Electrochemical Impedance Spectroscopy in Depth

Introduction

Electrochemical impedance spectroscopy, or EIS, perturbs a cell slightly at a series of frequencies and measures how its current lags the applied voltage. Fast electrical and interfacial processes dominate at high frequency; slower transport and reaction processes may appear at lower frequency. Equivalent circuits help organise the responses. They are models of physical pathways, not literal photographs of an electrode, so fitted resistors and capacitors must be interpreted with independent evidence.

Core explanation

For a small sinusoidal perturbation around a steady operating point, complex impedance is Z(ω) = Ṽ(ω)/Ĩ(ω) , where ω = 2πf. A resistor has real impedance R. An ideal capacitor has Z C = 1/(jωC) , with j² = −1; its response shifts phase because charge accumulates before current and voltage reach their maxima. A standard Nyquist plot uses Z′ on the horizontal axis and −Z″ on the vertical axis. Frequencies must be labelled or shown separately because position on a Nyquist curve alone does not reveal measurement frequency.

In a simplified charge-transfer model, a solution resistance R s is in series with a parallel combination of double-layer capacitance C dl and charge-transfer resistance R ct. Its impedance is Z = R s + R ct/(1 + jωR ctC dl) . Under ideal conditions its Nyquist curve is a semicircle: the high-frequency real-axis intercept is R s, and the low-frequency intercept is R s + R ct. The arc diameter is R ct. The frequency at its top is f peak = 1/(2πR ctC dl). Real arcs are often depressed because surface roughness, distributed reaction rates and non-ideal capacitance produce a range of time constants.

For a sufficiently simple one-electron, activation-controlled interfacial process near equilibrium, R ct = RT/(nFi₀) if i₀ is the exchange current for the whole electrode under the specified convention; using exchange current density requires area-normalised resistance. This relation connects a small-signal EIS parameter with Butler–Volmer kinetics. It is not valid if a measured arc is actually caused by a film, contact resistance or another process. Temperature and operating potential must be specified.

Diffusion can add a frequency-dependent Warburg response. For ideal semi-infinite linear diffusion, Z W is proportional to (1−j)/√ω, giving roughly equal positive real and negative imaginary components. A Nyquist plot then has an approximately 45° tail in the corresponding frequency region. Finite-length pores or bounded diffusion bend away from that infinite-diffusion ideal at low frequency. A 45° line can also arise from distributed porous-electrode transmission-line behaviour, so interpreting it as one diffusion coefficient without geometry and model checks is unsafe.

Circuit selection should follow the electrode physics. A model with many flexible elements may fit almost any curve but yield nonunique parameters. Check fit residuals, frequency range, perturbation amplitude, stability during the scan and whether parameters change plausibly with temperature or concentration. EIS is a local linear technique: too large a perturbation or a drifting cell violates the assumptions behind one fixed impedance spectrum.

Step-by-step reasoning

Choose a steady bias and small perturbation amplitude. Inspect high-frequency intercept, arcs and low-frequency trends. Start with the simplest physically motivated circuit, then calculate R ct from a well-resolved arc diameter if justified. Use f peak to infer C dl only for the ideal one-time-constant circuit. Test whether a low-frequency tail follows semi-infinite diffusion or a finite/porous geometry. Compare fitted values with independent DC kinetics and device dimensions.

Visual explanation

Draw R s in series with parallel R ct and C dl. Below it draw a Nyquist semicircle beginning at R s and ending at R s + R ct; label increasing frequency toward the left intercept. Append a 45° low-frequency tail for ideal diffusion, then draw a second depressed arc to remind readers that real surfaces are not perfect capacitors. Place the Bode plot beside it so frequency remains visible.

Real-world analogy

Tap a water-filled pipeline at different rhythms. At rapid tapping, the immediate pipe resistance dominates; at slower rhythms, storage in flexible sections and the time for water to redistribute become visible. This parallels multiple response times in EIS. The analogy cannot replace the complex-number phase calculation or distinguish electrochemical diffusion from fluid flow by itself.

Real-world example

A researcher records EIS for an electrode before and after cycling a battery cell. The high-frequency intercept changes little, but a mid-frequency arc grows. An interpretation is increased interfacial resistance, perhaps from a surface film or slower charge transfer. To distinguish those possibilities, the researcher examines potential dependence, film chemistry and a control electrode. Labelling every arc “R ct” without this evidence would overstate what EIS alone establishes.

Why?

Why can EIS separate processes? Processes with different characteristic times respond differently as frequency changes. Why is a small perturbation used? It keeps the response near one operating point, where linearisation is meaningful. Why can a simple circuit fail for a porous electrode? Reaction and transport are distributed through a three-dimensional network rather than located at one uniform interface.

Common misconception

The diameter of any Nyquist arc is not automatically an intrinsic electron-transfer resistance. Overlapping films, grain boundaries and porous paths can create arcs. A Warburg-like tail is also not proof of one unique diffusion coefficient. Circuit parameters are conditional on geometry, bias, temperature and chosen model.

Worked example

Question: An ideal Nyquist semicircle begins at 5 Ω and ends at 45 Ω on the real axis. Its top occurs at 100 Hz. Find R ct and C dl for the simple R s + (R ct parallel C dl) circuit.

Reasoning: R ct is the diameter, 45 − 5 = 40 Ω. At the peak, 2πfR ctC dl = 1, so C dl = 1/(2π × 100 s⁻¹ × 40 Ω) ≈ 3.98 × 10⁻⁵ F. These values describe the assumed lumped circuit, not necessarily a unique physical decomposition.

Answer: R ct = 40 Ω and C dl ≈ 40 µF.

Quick check

1. On the ideal Nyquist arc, what does the high-frequency real-axis intercept represent? Answer: The series or solution resistance R s in the specified simple circuit.

Exam focus

Write complex impedance with correct capacitor phase and state the chosen circuit. For a semicircle, label both intercepts and take their difference for R ct. Distinguish an ideal semi-infinite Warburg tail from finite diffusion and transmission-line effects, and mention that fit quality alone does not establish mechanism.

Advanced insight

Kramers–Kronig consistency tests can check whether an impedance dataset behaves like a causal, stable, linear response over the measured range. Passing them does not tell which equivalent circuit is correct, but failing them can expose drift or nonlinearity. Constant-phase elements often improve fits to depressed arcs yet introduce parameters that should not be casually reported as a literal double-layer capacitance. Converting them to an effective capacitance requires a stated model and assumptions.

Summary

EIS probes frequency-dependent voltage–current response around an operating point. A simple Randles-type arc can reveal series resistance, charge-transfer resistance and an ideal capacitance when its assumptions hold. Diffusion may produce a Warburg-like tail. Real electrodes often have distributed and overlapping processes, so circuit assignments need physical controls and model-aware interpretation.

Practice questions

1. An ideal arc intercepts the real axis at 3 Ω and 23 Ω. What is R ct in the simple circuit? Answer: R ct = 23 − 3 = 20 Ω.

2. What is the phase relation for an ideal capacitor in impedance form? Answer: Z C = 1/(jωC) = −j/(ωC), so voltage and current are phase-shifted by 90° under the ideal convention.

3. Why might a low-frequency 45° feature not prove semi-infinite molecular diffusion? Answer: Distributed ionic and electronic pathways in a porous electrode can give a similar transmission-line response.

4. What condition is violated if a battery's state of charge drifts significantly during one EIS scan? Answer: The assumed stationary operating point is lost, so one time-invariant linear impedance cannot describe the entire scan.

Sources: Gamry Instruments, Basics of Electrochemical Impedance Spectroscopy; BioLogic, EIS guide.