1. 1 mV/s is not a fast scan for a common CV. Try more than 50 mV/s.

2. During the long scan (20 h), the sample is contaminated by the leakage of the filling solution from the REF electrode. I suggest that the unexpected peaks are for: AgCl(s) + e– = Ag(s) + Cl–(aq)

Thanks for the suggestion - I hadn't really considered that but it does seem like a strong possibility. Here are some other pieces of information:

1) I don't really observe these peaks developing if I just keep the devices immersed in the electrolyte for the same amount of time (> 20h), and then run another 'fast' (I realized the scan rate is actually 10 mV/s not 1 mV/s) scan. Going with the AgCl leakage hypothesis, would you expect the size of the peaks to depend on whether the electrode had been under voltage cycling previously?

2) The size of the peaks also seems to depend on the bias that I cycle between during the long conditioning. For example if I just cycle in the positive (anodic) range, the peak size is observed to decrease. If I cycle in the cathodic range, they increase.

***

Application wise, I do need the long scans (at least right now). Would you have any suggestions for alternative reference electrode systems that might be better? Could I get away with maybe a Pt pseudo reference? It would probably be useful to test with another reference as a control at least.

Thanks a lot for your thoughts!

Add a salt bridge to your REF electrode, containing only KCl solution (inside the electrode, KCl solution is saturated with AgCl).

The peak size depends on the relative amounts of the oxidised and the reduced forms of the contaminants (commonly visible in the first sweep).