Original paper: doi: 10.1002/anie.201302577<p>From their paper it is mentioned they have a 40 nanolitre/min flow rate (25% desalination rate, 99% is considered safe to drink water).<p>They use an electric pole to generate an ion depletion zone, their problem right now lies in the severely limited flow rate and desalination rate. From my understanding the 25% desalination rate does not compound linearly and decreases in efficiency as the concentration of salt ions decreases.<p>Additionally their flow rate is 0.4 microlitres per minute, this would equate to needing 625 000 channels for one pass only to get 250mL / min. Scaling for microfluidics isn't simply using a larger pipe size, microfluidic devices largely operate with minimal forces and a Reynolds number of 1, that doesn't hold as you get larger.<p>The other big problem here is that this requires pressure driven flow, to do that they made two reservoirs of uneven height on opposite ends to drive flow. Again this is very electrically cheap (picowatts) when dealing with such small and perfect flow conditions.<p>The biggest problems they need to address are the low filtration rate and the low flow rate, it does not seem like there is a simple answer to the first, they do state that they are conducting a larger scale experiment that they will publish later.<p>I largely suspect that the larger scale experiment will fail in efficiency as another group (doi:10.1038/nnano.2010.34) who also uses a ion depletion zone (albeit by a different mechanism) found after publishing that they were orders of magnitude (10^3) too low in their predictions.<p>If anything this will perhaps be better than small, portable RO setups but not replace large factories which are actually pretty efficient.