Conductivity, voltammetry and impedance evidence

Each method perturbs a different response

Conductivity measurements infer transport with a geometry and contact model. Voltammetry varies potential and records current. Impedance applies a small oscillatory perturbation and resolves amplitude and phase over frequency around a specified state.

A lower measured resistance, a larger sweep current and a smaller impedance arc are different observations. Before comparing materials, identify what was controlled, where voltage and current were measured, and the specimen’s state. The examples here use three separate fictional records; they are not three tests on one real material.

The first record is a uniform specimen region at steady current. The second controls an electrode potential during a ramp. The third measures a small periodic response about a fixed operating state. None of their supplied numbers is a real material rating, chemical identification or device qualification.

Match voltage, current and geometry to one material region

Consider a hypothetical stationary, homogeneous gauge region of length L = 5.00 mm and cross-sectional area A = 2.00 mm² = 2.00 × 10⁻⁶ m². Its temperature and state stay fixed. Assume a uniform ohmic current, negligible voltage-sense current and no parallel leakage. Current is driven through outer contacts; voltage is sensed between inner points that span exactly L. These are separate connections, not a guarantee that every four-terminal measurement is valid.

For this region, J = I/A and E = V_bulk/L. If J = σE, then R_bulk = V_bulk/I = L/(σA), hence σ = L/(R_bulkA). R has units Ω; σ has units S/m, where S = Ω⁻¹. A bulk potential difference must span the same length and current cross-section used in the equation. Conductors, insulators and semiconductors develops that geometry and response distinction.

Supply signed observations at the same fixed voltage-sense polarity: V_obs,+ = +102.0 mV at I = +100.0 µA, and V_obs,− = −98.0 mV at I = −100.0 µA. Use the explicit local model V_obs = IR_bulk + V_off, with one unchanged additive offset during both observations. Subtracting gives R_bulk = (V_obs,+ − V_obs,−)/(2I₀) = 1000 Ω; averaging gives V_off = 2.00 mV. The recovered σ is 2.50 S/m. Reversal cancels that common offset, not every possible error.

In a separate outer-terminal comparison, suppose leads and contacts add a stipulated constant 200 Ω in series outside this same 5.00 mm gauge. The offset-corrected terminal ratio is then 1200 Ω. Using it as R_bulk would give an apparent 2.08333 S/m even though the actual supplied bulk conductivity remains 2.50 S/m. An improved outer contact could change the terminal reading without changing the material region.

The uniform-path and unchanged-offset assumptions need evidence. Reversal can fail to cancel a changing offset; current can alter temperature or composition, and sensing contacts can sample a different field. Conductivity alone also does not identify the carrier species. Ionic, electronic and mixed transport distinguishes those contributions, while From measured transport to device behaviour follows the complete measurement path. No van der Pauw or Hall procedure is prescribed here.

A voltammogram plots current against controlled electrode potential

In voltammetry, the working electrode potential E is controlled relative to a stated reference electrode while current is recorded. In a representative three-electrode arrangement, current flows between the working and counter electrodes; the high-impedance reference connection monitors potential with negligible current in the ideal model. E is not automatically the full working-to-counter cell voltage. Reference identity and placement, electrolyte voltage drop and instrument control matter.

A linear sweep changes E at a declared scan rate v = dE/dt, in V/s. A return sweep reverses the sign of v. The voltammogram’s horizontal axis is E versus the reference, and its vertical axis is signed current I, or a stated area-normalised current density. Time is implicit in the sweep programme; two points at the same E can occur on different scan branches with different histories.

Current can include interfacial charge rearrangement, electron-transfer reactions, transport and unwanted paths. A peak, onset or loop does not by itself identify a chemical reaction or a material’s stable operating window. Interfaces, double layers and charge-transfer kinetics develops the processes; here the task is to read a bounded output before assigning one.

For the next fictional record, stipulate that the controlled E also represents the intended local potential coordinate, with no unaccounted solution or lead drop, no changing composition or temperature, and no extra current path. Measurements that fail those conditions require a different interpretation. The ramp and current relation are supplied models, not a prescription for a real electrochemical experiment.

Scan direction separates two supplied current terms

Use the original teaching law I(E,v) = g(E − E₀) + Cv. Here E₀ = 0.200 V is the supplied zero-current coordinate of the first term, not an identified standard reaction potential. The constant g = 20.0 µS describes an instantaneous conductive contribution; C = 100 µF is a constant capacitance. Since a constant capacitor has charge change dQ = C dE, its charging current is dQ/dt = C dE/dt = Cv. No microscopic identity is assigned to either term.

Sweep E from 0.100 to 0.300 V at v = +0.0500 V/s, then back at −0.0500 V/s. Each branch takes 4.00 s. Within each ideal branch, Cv = +5.00 or −5.00 µA. The conductive term ranges from −2.00 to +2.00 µA. Thus the forward branch runs from +3.00 to +7.00 µA and the reverse branch from −7.00 to −3.00 µA when plotted with E increasing left to right. The return is traversed from right to left. Ideal instantaneous turning points are not instrument-resolved transients.

At any common E on these branches, their average is [I_forward + I_reverse]/2 = g(E − E₀); their difference is I_forward − I_reverse = 2C|v|. At E = 0.200 V the currents are +5.00 and −5.00 µA, so C = (10.0 µA)/(0.100 V/s) = 100 µF. At E = 0.300 V they are +7.00 and −3.00 µA; the 2.00 µA average recovers g = 20.0 µS over a 0.100 V change.

This separation works because the conductive term is stipulated identical on forward and reverse branches, C is constant and the two branches share the same local coordinate and state. Doubling |v| doubles the charging term, not necessarily every part of current. In a real specimen, adsorption, diffusion, reactions, changing capacitance and uncompensated voltage can invalidate the simple decomposition. A larger separation or a peak is not automatically more desired reaction current.

Impedance retains both an amplitude ratio and a phase

Now begin a separate hypothetical linear periodic measurement. Write δv(t) = V_a cos(ωt) for voltage relative to a specified bias, and δi(t) = I_a cos(ωt − φ) for current relative to its bias. V_a and I_a are positive peak amplitudes, f is cycles per second in hertz and ω = 2πf is angular frequency in radians per second. The signed impedance phase is φ = phase(voltage) − phase(current). Negative φ means current leads voltage in this convention.

Impedance magnitude is |Z| = V_a/I_a, in ohms. The same ratio follows from root-mean-square (RMS) amplitudes only if both use RMS; mixing peak voltage with RMS current introduces a √2 error. A phase and an amplitude ratio at a single frequency are not a complete spectrum. The bias, temperature, specimen history, perturbation amplitude and elapsed time belong with the record.

For compact circuit reasoning write Z = Z′ + jZ″, where j is the imaginary unit, j² = −1. The real part Z′ represents the in-phase voltage/current relation and the imaginary part Z″ encodes the quadrature component. This uses complex sinusoidal amplitudes with time dependence exp(jωt), whose real part is cos(ωt). It is notation for amplitude and timing, not an additional physical current. Then |Z| = √(Z′² + Z″²) and φ is the angle of Z.

A passive ideal resistor has Z_R = R. For a capacitor, i = C dv/dt means current leads its voltage by a quarter cycle, giving Z_C = 1/(jωC) = −j/(ωC). These relations describe settled sinusoidal responses. They do not imply that every real electrode is one ideal capacitor or that a chosen millivolt amplitude is always small enough. Local linearity and a stable operating state must be assessed.

Read one labelled resistor–capacitor spectrum

Use a separate original circuit with a series resistor R_s = 100 Ω and a parallel branch containing R_p = 400 Ω and C_p = 25.0 µF. All are ideal positive constants. The circuit is an illustrative one-relaxation model, not an identified electrode or a complete electrochemical cell. Its parallel branch has a common voltage v_p and currents that add: i = v_p/R_p + C_p dv_p/dt. The series resistor carries their total.

Hats denote complex amplitudes of the corresponding currents and voltages. For sinusoidal amplitudes, the branch relation becomes Î = (1/R_p + jωC_p)V̂_p. The total voltage is R_sÎ + V̂_p, so Z(ω) = R_s + R_p/(1 + jωR_pC_p). Define τ = R_pC_p = 0.0100 s and x = ωτ. Algebra gives Z′ = R_s + R_p/(1 + x²) and Z″ = −R_p x/(1 + x²). The products and units agree: ΩF = s and x is dimensionless.

A Nyquist plot uses Z′ horizontally and, here, −Z″ vertically, both in Ω. The minus sign is part of the declared axis convention. For this model, [Z′ − (R_s + R_p/2)]² + (−Z″)² = (R_p/2)², so the positive-height arc has centre (300, 0) and radius 200 Ω. Increasing frequency runs from the low-frequency model limit (500, 0) towards the high-frequency limit (100, 0). Finite-frequency data do not reach those exact limiting endpoints.

The arc height is greatest at x = 1, where f_p = 1/(2πτ) = 15.9155 Hz and Z = 300 − j200 Ω. The magnitude is 360.555 Ω and φ = −33.6901°, not −45°, because R_s also contributes. With the supplied 10.0 mV peak voltage, current peak amplitude is 27.7350 µA and leads by 33.6901°. Its peak occurs earlier within a cycle; it does not precede application of the excitation.

Bode-style panels keep frequency visible: one plots |Z| against f and another plots φ against f. The figure uses logarithmic frequency spacing and separately labelled vertical scales. At low frequency the capacitor admits little current and the response approaches R_s + R_p. At high frequency it provides an easier parallel route and the response approaches R_s. These are the circuit’s limits, not proof that a measured high-frequency intercept is always a specimen’s bulk resistance. A fitted R_s can include other unresolved series contributions.

Test the response before interpreting a fitted circuit

Classical impedance describes a local linear, time-invariant response about the chosen operating state. Linearity means proportional response and superposition within the tested range; time invariance means the response law does not change while the record is collected. A stable nonzero bias is possible; zero DC current is not a universal requirement. Long low-frequency measurements can be especially sensitive to changing state.

In a supplied amplitude check at the arc-height frequency, increasing peak voltage from 10.0 to 20.0 mV gives current peaks 27.7350 and 55.4700 µA with unchanged phase. That is consistent with local linearity over this check. If the second current were instead 45.0 µA, a single amplitude-independent Z would not describe both exactly. Noise, instrument limitations and changed state also need examination; one failed ratio does not identify its microscopic cause.

Repeat a frequency after the sweep or compare records at controlled elapsed times. A changing response challenges stationarity, even if every individual trace looks nearly sinusoidal. New harmonics or amplitude-dependent estimates challenge the simple linear model. Passing a few checks supports the stated domain rather than proving validity under every bias, temperature, ageing history or frequency.

A measured arc may combine bulk, grain-boundary, contact, distributed geometry and interfacial responses. The physical assignment needs controlled changes or independent structural, geometric and chemical evidence. The same carrier/condition caution applies to conductivity and voltammetry: use the actual measurement boundary, not just the technique’s name.

A fit is not a unique mechanism

Equivalent circuits can describe features without proving a particular microscopic origin. Check linearity, stationarity, reference electrodes and independent structural or chemical evidence. Two circuits may fit one frequency range similarly yet imply different processes.

Nonuniqueness can remain even with a perfect full spectrum. Replace the single R_p parallel C_p branch by two such branches in series: R₁ = 150 Ω with C₁ = 66.6667 µF, and R₂ = 250 Ω with C₂ = 40.0 µF. Here C₁ is the rounded display of the exact supplied value τ/150, and C₂ = τ/250. Each pair has the same τ = 0.0100 s. Their impedances add to 150/(1 + jωτ) + 250/(1 + jωτ) = 400/(1 + jωτ). Retaining R_s = 100 Ω therefore reproduces the entire original Z(ω).

One interpretation might assign the one relaxation to a single region; another could assign two coincident relaxations to distinct regions. The algebra supplies neither physical identity. Measuring more frequencies cannot distinguish these exactly identical model families. A controlled perturbation that separates the regions’ responses, spatial evidence or an independently justified mechanism is needed.

Conversely, one frequency supplies much less information than a spectrum. At ω = 100 rad/s, a series combination of 300 Ω and 50.0 µF also gives 300 − j200 Ω. At ω = 10 rad/s it gives 300 − j2000 Ω, unlike the original 496.040 − j39.6040 Ω. A new frequency can distinguish those particular alternatives. The right next observation depends on which ambiguity remains.

Report what was measured, what the circuit assumes and which parameters remain unresolved. Calibration, uncertainty and repeatability develops offsets and the measurement model; Charge, chemical state and electrochemical potentials distinguishes local/reference coordinates from complete-cell energy and voltage. A descriptive fit is evidence about a stated response, not proof of safe service or a selected chemical process.

Worked examples

Try the reasoning

1. Why keep perturbations small for linear impedance analysis?

   Show the answer and reasoning

   To remain near a linear response about the state.

   Large perturbations can activate different behaviour.

2. Can a drifting cell violate a stationary model?

   Show the answer and reasoning

   Yes.

   The state changes while frequencies are being measured.

3. Keep a true 1000 Ω gauge and ±100.0 µA, but let the offset be +2.00 mV for the positive observation and +6.00 mV for the negative one. What resistance would the common-offset formula report? Is reversal alone enough?

   Show the answer and reasoning

   It reports 980 Ω, 2.00% below the supplied true value. Reversal alone does not cancel the changing offset.

   The observations are +102.0 and −94.0 mV. Their difference is 196.0 mV; divide by 200.0 µA. The original cancellation required one offset, not two different states.

4. A different voltage-sense span of 2.50 mm gives 50.0 mV at 100.0 µA in the same uniform material and area, after valid offset correction. Find σ. What if the old 5.00 mm is mistakenly retained?

   Show the answer and reasoning

   Correct σ remains 2.50 S/m. The mismatched old length would report 5.00 S/m.

   The shorter-span resistance is 500 Ω. Its own L/(RA) gives 2.50 S/m; pairing its voltage with twice the actual length doubles the reported value without changing the material.

5. In another fictional matched-branch record at E − E₀ = +0.100 V and |v| = 0.0200 V/s, forward current is +8.00 µA and reverse current +2.00 µA. If the same two-term model is supplied, find C and g. May both branches be positive?

   Show the answer and reasoning

   C = 150 µF and g = 50.0 µS. Both are positive because the +5.00 µA conductive term exceeds the ±3.00 µA charging term.

   Difference 6.00 µA divided by 0.0400 V/s gives C. Average 5.00 µA divided by 0.100 V gives g. Sweep direction and sign of total current are different quantities.

6. A new ideal model has the same R_s = 100 Ω and τ = 0.0100 s but R_p = 200 Ω. Find C_p and the Nyquist low-frequency limit, high-frequency limit and greatest-height point. Did the unchanged characteristic frequency fix R_p and C_p separately?

   Show the answer and reasoning

   C_p = 50.0 µF; limits are 300 and 100 Ω; the highest point is (Z′, −Z″) = (200, 100) Ω. The frequency alone did not fix the two parameters.

   τ/R_p gives C_p. The arc width and height depend on R_p even though f_p = 1/(2πτ) is unchanged. Its phase at that point is arctan(−100/200) ≈ −26.5651°.

7. The original model and a series 300 Ω/50.0 µF model agree at ω = 100 rad/s. Compare them at ω = 10 rad/s and choose whether a second frequency can distinguish these alternatives.

   Show the answer and reasoning

   Original: about 496.040 − j39.6040 Ω. Series model: 300 − j2000 Ω. These particular alternatives are distinguishable at the second frequency.

   The original uses x = ωτ = 0.100. The series capacitor has imaginary part −1/(ωC) = −2000 Ω. This differs from the exact all-frequency degeneracy of the two coincident parallel-pair models.

8. At the same nominated state and frequency, 10.0 mV peak gives 27.7350 µA, but 20.0 mV gives 45.0 µA rather than 55.4700 µA. A repeat at the first amplitude later gives a changed phase too. What has failed to be established?

   Show the answer and reasoning

   An amplitude-independent, time-invariant response over those observations has not been established.

   Proportional scaling and repeatable phase are predicted by the chosen local LTI model. The discrepancies call for amplitude, state, timing, noise and instrument checks. They do not uniquely identify a reaction or validate an arbitrary universal small-amplitude threshold.

What to carry forward

• State the perturbation and measurement model.
• Separate descriptive fits from mechanism identification.
• Use the voltage span and cross-section that actually define the bulk region.
• Scan branch, rate and reference coordinate matter to a voltammogram’s current.
• Impedance requires a declared amplitude/phase convention and frequency, not only one resistance-like number.
• A well-labelled ideal arc can identify a conditional response without identifying a unique physical region.