Drifting and Biasing Potentials in the Electrodes Make Streaming Potential Measurements Difficult
A few weeks back I had finished my streaming potential/electro-osmotic pump setup. It looks like this:
For streaming potential measurements, we pressurize a tube filled with 100mM KCl that connects to the membrane and measure the voltage drop across the membrane. When I first applied a pressure to the setup, the voltage drop did seem to be dependent on the applied pressure, but after that first time the voltage drop seems insensitive to the applied pressure. Specifically, when I put the next pnc-Si chip in the setup and connected the leads to a voltmeter, but before I applied pressure, there was a voltage difference of 6 mV between the leads, which seemed to decay non-linearly towards 0 volts. Pressurizing or depressurizing the system seemed to do nothing to affect this decay or inherent bias.
I assumed this weirdness was due to the electrodes, and so I made several fresh electrodes. The electrodes are formed by taking thin silver wire and wrapping ~10 turns around a small nail. The coil is removed from the nail and dipped into a thoroughly mixed commercial Ag/AgCl ink solution, then placed into the oven to dry. This results in a fairly sizable blob that hides the shape of the coil. Next, a hole is drilled through a nylon screw, the silver wire is drawn through so that the coil is against the bottom of the screw, and PDMS is used to seal the hole to prevent leaks. Once the PDMS hardens, Teflon tape is wrapped around the screw threads and the electrodes are then screwed into the device.
Unfortunately, new electrodes didn’t make the problem go away. Putting a fresh pnc-Si chip in with new electrodes gave a similarly decaying potential that was insensitive to pressure. I next put a 3-slot chip with all it’s windows blown out into the setup and measured a 
There, the voltage drop was likewise a consistent 3mV, even after removing the electrodes from the beaker and touching them together and replacing them into the electrolyte solution. The resistance was 0.48k
If there are not any EEs that wander through the lab these days, I would be happy to help you with this. Single-digit millivolts are actually fairly difficult to measure consistently. I would look at the grounding of your setup, the connections of dissimilar metals, wire loops that can couple EMF signals, etc. I assume that your equipment is well-calibrated and in good repair?
Drop me an email if you want me to stop by the lab to provide my 2-cents….