A genuine photocurrent step settles within 0.1–1 s. A signal that decays more than 20 % between 5 s and 30 s, or changes shape with scan rate or chopping frequency, is an artifact. The usual culprits are cell capacitance (10–100 µF/cm2, an RC of 1–10 ms), the shutter or LED ramp (50–300 ms) and series resistance above 10 Ω.
Every feature in a step response belongs to a process with a known timescale. Reading the transient without that map is guessing.
| Time constant | Process | How it appears | How to confirm |
|---|---|---|---|
| 1–100 µs | Surface recombination, trapping | Spike lost inside the recorder's response | Pulsed LED or transient absorption |
| 0.1–10 ms | Cell RC = Rs × Cdl | Exponential rise, no matching light-off tail | EIS: read Rs and Cdl, multiply |
| 50–300 ms | Shutter, chopper or LED driver ramp | Rise mirrors the lamp, not the sample | Photodiode trace beside the electrode |
| 0.1–10 s | Ion diffusion, bubble growth, thermal drift | Monotonic drift, present in the dark too | Dark step after 30 min equilibration |
| >10 s | Back reaction, surface-state filling, degradation | Plateau falls across cycles | Three on/off cycles compared |
A 50 ms shutter convolved with a 5 ms RC looks like a 55 ms rise. That is arithmetic, not chemistry, and no fitting of the rise improves it.
Capacitive charging. A nanostructured film with a roughness factor of 50–200 carries 10–100 µF/cm2 of geometric area. Charging that capacitance to 0.5 V at 1 mA/cm2 takes 5–50 ms, and the charging current is a spike on top of the faradaic current. It decays exponentially with τ = RsCdl, which is why the spike shrinks when the electrolyte conductivity rises and grows when the film gets thicker.
Light-source ramp and light soaking. Xenon shutters take 30–100 ms with mechanical bounce; constant-current LED drivers often ramp over 100–300 ms to protect the diode. Separately, some films gain 10–40 % photocurrent over 2–10 min as surface states fill. That increase is real but extrinsic: record it as an activation period and quote the stabilised value with its timestamp.
Series resistance and bubbles. Ten ohms of electrolyte resistance at 1 mA/cm2 costs only 10 mV, yet it stretches the RC and rounds the step edge. Oxygen bubbles of 0.1–1 mm nucleating on the surface block 5–20 % of the illuminated area within minutes at 1 mA/cm2, producing a decay that looks like recombination but disappears when the cell is stirred or the electrolyte is sparged.
A fourth, quieter cause is instrument autoranging: a current amplifier switching from the microamp to the milliamp range inserts a step of its own, usually within the first 100 ms.
Sampling must be at least ten times faster than the fastest feature you intend to interpret. A 1 s transient logged once per second cannot be analysed at all, and a 10 Hz low-pass filter built into the potentiostat adds 16 ms of rise time that will be fitted as carrier dynamics if nobody checks.
Use a current-to-voltage converter with a rise time below 1 ms, disable software averaging during the step, and record the lamp trigger on a second channel so the delay between trigger and light is measured rather than assumed. Ground the cell at one point only, shield the cable, and subtract a dark baseline recorded immediately before the step.
State the sampling rate, the filter setting, the delay between trigger and illumination, and the time at which the plateau is read. Give the transient as a normalised curve plus the stabilised current density in mA/cm2. When the decay carries a physical claim, show the photodiode trace and one control that removes the artifact, because a reviewer cannot distinguish the two from the curve alone.
Thin compact films reach a plateau in 0.1–1 s once the cell RC has charged. Nanoparticle films with high capacitance need 1–5 s, and light soaking can add minutes. Report the value at a fixed time, for example 60 s after illumination, and say so.
Most often the spike is capacitive charging plus surface recombination, and the decay is the approach to steady state. If the spike amplitude grows with chopping frequency while the plateau stays constant, capacitance dominates and the decay is not evidence of poor transport.
Not on its own. Compare the rise with the photodiode trace and with the RC measured by EIS. Transport limitation shows up as a photocurrent that falls with thickness and with light intensity, whereas an instrument-limited rise is identical at every thickness.
Both, for different purposes. Steady-state current densities support activity comparisons; transients support statements about dynamics, and only when the acquisition is fast enough to resolve them and the lamp has been characterised.
