Diffusion Semi-predictive

One can state that a diffusion model, which allows only such calculations and extrapolations of D, exhibits only correlative and semi-predictive capabilities. All classical diffusion models presented in this section fall into this category. 

Having mentioned the correlative capabilities of this model, one can consider its semi-predictive abilities. It was mentioned that a number of diffusion data taken from a limited range of penetrant concentrations are required to calculate two of the parameters of the model. Once these parameters have been determined, one can make theoretical predictions for diffusion coefficients over a wider range of penetrant concentration or temperature variation. This is a critical test for any theoretical model,... 

The above discussion has tacitly assumed that it is only molecular interactions which lead to adhesion, and these have been assumed to occur across relatively smooth interfaces between materials in intimate contact. As described in typical textbooks, however, there are a number of disparate mechanisms that may be responsible for adhesion [9-11,32]. The list includes (1) the adsorption mechanism (2) the diffusion mechanism (3) the mechanical interlocking mechanism and (4) the electrostatic mechanism. These are pictured schematically in Fig. 6 and described briefly below, because the various semi-empirical prediction schemes apply differently depending on which mechanisms are relevant in a given case. Any given real case often entails a combination of mechanisms. 

Adsorption equilibrium of CPA and 2,4-D onto GAC could be represented by Sips equation. Adsorption equilibrium capacity increased with decreasing pH of the solution. The internal diffusion coefficients were determined by comparing the experimental concentration curves with those predicted from the surface diffusion model (SDM) and pore diffusion model (PDM). The breakthrough curve for packed bed is steeper than that for the fluidized bed and the breakthrough curves obtained from semi-fluidized beds lie between those obtained from the packed and fluidized beds. Desorption rate of 2,4-D was about 90 % using distilled water. 

The transition from single- to many-chain behavior already becomes obvious qualitatively from a line shape analysis of the NSE spectra (see Fig. 60) [116]. For dilute solutions (c = 0.05) the line shape parameter (3 is equal to about 0.7 for all Q-values, which is characteristic of the Zimm relaxation. In contrast, in semi-dilute solutions (e.g. c = 0.18), ft-values of 0.7 are only found at larger Q-values, whereas P-values of about 1.0, as predicted for collective diffusion [see Eq. (128)] are obtained at small Q-values. A similar observation was reported by [163]. 

If the pore-mechanism applies, the rate of permeation should be related to the probability at which pores of sufficient size and depth appear in the bilayer. The correlation is given by the semi-empirical model of Hamilton and Kaler [150], which predicts a much stronger dependence on the thickness d of the membrane than the solubility-diffusion model (proportional to exp(-d) instead of the 1 Id dependence given in equation (14)). This has been confirmed for potassium by experiments with bilayers composed of lipids with different hydrocarbon chain lengths [148], The sensitivity to the solute size, however, is in the model of Hamilton and Kaler much less pronounced than in the solubility-diffusion model. 

Figure 17. Comparison of Ju versus t plots predicted by different submodels for a system with one type of site (adsorbing and internalising) linear isotherm with dSS approximation (O) applying equation (43) with A)i = 5.2 x 10 6 m Langmuirian isotherm with dSS approximation (continuous line) applying equation (46) Langmuirian isotherm with semi-infinite diffusion (dotted line) by numerically solving integral equation (7)). Other parameters c(, = 5x 10-4 mol m-3, Dm = 8 x 10 10 m2 s-1, Kn = 2 x 10-5 m, k — 5 x 10 4 s 1, ro = 1.8 x 10 6 m, r0 + <5M = 10 5 m, KM — 2.88x 10 3mol m 3, Tmax = 1.5 x 10-8 mol m-2...

Diffusion coefficients may be estimated using the Wilke-Chang equation (Danckwerts, 1970), the Sutherland-Einstein equation (Gobas et al., 1986), or the Hayduk-Laudie equation (Tucker and Nelken, 1982), which state that Dw values decrease with the molar volume (Vm) to the power 0.3 to 0.6. Alternatively, the semi-empirical Worch relation may be used (Worch, 1993), which predicts diffusion coefficients to decrease with increasing molar mass to the power of 0.53. These four equations yield very similar D estimates (factor of 1.2 difference). Using the estimates from the most commonly used Hayduk-Laudie equation... 

Industrially relevant consecutive-competitive reaction schemes on metal catalysts were considered hydrogenation of citral, xylose and lactose. The first case study is relevant for perfumery industry, while the latter ones are used for the production of sweeteners. The catalysts deactivate during the process. The yields of the desired products are steered by mass transfer conditions and the concentration fronts move inside the particles due to catalyst deactivation. The reaction-deactivation-diffusion model was solved and the model was used to predict the behaviours of semi-batch reactors. Depending on the hydrogen concentration level on the catalyst surface, the product distribution can be steered towards isomerization or hydrogenation products. The tool developed in this work can be used for simulation and optimization of stirred tanks in laboratory and industrial scale. 

Occasionally (e.g., thin-layer electrochemistry, porous-bed electrodes, metal atoms dissolved in a mercury film), diffusion may be further confined by a second barrier. Figure 2.7 illustrates the case of restricted diffusion when the solution is confined between two parallel barrier plates. Once again, the folding technique quickly enables a prediction of the actual result. In this case, complete relaxation of the profile results in a uniform finite concentration across the slab of solution, in distinct contrast to the semi-infinite case. When the slab thickness t is given, the time for the average molecule to diffuse across the slab is calculable from the Einstein equation such that... 

The concentration-distance profiles predicted by Equation 1.48 and resulting in the current in Equationl.49 are given in Fig. 1.13. A complete expression for the current that results from semi-infinite diffusion is obtained as follows, which in fact is also called the Cotrell equation for a planar electrode ... 

The basic strategy in the application of electroanalytical methods in studies of the kinetics and mechanisms of reactions of radicals and radical ions is the comparison of experimental results with predictions based on a mechanistic hypothesis. Thus, equations such as 6.28 and 6.29 have to be combined with the expressions describing the transport. Again, we restrict ourselves to considering transport governed only by linear semi-infinite diffusion, in which case the combination of Equations 6.28 and 6.29 with Fick s second law, Equation 6.18, leads to Equations 6.31 and 6.32 (note that we have now replaced the notation for concentration introduced in Equation 6.18 earlier by the more usual square brackets). Also, it is assumed here that the diffusion coefficients of A and A - are the same, i.e. DA = DA.- = D. 

In an early study by Schery and Gaeddert (1982), an accumulator device was used to measure the effect of atmospheric pressure variations on the flux of Radon (222Rn), an inert radioactive element with a half-life of 3.8 days, from the soil. Fluxes measured by the accumulator were compared with predictions for flow-free diffusion from a model developed by Clements and Wilkening (1974), which applies Fick s law. A mean 222Rn-flux enhancement of about 10%, with a high value of 20%, due to cyclic atmospheric pressure variations was observed. However, the device s effectiveness was limited by back diffusion from the accumulator to the subsurface, leading the authors to view the flux values as semi-quantitative. 

The model predicts the yield of S Xg = 2S/ (2 S + R) at the end of the reaction (when all B is consumed). It was developed for batch and semi-batch reactors (119, 120), and later extended to continuous stirred reactors via a somewhat complicated procedure (121-112). Some criticism may be adressed, to the MIRE-model, in spite of its great interest arbitrary choice of spherical shape, assimilation of R to half the Kolmogorov microscale (which is not obvious as we have seen above) and above all, assumption that the initial reactant in the particle cannot diffuse outside, which creates an unwanted dissymmetry between A and B when V = Vg. 

In this way, the diffusion/reaction equations are reduced to trial and error algebraic relationships which are solved at each integration step. The progress of conversion can therefore be predicted for a particular semi-batch experiment, and also the interfacial conditions of A,B and T are known along with the associated influence of the film/bulk reaction upon the overall stirred cell reactor behaviour. It is important to formulate the diffusion reaction equations incorporating depletion of B in the film, because although the reaction is close to pseudo first order initially, as B is consumed as conversion proceeds, consumption of B in the film becomes significant. 

Theory (Odijk et al., 1977/79, Mandel et al., 1983/86) predicts that in the dilute state (c < c ) most of the parameters of the solution (intrinsic viscosity, diffusivity, relaxation times) will be functions of the molar mass, but not of the polymer concentration. In the so-called semi-diluted solution state the influence of the polymer concentration (and that of the dissolved salts) becomes very important, whereas that of the molar mass is nearly absent. Experiments have confirmed this prediction. 

Further experiments have been conducted to confirm whether or not the presumed diffusion layer and its thickness, 8, as estimated from (95) corresponds to physical reality. First AC impedance spectroscopy has been used to find the frequency response of the real and imaginary components of the cell impedance and compared with the theoretical prediction for diffusion across a thinned diffusion layer. At very high AC frequencies, where the AC perturbation had insufficient time to probe to the edge of the diffusion layer, effectively the response expected for semi-infinite diffusion was seen ( Warburgian behaviour ). At lower AC frequencies, as expected, the cell impedance was greatly reduced in the presence of ultrasound. Moreover, not only was the quantitative behaviour as predicted theoretically... 

Besides the mathematical improvements, the atmospheric model has been adapted to a semi-Lagrangian formulation. By following selected air masses, we avoid commitments of large quantities of memory and the incursions of artificial diffusion errors. Most important, we do not end up with stacks of computer printout that relate to regions where there are no measurements. Also, predictive calculations will become more useful, but our present levels of resources and sophistication demand that effort be concentrated on verification. Only in this way can the confidence be built that is needed for applying modeling techniques to implementation planning. 

---