J. Phys. Chem. 1983,.87, 5172-5176
5172
from the data analysis, along with values of kqo for quenching of the various probes in ethanol-water mixtures2OS2land for quenching of the fluorescence of sodium 4-(l-pyrenyl)butyrate (Na+PBA-) a t high ionic strength. In view of typical estimates of P (0.37 M-1)639band Fa, (0.33lcfor a probe solubilized predominantly in the region of the micelle-water interface) the similarity between the magnitudes of Eq and k,O for quenching by Br-, I-, and (when the more efficient quenching of Na+PBA- is taken into account) IAc- points to little or no intrinsic effect (F, ca. 1)of the micellar environment on the quenching efficiency.
Conclusions The data presented here demonstrate that, in the limit of fast counterion redistribution, the quenching of the fluorescence of micelle-solubilized aromatic hydrocarbons is a direct function of the local concentration of counterionic quencher(s) a t the micelle surface. For all the systems investigated here, the observed quenching behavior can be reproduced over a wide range of detergent and added common and foreign salt concentrations by using a simple pseudophase ion-exchange formalism which assumes constant ion-exchange selectivity _coefficients and constant intrinsic quenching efficiencies (kqvalues). From the agreement between model and experiment, one can conclude that, in these systems, the average solubilization site(s) of the probe and the average relative quencherprobe geometry in the micellar pseudophase are insensitive to variations in the overall counterionic content, both a t the micellar surface and in the intermicellar aqueous phase.
On the other hand, the fact that the observed quenching can be rationalzed by using a model which treats the micelles (mathematically) as a separate pseudophase (as opposed to an ensemble of individual micellar entities) necessarily implies that the quenching data in these systems are inherently uninformative as regards the details of the intramicellar quenching process (i.e., for distinguishing between quenching a t an unperturbed micellewater interface vs. probe-induced water channels, vide Introduction). Acknowledgment. F.H.Q. is a senior research fellow of the Conselho Nacional de Desenvolvimento Cientifico e Tecnolbgico (CNPq), Brasilia, supported in part by the Departmento de Bioquimica with funds from the Financiadora de Estudos e Projetos (FINEP). N.B. and L.M. acknowledge partial leaves of absence from the Instituto de Biociencias, Letras e Ciencias Exatas, UNESP, Ssio Jose do Rio Preto, SP, during the course of graduate studies. We thank Herga Indiistrias Quimicas, Rio de Janeiro, for generous donations of CTAC1. This collaboration was stimulated by travel funds from PNUD-UNESCO (RLA 024); the research in SBo Paulo was supported by the Funda@io de Amparo ii Pesquisa do Estado de Ssio Paulo (FAPESP 79/0272 to F.H.Q.). Registry No. NaNO,, 7631-99-4; NaBr, 7647-15-6; NaC1, 7647-14-5;S202-, 14383-50-7;Na2S203,7772-98-7;Br-, 24959-67-9; I-, 20461-54-5; IAc-, 152-34-1; Na+PBA-, 63442-80-8; CTAC1, 112-02-7; CTAB, 57-09-0;CTAS, 67355-36-6; CTAT, 82209-37-8; ethanol, 64-17-5;biphenyl, 92-52-4; naphthalene, 91-20-3; pyrene, 129-00-0; perylene, 198-55-0;benzo[ghi]perylene, 191-24-2.
X-ray Absorption Fine Structure Investigation of V,O,-TiO, Support
Catalysts. 1. The Titania
R. Kozlowskl,+ R. F. Pettifer," Department of Physics, University of Warwick, Coventry CV4 7AL, England
and J.
M. Thomas
Department of Physical Chemistry, University of Cambridge, Cambridge CB2 IEP, England (Received: January 26, 1983)
X-ray absorption fine structure (EXAFS) measurements have been performed on crystalline TiO, in the anatase form and on highly dispersed TiOz prepared by hydrolysis of titanium butoxide. The dispersed material had a mean particle radius R, = 35 A in which 30% of the titania octahedra lie on the surface. EXAFS analysis shows that the dispersed material is in the form of anatase with coordination distances identical with those of the crystalline material, aR < 0.005 A, for all coordinations involving the coupling of three octahedra. The variance of shell radii in the dispersed material shows a progressive increase with respect to the crystal as a function of the mean shell radius. Correlation to within 0.1 A of atomic position is lost beyond distances involving greater than three-octahedra coupling. The results are explained on thehasis of increased compliance of the surface over that of the bulk. No major surface reconstruction is observed.
I. Introduction Vanadium oxide based catalysts have long been emplayed in the selective oxidation of hydrocarbons. particular, the studies have concentratedon their use in
the industrially important oxidation of o-xylene to phthalic anhydride, the precursor of anthraquinone, a n d several useful esters. Wainwright and Foster1 reviewed the recent literature concerned with this reaction, placing emphasis on the influence of supports and promoters in catalyst
On leave from the Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Krakow, Poland.
(1) M. S. Wainwright and N. R. Foster, Catal. Reu., 19, 211 (1979).
0022-3654/83/2087-5 172$01.50/0 0 1983 American Chemical Society
EXAFS of V,05-Ti0,
performance. It can be concluded that by far the most successful support is anatase (Ti0.J. For the catalyst containing 84% Ti and 6% Vz05,promoted with KzS04 and Sb203, phthalic anhydride selectivity of up to 85% could be obtained.2 The performance of the catalyst can be further improved if close to monolayer amounts of Vz05 are laid down in coherent contact with the up port.^ The TiO2-V20, is a classic example of support enhancement of the active phase. In part 2 of our paper4 we will review the studies attempting to establish the structural basis of the succcess of this particular combination of oxides. Here, the technique of X-ray absorption fine structure (EXAFS) has been used to study the surface structure of the anatase. We present the results for anatase in two forms: highly crystalline and with very fine particle size. In the case of the very fine particles a large fraction of titanium sites are located on or close to the surface and thus we can infer from a bulk EXAFS experiment some of the properties of the surface. The results of an EXAFS study of crystalline anatase and rutile have been recently published by Vlaic et a1.j In the following section we describe the material production, techniques used, and the methods used to obtain the fine structure. Comments are given on some of the experimental aberrations which are present and which may affect the results. The data are analyzed in section I11 by using two techniques which are based on Fourier transform and curvefitting analysis. Finally, our results are discussed in terms of the possible mechanisms which may be responsible for the differences between surface and bulk anatase.
11. Experimental Section Materials. The crystalline anatase chosen was a commercial Fluka product (specific area 7 m2/g). The material exhibited a particle size >lo00 A as assessed by X-ray diffraction line broadening. Thus, the probability of excitation of a K-shell electron close to the surface is much less than that of the bulk. The anatase of the very fine particles was produced by hydrolysis of titanium(1V) butoxide (Alfa Chemicals) in water at 70 "C, followed by filtering the obtained precipitate, drying it at 70 "C, and calcining a t 400 "C for 5 h. The stability of the resulting ultrafine material was assessed by X-ray diffraction and differential microcalorimetry (DSC). The results clearly indicate that the material undergoes crystallization in the region 500-550 "C, which is well above our calcination temperature. The particle size of the material treated at 400 "C was assessed from the breadth of the strongest (101) diffraction line by using the Scherer equation and also by BET and electron microscopy. X-ray diffraction yielded a value of 60 f 20 A for the mean crystal diameter. Electron micrographs showed regular grains of an average diameter 60 A. BET results gave 180 m2/g, which corresponds to a particle diameter of 80 A. However, we expect the distribution of praticle size to be asymmetric, which could easily accommodate the differences in those estimates. In particular, the filtering technique employed probably ensures a ~
~~
(2) M. S. Wainwright and T. W. Hoffman, Can. J. Chen. Eng., 55,557 (1977). (3) G. C. Bond and K. Bruckman, "Faraday Discussion No. 72, Selectivity in Heterogeneous Catalysis", Royal Society of Chemistry, London, 1982. (4)R. Kozlowski, R. F. Pettifer, and J. M. Thomas, J. Phys. Chen., following article in this issue. ( 5 ) G. Vlaic, J. C. J. Bart, W. Cavigiolo, S. Mobilio, and G. Navarra, 2. Naturforsch. A, 36, 1192 (1981).
The Journal of Physical Chemistry, Vol. 87, No. 25, 1983 5173
25
I'
W I0
A
15 -
A i
20 -
I i
-
The Journal of Physical Chemistry, Vol. 87, No.
5174
25, 1983
TABLE I: Most Important Interatomic Distances from Titanium in the Cluster of Anatase Structure Including Four TiO, Octahedra
1800 -
9 E
E"
1600
i
1400
-
/-I I /
1
1I
1
\1
Bond
Bond Structure
1000 BO0
-
600
-
400
-
20c
+J.'
0 0.5
1.0
2.0
1.5
2.5
3.0
3.5
4.0
4.5
5.0
'
length
type
-
1200 -
0.0
Kozlowski et al.
(8,
Effective
coordination number'
1
1
1
AU2 (~103)~
~
Ti-0
4 x 1.93 2x 1.98
6.0
12
Ti-Ti
4 X 3.04
3.7
2.0
Ti -Ti
4 x 3.78
3.7
40.0
Ti-0
8~3.86
7.3
3.2
Ti - o
8~4.25 8~4.27 4r4.75
7.3 7.3 3.6
62
Ti -Ti
8X4.85
5.5
J
where Nj, Rj, and u,2 are the number, radius, and mean square relative displacements of the jth-shell atoms relative to the excited titanium atom and k is the photoelectron wave vector A,(k) and cy,&) are amplitude- and phase-dependent functions which to a first approximation are dependent on the type of atom in the jth shell (in our case this can be either Ti or 0). Stern et al.' showed that, if a&) is roughly linear as a function of photoelectron wave vector, then a Fourier transform of x with respect to 2k yields peaks located at Rj + a where a is an average of 2(da/dk) over the spectrum. The modulus of the Fourier transform IFT(R)I is shown in Figure 2 for the two forms of anatase. These curves may be understood with reference to the crystal structure of anatase. The anatase structure can be seen as a framework of distorted TiOz octahedra; the two Ti-0 bonds along the c axis are longer (1.98 A). The dominating elements in the structure are infinite double chains of edge-linked octahedra parallel to the a and b axes. Each octahedron belongs to two chains and shares thus four edges with other octahedra. Tables I lists the most important interatomic distances from titanium in the cluster of anatase structure including four Ti06 octahedra. (6)A.Balzarotti, F.Comin, L. Incoccia, M. Piacentini, S. Mobilio, and A. Savoia, Solid State Commun., 35, 145 (1980). (7) E. A. Stern, D. E. Sayers, and F. W. Lyttle, Phys. Rev. B , 11,4836 (1975). ~~
(8)M. Horn, C. F. Schwerdtfeger, and E. P. Meagher, Z . Kristallogr., 136, 273 (1972).
I
9.3 7'2
a Owing t o the presence of the surface not all coordination shells would be fully occupied. With the assumption of spherical particles of radius R g = 35 A , the fractional occupation F c of a coordination shell at a radius Rj, compared with the bulk material can be shown to be F c = ' / z { 1 f [(Rg- Rj)/Rg13 + 3(2Rg - Rj)Rj(2Rg 3Rj)/(8Rg3)}. This correction is not applied for the oxygen atoms lying in the first octahedron surrounding titanium. A o j 2 = u 2 j 2 - o , j 2 , where o j 2 is the mean square relative displacement of the jth-shell atoms relative to the excited titanium atom; 1 and 2 denote crystalline and fine-size anatase, respectively.
Clearly the first peak of IFT(R)I may be attributed to the first Ti-0 coordination sphere at 1.96 b, (the average value of Ti-0 distance in an octahedron). The difference between the observed Fourier transform peak position and the real absorberscatterer distance of 1.96 b, is about -0.4 A. This represents the value a = 2(da/dk) by which all transform peaks corresponding to Ti-0 must be shifted to yield interatomic distances. The next two peaks are a complex result of the overlap of four different features between 2.5 and 4.5 A. The next peak corresponds mainly to the Ti-Ti contribution at 4.85 b, arising from coupling the fourth octahedron to a three-octahedra cluster. It is clearly present in the crystalline anatase spectrum but almost entirely missing in that of ill-crystalline material. Moreover, all peaks in the Fourier transform of the fineparticle anatase spectrum are reduced in amplitude with respect to the crystal. The origin of this reduced intensity in the Fourier transform can be investigated further by isolating the fine structure from a single shell by Fourier techniques to yield xj,mpand Pj(k) separately (eq 2b, Stern et al.7). Identifying x ~from, eq ~2a we ~ find~ that a plot of In [ (xjlamp)2/ (xj,amp)l]against k 2 should yield a straight line passing through the origin with a slope - 2 h j 2 where Auj2 = uzj2- ulj2 is the difference in mean square relative displacement, 1 and 2 denoting crystalline and fine-size anatase, respectively. The above analysis relies on the transferability of the amplitude function A,&) which
EXAFS of V,O,-TiO,
The Journal of Physical Chemistry, Vol. 87, No. 25, 1983
5175
TABLE 11: Fitted Amplitude and Phase Parameters for Ti-0 and Ti-Ti Established from Crystalline Anatase parametersa
Ti-0
Ti-Ti 1.681 X lo-' 106.40 X -9.06 -0.857 -1.006 0.0114
0.667 26.26 X 1.331 2.433 -1.907 0.0866
Cl
CZ c3 a0 a1
a2
Defined in eq 3.
t
-'.O
0
20
40
80
60
IO0
120 k2
x
Figure 3. Logarithmic ratio of (1) and fine-size (2) anatase vs.
k'.
140
160
I60
(E-*)
(defined in eq 2) for crystalline
should hold as the chemical composition of the materials studied here is the same. Further, we assume that N, and Rj are the same, again a realistic assumption in the present case. Finally, the form of the relative mean square displacement term is derived under the assumption that the pair correlation function is of Gaussian form. The above amplitude analysis has been performed for the first shell of Ti02precipitate against Ti02 crystalline and is shown in Figure 3. Error analysis has been performed on the data indicated by the vertical error bars. The errors were calculated on the assumption that the noise has a white spectrum. Increase in error at high k 2 values merely reflects the decaying amplitude of the fine structure. It can be observed that the plot has a clear k2 dependence which is consistent with Auf 1.2 X A2. The plot however projects to k2 = 0 a t a point indicating that an overall reduction of fine-structure amplitude has occurred compared with that of the crystalline material. This we ascribe to a difference in homogeneity of our prepared films as it is consistent with the variation of amplitudes found in a different preparation earlier. Further, a comparison of the phases p ( k ) yields a value for the difference in mean first shell bond length between specimens of