HM Medical Clinic

 

Microsoft word - chapter 12.metal.oxides

Chapter 12 The Anisotropy of Metal Oxide Surface Properties
G.S. Rohrer
Department of Materials Science and Engineering, Carnegie Mellon University,Pittsburgh PA, 15213-3890, U.S.A.
The surface properties of metal oxides influence the rates of heterogeneouschemical reactions, the growth of heteroepitaxial films, and the sintering ofparticles during the consolidation of ceramics. Just as certain bulk properties ofcrystalline materials are anisotropic, surface properties also depend onorientation. The dependence of surface properties on orientation can berationalized by recognizing that the atoms on crystallographically distinct facetshave different coordination environments, as illustrated schematically in Fig. 1.
This same figure also illustrates that in binary and more complex materials,surfaces with identical orientations can be terminated by different atomic layerswith distinct compositions. Throughout this chapter, we shall take the"character" of a surface to be defined by its orientation, {hkl}, and its atomictermination layer.
The relationship between surface properties and surface character allows the macroscopic properties of surface active materials to be manipulated bycontrolling the distribution of the exposed surfaces. For particulate materials,this distribution is determined by the crystal habit. The character of surfacesexposed by a dense polycrystal can be controlled by the introduction of somecrystallographic texture and thin film materials can be oriented throughheteroepitaxial seeding. While techniques for structural control are generallyavailable to the material scientist, what is less frequently known is how thephysiochemical properties of a material vary with surface character. In thischapter, efforts to measure the surface properties of oxides as a function ofcharacter will be described. Section two of this chapter reviews some specificexamples of the relationship between surface character and properties that havebeen derived from studies of particulate oxides, single crystals, and thin films.
In section three, the factors that determine surface morphology and the range ofachievable orientations are described. The factors that influence surfacestoichiometry and the composition of the termination layer are described in Fig. 1. Schematic illustration of a two-dimensional metal (M) oxide (O) crystal. Note thatatoms on the {10} type faces have different numbers of nearest neighbors than the atoms onthe {11} faces. Furthermore, the (11) surface is terminated by M atoms while the (1 1)surface is terminated by O atoms.
section four. Two specific examples illustrating anisotropic reactivity arereviewed in sections five and six, which deal with partial oxidation reactions onMoO and photochemical reactions on TiO , respectively.
2. SURFACE CHARACTER-PROPERTY RELATIONSHIPS
2.1 Particulate oxidation catalysts
One clear piece of evidence for surface character-property relationships lies in
the observation that particulate catalysts with the same composition, but
different crystal habit, have different reactivities. This phenomenon, sometimes
referred to as a catalytic anisotropy or a structure-sensitive catalytic reaction, has
been observed in a variety of Mo, V, and W containing materials [1]. Because
the catalytic anisotropy of α-MoO has been studied in relatively greater depth,
this exemplary case is described below [1-10].
The orthorhombic (α) phase of MoO has the highly anisotropic layered structure illustrated in Fig. 2 [11]. Because of the weak, secondary bondingbetween the sheets of corner and edge-sharing MoO octahedra perpendicular to [010], the crystals are micaceous. The most prominent facet on the growth formis {010}. The crystals are bounded laterally by {001} and {hk0} facets, wherethe h/k ratio is about 2. In some case, the {001} and {h k0} facets may bereplaced by {101} and {100} facets, respectively.
The Mo atoms on the (010) surface have their full complement of six oxygen nearest neighbors. However, termination of the bulk structure in other Fig. 2. The structure of α-MoO3 crystals. (a) The structure is built of double octahedrallayers stacked along [010]. The spacing between adjacent double layers is 6.9 Å. (b) Aprojection on the (010) plane shows two superimposed rectangular nets of corner-sharingoctahedra. The shaded net is below the unshaded one. (c) A typical growth form of α-MoO3. All indexing is with respect to space group Pbnm.
directions leaves the surface Mo atoms on lateral facets with fewer than sixneighboring O. Investigations of molybdenum trioxide's catalytic anisotropieshave been based on the hypothesis that this difference in coordination numberwill lead to differences in the catalytic properties of the different facets. Thishypothesis has been experimentally tested in the following way. Chemicallyidentical particulate samples were prepared in different ways so that theyexhibited different characteristic particle habits and aspect ratios. Under theassumption that all of the particles in a given sample have approximately thesame shapes, the fractional contribution that each type of facet makes to the totalsurface area of each specimen was determined by microscopic inspection. Therole of each facet in a given reaction can then be inferred by comparing theproperties of specimens with different distributions of external surfaces.
Based on such studies, it has been concluded that the lateral facets are active for the oxidation of methane to formaldahyde [10] and the oxidativeammonalysis of toluene [7], while the (010) facet is active for the conversion ofmethanol to formalahyde [1]. Studies of the oxidation of propene to acroleinillustrate that it is not always easy to relate overall activities or selectivities to thepresence of a single face [3, 5, 8, 9]. Since the overall reaction is composed ofseveral elementary steps, it is possible that different steps occur on differentfacets. For example, it has been proposed that the mechanism for the oxidationof propene to acrolein begins with the activation to an allyl intermediate on alateral facet and ends with the addition of O on a basal facet [5]. The (210)facet, which is thought to consist of terraces with (010) character and ledges with(100) character, should be able to perform both elementary steps. Thisexplanation has been used to rationalize the observation that the (210) surface isespecially active for the conversion of propene of acrolein [9]. Using similar experimental techniques, the functionality of specific facets in other layeredmaterials, including vanadium pentoxide [12] and vanadyl pyrophosphate [13],has been identified.
While the knowledge gained from these studies is valuable, a number of limitations should be mentioned. First, the conclusions rest mainly on theobserved macroscopic shape of the crystals and little is known about themorphology and structure of the individual facets on the particles. Furthermore,the surface morphology of the catalyst is expected to change during reactions, asstoichiometry compensating defects are continuously created and annihilated.
Finally, it should be noted that conclusions from studies of particulate materialsare not always consistent with observations from other experiments. Forexample, by comparing the properties of different particulate samples, Tatibouëtand Germain [2] identified the (010) facet of MoO as being responsible for the conversion of methanol to formaldehyde while Farneth et al. [14] concluded, onthe basis of temperature programmed desorption studies, that this facet waslargely inactive for the partial oxidation of methanol. We shall see in section 5that such discrepancies might be explained by morphological changes that occuron the surface of the oxide while in service.
2.2 Single crystal surfaces
Perhaps the most convincing demonstrations of structure sensitivity come from highly controlled single crystal experiments in ultrahigh vacuum [15-18].
For example, Vohs and Barteau [15] compared the ability of the ZnO(0001) andZnO(000 1) to abstract protons from alkynes. While the alkynes (C H , CH CCH, C H CCH) decomposed on the (0001) Zn-terminated surface, the (000 1) O-terminated surface was observed to be inactive for dehydrogenation.
Barteau's [16-18] group conducted parallel studies on TiO surfaces in UHV.
Flat TiO (001) orientated crystals can be induced to facet into nonplanar configurations terminated by {101} or {114} facets, depending on the thermaltreatment. This phenomenon was exploited to show that when either methanol[16] or acetic acid [17] was reacted with TiO , the distribution of products was governed by whether the crystal was terminated by {001}, {101}, and {114}surfaces. While the effect of orientation was convoluted with the level ofsurface oxidation, it was still possible to draw some conclusions about theproperty-character relationships for these surfaces. For example, the four-coordinate Ti4+ cations found on the {114} faceted surface are capable ofcoupling carbon containing species to form higher order products (for example,dimethyl ether from methanol and acetone from acetic acid) while the five-coordinate cations on the {101} faceted surface initiate only dehydration anddehydrogenation reactions. The rationale for this difference is that the lowercoordinate cations on the {114} surface are able to adsorb both of the necessary reactants to a single site while the cations on the {101} surface can adsorb onlyone [18].
2.3 Thin films
As noted briefly in the introductory section of this chapter, specifying only the orientation of a surface can sometimes lead to an ambiguity with respectto the chemical composition of the termination layer. Consider, for example, acrystal with the perovskite structure in the (100) orientation (see Fig. 3). Thestructure can be terminated either on the AO plane or on the BO plane. The composition of the termination layer has been studied most extensively forSrTiO , which is commonly used as a substrate for the growth of superconducting cuprate phases such as YBa Cu O . SrTiO (100) substrates annealed at about 1000 °C exhibit two characteristic types of terraces. Terraceswith smoothly curved edges are associated with SrO terminated surface; TiO2terminated terraces have polygonized edges made of <010> and <001> stepsegments [19]. It has been demonstrated that it is possible to prepare surfacesthat have a uniform termination layer. TiO -terminated (100) surfaces can be prepared by chemical etching [20, 21] and SrO surfaces can be prepared bydepositing a monolayer thick film of SrO [22, 23] or Sr TiO [24]. It has been observed that the SrO and TiO surfaces stabilize different polytypes of SrCuO during thin film deposition [24] and that YBa Cu O films are less susceptible to decomposition when deposited on SrO terminated substrates [23].
While SrTiO is the most well-studied example, the influence of termination layer on film growth has been demonstrated to occur in othersystems as well. For example, [0001] oriented sapphire can be terminated bythree distinct planes. One is oxygen terminated and the other two are Alterminated. Bench et al. [25] have observed that sapphire (0001) surfaces heated Fig. 3. Schematic illustration of the perovskite structure. Note that surfaces perpendicularto (001) can be terminated by and AO or a BO layer.
Fig. 4. Scanning electron microscope image of Cu O deposits on the Al O (0001) surface.
The different particle sizes and morphologies correlate to different surface compositions[25].
for 6 h in air at 1400 °C exhibit a combination of these possible terminations ona sample with a constant average orientation. When Cu O islands were deposited on these surfaces, two distinct island sizes and morphologies wereobserved that correlated to differences in the terrace termination (see Fig. 4).
When TiO was deposited on these surfaces, one of the terrace types promoted the growth of an unanticipated (and unidentified) phase that was suspected to bea reduced form of titania or an Al-Ti-O ternary phase.
The results cited above were selected to clearly illustrate the influence that the character of a surface has on its properties. To be more specific, the abilityof an oxide surface to catalyze specific reactions or nucleate specific phases isknown to depend on both the surface orientation and the termination layer. Wenow turn to the question of what determines the surface character and whatcharacters are reasonably attainable.
3. FACTORS DETERMINING SURFACE MORPHOLOGY
In equilibrium, the morphology of a surface is determined by the anisotropy ofthe surface energy. While the surface morphologies that we observe in mostpractical situations are not in global equilibrium, local equilibrium can beachieved at the intersection of two facets or at the point were a grain boundaryintersects the surface. Observations of surfaces in local equilibrium give us the opportunity to determine the anisotropy of the surface energy. Conversely,knowledge of the surface energy anisotropy allows us to specify possiblemorphologies. The two important conclusions from this section are that becauseof anisotropy, it is usually not possible to realize all surface orientations and thatexperimental observations can be used to determine the range of allowableorientations.
The equilibrium shape of a free particle is determined by the surface energy, which for crystalline solids is usually anisotropic. The microscopicorigin of this anisotropy can be understood if one considers that the workrequired to create a new surface should (ignoring relaxations or reconstruction)increase with the number of pair-wise interatomic bonds that must be broken.
Those surfaces with a higher density of broken bonds will have a higher relativeenergy. Therefore, we can assume the surface energy is a function of the surfacenormal, γ ( ˆn), and for any element of the crystal surface, dA, the surface energyis γ ( ˆn)dA, where ˆn is normal to the element dA. For a single crystal with fixedvolume in thermodynamic equilibrium, the shape is that which minimizes thetotal surface energy, E [26]: For an isotropic material, the equilibrium shape is a sphere. For an anisotropicmaterial, the equilibrium shape will depart from sphericity so thatcrystallographic orientations with relatively low values of γ will be exposed.
Based on Eq. 1, it has been demonstrated that for a crystal in equilibrium, theratio of the energies of two orientations (γ and γ ) is equal to the ratio of the perpendicular distances from the center of the crystal to the facet (l and l ) [27]: In other words, the lower the surface energy of the facet, the closer the facet is tothe center of the crystal and the greater the surface area of the facet. This is thebasis of the Wulff [28] construction, which specifies the relationship between thesurface energy of a crystal and its equilibrium shape. To apply the Wulffconstruction, we begin by imagining that a normal vector represents each distinctcrystallographic orientation and the magnitude of each normal vector is equal tothe surface energy of the orientation that it represents. Further, we assume thatthe vectors share a common origin. If a perpendicular plane is situated at the endof each vector, then the inner envelope of the planes defines the equilibriumcrystal shape.
The Wulff construction is important because it allows the relative energies of different observed surfaces can be determined by direct microscopicmeasurement of crystals (or voids within crystals) with an equilibrium shape.
By measuring the distances from the center of a crystal shape to the variousfacets (li), the relative surface energies (γi) can be computed using Eq. 2. Whilethis method has been applied to many crystals, applications to oxides have beenless frequent [29-31]. Most recently, two groups reported results for Al O (see Table 1) [32, 33]. Underpinning such studies is the assumption that the particle(or small cavity) is in equilibrium. To reach this state, it is necessary to havesufficient mass transport and surface attachment kinetics and, in the case offaceted particles, step producing defects that reduce the barrier for the nucleationof new layers [34].
Table 1.
Experimentally measured surface energies for Al O That the data in Table 1 are not in agreement illustrates the difficulty of making such measurements. The constraint of global equilibrium is particularlychallenging. An alternative method for experimentally assessing the anisotropyof the surface energy has been developed based on the assumption of localequilibrium between free surfaces and grain boundaries at thermal groovesformed by surface diffusion [35]. The results of this measurement are illustratedin Fig. 5. While the observed anisotropy is not large, the effects on surfacemorphology are significant. Furthermore, it should be pointed out that there aregeometric limits to observable anisotropies for nonperpendicular surfaces inequilibrium experiments. For example, if the energy of the (110) surface of acubic crystal were more than √2 times the energy of the (100) surface, then thesurface would lower its energy by faceting into (100) and (010) surfaces thatwould meet along lines in the [001] direction and form ridges. Therefore, theenergy of the (110) orientation has an upper limit of √2γ . Similarly, if the energy of the (111) surface were more than √3 times the energy of the (100)surface, it would lower its energy by faceting into trigonal pyramids bound by(100), (010), and (001) facets. In fact, at lower temperatures where theanisotropy of the surface energy is expected to be higher, the MgO(111) surfacehas been observed to break up into (100) facets [36]. Calculations for MgO angle, °
Fig. 5. Plot of the relative surface energy of MgO in air at 1400 °C, around the perimeter ofthe unit triangle, from (100) to (111), then to (110), and back to (100).
suggest an anisotropy so high that only {100} facets should be observed [37-40].
While this is consistent with relatively low temperature vacuum experiments, itis not consistent with higher temperature observations in air [35, 41].
In the present context, where our interest is in controlling surface character, it is important to note that surface energy anisotropy frequently leadsto missing crystallographic orientations. For example, consider the two- n 1
n c
n
n
n 2
n c
n 2
Fig. 6. (a) A hypothetical two-dimensional equilibrium crystal shape. A continuous rangeof orientations exists between ˆn1 and ˆnc, the complex facet. Between ˆnc and ˆn2 , there is arange of missing orientations ( ˆnm is an example). (b) The surface of a crystal whoseorientation is macroscopically constrained to a missing orientation will break up into facetswith orientations ˆnc and ˆn2 .
dimensional Wulff shape illustrated in Fig. 6. The surfaces of constrainedcrystals with orientations between ˆ n will be flat, as this will minimize the total surface energy. However, surfaces with orientations between ˆ will be unstable with respect to faceting and in equilibrium, the surface will becovered with tent-shaped ridges bounded by surfaces with orientations ˆ n . The relative proportion of the two surfaces will depend on how far the orientation is from either of the stable orientations. If it is very close to ˆ the surface will be composed of large terraces with small steps whoseorientations are ˆ n . On the other hand, if the orientation is half way between the two stable surfaces, the surface will be composed of equal parts of the twoorientations.
The appearance of facets on surfaces with unstable orientations is one common manifestation of anisotropy. Surfaces with macroscopic orientationsthat are missing from the equilibrium shape can be prepared at low temperatureby polishing. When heated, such surfaces form microscopic facets whilemaintaining their global orientation. The ranges of stable and unstableorientations can be mapped out by examining the surfaces of individual grains ina polycrystalline ceramic that has been polished and then annealed at hightemperature. Data for the orientation stability of MgO at 1400 °C in air areshown in Fig. 7. The orientations of more than 100 grain surfaces weredetermined by electron backscattered diffraction. AFM images of the samesurfaces were used to discriminate flat from faceted surfaces (see Fig. 7).
Fig. 7. An orientation stability map for magnesia at 1400 °C. Each point corresponds to theorientation of an observed grain. Faceted orientations are marked with an x and smoothorientations are marked with an empty square. An AFM image of a typical faceted surface Because of the relatively high energy of the polar (111) surface, this orientationand orientations within about 20 ° decompose into lower energy complex facets[35].
The relative stability of the different orientations is conveniently represented on a stereographic projection, as proposed by Cahn and Handwerker[42]. These orientation stability diagrams are composed according to an analogywith isothermal sections of ternary phase diagrams. Ranges of orientations thatare flat (or form part of a continuously curved surface on a particle) arerepresented as a shaded region and are analogous to a solid solution.
Orientations that break up into two separate surfaces are placed on tie lines thatconnect the two stable facets, in analogy with two-phase regions; orientationsthat break up into three facets make up empty polygons whose vertices are stable Fig. 8. (a) Orientation stability diagram for SrTiO at 1200 °C, in air. There are a range of stable orientations near {100}. The {110} (black circles) and {111} (white triangles)orientations are also stable. Typical AFM images of surfaces from (b) a stable region, (c) atwo facet region, and (d) a three facet region are illustrated below. Each shows a 5µm x5µm topographic image.
orientations. As an example, Fig. 8 illustrates an orientation stability diagramfor SrTiO annealed at 1200 °C in air for 6 h. The observations indicate that the {100}, {110}, and {111} surfaces are stable under these conditions. In addition,there is an extended range of stable surfaces vicinal to {100}. We can thereforesay that the stable shape of SrTiO (after heating for 6 h in air at 1200 °C) has continuously curved surfaces at each of the cube faces which meet flat {110}and {111} surfaces at sharp edges. In practice, it is much easier to deriveinformation on the equilibrium crystal shape from observations of orientationstability than from small crystals or cavities. This is because orientation stabilitymeasurements require only local equilibrium at the points where facets meet,whereas global equilibrium is required for equilibrium crystal shapemeasurements. However, because only a portion of the equilibrium form isobserved at any time, the relative energies of different continuously curvedregions of the equilibrium crystal shape can not be determined.
It is, however, possible to estimate the relative energies of any curved and flat regions of the equilibrium form that intersect [43]. Consider the examplediscussed above where and orientation ˆ n breaks up into to other orientations, n . Under the assumption of local equilibrium at the intersection n , then the Herring [44] equation gives the following relation: θ 1 ∂γ c γ c ∂θ If the anisotropy is small enough that the differential term can be ignored, thenthe ratio of the surface energies is simply related to the cosine of the angle atwhich they meet. Note that one criterion for the application of Eq. 3 is that thefacets are in local equilibrium. For this to be true, there must be no net growthor evaporation. If this is true, then thermal grooves at surface-grain boundaryintersections should maintain a quasistatic profile and increase in width with theone quarter power of time [45].
4. FACTORS INFLUENCING SURFACE STOICHIOMETRY
The factors influencing the surface composition of pure materials are describedin this section. While the factors influencing surface morphology and surfacecomposition are treated in different sections, it should be emphasized that theyare closely related. We begin by considering the solid-vapor equilibrium andhow this influences surface and bulk stoichiometry. The equilibrium defectstructure, like the equilibrium morphological structure, might not always beobtained. Regardless, it remains useful to know the destination state of a surfacewhose composition is evolving at elevated temperature. The section concludes with a brief consideration of the composition of oxide surfaces produced at lowtemperature by cleavage.
4.1 The solid-vapor equilibrium
We begin by noting that the bulk stoichiometry of an oxide is a function of the temperature and composition of the surrounding atmosphere [46]. For ahypothetical metal oxide (MO), we can write the appropriate defect chemicalreactions in Kröger-Vink notation and equilibrium constants for congruentevaporation (k ), the production of vacancies by evaporation (k and k ), and Shottkey defect formation (k ): p p 1/2 = k p [V *] = k p 1/2[V *] = k For the purpose of simplicity, the vacancies are taken to be neutral and we do notconsider the Frenkel or anti-Frenkel processes. In both cases, extending theanalysis does not alter the important parts of the results. From Eqs. 5-7, it can beshown that: [V *] = k p -1/2 [V *] = (k /k )p 1/2 The equations are represented schematically in Fig. 9. There are several important conclusions to draw from this simple analysis. First, the defectconcentration is a function both of temperature (through the equilibriumconstants) and the partial pressure of oxygen. Second, there is a singlecomposition of the gas phase where the compound is stoichiometric.
Furthermore, the range of accessible compositions is limited by either theappearance of new phases or by experimental limitations. For example, theranges of stoichiometry available to Fe O are limited at high and low partial pressures of oxygen by the formation of Fe O and FeO, respectively. For TiO , the appearance of the Magnéli phases at low partial pressures of O limit the maximum concentration of oxygen vacancies at 1000°C to x = 0.008 in TiO2-x[47]; the upper limit of oxygen pressures is determined only by the strength ofthe vessel that contains the experiment.
Fig. 9. Schematic graph illustrating how the composition of the surrounding atmosphereinfluences the stoichiometry of a metal oxide.
If we take the total pressure to be p + p , we can use Eqs. 4-6 to show that the total pressure is: p = p + k By minimizing Eq. 10 with respect to p , it can be shown that there is a minimum partial pressure of oxygen at which the compound can be inequilibrium. In this case: As long as the partial pressure of O is greater than p , the stoichiometry of the compound will adjust to equilibrate with its surroundings. Below thisminimum, some of the crystal will have to sublime to establish the minimumvapor pressure. In a dynamic vacuum, where subliming material is constantlypumped away, this is clearly not possible. Under such conditions, thestoichiometry of the compound is determined by kinetic factors and it might notbe possible to conduct experiments with reproducible stoichiometries.
While the preceding statements apply to the bulk of the crystal, the surface of the crystal is subject to the same constraints with some additional complexity.
Specifically, because different point defects are likely to have different freeenergies of formation, there is an enhanced concentration of one of the twodefects in the near surface region and a compensating layer of charge on the surface [48, 49]. Suppose, for example, that it is easier to create O vacanciesthan M vacancies. The enhanced concentration of O vacancies that would resultfrom this situation would lead to a net positive charge in the near surface regionthat would be have to be compensated by negative charge on the surface. It hasbeen pointed out that the amount of charge that it is possible to store in thesubsurface region will be limited by the capacity of the surface to accommodatethe compensating charge [50]. As with the bulk situation, equilibrium surfacedefect concentrations will be established only at a fixed and sufficiently highpartial pressure of oxygen.
One of the more well-documented cases that illustrates the effect of the annealing atmosphere on the surface composition concerns the Al O (0001) surface [51]. After heating in air, 1 X 1 LEED patterns are observed. Hightemperature annealing (1350 °C) in UHV leads to the loss of O and theformation of an Al terminated surface that has a (√31 X √31)R±9° LEEDpattern. The model for this reconstruction is two pure and nearly perfect Alplanes on an O terminated Al O (0001) surface. The same structure can be produced by depositing Al metal from the vapor phase onto an Al O (0001) surface with the 1 X 1 structure [52, 53]. Therefore, and alternate way to viewthis reconstruction is that the Al O (0001) surface is decomposing during high temperature annealing in a low partial pressure of oxygen atmosphere to formAl, which adopts a low energy epitaxial structure. Similarly, Fe O (0001) surfaces are found to be enriched in metal and transform to Fe O (111) and FeO(111)-like reconstructions after being ion bombarded, annealed in UHV, andthen heated in 1 x 10-6 mbar of O [54, 55]. As in the case of the Al O (0001) surface, the observed restructuring of the surface is consistent with O loss in alow p environment. One certain implication of the solid-vapor equilibrium is that when an oxide is ion bombarded or heated in UHV, the resulting bulkstoichiometry is neither well defined nor likely to be reproducible. This is alikely explanation for the diverse range of observations that have been reportedfor the TiO (110) surface which include not only a (1 X 1) structure [56-62], but also structures related to Magnéli phases [63,64], and (1x2) [56,57,59,61], (1x3)[61], (2x2) [59], c(2x1) [56], and (2x3) [65] structures. Despite the difficulty ofproducing oxides with reproducible bulk stoichiometries in vacuum, it must beemphasized that kinetically stable surfaces with reproducible structures havebeen prepared in the UHV environment. The creation of the TiO (110) (1x1) surface in UHV is an excellent example [62]. Not surprisingly, STM images ofthe surface illustrate that it has a remarkably high concentration of oxygenvacancies.
Surfaces prepared by cleavage at ambient temperature, if not annealed or ion bombarded in vacuum, are likely to have relatively well defined andreproducible stoichiometries. While not generally in equilibrium, they should bekinetically stable at room temperature. However, the composition of such a surface is not always easy to specify. In very simple situations, cleavage canproduce two identical surfaces and there is no ambiguity about the surfacetermination. In other cases, cleavage leads to two distinct surfaces. It is alsopossible that there are multiple possible cleavage planes. Consider, for example,the V O structure shown in Fig. 10 [66]. Cleavage at the points labeled A, B, or C lead to distinct terminations. The analysis of STM and AFM images of thesurface and comparison to calculated electron density maps are consistent withthe conclusion that cleavage occurs at C. Based on studies of tungstates,molybdates, and vanadates, it is possible to make the general statement thatoxides cleave at the longest, weakest set of bonds perpendicular to the cleavageplane [66-70].
Fig. 10. (a) Idealized polyhedral projection of the V O structure, showing the three possible planes of cleavage. (b) An STM image of the plane, identified as C.
5. REACTION ANISOTROPIES ON MoO SURFACES
In section 2 of this chapter, surface character-property relationships for thecatalytic properties of MoO were described. These relationships were deduced largely from experiments on particulate materials. In this section, AFMexperiments are described that were designed to discriminate the functions of thebasal and lateral facets by direct observation. Experiments were conducted onsingle crystals surfaces with a clearly defined orientation and composition thatwere treated in a reactor system capable of reproducing the range oftemperatures and environmental conditions relevant to the oxidation of alcohols[71-75]. Under these conditions, both the morphology and the stoichiometry ofthe surface evolved in a complex way. Room temperature AFM imaging at theconclusion of the treatment was used to determine how the surface structurechanged.
The AFM images in Fig. 11 show a variety MoO (010) surfaces before and after reactions with methanol [73, 74]. The images in Fig. 11a and b showthe surface structure before the reaction. Cleavage surfaces, such as thoseillustrated in Fig. 11a, are characterized by atomically flat terraces separated bysteps that have heights that an integer multiple of about 7 Å. This characteristicheight is half the unit cell repeat length along [010] and corresponds to thedistance between the van der Waals gaps which separate the adjacent layers ofthe structure (see Fig. 2). The steps on cleavage surfaces are most frequentlyaligned along [001]. The configuration of steps on a growth surface (see Fig.
11b) is determined by the positions of the screw dislocations. Defects such asthe one near the center of the image in Fig. 11b provide a continuous source ofsteps that facilitates crystal growth. In the reactivity studies described below,there was no detectable difference in properties of the growth surfaces and thecleavage surfaces.
When MoO (010) surfaces are reacted with air-N -MeOH mixtures at 300 ˚C, loops of steps surrounding shallow pits nucleate on the previously flatterraces (see Fig. 11c-e). The pits in these images are always bounded by halfunit steps and are therefore only 7 Å deep. While the pitting process wasobserved in all of the air-N -MeOH mixtures tested (air:N ratios between 100:0 and 02:98), the shapes of the pits and their evolution depended on thecomposition of the mixture. Under oxygen rich conditions, the pits had arectangular or ovular shape, and were always elongated along <001>. As the airconcentration is reduced, the pits elongate along <100> and assume a triangularhabit (see Fig. 11e). The pits are only formed when some methanol is includedin the feed. When air is completely eliminated from the feed, H MoO precipitates (the white contrast in Fig. 11f). These precipitates are formed whenthe hydrogen liberated during the chemisorption of the alcohol reacts with theMoO to form the molybenum bronze, rather than reacting with lattice O to form water [73]. The same reaction occurs when MeOH is replaced by higher orderalcohols.
Fig. 11. AFM images of the MoO (010) surface. The arrows show the direction of the [001] axis. A cleavage (a) and growth (b) surface before the reaction. (c) After 2 h at 300°C in air saturated with methanol. (d) After 2 h at 300 °C in a 50:50 air : N mixture saturated with methanol. (e) After 2 h at 300 °C in a 30:70 air : N mixture saturated with methanol. (f) After 6 min at 330 °C in N saturated with methanol.

The most plausible explanation for the formation of the pits during the reactions with the methanol-air mixtures is that the surface methoxide speciesthat form during the initial dissociative chemisorption of methanol are capable ofdesorbing as molybdenum-oxide-methoxide molecules, which are then carriedaway in the flowing gas. The observation that crystalline Mo O (OCH ) forms at the reactor exhaust supports this mechanism for evaportation [74]. Theseobservations, made far from equilibrium, illustrate that the morphologicalstructure of the MoO surface in service is continuously evolving and depends on the composition of the gas feed. Of particular interest is the generation ofstep edges on the surface. As discussed in section 2, the step edges provideundercoordinated Mo sites (similar to those found on the lateral facets of the Fig. 12. AFM images of defect related pits that form on the MoO (010) surface. In each case, the arrow indicates the [001] direction. (a) After heating for 3.5 h in air at400 °C, shallow pits form at surface-dislocation intersections. (b) After 8 min at 400°C in a 90:10 N :H mixture containing 20 ppm water. (c) After 5 min at 400 °C in a 90:10 N :H mixture containing 20 ppm water. (d) After 8 min at 400 °C in a 90:10 N :H mixture containing 20 ppm water, followed by a 3.5 h anneal in air at the same crystal) that are thought to have a different function than the basal sites.
It has been observed that the elastic strain associated with extended crystal defects also promotes the formation of pits in the (010) surface when heated toapproximately 400 °C in the presence of water vapor. It is therefore possible togenerate controlled distributions of undercoordinated Mo to test their relativeimportance in the reaction with methanol. There are two primary crystal defectsthat create enough strain to nucleate a pit during reactions with water vapor:dislocations and crystallographic shear planes. Several examples of pits that canbe formed at the sites where defects intersect the surface are illustrated in Fig.
12. Water vapor is known to enhance the sublimation of MoO through the formation of volatile molybdenum oxyhydroxides such as MoO (OH) and is thought to play a similar role in the formation of pits near the points wheredefects intersect the surface [75].
For crystals heated to 400 °C in oxidizing atmospheres containing water vapor, pits form only at the points where screw dislocations intersect the surface.
The image in Fig. 12a shows decending spiral steps into the etch pit formedaround a dislcoation with a Burgers vector of approximately 14 Å, which isequal to the lattice spacing along [010]. By examing dislocations with differentBurgers vectors, it was concluded that the amount of material removed in a fixedtime period was proportional to the stored elastic strain around the defect.
Because the average dislocation density was only 1.5 x 104/cm2, such pits arerather rare. However, in reducing atmospheres, the wide spread formation ofcrystallographic shear planes (defects that accommodate oxygen loss) leads topits densities up to four orders of magnitude greater [71]. A surface with a highdensity of small pits (black contrast) is shown in Fig. 12b. Higher resolutionimages show that 1.5 Å high steps that are characteristic of surface/CS planeintersections invariably intesect the pits (see Fig. 12c). When a surface such asthe one shown in Fig. 12b is reheated in an oxidizing atmosphere, the CS planesare eventually annihilated and and the remaining pits reorient to form aciculartrenches along the <001> direction, as illustrated in Fig. 12d.
Because pit formation is activated by elastic strain from extended defects (which can be deliberately introduced), we can exert some degree of control overthe size and density of the pits and, therefore, the available population ofundercoordinated Mo. To test the influence that these sites have on thechemisorption of alcohols, a series of model surfaces were created with differentdensities of undercoordinated Mo associated with pits and then reacted withmethanol saturated N at 330 °C. These conditions lead to the formation of the H MoO phase, which was used as a quantitative and local indicator of the dissociative chemisorption reaction. The images in Fig. 13 illustrate selectedresults from these experiments. In Fig. 13a and b, it is clear that the acicularH MoO precipitates are growing from the lateral walls of the surface pits [75].
The images in Fig. 13c and d show that when the density of the pits increases and the distance between adjacent pits is small compared to the size ofthe precipitates, the preferential nucleation is not obvious. However, the totalamount of the hydrogen bronze phase produced still correlates with the total areaof the pits. Both crystals in Fig. 13 c and d were reduced for 5 min at 400 ˚C informing gas to form the pits. One was then oxidized in ambient air at 410 ˚C for1 h so that the pits could grow, while the other was not. As a results, the pitsshown in Fig. 13c are all less than 200 Å deep while those on the surface thatreceived the oxidation treatment are all greater than 500 Å deep (see Fig. 13d).
Fig. 13. AFM topographs of MoO3(010) surfaces. (a) A crystal that was reduced for 10 minat 400˚C in forming gas and then oxidized for 2 hr at 400˚C in the ambient prior to beingreacted for 20 min at 330 ˚C in N2 saturated with MeOH. The black-to-white contrast is100 Å. (b) A surface that was reduced in forming gas at 400˚C for 10 min, oxidized for 3 hrat 410˚C in air, and then exposed to N2 saturated with MeOH for 15 min at 330 ˚C. Theblack-to-white contrast in the image is 100 Å. (c & d) comparison of precipitation onsurfaces with different pit sizes (see text). Each was reacted for 10 min at 330 ˚C with N2saturated with MeOH. The black-to-white contrast in (c) is 200 Å, while that in (d) is 400Å.
Both surfaces were then reacted for 10 min at 330 ˚C in N saturated with MeOH at room temperature; the precipitates on the surface with the larger pits (Fig.
13d) are larger and more numerous than those on the surface with more shallowpits (Fig. 13c). When the total precipitate volume on each surface is estimated,the results indicate that the surface with the larger pits produced four times asmuch of the H MoO phase. It should be pointed out that the stoichiometry of these two surfaces was not the same. The one that was heated in air was moreoxidized than the other and, therefore, had fewer surface oxygen vacancies.
Since surface oxygen vacancies are also sources of undercoordinated Mo, onemust consider the possibility that they compete with the lateral facets aschemisorption sites. However, when the upper limit of the oxygen vacancyconcentration (10-4) is compared to an estimate of the number ofundercoordinated Mo associated with the lateral walls of pits, one finds that thesurface oxygen vacancy concentration is not a significant factor [75]. This isconsistent with the observations that the sample with fewer oxygen vacancies(but deeper pits) formed more of the hydrogen bronze phase.
In summary, the results presented in this section indicate that at the temperatures and pressures relevant for partial oxidation reactions, it is the{h0k} surfaces that are active for the dissociative chemisorption of alcohols.
Furthermore, the morphological structure of the catalyst particles is continuouslyevolving during the reactions, introducing a higher than anticipated density ofundercoordinated Mo on {h0k} surfaces while the catalyst is in service. Whileproposed mechanisms for partial oxidation usually concentrate on the exchangeof O between the catalyst and the gas phase, the current results suggest that theexchange of H and Mo may also be important.
5. THE ORIENTATION DEPENDENCE OF THE PHOTOCHEMICAL
REACTIVITY OF TiO2
The photochemical properties of titania surfaces are of interest for severalreasons. They determine the stability of pigmented paint systems [76], the rateat which pollutants can be degraded in systems designed to purify air and water[77], and are the root cause of poorly understood phenomena such as waterphotolysis [78] and "super hydrophilicity" [79]. Using thin rutile epilayers withfive low index orientations, it has been shown that the relative rates ofphotochemical reactions catalyzed by titania depend on the surface orientation[80]. In this chapter, experiments used to map the complete orientationdependence of the relative photochemical reactivity of TiO are described [81- 83]. In this case, the relevant reactions are carried out at room temperature andthis gives us the opportunity to fix both the surface morphological structure andstoichiometry. For the reactions described here, all of the surfaces were annealed in air for several hours at 1200 °C to define the stoichiometry andmorphological structure.
The challenge associated with measuring the photochemical properties of titania as a function of orientation lies in the number of required measurements.
Each possible surface orientation is defined by two independent variables. Ifeach orientation parameter can be measured to within ± 3°, then for a tetragonalmaterial such as rutile, there are 436 distinct surface orientations. It is clear thatthe preparation and analysis of oriented single crystal surfaces would beprohibitively tedious. This challenge can be overcome by using polycrystallinespecimens. Assuming a 50 micron grain size, a 5 mm2 area of a polished sectionthrough a random microstructure reveals approximately 2000 grain surfaces.
Single grain orientations were determined based on electron backscattereddiffraction patterns (EBSPs) obtained in a scanning electron microscope (SEM).
For unfaceted grains, the index of the surface plane can be determined directlyfrom the EBSP. Faceted grains, however, are terminated by more than onesurface plane as descibrd in section 3. In such cases, atomic force microscopy(AFM) was used to measure the inclination of each facet with respect to thesurface normal; when these data are combined with knowledge of the grainorientation from the EBSPs, the facet planes can be indexed.
To make a local measurement of the photochemical activity of each grain, we used a well established probe reaction (the reduction of aqueous Ag+ to Ag0)that deposits metallic silver on the surface as a reaction product [84-85]. Theamount of silver deposited on each grain's surface during a given reaction, whichis determined from atomic force microscopy (AFM) images, is taken to be aquantitative indicator of the grain's relative photochemical reactivity. Thereactivity can then be correlated to surface orientation and/or the relative area ofeach facet on the surface.
The AFM images in Fig. 14 illustrate how this technique is used to determine the anisotropy of the reactivity [81]. The rutile polycrystal used inthis experiment was polished and then annealed in air at 1473 K for four hours toheal polishing damage. Note that grain boundaries and pores appear in theimages as black contrast. Figure 14a shows several contiguous grains in thespecimen, Fig. 1c shows a triple grain junction, and Fig. 14e shows the stepstructure on the surface of a single grain. The sample was immersed in aqueousAgNO3, exposed to ultraviolet (UV) light, and the same areas were located andimaged again. The results are shown in Fig. 14 (b,d,f).
Note that after the reaction, heterogeneous white contrast (corresponding to elevated regions) is apparent in the images. On the basis of energy dispersiveX-ray spectroscopy conducted in an SEM, there is a one-to-one correlationbetween the elevated regions of white contrast and high concentrations of Ag.
Fig. 1b illustrates that some surfaces appear to be almost completely coveredwith silver, while others have very few deposits. On the basis of these observations, three types of grains (illustrated in Fig. 14d) are distinguished:very reactive (upper left), moderately reactive (lower left), and relativelyunreactive (right). The very reactive grains are uniformly covered with silver sothat the underlying surface structure can no longer be distinguished. This occurswhen there are more than 4 Ag islands per µm2. The moderately reactive grains Fig. 14. AFM images of polycrystalline rutile surfaces before (a,c,e) and after (b,d,f) the photochemical deposition of silver (white contrast).
have easily distinguishable silver islands with a density between 2 and 4 per µm2, while the morphology of the TiO surface remains visible in the background. The relatively unreactive grains have an inhomogeneousdistribution of Ag and there are less than 2 silver islands per µm2. While parts ofunreactive grains in the vicinity of grain boundaries, residual polishing scratches,pores, and other defects do show deposited silver, the flat regions of the surfaceare indistinguishable before and after the reaction. It is the inhomogeneousdistribution of silver that distinguishes the relatively unreactive grains from themoderately reactive grains.
Based on knowledge of the grain orientation and surface inclinations measured by AFM, it is possible to index all of the grain surfaces. Thedependence of the reactivity on the grain and surface orientation is illustratedsimultaneously in Fig. 15, which shows an inverse pole figure that combinesphotochemical reactivity and surface orientation stability data. As discussed insection 3, the areas with the constant gray shading are regions where the surfaceorientation is stable and the grain has a flat surface. In such areas, the surfaceorientation and the grain orientation are the same. The areas filled by tie linesindicate orientations that break up into two facets. In this case, the two stablesurface facets are found at the ends of the tie lines. The white areas showorientations that break up into three stable surfaces whose orientations are foundat the vertices of the surrounding triangle.
Based on the results in Fig. 15, we can make the following statements.
First, the most reactive grains have orientations near {101}. Second, with theexception of a few outliers, the grains that have high reactivity have surface > 4 deposits/cm2 > 2 deposits/cm2 ≤ 2 deposits/cm2 Fig. 15 Inverse pole figure summarizing reactivity and orientation stability data for orientations that are unstable and break up into two or three separate facets.
Furthermore, one of the stable facets is always {101}. Further analysis indicatedthat the relative amount of silver on surfaces containing {101} facets correlateswith the fractional {101} area. Based on this result, we conclude that the {101}surface is the most reactive and this is the key microstructural feature for thedesign of high reactivity materials. This conclusion allows microstructuresoptimized for high reactivity to be specified. Planar structures should beoriented such that the surface normal breaks up into complementary {101} facets(the surface area of a faceted structure is greater than a single {101} type facet).
Particulate catalysts should have a pseudo-octahedral habit with {101} planesbounding the crystal.
The observation that the most photochemically reactive surface orientations all contain {101} facets suggests that the enhanced reactivity is notassociated specificially with the bulk crystal orientation, but is a property of thisparticular surface plane. It is likely that there are special atomic configurationson this plane that either alter the efficiency with which photogenerated carriersare trapped at the surface or the rate at which they are transferred across thesolid-liquid interface. Based on bulk geometry alone, there is nothing that setsthe {101} plane apart from less reactive surfaces. On this surface, Ti cations arecoordinated by five O, as they are on the (100) surface (which is inert). In theabsence of higher resolution microscopy results, it is not possible to say ifspecial molecular configurations are created by reconstruction or defectformation.
The surface properties of metal oxides are anisotropic and because of this, thereis a potential to control the properties of surface active materials by manipulatingthe distribution of exposed surface characters. Under equilibrium conditions, thedistribution of surface characters is determined by the anisotropy of the surfaceenergy and the solid-vapor equilibrium. While one is not necessarily limited toequilibrium configurations, materials at sufficiently high temperatures willevolve in this direction and significant morphological and stoichiometricchanges can occur. By systematically evaluating the properties of surfaces withdifferent characters, it is possible to define optimized morphological structures.
The author acknowledges the support of the US National Science Foundationunder grant DMR 0072151. The work was supported in part by the MRSECprogram of the National Science Foundation under award number DMR-0079996. Contributions to the research described in this chapter from Richard L. Smith, Jennifer B. Lowekamp, David M. Saylor, Jennifer L. Giocondi, P.A.
Morris Hotsenpiller, J.D. Bolt, and W.E. Farneth are also gratefullyacknowledged. The author thanks Prof. C.B. Carter for the permission to useFig. 4.
[1] J.E. Germain in: M. Che, G.C. Bond, (Eds.), Adsorption and Catalysis on Oxide Surfaces,Elsevier Science Publishers, B.V., Ansterdam, 1985, p. 355.
[2] J.M. Tatibouët, J.E. Germain, J. Catal., 72 (1981) 375.
[3] J.C. Volta, J.M. Tatibouët, J. Catal., 93 (1985) 467.
[4] J.M. Tatibouët, Ch. Phichitkul, J.E. Germain, J. Catal., 99 (1986) 231.
[5] K. Brückman, R. Grabowski, J. Haber, A. Mazurkiewicz, J. Sloczynski, T. Wiltowski, J.
Catal., 104 (1987) 71.
[6] K. Brückman, J. Haber, T. Wiltowski, J. Catal., 106 (1987) 188.
[7] A. Anderssen, S. Hansen, J. Catal., 114 (1988) 332.
[8] B. Mingot, N. Floquet, O. Bertrand, M. Treilleux, J.J. Heizmann, J. Massardier, M. Abon,J. Catal. 118 (1989) 424.
[9] M. Abon, J. Massardier, B. Mingot, J.C. Volta, N. Floquet, O. Bertrand, J. Catal., 134(1992) 542.
[10] M.R. Smith, U.S. Ozkan, J. Catal., 141 (1993) 124.
[11] L. Kihlborg, Arkiv Kemi, 21 (1963) 471.
[12] M. Gasior, T. Machej, J. Catal., 83 (1983) 472.
[13] E. Bordes, Catal. Today, 6 (1992) 21.
[14] W.E. Farneth, F. Ohuchi, R.H. Staley, U. Chowdhry, A.W. Sleight, J. Phys. Chem., 89(1985) 2493.
[15] J.M. Vohs, M.A. Barteau, J. Phys. Chem., 91 (1987) 4766.
[16] K.S. Kim, M.A. Barteau, Surf. Sci., 223 (1989) 13.
[17] K.S. Kim, M.A. Barteau, J. Catal., 125 (1990) 353.
[18] M.A. Barteau, J. Vac. Sci. Technol. A, 11 (1993) 2162.
[19] J. Fompeyrine, R. Berger, H.P. Lang, J. Perret, E. Mächler, Ch. Gerber, J.-P. Locquet,Appl. Phys. Lett., 72 (1998) 1697.
[20] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara, O. Ishiyama, T.
Yonezawa, M. Yoshimoto, H. Koinuma, Science, 226 (1994) 1540.
[21] G. Koster, B.L. Kropman, G.J.H.M. Rijnders, D.H.A. Blank, H. Rogalla, Appl. Phys.
Lett., 73 (1998) 2920.
[22] H. Tanaka, H. Tabata, T. Kawai, Thin Solid Films, 342 (1999) 4.
[23] M. Kawasaki, O. Ohtomo, R. Tsuchiya, J. Nishino, K. Koinuma, Mat. Res. Soc. Symp.
Proc., 474 (1997) 303.
[24] P.A. Salvador, B. Mercey, O. Perez, A.M. Haghiri-Gosnet, T.-D. Doan, B. Raveau, Mat.
Res. Soc. Symp. Proc., 587 (2000).
[25] M.W. Bench, P.G. Kotula. C.B. Carter, Surf. Sci., 391 (1997) 183.
[26] C. Herring, in: R. Gomer C.S. Smith. (Eds.), Structure and Properties of Solid Surfaces,The University of Chicago Press, Chicago, 1952, p. 5.
[27] P. Curie, Bull. Soc. Min. de France, 8 (1885) 145.
[28] G. Wulff, Z. Krist., 34 (1901) 449.
[29] R.S. Nelson, D.J. Mazey, R.S. Barnes, Philos. Mag., 11 (1965) 91.
[30] J.C. Heyraud, J.J. Métois, Acta Met., 28 (1980) 1789.
[31] Z.Y. Wang, M.P. Harmer, Y.T. Chou, J. Am. Ceram. Soc., 69 (1986) 735.
[32] J.-H. Choi, D.-Y. Kim, B.J. Hockey, S.M. Wiederhorn, C.A. Handwerker, J.E. Blendell,W.C. Carter, A.R. Roosen, J. Amer. Ceram. Soc., 80 (1997) 62.
[33] M. Kitayama, A.M. Glaeser, Key Engineering Materials, 159-160 (1999) 193.
[34] W.W. Mullins, G.S. Rohrer, J. Amer. Ceram. Soc., 83 (2000) 214.
[35] D.M. Saylor, D.E. Mason, G.S. Rohrer, J. Amer. Ceram. Soc., 83 (2000) 1226.
[36] V.E. Henrich, Surface Science, 57 (1976) 385.
[37] P.W. Tasker, D.M Duffy, Surf. Sci., 137 (1984) 91.
[38] M. Causa, R. Dovesi, C. Pisani, C. Roetti, Surf. Sci., 175 (1986) 551.
[39] W.C. Mackrodt, Phys. Chem. Minerals, 15 (1988) 228.
[40] J. Goniakowski, C. Noguera, Surf. Sci., 319 (1994) 68.
[41] M. Gajdardziska-Josifovska, P.A. Crozier, J.M. Cowley, Surf. Sci. Lett., 248 (1991)L259.
[42] J.W. Cahn, C.A. Handwerker, Materials Science and Engineering A,162 (1993) 83.
[43] A.J.W. Moore, Acta Met., 6 (1958) 293.
[44] C. Herring in: W.E. Kingston (Editor), The Physics of Powder Metallurgy, McGraw Hill,New York, 1951, p. 79.
[45] W.W. Mullins, J. Appl. Phys., 28 (1957) 333.
[46] W. van Gool, Principles of Defect Chemistry of Crystalline Solids (Academic Press, NewYork, 1966) p. 44.
[47] P. Kofstad, Nonstoichiometry, Diffusion, and Electrical Conductivity in Binary MetalOxides (Robert E. Krieger Publishing Company, Malabar, FL, 1983) p. 139.
[48] J. Frenkel, Kinetic Theory of Liquids (Oxford University Press, New York, 1946) p. 36.
[49] K.L. Kliewer, J.S. Koehler, Phys. Rev., 140 (1965) A1226.
[50] R.B. Poeppel, J.M. Blakely, Surf. Sci., 15 (1969) 507.
[51] M. Gautier, G. Renaud, L.P. Van, B. Villette, M. Pollak, N. Thormat, F. Jollet, J.-P.
Duraud, J. Amer. Ceram. Soc., 77 (1994) 323.
[52] T.M. French, G.A. Somorjai, J. Phys. Chem., 74 (1970) 2489.
[53] M. Vermeersch, R. Sporken, Ph. Lambin, R Caudana, Surf. Sci., 235 (1990) 5.
[54] N.G. Condon, P.W. Murray, F.M. Leibsle, G. Thornton, A.R. Lennie, D.J. Vaughan,Surf. Sci., 310 (1994) L609.
[55] N.G. Condon, F.M. Leibsle, A.R. Lennie, P.W. Murray, D.J. Vaughan, G. Thornton,Phys. Rev. Lett., 75 (1995) 1961.
[56] M. Sander, T. Engel, Surf. Sci. Lett., 302 (1994) L263.
[57] H. Onishi, Y Iwasawa, Surf. Sci. Lett., 313 (1994) L783.
[58] A. Szabo, T. Engel, Surf. Sci., 329 (1995) 241.
[59] P.W. Murray, N.G. Condon, G. Thornton, Phys. Rev. B, 51 (1995) 10989.
[60] S. Fischer, A.W. Munz, K-D. Schierbaum, W. Göpel, Surf. Sci., 337 (1995) 17.
[61] A. Berkó, F. Solymosi, Langmuir, 12 (1996) 1257.
[62] U. Diebold, J.F. Anderson, K.-O. Ng, D. Vanderbilt, Phys. Rev. Lett., 77 (1996) 1322.
[63] G.S. Rohrer, V.E. Henrich, D.A. Bonnell, Science, 250 (1990) 1239.
[64] G.S. Rohrer, V.E. Henrich, D.A. Bonnell, Surf. Sci., 278 (1992) 146.
[65] M. Wagner, D. A. Bonnell, M. Rühle, Appl. Phys. A, 66 (1998) 1165-70.
[66] R.L. Smith, G. S. Rohrer, K.S. Lee, D.-K. Seo, M.-H. Whangbo, Surf. Sci., 367 (1996)87.
[67] R.L. Smith, W. Lu, G.S. Rohrer, Surf. Sci., 322 (1995) 293.
[68] G.S. Rohrer, W. Lu, R.L. Smith, A. Hutchinson, Surf. Sci., 292 (1993) 261.
[69] W. Lu, N. Nevins, M.L. Norton, G.S. Rohrer, Surf. Sci., 291 (1993) 395.
[70] R.L. Smith, G.S. Rohrer, J. Solid State Chem., 124 (1996) 104.
[71] R.L. Smith, G.S. Rohrer, J. Catalysis, 163 (1996) 12.
[72] R.L. Smith, G.S. Rohrer in: K.S.Ramesh, M. Misono, and P.L. Gai (Eds.), CatalystMaterials for High-Temperature Processes, American Ceramic Society, Westerville, OH,1997. p. 139.
[73] R.L. Smith, G.S. Rohrer, J. Catalysis, 173 (1998) 219.
[74] R.L. Smith, G.S. Rohrer, J. Catalysis, 180 (1998) 270.
[75] R.L. Smith, G.S. Rohrer, J. Catalysis, 184 (1999) 49.
[76] H.G. Völz, G. Kaempf, H.G. Fitzky, A. Klaeren in: S.P. Pappas, F.H. Winslow, (Eds.),Photodegradation and Photostabilization of Coatings, American Chemical Society,Washingtion D.C., 1981. p. 163.
[77] A. Wold, Chem. Mater., 5 (1993) 280.
[78] A. Fujishima, K. Honda, Nature, 238 (1972) 37.
[79] R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M.
Shimohigoshi, T. Watanabe, Nature, 388 (1997) 431.
[80] P.A. Morris Hotsenpiller, J.D. Bolt, W.E. Farneth, J.B. Lowekamp, G.S. Rohrer, J. Phys.
Chem. B, 102 (1998) 3216.
[81] J.B. Lowekamp, G.S. Rohrer, P.A. Morris Hotsenpiller, J.D. Bolt, W.E. Farneth, J. Phys.
Chem. B, 102 (1998) 7323.
[82] J.B. Lowekamp, G.S. Rohrer, P.A. Morris Hotsenpiller, J.D. Bolt, W.E. Farneth, to bepublished.
[83] J.B. Lowekamp, Ph.D. Thesis, Carnegie Mellon University, 1999.
[84] W.C. Clark, A.G. Vondjidis, J. Catalysis, 4 (1965) 691.
[85] J.-M. Herrmann, P. Pichat, J. Catalysis, 113 (1988) 72.

Source: http://www.f-16.net/forum/download/file.php?id=9259