The Coulomb efficiency of a device with a microporous, non-selective membrane depends fundamentally on how fast you charge/discharge it, as the device self-discharges while it runs. The faster you charge/discharge, the higher the CE will be.

In the case of the photopaper device, it will be in the 85-90% range when charging to high SOC values at 20mA/cm2. The big advantage is that microporous membranes are really cheap and they still work even if dendrites pierce them. I must me clear that photopaper is meant as a DIY demonstration, a commercial unit would never use that but a polyethylene microporous separator - such as Daramic - with these memrbanes the CE and EE tend to be higher.

There are much cheaper membranes; ESS for example uses a membrane that is used by lithium ion batteries (I think) and thus is commonly available and very inexpensive.