J . Phys. Chem. 1989, 93, 7053-7054
7053
Evidence for Cluster Sites on Catalytlc Alumlna Arunabha Datta Alchemie Research Centre, P.O. Box 155, Thane-Belapur Road, Thane 400 601, India (Received: May 31, 1989)
The nature of surface sites created on catalytic alumina through dehydroxylation has been investigated by TGA, FTIR, and XPS studies. There appears to be evidence for the formation at high temperatures of "cluster" sites which involve considerable redistribution of charge on the aluminum and oxygen atoms.
Introduction Alumina has been widely used as a solid acid catalyst for isomerization, dehydration, cracking, and other reactions.' However, the nature of the active sites on the alumina catalyst is still not well characterized. Current models for the alumina surface propose that there are a t least five different types of "structural" or "isolated" hydroxyl groups present on the surface with each of these hydroxyls having a different configuration.2 It is also speculated that the active sites on catalytic alumina are created by the dehydroxylation of these hydroxyls either by combining amongst themselves or through self-condensation. The actual occurrence and relative concentration of each of these hydroxyls on a particular alumina surface will depend, however, upon the relative contributions of the different crystal faces (1 lo), (IOO), or (1 11) to the total surface layer. This in turn is dependent upon the detailed preparation conditions of each alumina sample. It is evident therefore that alumina samples prepared under different conditions will have different structural hydroxyls to start with and should therefore give rise to different types of surface sites on dehydroxylation at various temperatures. in fact, our FTIR of gas adsorption on alumina do indeed indicate the presence of different active sites on an alumina sample activated a t 700 OC as compared to one activated at 400 O C . The present work therefore is aimed at investigating, through FTIR, TGA, and XPS studies, the nature of the different surface sites created on an alumina sample by dehydroxylation of various hydroxyls at different temperatures. Experimental Section The alumina sample was the Kaiser S-201Claw catalyst. The TGA was done on a Perkin Elmer 7 series thermal analysis system. The experiment was done in a N, atmosphere using a heating rate of 20 OC/min but maintaining the temperature constant for 30 min each time at 120,200,300,400,500,600, and 700 OC. The XPS spectra were recorded on a VG ESCA-3 I1 spectrometer. After the samples were heated to the desired temperature they were cooled to room temperature before the spectra were recorded. The C 1s line (285.0 eV) recorded at each temperature was taken as the internal standard. The FTIR spectra were recorded on a Nicolet Model 7199 spectrometer, after heating under vacuum Torr) in an in situ cell described earliere3 ( Results and Discussion TGA and FTIR. The TGA pattern (Figure 1) of the alumina sample shows a progressive loss of water up to 700 O C . The maximum loss of 9.8% at 120 O C is due to the removal of adsorbed water whereas subsequent weight loss is due to the removal of water through dehydroxylation. It is evident that the dehydroxylation takes place in stages since the time of 30 min at each temperature should be enough to remove all water through de(1) Tanabe, K. In Heterogeneous Catalysis; Catalysis by Novel Solid Strong Acids and Superacids; Shapiro, B.L., Ed.; Texas A & M University Press: College Station, TX, 1984; pp 71-94. (2) Knozinger, H.; Ratnasamy, P. Catal. Rev. Sei. Eng. 1978, 17, 31. (3) Datta, A.; Cavell, R. G.;Tower, R. W.; George,Z. M.J . Phys. Chem. 1985, 89, 443. (4) Datta, A.; Cavell, R. G. J . Phys. Chem. 1985, 89, 450, 454.
0022-3654/89/2093-7053$01.50/0
hydroxylation at that particular temperature. It can be seen from the FTIR spectra of the O H stretching region (Figure 2) that up to 300 O C there are closely spaced hydroxyls which are hydrogen bonded as evident from the strong absorption in the 37003400-cm-' region. Only on heating to 400 OC does one observe well-defined bands (at 3675 and 3750 cm-I) due to isolated hydroxyl groups. On heating to 500 O C and above, however, there is a pronounced and progressive decrease in the intensity of the band at 3680 cm-l, so much so that this band disappears almost completely at 700 O C and the 3750-cm-l band, which correspondingly increases in intensity, becomes the predominant one. It can be speculated therefore that, up to 300 OC, dehydroxylation leading to the formation of a coordinatively unsaturated (cus) aluminum (Lewis acid site) and a cus oxygen (Lewis base site) results from interaction between hydroxyls that are hydrogen bonded. On the other hand, at 400 OC, the hydroxyls involved in the dehydroxylation process are widely separated (not hydrogen bonded) and therefore give rise to strained cus aluminum and oxygen sites. In may be noted that alumina is known to develop catalytic activity only on activation at about 400 OC,indicating that it is only the strained cus sites that are catalytically active. On heating to 500 OC and above, however, since only the 3680cm-' band decreases in intensity it would appear that dehydroxylation at these temperatures results primarily through the selfcondensation of hydroxyls responsible for the 3680-cm-l band. It has in fact been postulated2 that this band is due to a hydroxyl attached to three other aluminum atoms in octahedral coordination. One could expect therefore that heating to 500 OC and beyond should produce OAls kind of clusters through the selfcondensation of the 3680-cm-l hydroxyls. XPS. The XPS spectra at different temperatures (Figure 3) indicate that on heating up to 400 OC there is not much change in the A1 2p BE (binding energy) which is in the region of 74.4-74.7 eV. On heating to 500 OC, however, the shift of 1.7 eV toward higher BE suggests a significant increase in the effective positive charge (and hence the Lewis acidity) on aluminum. At the same time, the 0 Is BE remains in the region of 531.5-531.7 eV upto 400 OC and shifts by 0.7 eV to lower BE at 500 OC, indicating increased negative charge (and higher Lewis basicity) on the oxygen. In fact an empirical relation between the binding energy and the effective charge on the oxygen atom has been suggested from Mulliken population data based on a b initio calculations and well-calibrated XPS data.5 With this relation, the effective charge on oxygen on the alumina surface changes from -0.57 at 400 O C to -0.73 at 500 OC. The significant 'shifts in the BE of A1 and 0 at 500 OC are, needless to say, very interesting and have not been reported before. Dehydroxylation itself should produce a small positive and negative shift in the BE of aluminum and oxygen, respectively, because of the transition from an AI-OH species to an A16+-0" species. In fact it would appear from the widths (at half-height) of the A1 2p peak (2.9 eV) and the 0 1s peak (3.6 eV) that they contain contributions from both hydroxyls as well as dehydroxylated species. However, it is significant that a pronounced shift in BE'S ( 5 ) Sundberg, P.; Larsson, R.; Folkesson,B. J. Electron Spectrosc. Relat. Phenom. 1988, 46, 19.
0 1989 American Chemical Society
7054
The Journal of Physical Chemistry, Vol. 93, No. 20, 1989
Letters
1
400
300
200
500
600
80
700
TemperotureI'Cl
75 70 BINDING ENERGY (BE.)
Figure 1. TGA pattern of alumina (the figures against the peaks in the first-derivative trace correspond to the percentage weight loss at that
531.7 531.0
particular temperature).
n 700'c
500'C 400' C
300' C
V W
540
535
530
BINDING ENERGY
525 (BE)
Figure 3. XPS of alumina at different temperatures.
I
,p
, \ ,
-1°C
4000
3800
3600
3400
Conclusion
WAVENUMBERS i cm-1)
Figure 2. FTIR spectra (0-Hstretching region) of alumina heated to
different temperatures. is observed only on heating to 500 OC, although dehydroxylation takes place even at lower temperatures. It is evident therefore that, on heating to 500 OC, new sites are created where there is considerable reorganization of charge on both the aluminum and oxygen atoms. SCF-MO studies6 done on OAl, clusters reveal that there is a significant change in the atomic charges of oxygen and aluminum on cluster formation. More recent quantum chemical studies of alumina' also indicate that the number of acidic aluminum atoms composing the acid site is very significant because of the cooperative enhancement of acidity by the other aluminum atoms coordinated to the oxygen atom. It is tempting to suggest therefore from the overall evidence that OAls like clusters are probably formed on the alumina surface on heating to 500 OC through the self-condensation of the 3680-cm-' hydroxyls. These sites can perhaps tentatively be described as follows: \ Al-/ -
\ / -
-AI
\ / \--/\ /
-Al--OH
/
\
I@-Ai-
/,-----
-77
\ /\
-4-
/ \
The fact that such sites are catalytically active is evident from our earlier s t ~ d i e s . ~ . ~
\d/
-AI
\a+/
-A-
\*+/ \d \ A+
-AI-0-AI-
w-\AI-d+/ \ / A,, / \ /\
( 6 ) Schwartz, M. E.; Quinn, C. M. Surf. Sci. 1981, 106, 258. (7) Kawakami, H.;Yashida, S. J . Chem. Soc.,Faraday Tram. 2 1986,82, 1385; 1985, 81, 1117, 1129.
The active sites of catalytic alumina have traditionally been classified as the hydroxyls (Brornsted site) and the cus aluminum (Lewis acid site) and oxygen (Lewis base site) ions produced on dehydroxylation. In the present case it is evident that hydroxyls such as the 3750- and 3680-an-' hydroxyls at 400 O C or the 3750and 3785-cm-' hydroxyls at 700 OC could act as the Brornsted sites. At the same time, strained cus aluminum and oxygen sites are also formed probably through the dehydroxylation of dissimilar but non-hydrogen-bonded hydroxyls. However, at temperatures of 500 OC and above, it would appear that dehydroxylation can only result from self-condensation of the 3680-cm-' hydroxyls giving rise to cluster sites. Although the exact description of such sites is still rather speculative, the extensive reorganization of charge during the formation of such sites is clearly evident. The extent of this charge redistribution, leading to changes of acidity and basicity of the aluminum and oxygen sites, is expected to change with the size of the "clusters" and hence with the nature of hydroxyl groups (Le., the preparation conditions) involved in the dehydroxylation process at high temperatures. The present study therefore raises the interesting possibility of tailoring the strength of the acidic and basic sites on catalytic alumina by controlling the type of surface hydroxyls through the preparation procedure. Such studies are in progress. Acknowledgment. Financial support for this work was provided by IC1 India Ltd. and is gratefully acknowledged.
