Fig. 3: Multi-pass thermodiffusive desalination for aqueous NaCl and seawater.

Both seawater level and brine level (double the seawater concentration) are considered. a The initial concentration of the NaCl in aqueous solution was 60,000 ppm. NaCl concentration in both the top and bottom streams are plotted for each pass. Both experimental results and numerical results are demonstrated. The errors in the experimental ΔC are standard deviations from PSI measurements. In each pass, the top stream with a lower salinity was accumulated and put into the next pass. The NaCl concentration in the top stream kept dropping, demostrating that thermodiffusion is scalable beyond a single pass. b The concentration drop is plotted as a function of number of passes recirculated for two different initial concentrations (30,000 and 60,000 ppm). On the left axis, the concentration drop C_drop achieved after four passes were 1200 ppm and 2000 ppm for an initial concentration of 30,000 and 60,000 ppm, respectively. On the right axis, the recovery rate is shown. There is a trade-off between C_drop and recovery rate. c Same experiment was performed using multi-ion seawater substitute. The concentration of different cations: Na⁺ (c1), Mg²⁺ (c2), Ca²⁺ (c3), and K⁺ (c4), was measured by two different ICP methods for the same set of samples collected in each pass. Linear fits were performed for the top stream concentrations to show the concentration reduction trend. The errors in ICP-AES measurements are the standard deviations between five replicates. The errors of the ICP-OES results were not available. Na⁺ is most abundant in quantity and provides most reliable results. For ICP-AES results, the C_top = (−190n + 14,200) ppm, i.e. C_drop = 190 ppm and C₀ = 14,200 ppm. For ICP-MS results, C_top = (−111n + 13,329) ppm. Linear fits are also applied to Mg²⁺ and Ca²⁺. For K⁺, however, no trendline is fitted due to lack of discernible trend.