Five electrochemical measurement mistakes

Five electrochemical measurement mistakes

Two laboratories can measure the same material and report activity numbers an order of magnitude apart without either of them making an experimental error.

The difference may not be the catalyst. It may be what the number was divided by, where on the potential axis it was taken, or what was assumed about the surface.

Electrochemistry reuses a small set of symbols — V, mA, mA cm-2, mV dec-1 — for quantities produced by completely different physics. That overlap is where the five mistakes below come from. Each appears regularly in published work, and each has a documented correction.

Key facts

  • Potential is a position; overpotential is a distance. Neither can be reconstructed from the other without the reference electrode and the equilibrium potential under your actual conditions
  • Ohmic drop is not intrinsic electrode overpotential and should be accounted for separately
  • OER benchmarking convention: η at 10 mA cm-2 is conventionally reported per geometric area [1] — a protocol convention, not a universal one
  • ECSA scales inversely with the specific capacitance you assume, and so does every activity number derived from it
  • Tafel slopes are coverage-dependent — the textbook mechanism assignments assume extreme coverage [2]
  • Frequently under-reported: whether the data was iR-corrected

Jump to a section

  1. Potential vs overpotential
  2. Current vs current density
  3. Geometric vs ECSA current density
  4. Tafel slope and reaction mechanism
  5. Exchange vs limiting current density
  6. Symbols and what each is referenced to
  7. Reporting checklist
  8. Frequently asked questions

1. Potential vs overpotential: not the same measurement

Potential is where the electrode sits relative to a reference electrode. Overpotential is how far it has been driven from equilibrium. They share the unit volt and nothing else.

The assumption: overpotential is the electrode potential on a different scale, so reporting one is as good as reporting the other.

Why it fails. Electrode potential (E) exists whether or not any net reaction is happening — open circuit potential and equilibrium potential are both electrode potentials. Overpotential is the deviation of the electrode potential from its equilibrium value, η = Eapplied − Eeq. One is a position, the other a displacement.

Reconstructing η from a reported E therefore requires more than the reference electrode: it requires the equilibrium potential for the relevant reaction under your actual conditions. That depends on the activities of the species involved and, for gas-evolving reactions, on partial pressures — not on pH alone.

The measured potential under current is not all overpotential. It can contain electrode kinetic overpotential, concentration polarisation, and uncompensated ohmic drop. The ohmic component is not intrinsic electrode overpotential — it is a property of the cell and the electrolyte — and should be accounted for separately rather than folded into a reported η [3].

PEM electrolyser polarisation curve alongside its activation, ohmic and concentration overpotential contributions plotted against current density
Figure 1. (a) Modelled polarisation curve for PEM electrolysis, cell voltage against current density. (b) The overpotential contributions that separate the actual cell voltage from the thermodynamic value: activation rises exponentially and dominates at low current density, ohmic scales linearly and overtakes it near 1 A cm-2, and concentration polarisation stays small across this range, rising only at higher current density. This is a decomposition of cell voltage, not of a single electrode's overpotential — the ohmic term is a property of the cell and electrolyte, which is why it is compensated rather than reported as part of η. The model is benchmarked against experimental data: the ohmic term agrees to within 2 %, the activation term to about 5 %, the latter attributed to approximations in exchange current density and reaction kinetics. Reproduced from Figure 20 of Vedrtnam, Kalauni and Pahwa [4], licensed under CC BY 3.0.

That component is where most reported overpotentials go wrong. Uncompensated solution resistance inflates the measured value in proportion to current, so an uncorrected number at high current density can be tens of millivolts too high. Anantharaj and Noda's treatment of iR correction covers how much of the drop to compensate and how to measure it [5].

The reference electrode itself is the other half of the problem. The same group tested Hg/HgO, Hg/Hg₂SO₄, Ag/AgCl and SCE references against the HER onset of Pt across acid, alkaline and neutral electrolytes, and found the apparent reference potential shifted from +78 mV to −74 mV at the extremes when an inappropriate reference was used for the pH [6]. That is larger than many of the activity differences being reported.

What to do instead. Report η with the reference electrode and its specified potential, the electrolyte composition and pH, the equilibrium potential assumed, and the iR correction applied. Convert to RHE for HER and OER work so data at different pH can be compared.

Reference electrode Approx. E vs SHE at 25 °C
Ag/AgCl, saturated KCl +0.197 to +0.198 V
Ag/AgCl, 3.0 M NaCl +0.210 V
SCE, saturated KCl +0.241 V
Hg/HgO, 1.0 M KOH +0.098 V
Hg/Hg2SO4, sat. K2SO4 +0.640 V

At 25 °C, the RHE conversion is E(RHE) = Emeasured + Eref + 0.05916 × pH, where Eref is your electrode's potential versus SHE. Tabulated values are a starting point, not a guarantee — reference potentials drift with filling solution concentration, temperature and age. Full conversion detail is in our Ag/AgCl potential and conversion guide, and the choice between reference chemistries is covered in our reference electrode selection guide.

Browse Beyond Battery reference electrodes →

2. Current vs current density: why more current is not a better catalyst

Current scales with electrode area; current density does not. Comparing raw current turns a geometry difference into an apparent activity difference.

The assumption: the electrode drawing more current at a given potential is the more active one.

Why it fails. Two electrodes cut from the same sheet, one 1 cm² and one 5 cm², draw 10 mA and 50 mA and have identical activity: 10 mA cm-2 each.

Stated that plainly it sounds too obvious to be a real error. Where it actually bites is loading. Increase catalyst loading tenfold on the same substrate and geometric current density rises substantially — not because the material improved, but because there is more of it.

For widely used OER benchmarking protocols, η at 10 mA cm-2 is conventionally reported using geometric electrode area. McCrory and co-workers selected that figure of merit because it approximates the current density of a 10 %-efficient solar-to-fuels device [1]. It is a protocol convention with a specific origin, not a universal electrochemical standard — practical water electrolysis is routinely evaluated at hundreds of mA cm-2 or more, where the balance between kinetic, ohmic and transport limitations is completely different.

Symbol Normalised by Answers
jgeo Geometric electrode area Product delivered per unit footprint — device relevant
jECSA Electrochemically active surface area Rate per unit real surface — material relevant
jmass Mass of active material Activity per unit of expensive metal — cost relevant

What to do instead. Put the normalisation basis in the axis label itself — mA cm-2geo, mA cm-2ECSA or A mg-1 — and state catalyst loading alongside any geometric figure. A sealed disk electrode with a defined exposed area removes the largest single source of error here; an area estimated from a droplet edge does not. Electrode geometry and cell design are covered in our practical guide to the three-electrode system.

Compare working electrode formats →

3. Geometric vs ECSA current density: what normalisation cannot fix

ECSA normalisation does not deliver a true intrinsic activity. It replaces one assumption — that geometric area is meaningful — with another: that you know the specific capacitance of your material.

The assumption: dividing by electrochemically active surface area removes loading and morphology, leaving the intrinsic activity of the material.

Why it fails. The usual route measures double layer capacitance Cdl from cyclic voltammograms at several scan rates in a window assumed to be non-faradaic, then divides by a specific capacitance Cs to get area.

The weak link is Cs. It is material-specific, usually unknown, and in practice one literature value gets applied to everything — in alkaline media, typically 40 µF cm-2 following the McCrory protocol [1]. Because ECSA scales inversely with Cs, every activity number derived from it inherits that assumption. Underestimate the area and you overestimate the activity, cleanly and invisibly.

The measurement of Cdl itself is also less robust than it looks. Morales and Risch showed that the choice of potential window, uncompensated resistance, scan rate range, data acquisition settings and fitting approach each change the extracted value, and set out a seven-step procedure for obtaining it consistently [7]. Anantharaj, Karthik and Noda go further, cataloguing the ambiguities in determining active sites and real surface area at monometallic interfaces and what constitutes best practice [8].

The assumption of a genuinely non-faradaic window is itself the problem for many of the materials people most want to measure. Porous, pseudocapacitive and surface-reconstructing OER catalysts may have no potential range where the response is purely capacitive. The practical check is the scan-rate dependence, i = avb: a value of b near 1 indicates capacitive control, while b near 0.5 indicates a diffusion-controlled, battery-type response — in which case the window you chose is not non-faradaic and the resulting Cdl is not an area measurement [9].

None of this is new. Trasatti and Petrii's IUPAC report argued that comparing electrode kinetics without normalising to real surface area is physically groundless, while stating just as plainly that several widely used normalisation methods are poorly justified [10]. Both halves of that sentence still apply.

ECSA method Best suited to Main limitation
Cdl from scan-rate CV Approximate surface-area comparison where a genuinely non-faradaic window can be identified Requires an assumed Cs; sensitive to window, scan rates and iR
Hupd stripping Pt-group metals with clean Hupd Vertex potential choice alone can change the extracted ECSA substantially [11]
CO stripping Alloyed or modified surfaces where Hupd is obscured Requires CO dosing, a clean baseline and known adsorption stoichiometry

What to do instead. Report jgeo always. Report jECSA alongside it with the method, potential window, scan rates and the Cs value with its source. Where no clean non-faradaic window exists, say so rather than assuming one.

4. Tafel slope and reaction mechanism: what the slope does not tell you

A Tafel slope is consistent with a mechanism; it does not prove one. The textbook assignments assume extreme coverage of adsorbed intermediates, and real slopes are coverage-dependent.

The assumption: 120 mV dec-1 means the Volmer step is rate-determining, 40 means Heyrovsky, 30 means Tafel recombination.

Why it fails. Those assignments come from a microkinetic treatment assuming θ close to 0 or close to 1. Real surfaces sit between those limits. Shinagawa, Garcia-Esparza and Takanabe showed that Tafel slopes are coverage-dependent across HER, HOR, ORR and OER, and concluded that while Tafel analysis remains a powerful tool for discussing rate-determining steps, oversimplified assumptions produce an inaccurate description of the surface [2].

Two further complications. Uncompensated resistance inflates the slope, so an uncorrected Tafel number is not mechanistically interpretable at all [5]. And the slope depends on the potential window it was fitted over — a value from a narrow or poorly chosen range can be produced almost at will. Anantharaj and Noda's review of how Tafel lines are actually constructed and read in the literature catalogues the resulting problems [12].


Figure 2. (a) Linear sweep voltammograms with stirring at a scan rate of 3.0 mV s-1 of 1.0 mM [Ru(bpy)₃]Cl₂ in 80 mM sodium phosphate at pH 7.2. [Ru₄POM]: 0 (black), 5 (red), 10 (blue), and 15 μM (green and orange). (b) Tafel plot in the range of potentials between 900 and 1200 mV, giving a slope of 125 mV dec-1. Read the axes before reading the slope. This is a homogeneous molecular system and the current is normalised to catalyst concentration rather than electrode area, so the plot is Tafel-like rather than a conventional log j analysis. The 125 mV dec-1 value sits close to the textbook 120 mV dec-1 that is routinely read as a one-electron rate-determining step — yet the authors show that a traditional Tafel treatment does not apply to this four-electron process at all, and that slow electron exchange at the electrode prevents standard electrochemical methods from producing a Tafel plot below 0.5 M ionic strength. A number in the expected range is not, by itself, evidence for the mechanism it is usually taken to indicate. Reproduced from Yin et al. [13], licensed under CC BY.

Related: activity extracted from fast transient sweeps rather than steady-state measurement tends to be exaggerated, which affects any Tafel fit taken from a rapid LSV [14].

What to do instead. iR-correct first. State the overpotential range the fit covers and the scan rate. Treat the slope as one line of evidence supported by coverage-aware kinetic analysis or spectroscopy, not as a mechanistic fingerprint. Reporting the slope across a stated range is explicit best practice [15].

5. Exchange current density vs limiting current density

Exchange current density describes charge transfer kinetics at equilibrium. Limiting current density is the mass transport ceiling. They are controlled by different physics and can differ by orders of magnitude.

The assumption: both are "the current", so a larger value of either means a faster electrode.

Why it fails. j0 describes the exchange rate at equilibrium, where η = 0 and the net current is zero — forward and reverse reactions proceed at equal rates. jL is a ceiling imposed by mass transport: reactant cannot reach the surface fast enough, so current stops responding to potential and the curve plateaus.

Exchange current density (j0) Limiting current density (jL)
Controlled by Charge transfer kinetics Mass transport to the surface
Located at η = 0, equilibrium The transport-limited plateau at high η
Changes with Catalyst, active area, temperature Rotation rate, viscosity, concentration, temperature
Obtained from Tafel extrapolation to η = 0 Read directly from the plateau

For a rotating disk electrode, the limiting current follows the Levich relationship and scales with the square root of the rotation rate. Koutecký–Levich analysis is the separate step that separates the kinetic and mass-transfer contributions to the measured current.

If a "kinetic" comparison is made where the curve has already flattened, what is being compared is hydrodynamics, not catalysis.

What to do instead. Establish which regime you are in before comparing. Under rotating disk control the rotation rate is part of the answer, not a nuisance parameter. A stir bar does not provide the well-defined, reproducible hydrodynamics that quantitative Levich or Koutecký–Levich analysis requires — cell geometry and electrode mounting matter as much as the potentiostat.

See electrochemical cell configurations →

6. Symbols and what each is referenced to

The right-hand column is the one that matters. These quantities share units; what they are measured against is what separates them.

Symbol Name Unit Referenced to
E Electrode potential V The stated reference electrode
η Overpotential V The equilibrium potential, E − Eeq
jgeo Geometric current density mA cm-2 Electrode footprint
jECSA Specific current density mA cm-2 Electrochemically active surface area
jmass Mass activity A mg-1 Mass of active material
j0 Exchange current density mA cm-2 Equilibrium, where η = 0
jL Limiting current density mA cm-2 The transport-limited plateau
Cdl Double layer capacitance mF Measured from scan-rate CV
Cs Specific capacitance µF cm-2 Assumed; material-specific
b Tafel slope mV dec-1 The stated fitting range
Ru Uncompensated resistance Ω Cell geometry and electrolyte

7. A reporting checklist

Reporting State alongside it
Potential Reference electrode, filling solution, temperature, conversion used
Overpotential Equilibrium potential assumed and the conditions it was calculated for; iR correction applied
Current density Normalisation basis in the axis label; catalyst loading
ECSA Method, potential window, scan rates, Cs value and its source
Tafel slope Fitting range, scan rate, iR correction

A broader critical survey of how these activity parameters are used, and misused, across the water splitting literature is worth reading alongside this list [16].

One item is deliberately missing: the counter electrode. Platinum counter electrodes can dissolve and redeposit on the working electrode, and a perfectly normalised number measured on a contaminated surface is still wrong. That is the subject of the next post in this series.

Frequently asked questions (FAQs)

What is the difference between potential and overpotential?

Potential is the electrode's position relative to a reference electrode and exists with or without a net reaction. Overpotential is the deviation of the electrode potential from its equilibrium value — the driving force for the reaction. Both are measured in volts, but only overpotential describes how far the electrode has been pushed from equilibrium.

Is ohmic drop part of the overpotential?

No. Uncompensated ohmic drop is a property of the cell and electrolyte rather than intrinsic electrode overpotential, and it should be accounted for separately. Because it scales with current, an uncorrected value at high current density can overstate the overpotential by tens of millivolts.

Should I report geometric or ECSA current density?

Report both. Geometric current density is the convention for benchmark figures such as the overpotential at 10 mA cm-2 and reflects device performance. ECSA-normalised current density supports comparison of intrinsic material activity, provided the specific capacitance used is stated and justified.

What specific capacitance should I use to calculate ECSA?

There is no universal value. 40 µF cm-2 is commonly applied in alkaline media following the McCrory benchmarking protocol, but specific capacitance is material dependent. State the value used and cite its source, because ECSA scales inversely with it and so does every activity number derived from it.

Does a Tafel slope of 120 mV/dec prove the Volmer step is rate-determining?

No. The standard assignments assume extreme coverage of adsorbed intermediates, and real Tafel slopes are coverage-dependent. A measured slope is consistent with a mechanism rather than proof of one, and it must be iR-corrected and reported with its fitting range to be interpretable.

Over what potential range should a Tafel slope be fitted?

There is no universal range, which is exactly why the range must be stated. A slope taken from a narrow or poorly chosen window can be made to land almost anywhere, so report the potential interval the fit covers, the scan rate, and whether the data was iR-corrected, and check that the fitted region is genuinely linear.

What is the difference between exchange current density and limiting current density?

Exchange current density describes charge transfer kinetics at equilibrium, where the net current is zero. Limiting current density is the mass transport ceiling, read from the plateau at high overpotential. They are controlled by different physics and can differ by orders of magnitude.

Do I need to iR-correct my data?

For any quantitative comparison, yes. Uncompensated resistance inflates measured overpotential in proportion to current and inflates Tafel slopes, so uncorrected values are not comparable between cells, electrolytes or laboratories.

Your potentiostat will not tell you which area you divided by

Every one of these five mistakes produces a clean-looking result. The CV is smooth, the Tafel plot is linear, the polarisation curve is well behaved, and the instrument reports no error. What has moved is the meaning of the axis.

State the reference and the conditions. State the normalisation basis. State the Cs and where it came from. State the fitting range. State whether it was iR-corrected. That is the difference between a number and a measurement someone else can use.

Related reading in this series

References

  1. C. C. L. McCrory, S. Jung, J. C. Peters, T. F. Jaramillo, Benchmarking Heterogeneous Electrocatalysts for the Oxygen Evolution Reaction, J. Am. Chem. Soc. 2013, 135, 16977–16987. doi.org/10.1021/ja407115p
  2. T. Shinagawa, A. T. Garcia-Esparza, K. Takanabe, Insight on Tafel slopes from a microkinetic analysis of aqueous electrocatalysis for energy conversion, Sci. Rep. 2015, 5, 13801. doi.org/10.1038/srep13801
  3. IUPAC, Compendium of Chemical Terminology (the "Gold Book"), entries for overpotential and concentration overpotential.
  4. A. Vedrtnam, K. Kalauni, R. Pahwa, Water Electrolysis Technologies and Their Modeling Approaches: A Comprehensive Review, Eng 2025, 6(4), 81. doi.org/10.3390/eng6040081 — source of Figure 1.
  5. S. Anantharaj, S. Noda, iR drop correction in electrocatalysis: everything one needs to know!, J. Mater. Chem. A 2022, 10, 9348–9354. doi.org/10.1039/D2TA01393B
  6. S. Anantharaj et al., The reference electrode dilemma in energy conversion electrocatalysis: "right vs. okay vs. wrong", J. Mater. Chem. A 2023, 11, 17699–17709. doi.org/10.1039/D3TA03145D
  7. D. M. Morales, M. Risch, Seven steps to reliable cyclic voltammetry measurements for the determination of double layer capacitance, J. Phys. Energy 2021, 3, 034013. doi.org/10.1088/2515-7655/abee33
  8. S. Anantharaj, P. E. Karthik, S. Noda, Ambiguities and best practices in the determination of active sites and real surface area of monometallic electrocatalytic interfaces, J. Colloid Interface Sci. 2023, 634, 169–175. doi.org/10.1016/j.jcis.2022.12.040
  9. Y. Jiang, J. Liu, Definitions of pseudocapacitive materials: a brief review, Energy Environ. Mater. 2019, 2, 30–37. doi.org/10.1002/eem2.12028
  10. S. Trasatti, O. A. Petrii, Real surface area measurements in electrochemistry, Pure Appl. Chem. 1991, 63, 711–734. doi.org/10.1351/pac199163050711
  11. S. Anantharaj, S. Noda, The importance of carefully choosing vertex potentials in hydrogen underpotential deposition, Mater. Today Energy 2023, 32, 101234. doi.org/10.1016/j.mtener.2022.101234
  12. S. Anantharaj, S. Noda, How properly are we interpreting the Tafel lines in energy conversion electrocatalysis?, Mater. Today Energy 2022, 29, 101123. doi.org/10.1016/j.mtener.2022.101123
  13. Q. Yin, Z. Xu, T. Lian, D. G. Musaev, C. L. Hill, Y. V. Geletii, Tafel Slope Analyses for Homogeneous Catalytic Reactions, Catalysts 2021, 11(1), 87. doi.org/10.3390/catal11010087 — source of Figure 2.
  14. S. Anantharaj, S. Kundu, S. Noda, Worrisome exaggeration of activity of electrocatalysts destined for steady-state water electrolysis by polarization curves from transient techniques, J. Electrochem. Soc. 2022, 169, 014508. doi.org/10.1149/1945-7111/ac47ec
  15. D. Voiry et al., Best practices for reporting electrocatalytic performance of nanomaterials, ACS Nano 2018, 12, 9635–9638. doi.org/10.1021/acsnano.8b07700
  16. S. Anantharaj et al., Precision and correctness in the evaluation of electrocatalytic water splitting: revisiting activity parameters with a critical assessment, Energy Environ. Sci. 2018, 11, 744–771. doi.org/10.1039/C7EE03457A

Latest Battery R&D News