A spherical model has been developed for a Clark electrode immersed in a flowing liquid medium. This model relates the sensitivity of the electrode to cathode radius, to electrolyte layer permeability and thickness, to membrane permeability and thickness and to the concentration boundary-layer thickness established by the flowing medium. The model was tested with sensitivity values obtained from a 1.3 cm diameter Teflon-covered electrode assembly which was placed in a 2.5 cm i.d. flow loop such that the centerline streamline stagnated at the cathode. The data set spanned a range of cathode radii from 65 to 677 μm, membrane thicknesses of 6 μm and 76 μm and a range of mean axial tube velocities ( V) from 0.5 to 62 cm s −1 corresponding to tube Reynolds numbers of 125–16,000. The boundary-layer thickness was assumed to be of the form (γ V − n ) and the constants γ and n were determined by a non-linear least-squares regression of the sensitivity data to the spherical model. For the purposes of this regression, the sensitivity values were grouped into a laminar and a turbulent data set and (γ, n) values were determined for each set separately. The parameters P e/ P m (electrolyte to membrane permeability ratio) and ω e (electrolyte layer thickness) were treated as constants, with their values set at the regressed values previously obtained from fitting the gas-phase sensitivity values. Here, γ and n were both treated as free parameters. The spherical model fits the experimental current values in the laminar and turbulent regions with average relative errors of 8.74% and 13.1% respectively. Residuals were randomly distributed over the range of cathode radii and the two membrane thicknesses used. The final regressed values of γ and n were 0.0307 and 0.518 respectively for the laminar region, and 0.0317 and 0.451 respectively for the turbulent region. The n value for the laminar region is in reasonable agreement with the theoretically predicted value of 0.5. For comparison, a one-dimensional planar model was used to correlate the laminar and the turbulent data sets. Here again the parameters P e/ P m and ω e were assigned those values previously obtained from fitting gas-phase sensitivity values by the planar model. The relative average errors, 34.7% and 37.6% for the laminar and turbulent regions respectively, were excessively large and the final regressed value of n, 0.387 for the laminar data set, was unrealistic with respect to the theoretical prediction.
Read full abstract