How Is Green Hydrogen Made?

Splitting water with electricity

Junior level — plain language, no maths

Hydrogen is not something you dig up. There are no hydrogen mines, because on Earth hydrogen is always stuck to something else - usually to oxygen, as water. So if you want hydrogen you have to take it back, and that costs energy. Hydrogen is not a fuel you find. It is a way of carrying energy you already had.

The cleanest way to make it is beautifully simple. Put two metal plates in water, connect a battery, and the water comes apart: bubbles of hydrogen stream off one plate and bubbles of oxygen off the other. And here is the lovely bit - you get exactly twice as much hydrogen as oxygen, because water is H₂O and every molecule has two hydrogens for every one oxygen. You can watch the formula happen in real time, in the size of the two columns of gas.

Burn that hydrogen later, or run it through a fuel cell, and the only thing that comes out is water again. Nothing else. That is why it is such an attractive idea, and why the word green is doing so much work: hydrogen is only clean if the electricity that split the water was clean. Today about 95% of the world's hydrogen is made from natural gas instead, and that process releases roughly ten kilograms of carbon dioxide for every kilogram of hydrogen.

The catch is that every step costs you. Start with 100 units of electricity and you get about 70 units stored as hydrogen. Squeeze it into a tank and you lose more. Turn it back into electricity in a fuel cell and you are down to around 35. A battery would have handed back 90. So hydrogen is not the answer to everything - it is the answer to the things a wire and a battery genuinely cannot do: making steel and fertiliser, fuelling ships and planes, and storing energy for a whole windless month rather than for a night.

Volts, overpotential and the price of a kilogram

Student level — the core equations

Splitting water costs a fixed amount of energy: \(\Delta H = 286\ \text{kJ/mol}\) for liquid water, of which \(\Delta G = 237\ \text{kJ/mol}\) must arrive as electrical work and the remaining \(T\Delta S\) can arrive as heat. Divide the free energy by \(nF\) and you get the reversible cell voltage, \(E_{\text{rev}} = 1.23\ \text{V}\). Divide the enthalpy instead and you get the thermoneutral voltage, 1.48 V - the point at which the cell needs no heat from outside and produces none.

That second number is the one that matters, because it turns efficiency into something you can read off a voltmeter: \(\eta = 1.48/V_{\text{cell}}\). Run a cell at 1.8 V and you are at 82% - every volt above 1.48 is heat, not hydrogen. Real systems lose another 5-10% to pumps, drying and power electronics, landing at 65-75%.

Why not simply run every cell at 1.5 V, then? Because the extra voltage is what drives the current. The polarisation curve \(V(j) = E_{\text{rev}} + \eta_{\text{act}} + jR\) rises with current density: activation overpotential, dominated by the sluggish oxygen evolution reaction, plus ohmic loss through the membrane. Push harder and you make more hydrogen from the same expensive stack, but each kilogram costs more electricity. The economic optimum is a trade between capital cost per kilowatt and electricity cost per kilogram, and cheap electricity always pushes you to run harder.

Production itself is pure Faraday: \(\dot n_{H_2} = I/2F\), two electrons per molecule, no exceptions and no efficiency term - the electrons that arrive make hydrogen. What varies is only the voltage you paid to move them. In practical units, a system at 70% needs about 56 kWh per kilogram, so at 50 €/MWh the electricity alone costs 2.8 € per kilogram, which is why the whole industry is a bet on cheap electricity and high running hours rather than on clever chemistry.

Overpotentials, the current-density optimum, and where hydrogen actually wins

Scholar level — full mathematical depth

01Two voltages, and the arithmetic of efficiency claims

Water splitting has two characteristic voltages and confusing them is how most public numbers go wrong. \(E_{\text{rev}} = \Delta G/nF = 1.229\ \text{V}\) at standard conditions is the minimum electrical work; the balance \(T\Delta S = 48.6\ \text{kJ/mol}\) may be supplied thermally, which is why a cell operated between 1.23 and 1.48 V cools its surroundings. \(V_{tn} = \Delta H/nF = 1.481\ \text{V}\) is the thermoneutral point. Efficiency quoted against HHV uses 1.48; against LHV it uses 1.25; a vendor may quote either, so a stack that is 82% on one basis is 69% on the other with no physical change whatsoever. High-temperature solid-oxide cells exploit the same thermodynamics honestly: raising \(T\) shifts more of \(\Delta H\) into the entropy term, so with a free heat source they can exceed 90% on electricity alone.

02Why the oxygen side is the problem

Both electrodes obey Butler-Volmer kinetics, with the Tafel form \(\eta_{\text{act}} = a + b\log j\) at appreciable overpotential. The hydrogen evolution reaction on platinum has an exchange current density near \(10^{-3}\ \text{A/cm}^2\) and is close to reversible; oxygen evolution is four-electron, involves O-O bond formation, and has exchange current densities orders of magnitude lower, so it contributes most of the activation loss. That single fact explains the materials problem: PEM electrolysers need iridium oxide anodes, and world iridium production is measured in single-digit tonnes per year. Alkaline cells avoid iridium at the price of lower current density and worse dynamic response; solid-oxide cells avoid both problems and inherit a materials-degradation one instead.

03The current-density optimum

Levelised hydrogen cost splits into a capital term that falls with current density and an energy term that rises with it: \(\text{LCOH} \approx \frac{C_{\text{cap}}\,\mathrm{CRF}}{8760\,\mathrm{CF}\,\dot m(j)} + \frac{p_e\,\text{SEC}(j)}{1}\). Since \(\dot m \propto j\) and \(\text{SEC} \propto V(j)\), the optimum sits where the marginal electricity penalty equals the marginal capital saving. Cheap electricity or a low capacity factor pushes the optimum to high \(j\); expensive electricity and cheap stacks push it back down. This is also why the electrolyser's own capacity factor is the hinge of every project: a stack running 2000 hours a year on surplus renewables has four times the capital burden per kilogram of one running 8000, which is the tension at the heart of pairing electrolysis with variable generation.

04Everything after the stack

Hydrogen's gravimetric advantage - 120 MJ/kg against 43 for diesel - is matched by a volumetric disaster: 0.09 kg/m³ at ambient conditions. Compression to 700 bar costs roughly 10-12% of the fuel's energy, liquefaction 30-35%, and both add capital. Carriers such as ammonia or liquid organic hydrides trade that for conversion losses at both ends. The molecule also leaks through materials that hold methane, embrittles many steels, and - a result only recently quantified - is itself an indirect greenhouse gas, extending methane's lifetime through OH depletion, with a hundred-year GWP now estimated near 11. A leaky hydrogen economy is not a harmless one.

05The efficiency ladder, and what survives it

Chain efficiency decides the use case, and the comparisons are not close. Electricity to hydrogen to a fuel-cell car: about 0.70 × 0.90 × 0.55 ≈ 35%, against 77% for charging a battery car directly. Hydrogen for domestic heating: 0.70 × 0.90 ≈ 63% against a heat pump delivering 300%. On that arithmetic, hydrogen loses wherever electrons can do the job. What remains is substantial anyway, because some jobs need a molecule rather than a current: ammonia and methanol synthesis, direct reduction of iron ore for steel, refining, aviation and shipping fuels, and seasonal storage where a cheap tank beats an expensive battery even at 35% round trip. The distinction is not ideological; it is whether the alternative is a wire.

06What actually determines the cost

At today's stack prices, electricity is 60-80% of the cost of green hydrogen, so LCOH is essentially a rearranged electricity price divided by an efficiency, plus a capital term that punishes idle stacks. That is why announced projects cluster where power is cheapest and most constant, why almost every credible cost curve is really a bet on the price of renewable electricity, and why the gap between announced and operating capacity has stayed enormous - the technology is not the constraint. The interesting engineering question for the next decade is not how to reach 85% efficiency but how to build a stack that tolerates being switched on and off with the wind without dying.