‹ drnahmkang.com
KMUTT Department of Chemical Engineering
AEM water electrolyser Single cell crash course · Chemical Engineering, KMUTT

Anion exchange membrane water electrolysis

One cell,
thirteen layers.

Two half reactions, a membrane between them, and eleven more layers whose job is to get water in, gas out and current through. Turn the model, take it apart, and click any layer for what it does and how thick it really is.

active area 5 cm² plates 100 × 100 mm Ti felt anode 2 mg cm⁻² Pt cathode drag or arrow keys to turn · click a layer

The two half reactions

Alkaline conditions,
so hydroxide is the carrier.

Water is reduced at the cathode and hydroxide is oxidised at the anode. OH⁻ moves through the membrane, electrons go the long way round through the supply.

Cathode · negative

Hydrogen evolution

4 H2O + 4 e → 2 H2 + 4 OH

Two electrons per H2. Fast on platinum: i0 near 10−3 A cm−2.

Membrane

Hydroxide transport

OH   cathode → anode

Fixed positive groups on the polymer pass the anion and block electrons. About 50 µm thick.

Anode · positive

Oxygen evolution

4 OH → O2 + 2 H2O + 4 e

Four electrons and one O to O bond. Slow: i0 six to eight decades lower than the cathode.

H2O → H2 + ½ O2
Gibbs energy+237kJ mol⁻¹ · ΔG°
Enthalpy+286kJ mol⁻¹ · ΔH°
Reversible voltage1.229 V25 °C, 1 bar · ΔG / 2F
Thermoneutral voltage1.481 VΔH / 2F

The hardware, as delivered

Eight bolts, two plates,
five sheets.

The cell parts laid out on the bench
Everything that goes into the cell. Two gold plated brass end plates with Durex silicone heaters bonded on, eight bolts, the titanium fibre felt and the 2 mg cm⁻² Pt carbon cloth GDE as 10 × 10 cm sheets, the graphite plate and the titanium plate with their machined fields and side fittings, and the PID controller.
Silicone gasket over the flow field plate
The gasket sets the active area. A 0.5 mm silicone sheet with a square window and eight bolt holes. The window is the 5 cm² that reacts; the machined serpentine sits inside it.
The assembled cell on the bench
Closed and plumbed. Four fittings, two per plate, one on each side face. Feed in low, product out high. The two leads at the top are the current terminals.
The cell connected to the Autolab potentiostat
The whole rig. Cell, gas separator and the Autolab potentiostat-galvanostat that drives it and records the curve.

Step one

The membrane
electrode assembly.

Five layers, about 0.8 mm thick in total. Build this first, on the bench, before anything is bolted together.

Step two

Into the cell.

The MEA goes between two gaskets, then two flow field plates, then two end plates with their heaters. Eight bolts, torqued in a star pattern.

The whole cell

Drag to turn it. Slider takes it apart.

arrows turn · + − zoom · [ ] apart · R reset

Run the cell feeds KOH into the lower port of each plate and brings gas out of the upper one, with the hardware solid. Cut away makes everything except the membrane assembly transparent, so the hydroxide crossing the membrane is visible as well.

Layer by layer

Bipolar plate and flow field

Four jobs, one plate.

A single cell has two of these, one each side. In a stack, one plate carries the anode field on its front face and the cathode field of the next cell on its back face. That is what bipolar means.

Job 1Carryelectrons in and out
Job 2FeedKOH across the area
Job 3Ventgas out through the ports
Job 4Sealkeep H₂ and O₂ apart

Titanium on the anode, graphite on the cathode

The anode sits above 1.5 V in hot alkali, which oxidises carbon. Titanium survives by growing a passive oxide, at the cost of contact resistance, so the interface is often coated with platinum or gold. The cathode is reducing, so graphite is safe there and cheaper to machine.

Six channel geometries

Yours is one choice
out of six.

Same plate, same 5 cm² window, six ways of cutting it. Turn the plate and switch the pattern.

The same six, flat

Cross section

Rib and channel

Under a rib there is electrical contact but no gas path. Over a channel there is a gas path but no contact. The porous layer bridges the two, so the ratio of rib width to channel width sets how far current travels sideways through the PTL before it reaches metal.

wide channelLow ΔPless contact area
wide ribLow IRcatalyst starves under it

Most designs land near equal widths, typically 1 to 2 mm each on a cell this size.

Theory

Why the cell needs
more than 1.23 volts.

Four terms, each one measurable on your own cell. Move the sliders and watch which term grows.

1 · Thermodynamics

Reversible voltage

Erev is the minimum voltage set by ΔG. It falls with temperature, because part of the energy can come from the surroundings as heat, and rises with product pressure. The thermoneutral voltage Etn is set by ΔH and does not move; above it the cell heats itself.

Erev = E° − (∂E/∂T)(T − 298) + (RT / 2F) · ln( pH2 · pO21/2 )     Etn = ΔH / 2F = 1.481 V

2 · Kinetics

Tafel slope and exchange current density

At equilibrium the reaction still runs, forwards and backwards at equal rate. That rate per unit area is the exchange current density i0, and it is the number that measures a catalyst. Above it, overpotential grows with the logarithm of current: the Tafel line. Where a line meets zero overpotential is that electrode's i0.

η = b · log10( i / i0 )     b = 2.303 RT / αF     i = i0 [ exp(αFη/RT) − exp(−(1−α)Fη/RT) ]

3 and 4 · Resistance and transport

Cell voltage

Ohmic loss is linear in current and comes mostly from the membrane. Concentration overpotential is flat until the current approaches iL, then rises steeply. Added to Erev and the two activation terms, they give the polarisation curve.

Vcell = Erev + ηact,anode + ηact,cathode + i·Rohm + ηconc     ηconc = (RT/2F) ln[ 1 / (1 − i/iL) ]

Starting values. αa = 1.65, αc = 0.73 and i0,c = 10−3 A cm−2 are the alkaline defaults in the Aspen Plus Electrolyzer block, so this page and the simulation start from the same numbers.

Limiting current

Gas fills the pores of the transport layer faster than liquid can refill them, the catalyst runs dry, and no extra voltage buys more current. On the curve it appears as a steep upturn.

iL = n F D cbulk / δ

It is a transport problem, so the fix is flow rate, temperature, gasket thickness or PTL structure, not a better catalyst.

Voltage efficiency

The thermoneutral voltage divided by the voltage paid. Quote it with the current density and the temperature, because a cell at 1.7 V and 0.2 A cm⁻² and a cell at 1.7 V and 2 A cm⁻² are not the same machine.

εV = 1.481 / Vcell

At the default settings this cell runs at 1.93 V and 77 % on that definition.

Sources

Aspen Plus Electrolyzer block

Every field is
a piece of hardware.

The block asks for about thirty numbers. Each one describes a specific layer of the cell you have just taken apart, and each one comes from somewhere: a supplier data sheet, a measurement, or a fit to your own curve.

Aspen Electrolyzer Specifications tab
Specifications. Alkaline water, stack only. The two half reactions the block writes are the same ones on the first page.
Aspen Electrolyzer Electrode tab
Electrode. The diagram on the right is your cell: ta and tc are the two porous layers, das and dcs the gaps the gaskets set, ts the membrane.

Click a part name to jump to that layer in the 3D cell. Where two values are shown they are anode / cathode.

Exchange current density and charge transfer coefficient rows
The two rows that matter most. Reference exchange current density and charge transfer coefficient, anode and cathode. Everything else can be looked up; these two have to be measured.
Aspen Electrolyzer Membrane tab
Membrane. Thickness, porosity, tortuosity and wetness factor together set the ohmic term. Fit them to the high frequency intercept of your impedance spectrum rather than guessing each one.

Where i₀ and α come from

One electrode
at a time.

A full cell gives the sum of both electrodes plus the ohmic drop, so it cannot separate them. Three electrode mode measures the working electrode against a reference that carries no current.

Reading the two numbers off

Sweep slowly, correct the potential for the uncompensated resistance, plot η against log i, and fit the straight part. The slope gives α, the intercept at η = 0 gives i₀. Repeat at three or four temperatures for the activation energy.

Notice how far the fit is extrapolated to reach η = 0. A 5 % error in the slope moves i₀ by a factor of two, which is why published values scatter so widely. Quote the slope alongside it.

1 · Polarisation

Step the current, wait for the voltage to settle, record. Gives the curve on the theory page and, with an iR correction, the Tafel region.

2 · Impedance

The high frequency intercept on the real axis is the ohmic resistance: membrane, contacts and plates together. That is the number the membrane parameters have to reproduce.

3 · Chronoamperometry

Hold the voltage and watch the current over hours. A slow decay is degradation; a noisy one is usually gas blocking a channel.

Next

What is still
to build.

The hardware, the theory, the Aspen map and the half cell are in. What is left is the measurement side.

Reading an impedance spectrum

A Nyquist plot you can drag, with the equivalent circuit next to it: which arc is charge transfer, which is transport, and where the ohmic resistance is read off.

A worked polarisation sweep

Real data from your own cell, with the iR correction and the Tafel fit done on screen, so the students see the arithmetic and not just the answer.

Stack scale up

The same cell repeated, bipolar plates shared between neighbours, and what changes: voltage adds, current does not, and the manifolds become the hard part.

Tell me which of the three you want first, and anything on this page that reads wrong for your students.