I am trying to optimise the Prussian blue layer deposited on screen-printed electrodes (DRP-110SWCNT, with a single-walled carbon nanotube working electrode), and I am consistently observing the following when comparing two electrodes, one prepared by electrodepositing PB using cyclic voltammetry (PB(el)), and the other prepared by dropcasting PB nanoparticles onto the working electrode (PB(drop-casted)).

Even though that both electrodes show comparable peak height in the CV, the responses to 10 uM hydrogen peroxide additions in neutral medium are vastly different. Namely, the electrodeposited PB shows higher response (preferable), but also very high noise. I have attached the graphs.

We did a lot of optimization, but the problem with high noise persists, limiting the possible LOD.

I was wondering if anyone would be able to have any explanation and/or suggestions on how to proceed? The layer will be used as a redox mediator in a biosensor.