Characterise the material in three electrodes, the device in two. In 0.5 M Na2SO4 across a 1 cm gap the series resistance is 15–20 Ω, so at 5 mA/cm2 a two-electrode cell hides 75–100 mV of iR drop plus the counter-electrode overpotential. A photoanode reading 3.2 mA/cm2 at 1.23 V vs RHE in three electrodes typically gives 2.4–2.9 mA/cm2 as a two-electrode device.
A potentiostat holds the working electrode at a set potential against the reference and passes current between working and counter. Almost no current flows through the reference, so the number you set is the number the semiconductor sees; the counter electrode can be as sluggish as it likes, its overpotential never enters the record.
A two-electrode cell has no such separation. The instrument controls one voltage, and everything in the current path sits inside that number: the thermodynamic potential, both overpotentials and the ohmic drop. That is not a defect; it is what a real device experiences, which is why efficiency claims belong to the two-electrode configuration.
| Property | Three-electrode | Two-electrode |
|---|---|---|
| Quantity controlled | Potential of the working electrode vs reference, ±1 mV | Cell voltage between the two electrodes |
| Current path | Working to counter; reference carries under 1 nA | Working to counter only |
| Counter-electrode overpotential | Not measured, not relevant | Included in the reported voltage |
| Uncompensated resistance | Measurable by EIS, correctable to below 5 mV | Included and not separable |
| Best used for | Absorber screening, Tafel slopes, IPCE at fixed potential, mechanism | Full device efficiency, tandems, 100 h stability, STH |
| Reported as | V vs RHE, with pH stated | Vcell, two-electrode, with both electrodes named |
| Typical penalty if confused | Current quoted at a bias the device cannot reach | 100–200 mV of apparent overpotential at 5–10 mA/cm2 |
Electrolyte conductivity sets the floor. A 0.5 M Na2SO4 solution sits near 0.055 S/cm at 25 °C, so a cell with 1 cm2 electrodes 1 cm apart has R = L/(σA) = 1/(0.055 × 1) ≈ 18 Ω. At 5 mA that is 90 mV, at 10 mA 180 mV — larger than most overpotential differences people argue about.
Measure it rather than estimate it. A potentiostatic impedance sweep at 100 kHz gives a high-frequency intercept that is the series resistance, typically 10–30 Ω in a compact cell and 50–100 Ω when the electrodes sit far apart. Then correct point by point: Ecorrected = Emeasured − i × Rs. Uncorrected, a 90 mV offset on a 60 mV/decade Tafel region shifts the apparent slope by roughly 15 %, enough to change the mechanistic conclusion.
A Luggin capillary 1–2 mm from the working electrode drops Rs below 5 Ω, but its tip must stay out of the illumination path: shadowing 10 % of the electrode area costs 10 % of the photocurrent, a worse error than the one it fixes.
There is no exact transformation, because the counter electrode is a different material with its own kinetics, but a four-step estimate works.
For a photoanode at 3.2 mA/cm2 and 1.23 V vs RHE under 100 mW/cm2, applying 0.6 V of bias gives ABPE = 3.2 × 0.63 / 100 = 2.0 %. In a two-electrode cell the same electrode at 1.23 V total delivers about 2.6 mA/cm2, so the estimate and the measurement usually bracket each other within 10–15 %. A wider gap points at the counter electrode before the semiconductor.
Any claim about a cell belongs to two electrodes: solar-to-hydrogen efficiency, tandem stacks, bias-free operation, and stability over 100 h. A three-electrode curve at a fixed potential says the photoelectrode survived, not that the device runs without a potentiostat. Reviewers ask for both, and the pair is stronger than either alone: three-electrode data explain why the material works, two-electrode data show that it works.
Two-electrode numbers are also sensitive to bubbles on the counter electrode, where a covered Pt surface adds tens of millivolts; flow keeps repeats within 5 %.
Not exactly. Subtracting the separately measured counter-electrode overpotential and the iR drop recovers an approximation good to 20–50 mV. Report the two-electrode curve as Vcell and keep the three-electrode curve beside it for potential-resolved discussion.
Because the cell voltage also pays for the counter electrode and the electrolyte. At 5 mA/cm2 in a 1 cm gap, 90 mV goes to iR and 30–50 mV to hydrogen evolution on Pt, so the photoanode sees 0.1–0.15 V less than the dial suggests; a bare carbon counter electrode can cost several hundred millivolts more.
Pt mesh or foil at least four times the working electrode area, in a separate compartment when the catholyte could contaminate the photoanode. In chloride electrolytes use carbon or platinised titanium; Pt dissolves measurably in concentrated chloride.
Only for a perfectly efficient device. Real single-absorber systems need 1.6–2.0 V for useful current, so a run at 1.23 V may sit in a regime the device never uses. Choose the voltage from the three-electrode J–V curve at the target current density, then hold it.
