Potential of Progressive and Disruptive Innovation-Driven Cost Reductions of Green Hydrogen Production

Green hydrogen from water electrolysis is a key driver for energy and industrial decarbonization. The prediction of the future green hydrogen cost reduction is required for investment and policy-making purposes but is complicated due to a lack of data, incomplete accounting for costs, and difficulty justifying trend predictions. A new AI-assisted data-driven prediction model is developed for an in-depth analysis of the current and future levelized costs of green hydrogen, driven by both progressive and disruptive innovations. The model uses natural language processing to gather data and generate trends for the technological development of key aspects of electrolyzer technology. Through an uncertainty analysis, green hydrogen costs have been shown to likely reach the key target of <$2.5 kg–1 by 2030 via progressive innovations, and beyond this point, disruptive technological developments are required to affect significantly further decease cost. Additionally, the global distribution of green hydrogen costs has been calculated. This work creates a comprehensive analysis of the levelized cost of green hydrogen, including the important balance of plant components, both now and as electrolyzer technology develops, and offers a likely prediction for how the costs will develop over time.


Methodology Economic Model
The levelised cost of hydrogen (LCOH) is calculated using a combined economic and voltametric model which takes the physical parameters of an idealised proton exchange membrane (PEM) electrolyser to generate operating information about the cell.The economic and voltametric model is defined in equations below.

NPV
The economic model is designed to calculate an NPV, using the yearly cash flow, adjusted for time in years (with a 20 year assumed lifetime), t , and a weighted average cost of capital (WACC) of 7.5%.This WACC indicates a moderate risk S1 .The NPV is dependent on the sale price of hydrogen, which is adjusted to give an NPV value of 0. An NPV of 0 indicates the lowest price the hydrogen could be produced for, and the process still be economically viable.
The cash flow is comprised of the net earnings less the depreciation, both in $ year -1 .

CF = [net earnings -depreciation](2)
The cash flow includes a year zero where the total capital and working capital expenditures are treated as the yearly outgoings.
Net earnings are the profit adjusted for tax by removing the depreciation, where the profit is the income less outgoings ($ day -1 ) multiplied by the working days per year, DPY, assumed to be 350.
The depreciation is calculated using the straight-line method whereby the total capital expenditure, CAPEX tot ($) is divided by the lifetime of the plant, in this case 20 years.

Income
The income is calculated from the mass flowrate of hydrogen produced, m H 2 (kg day -1 ), minus losses, loss H 2 , and then multiplied by the hydrogen sale price per unit mass, p H 2 ($ kg -1 ).
The flowrate of hydrogen production can be calculated using the total current, I (A), Faradaic efficiency, FE (%), molecular weight of hydrogen, MW H 2 (kg mol -1 ), number of electrons per mol of product, n e , and Faraday's constant, F (C s -1 ).The total current is calculated from the voltametric model and the faradaic efficiency is assume at 97%.The remaining parameters in equation 7 are fundamental to the reaction.
The hydrogen losses are dependent on molecular flux across the membrane, Q H 2 (mol m -2 s -1 ), the molecular weight, and the membrane area being considered, A (m 2 ).
Molar flux is defined in Eq. 9, where C H 2 & C l are the molar density of hydrogen and water (mol m -3 ), n p is the electro-osmotic coefficient (-), Γ H 2 is the water solubility of hydrogen (mol mol -1 ), D H 2 is the diffusivity of hydrogen in solid polymer electrolyte (m 2 s -1 ),   is the membrane thickness (m), and T is the temperature (K).

Outgoings
Outgoings represents the negative cash flow from expenses such as utilities and replacement cell parts.These expenses are collectively referred to as operating expenditure or OPEX i ($ day -1 ) where The OPEX water is the product of the purchase price of water, p water ($ kg -1 ), and the mass flow rate of water, m water (kg s -1 ), which in turn is calculated as a required ratio for the hydrogen production rate.
OPEX water = p water m water ( 18) The OPEX main is assumed to be 2.5% of the total capital expenditure, CAPEX tot ($), per year.
OPEX main = CAPEX tot × 2.5% year -1 (20) The OPEX comp is the electrical cost of operating the compressor, calculated by multiplying the compressor electrical power, Power comp (kW), by the price of electricity.
OPEX comp = Power comp p elec (21) The compressor power is derived from Khan et al.S3 , and composed of the number of stages, N (-), the ratio of specific heats, Cr (-), the compressibility factor, z (-), the isentropic efficiency, η isen (-), inlet temperature, T in (K), the molar gas flowrate,   (mol s -1 ), the universal gas constant, R (J mol -1 K -1 ), the pressure in and out, P out & P in (Pa), and the compression ratio, CR (-).

CAPEX
The capital expenditure, CAPEX tot ($), is comprised of the capital cost of the cell parts, compressor, and total plant cost, cell, comp, plant, all in $.
The CAPEX cell is composed of each of the cell material costs, p membrane , p catalyst , p PTL , p plates all in $ m -2 , multiplied by the considered cell area.

Balance of cell and plant
The balance of cell and plant are defined separately from the rest of the plant capital costs and do not depend on cell CAPEX so that if the cell materials change, the balance of plant will remain independent.
The CAPEX BOC is the balance of cell components, endplates, gaskets, and miscellaneous costs, per unit area, multiplied by the cell area.
The CAPEX BOP is the sum of the balance of plant units' costs, for the compressor, water purifier, pressure swing adsorption (PSA), thermal management, and piping & control.Thermal management and piping & control are taken as constants, each equal to half the PSA cost.
The thermodynamic voltage is defined as the reversible cell and pH potentials, where the reversible cell voltage is 1.23 V at STP, and the pH potential is defined in Eq. 29 & 30.
∆ ℎ =  0 + ∆  (29) Due to the cathode product being pure H2 which has no pH value, the delta pH value is assumed to be zero.
The resistance voltage is the voltage required to overcome resistance generated by the membrane,   , with the gap and junction resistances,   &   , being neglected in a PEMWE cell.Membrane resistance is determined by Eq. 32 & 33 S5 .
The   is comprised of the membrane thickness multiplied by the current density,  (A m -2 ), and divided by the membrane conductivity,   (Ω -1 m -1 ), where  is the humidification degree (-).
= ∆  + ∆   +  ℎ, +    (34) The conversion and mass transfer potentials are calculated as shown in Eq. 35 & 36, using the molar density of hydrogen, oxygen, and water,   2 ,   2 ,   2  , all in (mol m -3 ).They represent the voltages required in order to activate the reaction and drive the mass transport of particles to the active surface of the catalyst materials.
The limiting current density,   (A m -2 ), represents the physical limitation of reactant to reach the surface reaction sites and hence the theoretical maximum value the current density can reach, and is assumed from literature values.
The electrode overpotentials represent the deviation from the theoretical half-cell potential of the electrochemical reactions and are highly dependent on operating conditions and system configuration, as literature values for reference data vary widely.
The cathode conversion overpotential,  ℎ, , is calculated in Eq. 37 where   is the conversion of water to hydrogen (-),   is the Tafel slope for either the cathode or anode (V).
Eq. 38 calculates the mass transfer overpotential, where  &   are the current density and limiting current density, respectively.
ℎ, = - ℎ  10 (1 -  )(37) The cathode or anode over potential can be calculated from Eq. 39, where  , is the anode or cathode reference over potential taken from literature, given at a reference current density  , (A m -2 ),  , is the anode or cathode kinetic overpotential.

Current density
Current density,   , can be assumed at a set value, which is useful for comparison between real world data where a set current density and subsequent voltage are often used to determine cell performance, and for the generation of polarisation curves.However, the current density can also be linked to the overpotential from the Butler-Volmer equation, where it becomes a function of physical parameters such as catalyst loading and specific surface area  The exchange current density is analogous to a rate constant, measured when the electrochemical reactions occur at equilibrium, where there is no net current or overpotential. 0, is the reference exchange current density measured at STP (A m -2 ),  is the catalyst specific surface area (m 2 kg -1 ), is the catalyst loading (kg m -2 ),   is the partial pressure of the gas species at either the anode or cathode (Pa),   is the activation energy for the electrode reaction (reduction or oxidation) (J mol -1 ), and   &   are the reference pressure and temperature (Pa & K).
Equation 41 is valid for the anode and cathode and can be further simplified to equation 43 & 44 S6 .

Figure S3 :
Figure S3: Historical literature data and resampling prediction interval.

Figure S4 :
Figure S4: Historical data, fitted learning curve, offset deployment curve with optimistic and pessimistic scenarios for a) Iridium loading b) Platinum loading c) Current Density and d) Relative membrane cost for order volume with 95% confidence interval e) global weighted LCOE with 95% confidence interval f) LCOH trends for base, optimistic, and pessimistic scenarios.

Figure S5 :
Figure S5: a) PEM Gap Analysis and Scenario breakdown for b) AEM, c) BPM, and d) seawater electrolysers.

Figure S6 :
Figure S6: a) Iridium Loading data with prediction and offset b) Heat map of likely technological development routes under emergency development measures.

Figure S8 .
Figure S8.Deployment cases and literature comparison.

Table 1 :
Base Scenario Parameter Inputs i represents the electricity, elec, the membrane-electrode assembly replacement, MEA, the plates and porous transport layer replacement, plates & PTL, process water purchase, water, and maintenance and compressor costs, main & comp.Therefore, the OPEX elec is simply the daily electrical cost, calculated from the cell current and voltage,   (V), multiplied by the price of electricity p elec ($ kWh -1 ).
The OPEX MEA is composed of the membrane and catalyst price per unit area, p membrane & p catalyst ($ m -2 ), multiplied by the cell area considered divided by the lifetime of the MEA, life MEA (day).The catalyst price is dependent on loading.OPEX plates and OPEX PTL are calculated similarly to OPEX MEA . S6 where  0, is the exchange current density, and  , &  , are the activity coefficients.