- Email: [email protected]
solid interface
Surface Science 151 (1985) 301-310 North-Holland, Amsterdam
TJ3E PHENOMENON
OF WE’JTING AT SOLID/SOLID
301
INTERFACE
J. HABER, T. MACHEJ and T. CZEPPE Ins&t& of
Cata!ysis
and Surface Chemistry Potish Academy of Sciences, Krakbw, Poland
Received 20 June 1984; accepted for publication 16 October 1984
When a V,Os crystallite is placed on an anatase pellet and heated at 823-923 K, vanadium ions migrate over the surface of anatase grains enveloping them in a thin overlayer. XPS, X-ray and EPR studies show that at 823 K a very thin layer is formed, its properties being strongIy modified by interaction with the anatase support. At 923 K, on top of this inner layer an outer layer migrates, whose properties are similar to V,Os. As in the same conditions no migration is observed on rutile, it is concluded that this phenomenon is a manifestation of wetting of one oxide by another oxide, the difference in the surface free energy being the driving force of the migration.
The characteristic feature of solid state reactions consists in that they are localized at the interface between solid substrate and solid product, which is called the reaction interface. In a powder mixture of two solids, the reaction interface is formed by inter~anular contacts. If, however, one of the solids migrates rapidly over the surface of crystallites of the second solid, the reaction interface may considerably increase in the first period of the reaction, resulting in its spontaneous acceleration. Surface migration is also of paramount importance in the preparation of highly dispersed oxide-on-carrier systems which are the basis of catalysts for many key industrial processes such as hydrodesulphurization, selective oxidation, hydrogenation, etc. It is also of importance in various surface processing procedures aiming at the modification of mechanical, physical and chemical properties of solid surfaces. It is usually considered that migration of one solid over the surface of another solid results from surface diffusion of constituents of the lattice under the driving force of chemical potential. Taking into account the relatively high values of the lattice energy of the oxides, one can - however - expect surface diffusion to be negligible in the temperature range usually applied in the preparation of oxide-on-carrier systems or in some solid state reactions in oxide systems. Thus, a hypothesis was advanced some time ago that an entirely different mechanism may be responsible for the rapid surface migration of solids [l-3], described as wetting of one solid by another solid. Such migration
0039-6028/85/$03.30 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
J. Haber et al. / Wetting at solid/solid inter/ace
302
is due to the operation of the forces of surface tension in full analogy to those operating at the liquid/solid interface. In order to study this phenomenon, a model system of V,O, supported on two polymo~~c modifications of TiO, - rutile and anatase - was selected, this system being known as an excellent catalyst in the oxidation of alkylbenzenes to aldehydes and anhydrides when anatase but not rutile is taken as a support. 2. Experimental
V,O, was prepared by thermal decomposition of NH,VO, in the stream of air at 773 K for 20 h. TiO, (anatase) and TiO, (rutile) were commercial preparations supplied by Zaklady Chemiczne “Police”. Pellets of V,O,, sintered at 958 K for 48 h were put on larger pellets of anatase and rutile, presintered at 1023 K for 48 h (fig. la). The two sets were then placed in the electric oven and heated in air at 923 K for 48 h or at 823 K for 24 days. The V,O, pellets were then separated from the support and small samples of the support, taken from the areas I and II of the pellets (see fig. lb) were analysed by X-ray diffraction, photoelectron spectroscopy and ESR spectroscopy. X-ray powder diffraction patterns were obtained with the DRON-ZLOMO (USSR) diffractometer using Cu Ka radiation. ESR spectra were registered at 77 K with an SE/X type spectrometer produced by the Technical University Wroclaw. A VOSO, .5H,O sample was used as the standard to estimate the number of spins. The photoelectron spectra were obtained with a VG ESCA 3 spectrometer equipped with an argon ion gun for sputtering. Al Ka radiation with an energy of 1486.6 eV was used for excitation. The samples were deposited onto the iron holder from the ethanol suspension. Some spectra were also registered with fragments of pellets as received. In both cases identical results were obtained. In order to obtain the concentration profiles in the subsurface layers, all samples were sputtered with argon ions for 5-15 min at the pressure of lop6 Ton, voltage of 3 kV and ionic current of 20-26 PA. The ionic beam was directed at the angle of 45” to the surface of the sample. “2 05
rt! anatase
@p’ or rutile I a
b
Fig. 1. Scheme of experiment: (a) arrangement of pelllets in the course of heating; (b) areas of sampling from Ti02 pellets after heating.
J. Haber et al. / Wetting at solid/solid
303
interface
Analysis was based on the most intensive doublet of vanadium V 2~s,~,r,~, which partially overlaps with the Kcu, and Kaq satellites of the oxygen peaks 0 1s. Thus, its was necessary to subtract these satellites, their positions and intensities being assumed to be 9.7 eV, 7.3% and 11.7 eV, 3.1% [4] from the 0 1s peak, its binding energy taken as 529.6 eV. The spectra were accumulated and analysed with a computer, an appropriate computer programme being used for spectra deconvolution. The position of the carbon peak varied for the different samples within 0.4 eV. Still larger shifts were observed after argon sputtering and the position of the carbon peak could not be used for calibration of the binding energy values. All spectra were thus standardized against the position of the 0 1s peak, taken as 529.6 eV, the procedure adopted by Lars and Anderson [5]. The binding energy values of the vanadium electrons in the different vanadium oxides determined by these authors are summarized in table 1. The areas under the peaks were taken as the measure of the peak intensities. Quantitative analysis of the surface composition, based on the effective cross section ratio calculated from the Scoffield [6] and Wagner [7] values to amount to 2.17, led to a higher oxygen-to-vanadium ratio than expected. Results based on the experimentally determined ratio 1.7 give an oxygen-to-
Table 1 Values of binding energies and FWHM of 0 Is, V 2p,,* and Ti 2p3,2 electrons in some vanadium and titanium oxides [4] Compound
0 1s
TiO,
529.6 2.0
458.5 1.7
Ti02, sintered, 1250°C
529.6 2.3
458.2 1.9
v205
V,O,, sintered, 1250°C
v*o,/V,o,,
330°C
V,O, + Ti02 mixture
V 2P1,l
529.6 1.7
516.6 1.6
529.6 1.7
516.6 1.6
529.6 1.6
516.3 2.5
529.6 1.8
515.6 4.5
529.6 2.0
515.4 4.5
529.6 1.9
516.6 2.1
Ti 2pw2
458.5 1.7
304
J. Haber et al. / Wetting at solid/solrd
interface
vanadium ratio of the reference V,O, sample in very good agreement with the expected values. The value of 1.7 was thus used in the calculations of the composition of the analysed samples.
3. Results Fig. 2 shows microphotographs of the pellets of rutile and anatase and their cross-sections after they have been heated at 923 K with V,O, on top. One can see in fig. 2a the black spot in the place where the V,O, pellet was located on the anatase pellet, and the cross-section reveals (fig. 2b) that V,O, penetrated inside the pellet to a depth of about 0.5 mm, the black colour resulting from the reduction of the vanadium ions to lower valence states. No traces of V,O, were detected in the case of the rutile pellet (figs. 2c and 2d). Similar results were obtained in the experiment carried out at 823 K, the only difference being
Fig. 2. The microphotographs of futile and anatase pellets and their cross-sections after heating at 923 K with V,O, on top; (a), (c) top view of anatase and Wile pellets. (b), (d) cross-section of anatase and rutile pellets.
J. Haber et al. / Wetting at solid/solid interface
30s
that a yellow spot instead of a black spot appeared in the anatase pellet, indicating that the vanadium ions are not reduced at this temperature. X-ray analysis of the samples extracted from the areas I and II (see fig. 1) of the anatase pellets after heating at 823 K (AV 823) and after heating at 923 K (AV 923) revealed in all cases the presence of anatase; only in area I of sample AV 923 the presence of rutile was visible. Basing on the method described in ref. [8] its amount was estimated to be 13 mol%. No traces of V,O, phase could be detected in any of the samples. Table 2 summarizes the results of the XPS analysis of samples of V,O, and area I of AV 823 and AV 923. Data are given for samples as received as well as after sputtering for different periods of time. As an example, fig. 3 shows the photoelectron spectra of V 2p and 0 1s from sample AV 923, and fig. 4 the same V 2p spectra after computer deconvolution. In the case of pure V,O,, only a single peak of 0 Is electrons and a single doublet of V 2p electrons were observed, irrespective of the time of sputtering. Table 2 Values of binding energies and intensities of V 2,,,, argon ions Sample
Argon sputtering (min)
Ti 2~3~1 Ea (eV) I
II
and Ti 2,,/,
Ti 2~~~2 Intensity (arb. units)
2)
I
I
V 2P3,l
II
v205
_
5
5+10 5+10+12 _
Heating 10 h at 373 K AV 823 I
5
AV923 I
5
5+5 5+5+10
458.3 2.0 458.2
458.1 2.2 458.1 2.8 458.0 2.7 458.2 2.9
_
90.0
455.1
76.8
14.4
591.0 455.4 3.5 455.5 3.1 455.7 3.6
lines for samples sputtered with
480.3
174.6
500.1
236.1
475.5
329.4
v 2p,,, Intensity (arb. units) II
I
II
_
516.8 2.0 515.7 3.8 515.6 4.0 515.4 4.2 515.4 4.2
-
516.8 2.2 515.9 2.8 515.7 2.1 515.4 1.9
_ _ _
516.8 2.6 514.9 2.4
-
18.7
-
4.0
515.2 1.9 514.4 3.2 514.1 2.0 514.3 2.4
132.7
26.8
101.3
128.7
65.9
50.7
13.5
2.8
J. Haber et al. / Wetting at solid/solid
306
interface
The binding energy values of V 2p,,, electrons are given in table 2. Comparison with the data quoted in table 1 indicates that at the surface of V,O, initially V 5+ ions are present. The ESR spectrum revealed that the sample contains a small amount (of the order of 0.1 at%) of V4+ ions which may be distributed in the bulk of the sample and therefore are not visible in the photoelectron spectra. Sputtering with argon ions results in the shift of the doublet and its considerable broadening. The binding energies, observed after first and second sputtering, correspond to those characteristic of V,O, and determination of the surface concentrations of vanadium and oxygen gives the value of their ratio 2,s equal to 2. Subsequent sputtering caused a further shift of the doublet, the binding energy value attaining that of V,O, and a value of O/V equal to about 1.9. No more changes were observed after heating of the sample in 10 -lo Torr at 473 K for 10 h. In the spectra of the V,O,-rutile system (samples from area I of RV823 and RV923) only the Ti 2p doublet and 0 Is peak were present, no peaks
455
L65
BINDING
515
ENERGY
525
535
ieV1
515 BINDING
520 ENERGY
525 IeVl
Fig. 3. XPS spectra of Ti 2p, V 2p and 0 1s electrons of sample AV 9231: (AR) as received, (A) 5 min, (B) 10 min and (C) 20 min of sputtering with argon ions. Fig. 4. V 2p and 0 Is spectra from fig. 3 after computer (- - -) V 2p (form II); (. . .) 0 1s satellites; ( -) the same meaning as in fig. 3.
deconvolution: (- + -) V 2p (form I); computer fitted curve. AR, A, B, C, have
.I. Haber et al. / Welting at solid/solid interface
307
corresponding to vanadium electrons being visible. This confirms the conclusions drawn from the microscopic and X-ray studies that after heating of the V,O,-rutile system, the pellets can be separated without any traces of the migration of V,O, into rutile pellet being observed. The V 2p doublet in the spectrum of the AV 823 sample is located on the slope of the 0 1s peak and is perturbed by its satellites. After their subtraction, a single symmetric doublet is obtained, which after sputtering becomes more diffused. The values of E, of the V 2p,,, electrons indicate that before argon bombardment only V5+ are present at the surface, and after the bombardment V3+ ions appear. The spectrum before sputtering contains the single Ti 2p doublet at E, characteristic of TiO,. After sputtering a shoulder appears on the lower energy side, hinting at some reduction of Ti ions due to argon bombardment. In the spectrum of the AV 923 sample “as received”, two doublets of V 2p electrons are visible, their intensity ratio being 5. From a comparison of their E, values with those quoted in table 1, it may be concluded that two types of vanadium ions are present at the surface of the sample: V5+ ions (marked I in table 2) and V3+ ions (marked II). The Ti 2p doublet and 0 1s peak are single. Argon sputtering causes the shift of both V 2p doublets to lower E, values and the appearance of a lower energy shoulder at the Ti 2p doublet, its intensity increasing with the time of sputtering. Changes of the total intensity of the V 2p,,, peaks and of the Ti 2p peak with time of sputtering are shown in fig. 5, whereas fig. 6 illustrates the intensities of the V 2p,,, peaks of the two forms of vanadium ions as well as their ratio as a function of the sputtering time. The
1v2PII ,
0 v2p
.IyzpI 0
V2pI
o Ti2p
SPUTTERING
TIME [mm]
Fig. 5. Changes of the total intensity argon ions for sample AV 923 (I).
SPUTTERING of V 2p and Ti 2p electron
Fig. 6. Intensities of V 2p 3,2 peaks of the two forms of vanadium of sputtering time.
TlME]min]
peaks with time of sputtering
with
ions and their ratio as a function
308
J. Haber et al. / Wetting at solid/solid
mterjme
intensity of the Ti 2p peak increases with sputtering time, whereas the total intensity of the V 2p,,, p eaks decreases practically to zero after 20 min exposure to argon bombardment. The slight increase of this intensity in the initial period of sputtering may be due to the removal of carbon deposit. The fact that the total intensity of the vanadium peaks does not decrease after the first sputtering indicates that the thickness of the surface layer of the vanadium ions is greater than the escape depth of the electrons. Using Penn’s method [9], the average escape depth h, of the V 2p electrons from different vanadium oxides could be estimated to be 1.3 nm. As 95% of the peak intensity is originating from the depth of 3h, [lo], it may be concluded that the thickness of the layer of the vanadium ions is greater than 3.9 nm.
4. Discussion We shall first discuss the results of the experiments, carried out at 823 K. From the data summarized in table 2 it may be concluded that the vanadium ions have migrated over the surface of TiO, Crystallites and form a layer of Vs+ ions. When bombarded with argon ions, this layer is reduced to a different state than that observed on reduction of V,O,. The value of E, observed after sputtering is much lower than the value corresponding to V,O, obtained after prolonged sputtering of V,O,. This indicates that the surface layer is very thin and that its properties are strongly modified by the interaction with the TiO, support, the V 5+ ions either being reduced by sputtering to a valence state lower than in V,O,, or assuming the coordination of a much higher ligand field. After the experiment in 923 K, the spectrum of the layer resulting from the migration of the vanadium ions over the surface of the TiO, crystallites reveals the presence of two doublets of V 2p electrons. They cannot be due to the formation of a partially reduced layer of V,O,, because on reduction of the latter, even to a considerable degree, only the shift of the V 2p doublet is observed but never a separation into two doublets. After sputtering, one of the doublets (form I) shifts due to the reduction in a manner identical to that characteristic for pure V,O,, whereas the second doublet (form II) shifts to the very low value observed in the experiment at 823 K and assigned to vanadium ions strongly interacting with the support. Indeed, a considerable increase of with the the intensity of the V 2p,,, p eak of form II after first sputtering simultaneous decrease of the intensity of V 2p,,, peak of form I confirms the conclusion that the layer of vanadium ions of form II is beneath that of form I, of the being in direct contact with TiO,. Thus, the following mechanism phenomena taking place in the course of the experiment carried out at 923 K emerges from the XPS analysis: Migration of vanadium ions over the surface of anatase grains results at first in the formation of a thin (probably mono-
J. Haber et al. / Wetting at solid/solid interface
309
molecular) layer whose properties are strongly modified by the interaction with the surface of anatase. At 923 K, the vanadium ions in this layer undergo reduction to lower valence state. It may be noted here that reduction of V,O, as a result of the contact with anatase was observed by Vejeux and Courtine [ll] on heating above 873 K. The appearence of lower valent vanadium ions strongly interacting with the surface of TiO* may be the cause of some reconstruction of the surface of anatase into the rutile structure, observed in X-ray diffraction patterns of samples heated at 923 K. Further migration of vanadium ions on top of the first layer results in the formation of a multilayer structure, the upper layers assuming the structure of V,O,. Indeed, ESR spectra reveal the presence of the line characteristic of V4+ ions in the lattice of V,O, [12,13], its intensity indicating that the concentration of these ions is much smaller than in the sample of pure V,O, heated at the same temperature. It is noteworthy that the XPS spectra of samples “as received” after experiments in both 823 K and 923 K show the presence of a Ti 2p peak of considerable intensity, although the layer of V,O, formed in the experiment at 923 K is - as discussed above - thicker than the escape depth of V2p electrons, the more of Ti 2p electrons, characterized by lower energy. This may be easily explained if it is assumed that vanadium ions migrate not over the whole surface of anatase crystallites, but only over certain crystal planes characterized by appropriate values of surface free energy. The fact that already after first sputtering considerable reduction of Ti4+ to Ti3+ ions is observed is in line with such an assumption, because reduction of the titanium ions at such an early stage of sputtering may be expected only if the free surface of TiO, is exposed to a beam of argon ions. It should be noted that Inomata et al. [14] observed in the IR spectrum after adsorption of ammonia on V,O,/anatase the appearence of a band characteristic for ammonia adsorbed on TiO,. The XPS spectrum registered directly from the anatase pellet at the position I was identical with that of the sample taken from the pellet, ground and deposited on the sample holder from suspension. This seems to confirm the conclusion that part of the TiO, surface remained free of V,O,, not because of the low concentration of the latter, but because of the difference in spreading of V,O, over the different crystal planes of anatase. Migration of vanadium oxide over the surface of anatase cannot be due just to the spreading under the influence of the concentration gradient because of its high mobility, as in such case identical migration would have taken place over the surface of rutile, which is not observed. The only other driving force which may cause the spreading of V,O, over the surface of anatase is the difference in surface free energy. A conclusion may thus be formulated that spreading of V,O, over the surface of anatase is the manifestation of a general phenomenon of wetting of one solid by another solid, fully analogous to wetting of solids by liquids. Such spreading will namely take place if the cohesion energy of anatase is greater than the appropriate sum of the cohesion
310
J. Haber et al. /
Wetting at solid/solid interface
energy of V,O, and its energy of adhesion to anatase. The migration of vanadium oxide will continue until the film attains such thickness at which its surface free energy does not depend any more on the presence of the TiO, support. Indeed, as discussed above, the outermost layers of the film show a behaviour similar to that of V,O, at variance with the inner layers, whose properties are strongly modified by the interaction with the TiO, support. As the values of the cohesion energy vary with the type of crystal plane, it may be expected that wetting will take place only on certain crystal planes of the given solid, which is indeed the case in the V,O,/anatase system. Even more pronounced may be the differences in the cohesion energy between various polymorphic modifications, as illustrated by the case of anatase and r-utile: vanadium oxide is wetting some crystal planes of anatase, but not those of rutile.
Acknowledgements The authors would like to express their thanks to Mr. A. Cape&i M. Eng. for the preparation of the computer programme and to Dr. M. Rusiecka for the registration of ESR spectra.
References [l] J. Haber and J. ZioIkowski, in: Proc. 7th Intern. Symp. on the Reactivity of Solids, Bristol, 1972 (Chapman and Hall, London, 1972) p. 782. [2] T. Bak and J. ZioIkowski, Bull. Acad. Polon. Sci., Ser. Sci. Chim. 20 (1972) 821. [3] J. Haber, in: Surface Properties and Catalysis by Non-Metals, Eds. J.P. Bonnelle, B. Delmon and E. Derouane (Reidel, Dordrecht, 1983) p. 1. [4] M.O. Krause and J.G. Ferreira, J. Phys. B8 (1975) 2007. [5] S. Lam and T. Anderson, J. Chem. Sot. Faraday Trans. I, 6 (1979) 1356. [6] J.H. Scoffield, J. Electron Spectrosc. Related Phenomena 8 (1976) 129. [7] C.D. Wagner, Surface Interface Anal. 3 (1981) 211. (81 R.A. Spurr and H. Hyers, Anal. Chem. 29 (1957) 750. [9] R.R. Penn, J. Electron Spectrosc. Related Phenomena 9 (1976) 29. [lo] J. Finster, P. Lorenz and A. Meisel, Surface Interface Anal. 1 (1979) 179. [ll] A. Vejeux and P. Courtine, J. Solid State Chem. 23 (1978) 93. [12] H. Takahashi, M. Shiotani, H. Kobayashi and J. Sohma, J. Catalysis 14 (1969) 134. [13] R.S. Mann and K.C. Khulbe, Bull. Chem. Sot. Japan 45 (1972) 2929. [14] M. Inomata, K. Mot-i, A. Miyamoto, T. Ui and Y. Murakami, J. Phys. Chem. 87 (1983) 754.
TJ3E PHENOMENON
OF WE’JTING AT SOLID/SOLID
301
INTERFACE
J. HABER, T. MACHEJ and T. CZEPPE Ins&t& of
Cata!ysis
and Surface Chemistry Potish Academy of Sciences, Krakbw, Poland
Received 20 June 1984; accepted for publication 16 October 1984
When a V,Os crystallite is placed on an anatase pellet and heated at 823-923 K, vanadium ions migrate over the surface of anatase grains enveloping them in a thin overlayer. XPS, X-ray and EPR studies show that at 823 K a very thin layer is formed, its properties being strongIy modified by interaction with the anatase support. At 923 K, on top of this inner layer an outer layer migrates, whose properties are similar to V,Os. As in the same conditions no migration is observed on rutile, it is concluded that this phenomenon is a manifestation of wetting of one oxide by another oxide, the difference in the surface free energy being the driving force of the migration.
The characteristic feature of solid state reactions consists in that they are localized at the interface between solid substrate and solid product, which is called the reaction interface. In a powder mixture of two solids, the reaction interface is formed by inter~anular contacts. If, however, one of the solids migrates rapidly over the surface of crystallites of the second solid, the reaction interface may considerably increase in the first period of the reaction, resulting in its spontaneous acceleration. Surface migration is also of paramount importance in the preparation of highly dispersed oxide-on-carrier systems which are the basis of catalysts for many key industrial processes such as hydrodesulphurization, selective oxidation, hydrogenation, etc. It is also of importance in various surface processing procedures aiming at the modification of mechanical, physical and chemical properties of solid surfaces. It is usually considered that migration of one solid over the surface of another solid results from surface diffusion of constituents of the lattice under the driving force of chemical potential. Taking into account the relatively high values of the lattice energy of the oxides, one can - however - expect surface diffusion to be negligible in the temperature range usually applied in the preparation of oxide-on-carrier systems or in some solid state reactions in oxide systems. Thus, a hypothesis was advanced some time ago that an entirely different mechanism may be responsible for the rapid surface migration of solids [l-3], described as wetting of one solid by another solid. Such migration
0039-6028/85/$03.30 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
J. Haber et al. / Wetting at solid/solid inter/ace
302
is due to the operation of the forces of surface tension in full analogy to those operating at the liquid/solid interface. In order to study this phenomenon, a model system of V,O, supported on two polymo~~c modifications of TiO, - rutile and anatase - was selected, this system being known as an excellent catalyst in the oxidation of alkylbenzenes to aldehydes and anhydrides when anatase but not rutile is taken as a support. 2. Experimental
V,O, was prepared by thermal decomposition of NH,VO, in the stream of air at 773 K for 20 h. TiO, (anatase) and TiO, (rutile) were commercial preparations supplied by Zaklady Chemiczne “Police”. Pellets of V,O,, sintered at 958 K for 48 h were put on larger pellets of anatase and rutile, presintered at 1023 K for 48 h (fig. la). The two sets were then placed in the electric oven and heated in air at 923 K for 48 h or at 823 K for 24 days. The V,O, pellets were then separated from the support and small samples of the support, taken from the areas I and II of the pellets (see fig. lb) were analysed by X-ray diffraction, photoelectron spectroscopy and ESR spectroscopy. X-ray powder diffraction patterns were obtained with the DRON-ZLOMO (USSR) diffractometer using Cu Ka radiation. ESR spectra were registered at 77 K with an SE/X type spectrometer produced by the Technical University Wroclaw. A VOSO, .5H,O sample was used as the standard to estimate the number of spins. The photoelectron spectra were obtained with a VG ESCA 3 spectrometer equipped with an argon ion gun for sputtering. Al Ka radiation with an energy of 1486.6 eV was used for excitation. The samples were deposited onto the iron holder from the ethanol suspension. Some spectra were also registered with fragments of pellets as received. In both cases identical results were obtained. In order to obtain the concentration profiles in the subsurface layers, all samples were sputtered with argon ions for 5-15 min at the pressure of lop6 Ton, voltage of 3 kV and ionic current of 20-26 PA. The ionic beam was directed at the angle of 45” to the surface of the sample. “2 05
rt! anatase
@p’ or rutile I a
b
Fig. 1. Scheme of experiment: (a) arrangement of pelllets in the course of heating; (b) areas of sampling from Ti02 pellets after heating.
J. Haber et al. / Wetting at solid/solid
303
interface
Analysis was based on the most intensive doublet of vanadium V 2~s,~,r,~, which partially overlaps with the Kcu, and Kaq satellites of the oxygen peaks 0 1s. Thus, its was necessary to subtract these satellites, their positions and intensities being assumed to be 9.7 eV, 7.3% and 11.7 eV, 3.1% [4] from the 0 1s peak, its binding energy taken as 529.6 eV. The spectra were accumulated and analysed with a computer, an appropriate computer programme being used for spectra deconvolution. The position of the carbon peak varied for the different samples within 0.4 eV. Still larger shifts were observed after argon sputtering and the position of the carbon peak could not be used for calibration of the binding energy values. All spectra were thus standardized against the position of the 0 1s peak, taken as 529.6 eV, the procedure adopted by Lars and Anderson [5]. The binding energy values of the vanadium electrons in the different vanadium oxides determined by these authors are summarized in table 1. The areas under the peaks were taken as the measure of the peak intensities. Quantitative analysis of the surface composition, based on the effective cross section ratio calculated from the Scoffield [6] and Wagner [7] values to amount to 2.17, led to a higher oxygen-to-vanadium ratio than expected. Results based on the experimentally determined ratio 1.7 give an oxygen-to-
Table 1 Values of binding energies and FWHM of 0 Is, V 2p,,* and Ti 2p3,2 electrons in some vanadium and titanium oxides [4] Compound
0 1s
TiO,
529.6 2.0
458.5 1.7
Ti02, sintered, 1250°C
529.6 2.3
458.2 1.9
v205
V,O,, sintered, 1250°C
v*o,/V,o,,
330°C
V,O, + Ti02 mixture
V 2P1,l
529.6 1.7
516.6 1.6
529.6 1.7
516.6 1.6
529.6 1.6
516.3 2.5
529.6 1.8
515.6 4.5
529.6 2.0
515.4 4.5
529.6 1.9
516.6 2.1
Ti 2pw2
458.5 1.7
304
J. Haber et al. / Wetting at solid/solrd
interface
vanadium ratio of the reference V,O, sample in very good agreement with the expected values. The value of 1.7 was thus used in the calculations of the composition of the analysed samples.
3. Results Fig. 2 shows microphotographs of the pellets of rutile and anatase and their cross-sections after they have been heated at 923 K with V,O, on top. One can see in fig. 2a the black spot in the place where the V,O, pellet was located on the anatase pellet, and the cross-section reveals (fig. 2b) that V,O, penetrated inside the pellet to a depth of about 0.5 mm, the black colour resulting from the reduction of the vanadium ions to lower valence states. No traces of V,O, were detected in the case of the rutile pellet (figs. 2c and 2d). Similar results were obtained in the experiment carried out at 823 K, the only difference being
Fig. 2. The microphotographs of futile and anatase pellets and their cross-sections after heating at 923 K with V,O, on top; (a), (c) top view of anatase and Wile pellets. (b), (d) cross-section of anatase and rutile pellets.
J. Haber et al. / Wetting at solid/solid interface
30s
that a yellow spot instead of a black spot appeared in the anatase pellet, indicating that the vanadium ions are not reduced at this temperature. X-ray analysis of the samples extracted from the areas I and II (see fig. 1) of the anatase pellets after heating at 823 K (AV 823) and after heating at 923 K (AV 923) revealed in all cases the presence of anatase; only in area I of sample AV 923 the presence of rutile was visible. Basing on the method described in ref. [8] its amount was estimated to be 13 mol%. No traces of V,O, phase could be detected in any of the samples. Table 2 summarizes the results of the XPS analysis of samples of V,O, and area I of AV 823 and AV 923. Data are given for samples as received as well as after sputtering for different periods of time. As an example, fig. 3 shows the photoelectron spectra of V 2p and 0 1s from sample AV 923, and fig. 4 the same V 2p spectra after computer deconvolution. In the case of pure V,O,, only a single peak of 0 Is electrons and a single doublet of V 2p electrons were observed, irrespective of the time of sputtering. Table 2 Values of binding energies and intensities of V 2,,,, argon ions Sample
Argon sputtering (min)
Ti 2~3~1 Ea (eV) I
II
and Ti 2,,/,
Ti 2~~~2 Intensity (arb. units)
2)
I
I
V 2P3,l
II
v205
_
5
5+10 5+10+12 _
Heating 10 h at 373 K AV 823 I
5
AV923 I
5
5+5 5+5+10
458.3 2.0 458.2
458.1 2.2 458.1 2.8 458.0 2.7 458.2 2.9
_
90.0
455.1
76.8
14.4
591.0 455.4 3.5 455.5 3.1 455.7 3.6
lines for samples sputtered with
480.3
174.6
500.1
236.1
475.5
329.4
v 2p,,, Intensity (arb. units) II
I
II
_
516.8 2.0 515.7 3.8 515.6 4.0 515.4 4.2 515.4 4.2
-
516.8 2.2 515.9 2.8 515.7 2.1 515.4 1.9
_ _ _
516.8 2.6 514.9 2.4
-
18.7
-
4.0
515.2 1.9 514.4 3.2 514.1 2.0 514.3 2.4
132.7
26.8
101.3
128.7
65.9
50.7
13.5
2.8
J. Haber et al. / Wetting at solid/solid
306
interface
The binding energy values of V 2p,,, electrons are given in table 2. Comparison with the data quoted in table 1 indicates that at the surface of V,O, initially V 5+ ions are present. The ESR spectrum revealed that the sample contains a small amount (of the order of 0.1 at%) of V4+ ions which may be distributed in the bulk of the sample and therefore are not visible in the photoelectron spectra. Sputtering with argon ions results in the shift of the doublet and its considerable broadening. The binding energies, observed after first and second sputtering, correspond to those characteristic of V,O, and determination of the surface concentrations of vanadium and oxygen gives the value of their ratio 2,s equal to 2. Subsequent sputtering caused a further shift of the doublet, the binding energy value attaining that of V,O, and a value of O/V equal to about 1.9. No more changes were observed after heating of the sample in 10 -lo Torr at 473 K for 10 h. In the spectra of the V,O,-rutile system (samples from area I of RV823 and RV923) only the Ti 2p doublet and 0 Is peak were present, no peaks
455
L65
BINDING
515
ENERGY
525
535
ieV1
515 BINDING
520 ENERGY
525 IeVl
Fig. 3. XPS spectra of Ti 2p, V 2p and 0 1s electrons of sample AV 9231: (AR) as received, (A) 5 min, (B) 10 min and (C) 20 min of sputtering with argon ions. Fig. 4. V 2p and 0 Is spectra from fig. 3 after computer (- - -) V 2p (form II); (. . .) 0 1s satellites; ( -) the same meaning as in fig. 3.
deconvolution: (- + -) V 2p (form I); computer fitted curve. AR, A, B, C, have
.I. Haber et al. / Welting at solid/solid interface
307
corresponding to vanadium electrons being visible. This confirms the conclusions drawn from the microscopic and X-ray studies that after heating of the V,O,-rutile system, the pellets can be separated without any traces of the migration of V,O, into rutile pellet being observed. The V 2p doublet in the spectrum of the AV 823 sample is located on the slope of the 0 1s peak and is perturbed by its satellites. After their subtraction, a single symmetric doublet is obtained, which after sputtering becomes more diffused. The values of E, of the V 2p,,, electrons indicate that before argon bombardment only V5+ are present at the surface, and after the bombardment V3+ ions appear. The spectrum before sputtering contains the single Ti 2p doublet at E, characteristic of TiO,. After sputtering a shoulder appears on the lower energy side, hinting at some reduction of Ti ions due to argon bombardment. In the spectrum of the AV 923 sample “as received”, two doublets of V 2p electrons are visible, their intensity ratio being 5. From a comparison of their E, values with those quoted in table 1, it may be concluded that two types of vanadium ions are present at the surface of the sample: V5+ ions (marked I in table 2) and V3+ ions (marked II). The Ti 2p doublet and 0 1s peak are single. Argon sputtering causes the shift of both V 2p doublets to lower E, values and the appearance of a lower energy shoulder at the Ti 2p doublet, its intensity increasing with the time of sputtering. Changes of the total intensity of the V 2p,,, peaks and of the Ti 2p peak with time of sputtering are shown in fig. 5, whereas fig. 6 illustrates the intensities of the V 2p,,, peaks of the two forms of vanadium ions as well as their ratio as a function of the sputtering time. The
1v2PII ,
0 v2p
.IyzpI 0
V2pI
o Ti2p
SPUTTERING
TIME [mm]
Fig. 5. Changes of the total intensity argon ions for sample AV 923 (I).
SPUTTERING of V 2p and Ti 2p electron
Fig. 6. Intensities of V 2p 3,2 peaks of the two forms of vanadium of sputtering time.
TlME]min]
peaks with time of sputtering
with
ions and their ratio as a function
308
J. Haber et al. / Wetting at solid/solid
mterjme
intensity of the Ti 2p peak increases with sputtering time, whereas the total intensity of the V 2p,,, p eaks decreases practically to zero after 20 min exposure to argon bombardment. The slight increase of this intensity in the initial period of sputtering may be due to the removal of carbon deposit. The fact that the total intensity of the vanadium peaks does not decrease after the first sputtering indicates that the thickness of the surface layer of the vanadium ions is greater than the escape depth of the electrons. Using Penn’s method [9], the average escape depth h, of the V 2p electrons from different vanadium oxides could be estimated to be 1.3 nm. As 95% of the peak intensity is originating from the depth of 3h, [lo], it may be concluded that the thickness of the layer of the vanadium ions is greater than 3.9 nm.
4. Discussion We shall first discuss the results of the experiments, carried out at 823 K. From the data summarized in table 2 it may be concluded that the vanadium ions have migrated over the surface of TiO, Crystallites and form a layer of Vs+ ions. When bombarded with argon ions, this layer is reduced to a different state than that observed on reduction of V,O,. The value of E, observed after sputtering is much lower than the value corresponding to V,O, obtained after prolonged sputtering of V,O,. This indicates that the surface layer is very thin and that its properties are strongly modified by the interaction with the TiO, support, the V 5+ ions either being reduced by sputtering to a valence state lower than in V,O,, or assuming the coordination of a much higher ligand field. After the experiment in 923 K, the spectrum of the layer resulting from the migration of the vanadium ions over the surface of the TiO, crystallites reveals the presence of two doublets of V 2p electrons. They cannot be due to the formation of a partially reduced layer of V,O,, because on reduction of the latter, even to a considerable degree, only the shift of the V 2p doublet is observed but never a separation into two doublets. After sputtering, one of the doublets (form I) shifts due to the reduction in a manner identical to that characteristic for pure V,O,, whereas the second doublet (form II) shifts to the very low value observed in the experiment at 823 K and assigned to vanadium ions strongly interacting with the support. Indeed, a considerable increase of with the the intensity of the V 2p,,, p eak of form II after first sputtering simultaneous decrease of the intensity of V 2p,,, peak of form I confirms the conclusion that the layer of vanadium ions of form II is beneath that of form I, of the being in direct contact with TiO,. Thus, the following mechanism phenomena taking place in the course of the experiment carried out at 923 K emerges from the XPS analysis: Migration of vanadium ions over the surface of anatase grains results at first in the formation of a thin (probably mono-
J. Haber et al. / Wetting at solid/solid interface
309
molecular) layer whose properties are strongly modified by the interaction with the surface of anatase. At 923 K, the vanadium ions in this layer undergo reduction to lower valence state. It may be noted here that reduction of V,O, as a result of the contact with anatase was observed by Vejeux and Courtine [ll] on heating above 873 K. The appearence of lower valent vanadium ions strongly interacting with the surface of TiO* may be the cause of some reconstruction of the surface of anatase into the rutile structure, observed in X-ray diffraction patterns of samples heated at 923 K. Further migration of vanadium ions on top of the first layer results in the formation of a multilayer structure, the upper layers assuming the structure of V,O,. Indeed, ESR spectra reveal the presence of the line characteristic of V4+ ions in the lattice of V,O, [12,13], its intensity indicating that the concentration of these ions is much smaller than in the sample of pure V,O, heated at the same temperature. It is noteworthy that the XPS spectra of samples “as received” after experiments in both 823 K and 923 K show the presence of a Ti 2p peak of considerable intensity, although the layer of V,O, formed in the experiment at 923 K is - as discussed above - thicker than the escape depth of V2p electrons, the more of Ti 2p electrons, characterized by lower energy. This may be easily explained if it is assumed that vanadium ions migrate not over the whole surface of anatase crystallites, but only over certain crystal planes characterized by appropriate values of surface free energy. The fact that already after first sputtering considerable reduction of Ti4+ to Ti3+ ions is observed is in line with such an assumption, because reduction of the titanium ions at such an early stage of sputtering may be expected only if the free surface of TiO, is exposed to a beam of argon ions. It should be noted that Inomata et al. [14] observed in the IR spectrum after adsorption of ammonia on V,O,/anatase the appearence of a band characteristic for ammonia adsorbed on TiO,. The XPS spectrum registered directly from the anatase pellet at the position I was identical with that of the sample taken from the pellet, ground and deposited on the sample holder from suspension. This seems to confirm the conclusion that part of the TiO, surface remained free of V,O,, not because of the low concentration of the latter, but because of the difference in spreading of V,O, over the different crystal planes of anatase. Migration of vanadium oxide over the surface of anatase cannot be due just to the spreading under the influence of the concentration gradient because of its high mobility, as in such case identical migration would have taken place over the surface of rutile, which is not observed. The only other driving force which may cause the spreading of V,O, over the surface of anatase is the difference in surface free energy. A conclusion may thus be formulated that spreading of V,O, over the surface of anatase is the manifestation of a general phenomenon of wetting of one solid by another solid, fully analogous to wetting of solids by liquids. Such spreading will namely take place if the cohesion energy of anatase is greater than the appropriate sum of the cohesion
310
J. Haber et al. /
Wetting at solid/solid interface
energy of V,O, and its energy of adhesion to anatase. The migration of vanadium oxide will continue until the film attains such thickness at which its surface free energy does not depend any more on the presence of the TiO, support. Indeed, as discussed above, the outermost layers of the film show a behaviour similar to that of V,O, at variance with the inner layers, whose properties are strongly modified by the interaction with the TiO, support. As the values of the cohesion energy vary with the type of crystal plane, it may be expected that wetting will take place only on certain crystal planes of the given solid, which is indeed the case in the V,O,/anatase system. Even more pronounced may be the differences in the cohesion energy between various polymorphic modifications, as illustrated by the case of anatase and r-utile: vanadium oxide is wetting some crystal planes of anatase, but not those of rutile.
Acknowledgements The authors would like to express their thanks to Mr. A. Cape&i M. Eng. for the preparation of the computer programme and to Dr. M. Rusiecka for the registration of ESR spectra.
References [l] J. Haber and J. ZioIkowski, in: Proc. 7th Intern. Symp. on the Reactivity of Solids, Bristol, 1972 (Chapman and Hall, London, 1972) p. 782. [2] T. Bak and J. ZioIkowski, Bull. Acad. Polon. Sci., Ser. Sci. Chim. 20 (1972) 821. [3] J. Haber, in: Surface Properties and Catalysis by Non-Metals, Eds. J.P. Bonnelle, B. Delmon and E. Derouane (Reidel, Dordrecht, 1983) p. 1. [4] M.O. Krause and J.G. Ferreira, J. Phys. B8 (1975) 2007. [5] S. Lam and T. Anderson, J. Chem. Sot. Faraday Trans. I, 6 (1979) 1356. [6] J.H. Scoffield, J. Electron Spectrosc. Related Phenomena 8 (1976) 129. [7] C.D. Wagner, Surface Interface Anal. 3 (1981) 211. (81 R.A. Spurr and H. Hyers, Anal. Chem. 29 (1957) 750. [9] R.R. Penn, J. Electron Spectrosc. Related Phenomena 9 (1976) 29. [lo] J. Finster, P. Lorenz and A. Meisel, Surface Interface Anal. 1 (1979) 179. [ll] A. Vejeux and P. Courtine, J. Solid State Chem. 23 (1978) 93. [12] H. Takahashi, M. Shiotani, H. Kobayashi and J. Sohma, J. Catalysis 14 (1969) 134. [13] R.S. Mann and K.C. Khulbe, Bull. Chem. Sot. Japan 45 (1972) 2929. [14] M. Inomata, K. Mot-i, A. Miyamoto, T. Ui and Y. Murakami, J. Phys. Chem. 87 (1983) 754.