Adsorbent amorphous oxides

An even more general and correspondingly less detailed atomic model of amorphous oxide surfaces has been called the Bernal surface (BS)[3, 21]. It is based upon the fact that many oxides and halides can be regarded as close-packed arrays of large anions with much smaller cations occupying interstitial (usually tetrahedral or octahedral) positions (see., e.g. Ref. [4]). In line with this point of view, the BS is a surface of a collection of dense randomly packed hard spheres, a sphere representing an oxide anion. The cations in interstitial positions between hard spheres are excluded from the simulation since they do not attract adsorbed molecules due to their small polarizability. Thus only the atomic structure of the oxide ions is considered. This is called the Bernal structure and has been used for modelling simple liquids and amorphous metals [15]. 

Thus, the surface of this amorphous carbon (which is a model of the surfaces of non-graphitized carbon blacks [23]) differs considerably from the surface of amorphous oxide and the main structural characteristics such as the C-C and 0-0 coordination numbers are also drastically different. Nevertheless, the adsorption properties of heterogeneous surfaces of various nongraphitized carbon blacks with respect to an inert adsorbate such as argon are not that drastically different and actually have many common features. We discuss these properties in the next section. Here we only use this fact to show that subtle structural differences of various models of amorphous oxide surfaces discussed above may be not that important for their adsorption properties in comparison to other factors such as indefiniteness of adsorption potential on oxide surfaces (see below). Because of its generality and in spite of its approximate character, the BS appears to be a convenient model for the computer simulation of adsorption on amorphous, and even more general (see Introduction) heterogeneous oxide surfaces. 

As soon as the expressions and constants in Eq. (4) are fixed one may proceed with calculation of the adsorption energy at a given point of adsorption space. However, to make such a calculation possible one has to know the positions of all the atoms of adsorbent relative to the given point. In other words, one has to know exactly the atomic structure of the adsorbent. This is what is in fact unknown for amorphous oxides. Although one can simulate the atomic structure at the surface of an amorphous oxide as described above, the reliability of the result can at present only be checked by comparison of prediction of adsorption properties with experimental data. But the calculation of adsorption properties (described below) includes, generally speaking, two unknowns the atomic structure of an adsorbent and the adsorption potential. This is the reason why the computer simulation of physical adsorption on amorphous oxides should be preceded by similar simulations on oxides with well defined crystalline structures. 

Physical adsorption on the (001) face of MgO also attracted considerable attention in recent years (see short review and references in Ref. [26]). It provides another opportunity to test methods of adsorption potential calculation which can be used later to simulate adsorption on adsorbents with less reliable atomic structure of surfaces like amorphous oxide. There is a large and rapidly changing electric field near the surface of MgO which should be much stronger than in silicalite due to small cations of Mg " " and larger ionicity of MgO in comparison to Si02. Thus calculations with polar and quadrupole molecules which were carried out on that surface (see Ref. [26] and references therein) necessarily employ methods which may useful for computer simulations on amorphous oxides. 

When one can evaluate the energy of an adsorbed molecule in an arbitrary position near the surface of amorphous oxide, one may begin to calculate the adsorption characteristics of that surface. It is simpler to do that when the concentration of adsorbed molecules at the surface is infinitely small, i.e., in the Henry s Law region. Even in this case, calculations are usually carried out only for spherically synunetric molecules like rare gases or spherical models of more complex molecules like CH4 and SFe, the only exception being Ref. [19] where SFe was represented by a six-center model. In all cases the interaction of an adsorbed molecule with the oxide was modeled by LJ-potentials, the induction interaction being implicitly taken into account by the values of the parameters of the potential. 

The purpose of calculating Henry s Law constants is usually to determine the parameters of the adsorption potential. This was the approach in Ref. [17], where the Henry s Law constant was calculated for a spherically symmetric model of CH4 molecules in a model microporous (specific surface area ca. 800 m /g) silica gel. The porous structure of this silica was taken to be the interstitial space between spherical particles (diameter ca. 2.7 nm ) arranged in two different ways as an equilibrium system that had the structure of a hard sphere fluid, and as a cluster consisting of spheres in contact. The atomic structure of the silica spheres was also modeled in two ways as a continuous medium (CM) and as an amorphous oxide (AO). The CM model considered each microsphere of silica gel to be a continuous density of oxide ions. The interaction of an adsorbed atom with such a sphere was then calculated by integration over the volume of the sphere. The CM model was also employed in Refs. [36] where an analytic expression for the atom - microsphere potential was obtained. In Ref. [37], the Henry s Law constants for spherically symmetric atoms in the CM model of silica gel were calculated for different temperatures and compared with the experimental data for Ar and CH4. This made it possible to determine the well-depth parameter of the LJ-potential e for the adsorbed atom - oxygen ion. This proved to be 339 K for CH4 and 305 K for Ar [37]. On the other hand, the summation over ions in the more realistic AO model yielded efk = 184A" for the CH4 - oxide ion LJ-potential [17]. Thus, the value of e for the CH4 - oxide ion interaction for a continuous model of the adsorbent is 1.8 times larger than for the atomic model. 

Once the model atomic structure of an amorphous oxide adsorbent is created, one may proceed to simulate physical adsorption on (or in) this material. The peculiarity of oxide adsorbents (compared to carbon adsorbents for example) is that one has to take account of the highly inhomogeneous electrostatic field at their surfaces. The problem of the reliable calculation of the effect this field upon the adsorption energy is not yet totally resolved. However, an effective adsorption potential is typically used in such situations, with parameters that are adjusted by, for example, fitting the calculations to the temperature dependence of the experimental Henry s Law constants. Such potentials generally give reasonable values for other simulated equilibrium and kinetic adsorption properties. There is even an indication that the effective parameters of the gas-solid adsorption potential are sometimes transferrable from one (oxide) adsorption system to another. 

Silicates—These chemically inert synthetic amorphous silica adsorbents have an affinity for polar contaminants. The surface area, porosity, and moisture content of the silica adsorbents provide them the capability of adsorbing secondary oxidation products (aldehydes, ketones), phosphatidic compounds, sulfur compounds, trace metals, and soap. Moisture functions to hold the pores open and aid in the attraction of the polar contaminants. Most of the synthetic silicas do not have significant direct adsorption capabilities for carotenoid or chlorophyll compounds, but the removal of the other impurities enhances the efficiency of the bleaching earths (Young, 1990). 

The first step of oxide-layer formation is oxygen adsorption (chemisorption). In the case of platinum, the process stops at this stage, and depending on the conditions, an incomplete or complete monolayer of adsorbed oxygen is present on the platinum surface. In the case of other metals, layer formation continues. When its thickness 5 has attained two to three atomic diameters, the layer is converted to an individual surface phase that is crystalline (more seldom, amorphous) and has properties analogous to those of the corresponding bulk oxides. 

Since most trace elements in soils are at parts per million levels, a separate compound may be not formed. Most likely, trace amounts of these trace elements and their compounds are adsorbed on the surfaces of clay minerals and various crystalline and amorphous Fe/Mn/Al oxides and hydroxides. Curtin and Smillie (1983) reported that the solubilities of Mn2+ and Zn2+ in limed soils were not consistent with the solubilities of any... 

Arsenate is readily adsorbed to Fe, Mn and Al hydrous oxides similarly to phosphorus. Arsenate adsorption is primarily chemisorption onto positively charged oxides. Sorption decreases with increasing pH. Phosphate competes with arsenate sorption, while Cl, N03 and S04 do not significantly suppress arsenate sorption. Hydroxide is the most effective extractant for desorption of As species (arsenate) from oxide (goethite and amorphous Fe oxide) surfaces, while 0.5 M P04 is an extractant for arsenite desorption at low pH (Jackson and Miller, 2000). 

---