Thermodiffusion and molecular diffusion in binaryn-alkane mixtures: experiments and numerical analysis

Abstract

New experimental data on the thermodiffusion and molecular diffusion coefficients of nine non-equimolar binary mixtures of nC 6 À nC i (i ¼ 10, . . . , 18) at 298 K and 1 atm have been reported. The data have been obtained using a combination of the thermogravitational column and the sliding symmetric tubes technique, respectively. It was found that with an increase in the disparity between the two components, the Soret effect becomes more dominant. Further, the stronger Soret effect weakens the molecular diffusion process. These trends are observed in the form of an increase in magnitude of the thermodiffusion coefficient and a decrease in the magnitude of the molecular diffusion coefficient. Numerical analysis was also performed using the principles of non-equilibrium thermodynamics. The model was able to predict the experimental data as well as the trends fairly accurately.

New algebraic expressions for thermodiffusion in binary n‐alkane mixtures

“…For comparison, in addition to the present models (Equations (2),(3) and (6)), the values from the algebraic model of Madariaga et al [6] (Equation (7)), the Linear Non-equilibrium Thermodynamic (LNET) model and the NN model are also plotted in the figure. The experimental data are the ones reported by Srinivasan et al [21] The experimental data show a decrease in the magnitude of the thermodiffusion coefficient as the difference in the molecular weights decrease (Figure 4). In other words, thermodiffusion is weaker as the similarity among the molecules increases.…”

“…In the n-decane series, components ranging from nC 5 to nC 20 were used as the second liquid. [6,19,[21][22][23][24][25][26] For each n-alkane series, a comparison is also included with a similar model of Blanco et al, [19] in which D T is given as: The comparison between the model predictions and the experimental data is shown in Figure 1a-d for the four n-alkane series.…”

“…| VOLUME 91, JUNE 2013 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 1171 | As discussed in detail by Srinivasan et al, [21] Q * i can be estimated using the relation:…”

“…To compare the performance of the different methods, thermodiffusion coefficients are calculated by these schemes for the nC 6 − nC i series investigated by Srinivasan et al [21] The thermodiffusion experiments for this series were performed at an average mixture temperature of 298 K and at atmospheric pressure with a temperature gradient of 6 K. The effect of the molecular weight difference on the value of D T in each mixture of this series is shown in Figure 4. For comparison, in addition to the present models (Equations (2),(3) and (6)), the values from the algebraic model of Madariaga et al [6] (Equation (7)), the Linear Non-equilibrium Thermodynamic (LNET) model and the NN model are also plotted in the figure.…”

“…As seen in Figure 4, this trend is well-predicted by the models proposed in this study. [21] The accuracy is further increased when the quadratic dependence on the composition difference is included in the model (Equation (3)). This is well within the experimental uncertainty of 5%.…”