Chapter 3: Kinetics — Overpotential and Tafel Analysis

Why a Reaction That Thermodynamics Permits Can Still Refuse to Happen, and How to Read the Price of Making It Hurry

📖 Reading Time: 20-25 minutes 📊 Difficulty: Beginner 💻 Code Examples: 0 📝 Exercises: 0

Video Lecture

This video covers the same content as the text below. Choose your preferred learning format.

🌐 EN | 🇯🇵 JP | Last sync: 2026-08-20

Materials Science Dojo > Electrochemistry Introduction > Chapter 3

Chapter 2 ended on a satisfying number. Water splits at \(1.23\) V; the Daniell cell delivers \(1.10\) V; the Nernst equation tells you exactly how those numbers shift when concentrations change. Thermodynamics answered the question is this reaction allowed, and how much voltage does it owe or demand? — and it answered completely, from tabulated quantities, without any reference to what the electrode is made of.

That completeness is the trap. Build a water electrolyser, apply \(1.23\) V across it, and essentially nothing happens. Not "a little happens" — nothing you would notice. To pull useful current you may need to push half a volt or more beyond the thermodynamic requirement, and how much more depends entirely on what the electrodes are made of, a variable that never appeared anywhere in Chapter 2.

This chapter is about that gap. The gap has a name — overpotential — and it is not a defect in the theory of Chapter 2. Thermodynamics never claimed to say anything about speed. It said what the final state costs, not how fast the system gets there, and those are genuinely independent questions. A boulder at the top of a hill will roll down; thermodynamics guarantees it. Whether it rolls down this afternoon or in ten thousand years is a question about the shape of the path, not about the height of the hill.

For electrochemistry, the shape of the path is where all the engineering lives. Every catalyst, every electrode material, every dollar spent on platinum or iridium is spent on kinetics. So it is worth being precise about what kinetics can and cannot buy.

3.1 Equilibrium Is Not a Promise of Speed

Start with what equilibrium actually means at an electrode, because the everyday reading of the word is misleading.

Put a zinc electrode in a zinc sulfate solution and leave it alone. Chapter 2 says a potential establishes itself — the equilibrium potential \(E_{\text{eq}}\) given by the Nernst equation. It is easy to picture this as a static situation: nothing is happening, the system has settled, the meter reads a number.

That picture is wrong, and the correct picture is the foundation of everything in this chapter.

At equilibrium, both directions of the reaction are running at full speed. Zinc atoms are leaving the metal and entering the solution as \(\text{Zn}^{2+}\), and \(\text{Zn}^{2+}\) ions are arriving at the surface and plating out as metal. Neither has stopped. What is true at equilibrium is only that the two rates are exactly equal, so the net change is zero and the external meter, which can only see the net, reads zero current.

\[ \text{Zn} \;\rightleftharpoons\; \text{Zn}^{2+} + 2e^- \]

Equilibrium is a balance, not a halt. And this immediately raises a question thermodynamics has no way to ask: how fast is the balanced traffic?

Two electrodes can sit at exactly the same equilibrium potential — thermodynamically indistinguishable, identical Nernst equation, identical \(\Delta G\) — and have wildly different amounts of traffic crossing the interface in both directions. One might have ions hopping across constantly; the other might have an interface so sluggish that only a trickle crosses. Both read the same voltage. Both are at equilibrium. They are not the same electrode, and the difference will dominate everything you try to do with them.

That difference is the subject of the next several sections. But first, the quantity that measures how hard you are pushing.

3.2 Overpotential: The Price of Going Somewhere

Overpotential is defined with almost embarrassing simplicity. It is the difference between the potential you actually apply to an electrode and the equilibrium potential of the reaction happening there:

\[ \eta \;=\; E_{\text{applied}} \;-\; E_{\text{eq}} \]

If \(\eta = 0\), you are sitting exactly at equilibrium and the net current is zero. Push \(\eta\) positive and you drive the oxidation direction; push it negative and you drive the reduction direction. The sign convention follows the reduction-potential convention from Chapter 1: positive \(\eta\) is anodic (oxidation, electrons leaving the solution into the electrode), negative \(\eta\) is cathodic (reduction).

The important thing about \(\eta\) is not the definition but the accounting. Overpotential is wasted energy. Every volt of it multiplies the charge you pass and comes out as heat instead of as product. An electrolyser running at \(1.23\) V of thermodynamic requirement plus \(0.6\) V of overpotential is converting roughly a third of its electrical input into warming up the room. Reducing overpotential is not a refinement — it is the main lever on the efficiency of every electrochemical device that exists.

And overpotential is not one thing. It is a sum of at least three physically distinct penalties, which have different causes, different cures, and different signatures.

📚 The Three Contributions to Overpotential

Contribution Physical cause How it depends on current How you reduce it
Activation \(\eta_{\text{act}}\) The charge-transfer step itself has an energy barrier at the interface Logarithmic — this is the Tafel behaviour of this chapter Better catalyst; more real surface area; higher temperature
Concentration \(\eta_{\text{conc}}\) Reactant is consumed at the surface faster than transport can replace it Small until you approach a limiting current, then it diverges Stir; raise concentration; thinner diffusion layer; flow cell
Ohmic \(iR\) Electrolyte, membranes, contacts and wires all have resistance Strictly linear in current More conductive electrolyte; shorter path; thicker contacts; electronic compensation

Three things are worth extracting from that table before we go on.

They add up. The total driving voltage a real cell needs is the thermodynamic minimum plus all three penalties, at both electrodes:

\[ E_{\text{cell}} \;=\; E_{\text{thermo}} \;+\; |\eta_{\text{anode}}| \;+\; |\eta_{\text{cathode}}| \;+\; iR \]

Chapter 5 builds exactly this stack for a water electrolyser and shows which term dominates.

They have different current dependences, which is how you tell them apart. This is more useful than it sounds. If you double the current and the extra penalty doubles exactly, you are looking at ohmic loss. If doubling the current costs a fixed number of millivolts regardless of where you started, you are looking at activation loss — a logarithm has that property and nothing else does. And if the penalty suddenly runs away to infinity as you push harder, you have hit a transport limit. A well-designed measurement separates the three by exploiting these signatures.

Only one of them is chemistry. Ohmic loss is a property of the cell's plumbing. Concentration loss is a property of the fluid mechanics. Activation overpotential is the only one that is a property of the reaction on that particular surface, and it is therefore the only one a catalyst can address. When someone says a catalyst "reduces the overpotential", they mean the activation term, and the rest of this chapter is about that term specifically.

3.3 Exchange Current Density: How Busy the Equilibrium Is

Return to the question Section 3.1 raised: at equilibrium, how much traffic is crossing the interface?

That traffic has a name. The exchange current density \(i_0\) is the magnitude of the one-directional current flowing in each direction when the electrode sits at equilibrium. Anodic and cathodic partial currents are each equal to \(i_0\); they cancel; the net is zero.

\[ i_{\text{anodic}} = i_0, \qquad i_{\text{cathodic}} = -i_0, \qquad i_{\text{net}} = 0 \]

The units are current per unit area — A/cm² is the usual choice — because what matters is the traffic per unit of interface, not the total, which would just tell you how big your electrode is.

Here is the intuition to keep. \(i_0\) measures how busy the equilibrium is. A large \(i_0\) describes an interface where the reaction is constantly running in both directions, a crowded two-way street with the traffic exactly balanced. A small \(i_0\) describes an interface where crossings are rare events — the same balance, but achieved by almost nothing happening in either direction.

Now the consequence, which is the whole reason the quantity matters. A busy equilibrium is easy to unbalance. If a huge two-way flow is already running, a small nudge of potential tilts it slightly and immediately produces a large net current. If the equilibrium is nearly dead, you have to push very hard before the net flow becomes measurable. So:

That is the operational definition of catalysis in electrochemistry. A catalyst is a surface with a large \(i_0\) for the reaction you care about.

📚 An Honest Note on the Numbers

Textbooks and papers tabulate exchange current densities for particular reactions on particular metals, and it is tempting to quote them. This series will not, for a reason worth understanding rather than just accepting.

Exchange current density is extremely sensitive to the surface. Two samples of the same metal, differing only in how they were polished, cleaned, or how long they have been sitting in air, can give \(i_0\) values that differ by orders of magnitude. Trace impurities in the electrolyte adsorb onto exactly the sites that do the chemistry. Crystallographic facet matters. The values reported in the literature scatter accordingly, and a specific number quoted without the full surface preparation attached is close to meaningless. This is also why serious electrocatalysis work pairs electrochemical measurements with surface-sensitive characterisation — XPS above all, covered in the Introduction to Spectroscopy series — rather than trusting the nominal composition of the electrode.

What is robust — and what you should carry away — is the span. Between an excellent catalyst and a poor one for the same reaction, \(i_0\) can differ by many orders of magnitude. That is the qualitative fact that drives the entire field, and Section 3.8 will compute exactly what such a span costs you in volts. The specific digits are somebody's careful measurement of somebody's specific surface, and they do not transfer.

3.4 The Butler–Volmer Equation

We now have the two ingredients: a measure of how busy the equilibrium is (\(i_0\)), and a measure of how hard we are pushing (\(\eta\)). The Butler–Volmer equation connects them to the net current.

\[ i \;=\; i_0 \left[ \exp!\left( \frac{(1-\alpha)\, n F \eta}{RT} \right) \;-\; \exp!\left( \frac{-\alpha\, n F \eta}{RT} \right) \right] \]

It looks forbidding. It is not, because it is really just two statements written next to each other.

Read it as two terms, one per direction. The first exponential is the anodic (oxidation) partial current. The second is the cathodic (reduction) partial current. The net current is their difference, because they carry charge in opposite directions. Everything else is bookkeeping about how strongly each direction responds to potential.

Why exponentials? Because the charge-transfer step has an activation barrier, and the rate of crossing a barrier depends exponentially on its height — the same Arrhenius logic that governs every thermally activated process in chemistry. What is special about an electrochemical barrier is that you can change its height with a knob on the front of an instrument. Shifting the electrode potential by \(\eta\) shifts the energy of the electron, which shifts the barrier for the reaction that moves it. Very few reactions in chemistry let you dial the activation energy continuously; this is the property that makes electrochemistry such a precise experimental science.

What is \(\alpha\)? The transfer coefficient, or symmetry factor, a number between 0 and 1 that says how the applied overpotential is shared between the two directions. If you apply \(\eta\), the barrier for one direction falls by the fraction \(\alpha\) of \(nF\eta\) and the barrier for the other rises by \((1-\alpha)\) of it. Physically, it encodes where along the reaction path the transition state sits: a value near \(0.5\) means the transition state is roughly halfway, so the potential helps the forward direction about as much as it hinders the reverse. \(\alpha = 0.5\) is the conventional starting assumption, and it is a modelling choice, not a measured constant of nature — real systems deviate, and multi-step mechanisms can produce effective values well outside the range a single-step picture allows.

Notice what the equation says at \(\eta = 0\): both exponentials equal 1, the bracket vanishes, and \(i = 0\). Equilibrium falls straight out. But \(i_0\) is still sitting in front, describing traffic that the net current cannot see.

📚 What Butler–Volmer Assumes

The equation is a workhorse, and like most workhorses it is honest about very little unless you ask. It assumes:

None of these caveats stops the equation from being the right first model. They do mean that a Butler–Volmer fit is a description of behaviour, not a proof of mechanism.

3.5 Two Limits: Straight Line and Straight Log

The Butler–Volmer equation is not especially convenient to work with as written. Its value is that in the two regimes that matter, it collapses into something you can read off a graph.

Small overpotential: the linear region. When \(|\eta|\) is small compared with the thermal scale \(RT/F\) — a few tens of millivolts at room temperature — both exponentials can be expanded as \(e^x \approx 1 + x\). The 1's cancel between the two terms, the \(\alpha\) and \((1-\alpha)\) sum to unity, and what remains is a straight line:

\[ i \;\approx\; i_0 \, \frac{nF}{RT} \, \eta \]

Current proportional to overpotential is Ohm's law in disguise, and the proportionality constant has the units of a conductance. Its reciprocal is called the charge-transfer resistance:

\[ R_{\text{ct}} \;=\; \frac{RT}{nF\, i_0} \]

This is a genuinely useful result. Near equilibrium, an electrochemical interface behaves like a resistor, and that resistor's value is inversely proportional to \(i_0\). It is why small-amplitude techniques — impedance spectroscopy above all — can extract \(i_0\) without ever driving the system far from equilibrium and without changing the surface in the process. Section 3.8 measures how far this approximation can be stretched before it breaks.

Large overpotential: the Tafel region. Push \(\eta\) far in one direction and one exponential grows while the other collapses. For large positive \(\eta\), the cathodic term becomes negligible and only the anodic one survives:

\[ i \;\approx\; i_0 \exp!\left( \frac{(1-\alpha) n F \eta}{RT} \right) \]

Take the logarithm of both sides and the exponential becomes a straight line:

\[ \log_{10}|i| \;=\; \log_{10} i_0 \;+\; \frac{(1-\alpha) n F}{2.303\, RT}\, \eta \]

That is the Tafel equation, and the plot of \(\log_{10}|i|\) against \(\eta\) is the Tafel plot. It is the single most-used graph in electrochemical kinetics, for a reason that is entirely practical: it turns a curve with two unknowns into a straight line whose slope gives you \(\alpha\) and whose intercept gives you \(i_0\). Both parameters, from one measurement, by drawing a line.

3.6 The Tafel Slope, and Where 118 mV Comes From

Electrochemists usually quote the Tafel relation the other way up, as volts per decade of current rather than decades of current per volt, because the resulting number is a directly meaningful quantity: how much extra overpotential does it cost to increase the current tenfold?

\[ \eta \;=\; a \;+\; b \log_{10}|i|, \qquad b \;=\; \frac{2.303\, RT}{\alpha n F} \]

The quantity \(b\) is the Tafel slope, in millivolts per decade. Put in \(R = 8.314\) J/(mol·K), \(F = 96485\) C/mol, \(T = 298.15\) K and \(\alpha = 0.5\) with \(n = 1\), and you get the number this chapter's code will confirm: about 118 mV/decade.

It is worth seeing where the pieces of that number come from, because the same pieces built the Nernst slope in Chapter 2.

So the famous 118 mV is not an experimental fact about any material. It is \(59 / 0.5\), a direct consequence of assuming that the applied potential is split evenly between the forward and reverse barriers. Change the assumption and the number changes with it, in a completely predictable way — Section 3.8 tabulates the mapping.

📚 Reading a Tafel Slope in the Wild

Because the slope depends on \(\alpha\) and on how many electrons precede the rate-determining step, a measured Tafel slope is often used as a mechanistic fingerprint: 118, 59, 39 and 30 mV/decade all correspond to recognisable combinations. This is a legitimate and widely used inference, and it is also easy to over-read. Three warnings:

The honest use of a Tafel slope is comparative: same cell, same electrolyte, same fitting window, different electrode materials. Then the differences mean something.

3.7 What a Catalyst Actually Changes — and What It Cannot Touch

We can now state the most important structural fact in electrochemistry, and it takes one sentence.

A catalyst changes kinetics. It cannot change thermodynamics.

Look back at the Butler–Volmer equation. Everything a catalyst influences — \(i_0\), \(\alpha\), and hence the Tafel slope — sits inside the expression for the current. The equilibrium potential \(E_{\text{eq}}\), which is the origin of the \(\eta\) axis, does not appear as an adjustable quantity at all. It is fixed by \(\Delta G\) of the overall reaction, and \(\Delta G\) is a difference between initial and final states. A catalyst is neither.

So, concretely:

This division of labour is why the field looks the way it does. Thermodynamics tells you which reactions are worth attempting and sets the target. Kinetics tells you how close to the target you can get, and is where essentially all research effort goes. The OER Computational Chemistry series takes precisely this frame and pushes it further: it computes the free energies of the adsorbed intermediates on a candidate surface and derives a theoretical overpotential from them, turning "how good is this catalyst" into a calculation that can be run before any material is synthesised.

There is one more consequence, and it is the quietly brutal one. Because the Tafel relation is logarithmic, improving a catalyst has diminishing returns in current and constant returns in voltage. Gaining one order of magnitude in \(i_0\) buys you exactly one Tafel slope of overpotential — around 118 mV in the symmetric case — no matter whether you started from a good catalyst or a terrible one. Conversely, a factor of ten is a very large chemical improvement to make, and it buys you a tenth of a volt. That arithmetic explains why overpotentials in industrial electrolysis have come down slowly and in increments, and why they have never gone to zero.

3.8 Hands-On: Recovering the Tafel Slope from Butler–Volmer

The code below does one honest thing: it generates current–overpotential data from the Butler–Volmer equation, throws away everything except the high-overpotential branch, fits a straight line in log space as an experimentalist would, and checks whether the slope that comes back is the one the algebra predicts. Along the way it measures how far the linear approximation can be trusted and computes what an order of magnitude of \(i_0\) is worth in volts.

Two of the inputs are chosen rather than measured, and this is stated in the code: \(\alpha = 0.5\) is the conventional symmetric assumption, and the exchange current density is a stand-in. The stand-in is harmless because \(i_0\) cancels out of the slope entirely — it sets the intercept, not the gradient. That is exactly why a Tafel slope is a more portable quantity than an exchange current density.

import numpy as np

# ---------------------------------------------------------------
# Butler-Volmer -> Tafel slope, recovered by fitting
#
# Inputs are only universal constants plus two CHOSEN model
# parameters (alpha and i0). Nothing here is an experimental
# measurement: i0 is a stand-in whose value cancels out of the
# slope entirely, which is exactly the point of the exercise.
# ---------------------------------------------------------------
F = 96485.0        # C/mol      Faraday constant
R = 8.314          # J/(mol K)  gas constant
T = 298.15         # K          25 degrees Celsius
n = 1              # electrons per elementary step

ALPHA = 0.5        # symmetry factor: chosen, not measured
I0 = 1.0e-3        # A/cm^2, a stand-in exchange current density

f = n * F / (R * T)
print("Step 1: the thermal voltage scale")
print(f"  RT/F            = {R * T / F * 1000:.2f} mV")
print(f"  f = nF/(RT)     = {f:.3f} 1/V")
print(f"  2.303 RT/F      = {2.303 * R * T / F * 1000:.2f} mV/decade   (Nernst slope)")
print()

# --- 2. The Butler-Volmer current-overpotential curve -----------
eta = np.linspace(-0.4, 0.4, 4001)           # V
i_anodic = I0 * np.exp((1.0 - ALPHA) * f * eta)
i_cathodic = -I0 * np.exp(-ALPHA * f * eta)
i_total = i_anodic + i_cathodic

print("Step 2: both directions, and their sum")
print(f"{'eta (mV)':>10} {'i_anodic':>14} {'i_cathodic':>14} {'i_total':>14} {'|back/fwd|':>12}")
print("-" * 68)
for e_mV in [0, 10, 25, 50, 100, 200, 300]:
    k = int(np.argmin(np.abs(eta - e_mV / 1000.0)))
    ratio = abs(i_cathodic[k] / i_anodic[k])
    print(
        f"{eta[k] * 1000:10.1f} {i_anodic[k]:14.4e} {i_cathodic[k]:14.4e} "
        f"{i_total[k]:14.4e} {ratio:12.3e}"
    )
print()

# --- 3. The linear (low-overpotential) regime -------------------
# For |eta| << RT/F, exp(x) ~ 1 + x and the two exponentials
# collapse to a straight line: i ~ i0 * f * eta. The slope of that
# line has units of a conductance; its reciprocal is the
# charge-transfer resistance Rct = RT/(nF i0).
print("Step 3: near equilibrium the curve is a straight line")
Rct = R * T / (n * F * I0)
print(f"  Rct = RT/(nF i0) = {Rct:.3f} ohm cm^2   (for the stand-in i0 above)")
for e_mV in [1, 2, 5, 10, 20, 50]:
    e = e_mV / 1000.0
    exact = I0 * (np.exp((1 - ALPHA) * f * e) - np.exp(-ALPHA * f * e))
    linear = e / Rct
    err = 100.0 * (linear - exact) / exact
    print(f"  eta = {e_mV:3d} mV: exact {exact:.4e}, linear {linear:.4e}, error {err:+6.2f} %")
print()

# --- 4. The Tafel (high-overpotential) regime -------------------
# Once one exponential dominates, log10|i| is linear in eta.
# Fit that branch and see what slope comes back.
mask = eta >= 0.15                            # anodic branch, well past the linear region
x = eta[mask]
y = np.log10(np.abs(i_total[mask]))
slope, intercept = np.polyfit(x, y, 1)        # decades per volt, and log10(i) at eta = 0

fitted_tafel_mV = 1000.0 / slope              # mV per decade
predicted_tafel_mV = 1000.0 * 2.303 * R * T / ((1.0 - ALPHA) * n * F)
i0_recovered = 10.0**intercept

print("Step 4: fit log10|i| vs eta on the high-overpotential branch")
print(f"  fit window            : eta = {x[0] * 1000:.0f} to {x[-1] * 1000:.0f} mV")
print(f"  fitted slope          : {slope:.4f} decades/V")
print(f"  fitted Tafel slope    : {fitted_tafel_mV:.2f} mV/decade")
print(f"  predicted 2.303RT/(anF): {predicted_tafel_mV:.2f} mV/decade")
print(f"  intercept -> i0       : {i0_recovered:.4e} A/cm^2 (input was {I0:.4e})")
print()

# --- 5. What the symmetry factor does to the slope --------------
print("Step 5: Tafel slope vs transfer coefficient (anodic branch)")
print(f"{'alpha_a':>9} {'mV/decade':>12}")
print("-" * 22)
for a in [0.2, 0.3, 0.4, 0.5, 0.6, 1.0, 1.5, 2.0]:
    print(f"{a:9.2f} {1000.0 * 2.303 * R * T / (a * F):12.1f}")
print()

# --- 6. What a catalyst buys you --------------------------------
# Same equation, same equilibrium potential, different i0.
# How much overpotential does it take to reach 10 mA/cm^2?
TARGET = 10.0e-3                              # A/cm^2, a conventional benchmark current
b = 2.303 * R * T / ((1.0 - ALPHA) * n * F)   # V/decade
print(f"Step 6: overpotential needed to reach {TARGET * 1000:.0f} mA/cm^2")
print(f"{'i0 (A/cm^2)':>14} {'eta needed (mV)':>18}")
print("-" * 34)
for i0_try in [1e-3, 1e-5, 1e-7, 1e-9, 1e-11]:
    eta_needed = b * np.log10(TARGET / i0_try)
    print(f"{i0_try:14.0e} {eta_needed * 1000:18.0f}")
print()
print(f"  every factor of 10 lost in i0 costs exactly one Tafel slope: {b * 1000:.0f} mV")
print("  the equilibrium potential never moved.")

Output:

Step 1: the thermal voltage scale
  RT/F            = 25.69 mV
  f = nF/(RT)     = 38.924 1/V
  2.303 RT/F      = 59.17 mV/decade   (Nernst slope)

Step 2: both directions, and their sum
  eta (mV)       i_anodic     i_cathodic        i_total   |back/fwd|
--------------------------------------------------------------------
       0.0     1.0000e-03    -1.0000e-03     0.0000e+00    1.000e+00
      10.0     1.2148e-03    -8.2315e-04     3.9170e-04    6.776e-01
      25.0     1.6267e-03    -6.1475e-04     1.0119e-03    3.779e-01
      50.0     2.6461e-03    -3.7791e-04     2.2682e-03    1.428e-01
     100.0     7.0020e-03    -1.4282e-04     6.8591e-03    2.040e-02
     200.0     4.9027e-02    -2.0397e-05     4.9007e-02    4.160e-04
     300.0     3.4329e-01    -2.9130e-06     3.4328e-01    8.486e-06

Step 3: near equilibrium the curve is a straight line
  Rct = RT/(nF i0) = 25.691 ohm cm^2   (for the stand-in i0 above)
  eta =   1 mV: exact 3.8926e-05, linear 3.8924e-05, error  -0.01 %
  eta =   2 mV: exact 7.7867e-05, linear 7.7848e-05, error  -0.03 %
  eta =   5 mV: exact 1.9493e-04, linear 1.9462e-04, error  -0.16 %
  eta =  10 mV: exact 3.9170e-04, linear 3.8924e-04, error  -0.63 %
  eta =  20 mV: exact 7.9828e-04, linear 7.7848e-04, error  -2.48 %
  eta =  50 mV: exact 2.2682e-03, linear 1.9462e-03, error -14.20 %

Step 4: fit log10|i| vs eta on the high-overpotential branch
  fit window            : eta = 150 to 400 mV
  fitted slope          : 8.4547 decades/V
  fitted Tafel slope    : 118.28 mV/decade
  predicted 2.303RT/(anF): 118.33 mV/decade
  intercept -> i0       : 9.9813e-04 A/cm^2 (input was 1.0000e-03)

Step 5: Tafel slope vs transfer coefficient (anodic branch)
  alpha_a    mV/decade
----------------------
     0.20        295.8
     0.30        197.2
     0.40        147.9
     0.50        118.3
     0.60         98.6
     1.00         59.2
     1.50         39.4
     2.00         29.6

Step 6: overpotential needed to reach 10 mA/cm^2
   i0 (A/cm^2)    eta needed (mV)
----------------------------------
         1e-03                118
         1e-05                355
         1e-07                592
         1e-09                828
         1e-11               1065

  every factor of 10 lost in i0 costs exactly one Tafel slope: 118 mV
  the equilibrium potential never moved.

Reading the result. Five observations, in increasing order of importance.

Try changing ALPHA to \(0.3\) and rerunning. The fitted slope in Step 4 comes back at \(84.49\) mV/decade against a predicted \(84.52\), matching \(2.303RT/(0.7F)\) rather than \(2.303RT/(0.3F)\), because the anodic branch is governed by \((1-\alpha)\) rather than \(\alpha\) — a sign convention that has tripped up generations of students and is worth deriving once, by hand, so it stops being a trap.

🎯 Exercise Problems

  1. Equilibrium, two ways. Two electrodes have the same equilibrium potential but exchange current densities differing by a factor of \(10^4\). Sketch both current–overpotential curves on the same axes, and then on a Tafel plot. State which features coincide on each plot and which do not, and explain what that implies about which measurement you would choose to compare two catalysts.

  2. Diagnosing an overpotential. You measure a cell at three currents and find that the extra voltage above thermodynamic is 120, 240 and 480 mV as the current is doubled twice. Which contribution dominates, and how do you know? Repeat the reasoning for the case where the three values are 120, 155 and 190 mV.

  3. The linear region, by hand. Using \(R_{\text{ct}} = RT/(nF i_0)\), compute the charge-transfer resistance for \(i_0 = 10^{-6}\) A/cm² at 25 °C and \(n = 1\). Then state, using the error table in Step 3, the largest overpotential at which you would trust a resistance measured this way to 1%.

  4. A slope that is too big. A student measures a Tafel slope of 210 mV/decade and concludes \(\alpha \approx 0.28\). Before accepting that, list three experimental artefacts that inflate an apparent Tafel slope, and describe one control experiment that would distinguish an artefact from genuine kinetics. (Section 3.6 and Chapter 4 both contain relevant material.)

  5. The value of a decade. Using the Step 6 relationship, compute how many orders of magnitude in \(i_0\) a catalyst developer would need to gain to reduce an overpotential from 400 mV to 250 mV, assuming a 118 mV/decade slope. Then argue whether it would be easier to obtain that same 150 mV by reducing the Tafel slope instead, and state what would have to change physically for the slope to move.

Summary

Thermodynamics says whether a reaction can happen and what it costs. It says nothing whatever about how fast. This chapter filled that gap.

At equilibrium an electrode is not still: both directions of the reaction run at full speed and cancel exactly, so the meter reads zero while traffic crosses the interface continuously. The size of that traffic is the exchange current density \(i_0\) — how busy the equilibrium is — and it is the single number that distinguishes a good electrocatalyst from a bad one. A busy equilibrium is easy to unbalance and yields large net current for a small push; a nearly dead one demands a hard shove. Between good and bad catalysts for the same reaction, \(i_0\) spans many orders of magnitude, and specific tabulated values are so surface-sensitive that they do not transfer between laboratories.

The push itself is the overpotential \(\eta = E_{\text{applied}} - E_{\text{eq}}\), and it is wasted energy in its entirety. It decomposes into three physically distinct penalties with three distinct current dependences: activation (logarithmic in current, the only one a catalyst can touch), concentration (negligible until transport runs out, then divergent) and ohmic \(iR\) (strictly linear, a property of the cell rather than the chemistry). Those different dependences are how a well-designed experiment tells them apart.

The Butler–Volmer equation ties \(i_0\) and \(\eta\) to the net current, and it reads as two exponentials — one per direction — whose difference is what the meter sees. Exponentials appear because charge transfer crosses an activation barrier whose height is set by the electrode potential, a knob that almost no other branch of chemistry offers. In the two limits it simplifies: near equilibrium it linearises into an effective resistor \(R_{\text{ct}} = RT/(nF i_0)\), and far from equilibrium one exponential dies and the survivor gives the straight line of the Tafel plot, \(\log_{10}|i|\) against \(\eta\), whose slope yields \(\alpha\) and whose intercept yields \(i_0\).

Our code generated Butler–Volmer data, kept only the branch above 150 mV, and fitted it in log space exactly as an experimentalist would. It recovered 118.28 mV/decade against the predicted \(2.303RT/(\alpha nF) = 118.33\) mV/decade, and recovered the input exchange current density to within \(0.2\)%. It showed the reverse partial current falling to \(2 \times 10^{-2}\) of the forward one at 100 mV and \(8 \times 10^{-6}\) at 300 mV — the Tafel regime creating itself — and it showed the linear approximation holding to \(0.63\)% at 10 mV but failing by \(14\)% at 50 mV.

Finally, the structural fact that organises the whole field: a catalyst changes \(i_0\) and the Tafel slope, and can never change \(E_{\text{eq}}\). Water will not split below 1.23 V on any electrode ever built. Our benchmark calculation priced the difference exactly: reaching 10 mA/cm² costs 118 mV when \(i_0 = 10^{-3}\) A/cm² and 1065 mV when \(i_0 = 10^{-11}\) A/cm², with every factor of ten in \(i_0\) worth precisely one Tafel slope. That logarithm is why catalyst progress is real, incremental, and never free.

Everything in this chapter assumed we knew the potential the interface actually felt. Chapter 4 examines that assumption and finds it needs defending. We look at the electric double layer that forms in the first nanometre of electrolyte, explain why a two-terminal measurement cannot control a single interface and why a three-electrode cell is therefore mandatory, survey the reference electrodes used in practice and why pH-varying work is reported versus RHE, and then read a cyclic voltammogram properly — what the axes mean, why peaks appear, and how to separate capacitive from faradaic current. We build a voltammogram from diffusion alone in NumPy and watch the classic duck shape appear from nothing but Fick's law and Nernst.

← Chapter 2: Electrode Potentials and Thermodynamics Chapter 4: The Electrochemical Interface →

Disclaimer

Back to Series Index Next Chapter →