and related to extralattice (A10)+ cations created from partial dealumination occurring during cationic exchange, heat treatment, or catalytic reaction. Presumably hexacoordinated, if we refer to ref 31, the aluminum atoms of these species are likely to be highly electron deficient, which could account for the specific UV frequency as experimentally observed. Theoretical calculations in regard to the electronic state of that type of aluminum would be worthwhile.
Conclusion Despite difficulties in correlating the two UV absorption bands to Al content, it seems very likely that the observed phenomena are related to charge transfer between A1 and 0 atoms. Two different types of A1-0 units would give respectively the two UV transitions: (i) One transition refers to Al-0 units belonging to the zeolite framework and is sensitive to dehydroxylation and dealumination,however mild, like the one induced by steaming treatments These species do not seem to be directly concerned with catalysis. Then, the acidity related to these units is likely to be weak, in agreement with the idea of weakly polarized units. (ii) The other transition would be related to extralattice structures, much more stable toward dealumination or dehydroxylation. Such species, highly polar and evidently strongly polarizing, are specifically detected in active catalysts and thereby can possibly be involved in the catalytic process. These units could act either directly, as suggested in ref 32, or via an inductive effect through the framework, initiated by their electrophilic character, for instance, increasing the acid strength of the usual Bronsted and Lewis structural acid sites. The weak intensity of the transition band T2 suggests that these species are in very low concentration in the zeolitic material. This statement agrees well with recent ideas%which assume that catalytic activity is due to a very small number of highly active sites. Acknowledgment. We thank Drs. D. Barthomeuf, M. V. Mathieu, and M. Primet for helpful discussions and comments. (31) Kuhl, G. H. ACS Symp. Ser. 1977, No. 40, 96; Proc. Int. Conf. Zeolites, 3rd 1973, 227. (32) Jacobs, P. A.; Beyer, H.K. J. Phys. Chem. 1979,83, 1174.
