PROCESSES FOR

HYDROGEN
AND ENERGY TRANSITION

Michele Ceccarelli
POLITECNICO DI MILANO, 2024
 PROCESSES FOR HYDROGEN

Summary
BASICS AND OVERVIEW ............................................................................................................ 2
Climate Change ........................................................................................................................... 2
Electrical Energy Storage ............................................................................................................. 3
ELECTROLYZERS ........................................................................................................................ 5
Thermodynamics ......................................................................................................................... 5
Kinetics ..................................................................................................................................... 13
Alkaline Electrolyzer ................................................................................................................. 17
Proton Exchange Membrane Electrolyzer .................................................................................. 21
Solid Oxide Electrolyzer ............................................................................................................ 24
Unitized Regenerative Fuel Cell Systems ................................................................................... 28
Solar Thermochemical Hydrogen Production ............................................................................. 32
HYDROGEN FROM COAL ........................................................................................................ 37
Carbon Capture Technolgies ..................................................................................................... 38
HYDROGEN FROM BIOMASSES.............................................................................................. 41
Gasifiers .................................................................................................................................... 43
HYDROGEN FROM NATURAL GAS ....................................................................................... 46
Steam Methane Reforming ........................................................................................................ 46
Methane Pyrolysis ..................................................................................................................... 50
Solar Thermal Reforming .......................................................................................................... 51
HYDROGEN SAFETY AND FLAME THEORY ........................................................................ 52
PHYSICAL METHODS FOR HYDROGEN STORAGE ............................................................. 58
Compressed Gaseous Hydrogen Storage .................................................................................... 60
Hydrogen Liquefaction .............................................................................................................. 61
METAL HYDRIDES .................................................................................................................... 66
Thermodynamics and Kinetics ................................................................................................... 67
Types of Hydrides ...................................................................................................................... 68
Applications .............................................................................................................................. 69
HYDROGEN PURIFICATION ................................................................................................... 69
Membrane Gas Separation......................................................................................................... 70
Pressure Swing Adsorption ........................................................................................................ 71
HYDROGEN TRANSPORT ........................................................................................................ 72
Hydrogen Pipelines.................................................................................................................... 72
Liquid Organic Hydrogen Carriers ............................................................................................ 73
Ammonia .................................................................................................................................. 74

1
 PROCESSES FOR HYDROGEN

BASICS AND OVERVIEW

The hydrogen is the lightest and smallest element in the periodic table, as well as the most abundant one
in the universe and in our planet. It accounts for the 75% of the mass of all normal matter, but
unfortunately it doesn’t exist in the gaseous form and freely on Earth. In the last years, hydrogen became
one of the most important resources to speed up the energy transition, namely the shift from an energy
system based on fossil fuels to one that produces very limited, if not zero, carbon emission, based on
renewable energy sources (RES).
The fuels are substances which, under appropriate conditions, reacts with an oxidant (typically oxygen)
releasing a large amount of energy. Combustion chemistry involves carbon, hydrogen, oxygen, sulphur,
and nitrogen, and it releases useful energy, such as heat and work:
1
𝐻2 + 𝑂2 → 𝐻2 𝑂
2
𝐶𝐻4 𝑆 + 3𝑂2 → 𝐶𝑂2 + 2𝐻2 𝑂 + 𝑆𝑂2
4𝑁𝐻3 + 3𝑂2 → 2𝑁2 + 6𝐻2 𝑂
3
𝐶𝐻4 + 𝑂2 → 𝐶𝑂 + 2𝐻2 𝑂
2
The standard combustion enthalpy (𝚫𝑯𝟎𝑪 ) is the change in enthalpy due to a combustion process of a
unit amount (mole, mass, volume) referred to a fuel species and in standard conditions (1 [atm] and 298
[K]). The heating value is the heat released by combustion of a unit amount of a fuel at constant pressure
and temperature. It’s a positive quantity and equal to the standard combustion enthalpy (negative) with
sign changed. Usually, two heating values are defined on the assumption on the physical state of water
as a product: higher heating value (HHV) if liquid, and lower heating value (LHV) if vapor.
A fossil fuel should present large availability, low cost, low toxicity (including products), high heating
value, high energy density, complete and safe combustion, safe storage and transport. To date, the most
abundant fuels on Earth are the fossil fuels (coal, oil, natural gas) as they provide 80% of the world
energy demand, originated from the progressive transformation of organic materials synthesized during
geological eras, starting from biologic processes and by using solar energy. The increase in the energy
consumption over the last century stems from the high population growth and the gross domestic product
(GDP), but the relation between population growth and the energy consumption is non-linear, especially
in developing countries.

Climate Change
The global warming is known as the rise in the global average temperature due to the increase of the
atmospheric concentration of greenhouse gases (GHG), due to the emission from fossil fuels combustion,
intensification of agriculture, and to land-use modification. The effect of human activities (anthropogenic
effect) led to the growth of the average global temperature and to the increase of the radiative forcing flux.
In this framework, the lifetime is defined as the time required by the atmosphere to move back to the
equilibrium conditions after the concentration of a greenhouse gas varies. The one of CO2 is estimated
between 50 and 200 years. The global warming potential is the temporal integrated ratio between the
instantaneous adsorption 𝑎𝑖 of the i-th substance due to the unit increase of its concentration 𝐶𝑖 and the
same adsorption due to the emission of an identical amount of 𝐶𝑂2 (𝐺𝑊𝑃𝐶𝑂2 = 1):
𝑡
∫0 𝑎𝑖 𝐶𝑖 (𝑡)𝑑𝑡
𝐺𝑊𝑃𝑖 = 𝑡
∫0 𝑎𝐶𝑂2 𝐶𝐶𝑂2 (𝑡)𝑑𝑡

The carbon footprint is the total emission of CO2 and of other GHG associated with all the activities of
a generic entity. It includes direct emissions, such as those resulting from fossil fuels combustion, as well
2
 PROCESSES FOR HYDROGEN

as the emissions required to produce the electricity associated with goods and services consumed. It’s
measured in [g CO2eq/kWh]. The equivalent CO2 is based on the collective contributions of CO2, CH4,
N2O, and other GHG evolved during a certain activity. It includes the energy content of the fuel and the
GWP of each emitted gas for a specific time horizon. The amount of GHG type produced per amount of
energy spent for the process is 𝐸𝑖 :
𝑁𝐸 𝑁𝐸
𝐺𝑊𝑃𝑖 𝑛𝑖
𝐸𝐶𝑂2𝑒𝑞 = ∑ 𝐺𝑊𝑃𝑖 𝐸𝑖 = ∑
𝑛𝑓𝑢𝑒𝑙 𝐿𝐻𝑉𝑓𝑢𝑒𝑙
𝑖 𝑖

The equivalent change in CO2 emissions is found using the relation:

𝐸𝐶𝑂2𝑒𝑞,𝑖 = 𝐺𝑊𝑃𝑖 𝐸𝑖

For instance, if a gas has a 100-year GWP of 30, this means that a 30-unit reduction in CO2 emissions is
required to achieve the same climatic effect over the next 100 years as a one-unit reduction in the
emissions of the given gas.
The main anthropogenic sources of CO2 are transportation, domestic heating, power production and oil
refining, but also deforestation contributes. The main natural sink of CO2 is dissolution and deposition
in the oceans’ depth, causing the acidification of water.
About 94 [Mt] of hydrogen were produced in 2021, mainly used for ammonia production, methanol
production, refineries. Its demand is mainly met by unabated fossil fuels, and it’s responsible for more
than 900 [Mt] of CO2 emissions: methane steam reforming, partial oxidation, autothermal reforming,
gasification of coal and higher hydrocarbons, and water electrolysis. To date, hydrogen can’t decarbonize
the society as less than 1% of the hydrogen produced is low-carbon H2. The EU commission has set an
upper limit of 3 [kg CO2/kg H2] for the carbon footprint of hydrogen produced from renewable sources
or low-carbon sources. In this framework, carbon capture utilization and storage (CCUS) is a group of
technologies that includes CCU and CCS processes.
The climate neutrality foresees achieving net zero GHGs emissions by balancing the emissions, such
that they are equal to (or smaller than) the emissions that are removed through the planet’s natural
absorption (→ no atmospheric CO2 accumulation). The intermediate EU target is a 50‐55% reduction in
GHGs emission by 2030 compared to the 1990 levels. The EU taxonomy has been established, namely
a detailed and hierarchical list of features for classifying environmentally sustainable activities. The
industrial installations have a cap that is set by the EU, limiting the GHG that can be emitted. This cap
is reduced over time, and the installations buy or receive emissions allowances.

Electrical Energy Storage

The hydrogen produced by electricity (electrolysis) can either be green or yellow. In the former case, the
electricity is entirely produced through renewable energy sources, in the latter case, the electricity comes
from the grid. The average EU grid is featured by 250 [g CO2/kWh], namely 8.5 [ton CO2/ton H2].

Renewable energy sources, such as solar and wind, are featured by variability, depending on daily and
seasonal patterns and local weather conditions, unpredictability, due to forecasting, low load factor (%
of power delivered by the source in relation to the installed power), low dispatchability, depending on
geography and infrastructures.

3
 PROCESSES FOR HYDROGEN

In principle, the electrical grid has no storage capacity on its own as the alternating current cannot be
stored. The stability refers to the voltage and frequency regulation to comply with the grid’s technical
limits within seconds. Supply to the grid must be matched to the draw from the grid instantaneously and
continuously, on a second-by-second basis. Balancing refers to load changes over minutes or days. System
operators rely on ancillary services, either power capacity reserves or flexible customers. A balanced grid
has a load‐following capability on a minute basis. Adequacy is the capacity required to meet peak demand
under extreme conditions and on a long‐term scale from months to years. Flexibility is the power that
can be ramped up or down over a given temporal period. Ramp rate unit is [MW/min].

The load following is capability of the energy storage of providing support in following the load changes
of electricity demand. The energy generation for consumption during the hours of peak demand is shifted
to off‐peak periods. Load levelling implies limiting the requirements during periods of high demand and
increasing the production of energy during off‐peak periods for immediate storage and subsequent use
during high demand periods. Seasonal storage requires high energy capacity and low self‐discharge rates.
Efficiency and power density are advantages. The selected technology should store energy for days to
months, compensate for a longer‐term supply disruption or for seasonal variability. The arbitrage is an
economic trade practice whereby electricity which is bought and stored at low price during low demand
is sold back to the market during high-priced demand.
Every device capable of discharging power is defined by three fundamental features: power rating, the
power output, measured in [MW]; discharge time, the time needed to provide energy; the energy
capacity, the energy stored in a device, measured in [MJ].
𝐸𝑛𝑒𝑟𝑔𝑦 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦
𝐷𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒 𝑡𝑖𝑚𝑒 =
𝑃𝑜𝑤𝑒𝑟 𝑟𝑎𝑡𝑖𝑛𝑔
The 96% of world electrical energy stored is represented by pumped hydroelectric storage, namely by
two reservoirs of water at different heights. The water coming down passes through a turbine that
generates electricity. However, it has geographical limitations due to water availability. Another way to
store electrical energy is the compressed air energy storage system, featured by a motor/generator,
multistage air compressor, turbine train, and auxiliaries for fuel storage. This is not carbon free as the air
expansion cooling is balanced by burning fuel. The energy is stored by storing air into a tight space at 40-
80 [bar]. The discharge time range from hours up to days, with efficiency between 40-50%. The sensible
heat thermal energy storage systems involve heating a mass of a liquid or a solid without phase change
or structural transformation. It requires large masses or volumes, abundant and inexpensive materials
with high specific heat, but solids are preferential. The latent heat TES in phase change materials use
the enthalpy change during one or more congruent phase transitions. They have limited storable heat and
energy density but require much smaller volumes and cheaper materials. Batteries, fuel cells, and

4
 PROCESSES FOR HYDROGEN

supercapacitors all feature two electrodes in contact with a liquid or solid electrolyte. In batteries and
fuel cells, electrical energy is stored by conversion into chemical energy via redox reactions at the
electrodes (anode, cathode). Batteries are closed systems, with the electrodes being the charge‐transfer
medium and playing an active role in the redox reaction as active masses: energy storage and conversion
occur in the same compartment. Fuel cells and electrolysers are open systems where the electrodes are
only charge‐transfer media and the active masses involved in the redox reactions are supplied to the cell
from the outside environment. In electrochemical capacitors (supercapacitors), energy may not be
delivered via redox reactions. By orientation of electrolyte ions at the electrolyte/electrolyte interface,
electrical double layers are formed and released, leading to a parallel movement of electrons in the
external wire, i.e., in the energy‐delivering process.

ELECTROLYZERS
Electrolyzers are devices able to split the water molecule into its elements, hydrogen and oxygen, upon
the application of electricity:
1
𝐻2 𝑂 → 𝐻2 + 𝑂2
2
𝑘𝐽
Δ𝐺𝑅0 (298 [𝐾]) = 237.19 [ ]
𝑚𝑜𝑙
𝑘𝐽
Δ𝐻𝑅0 (298 [𝐾]) = 285.19 [ ]
𝑚𝑜𝑙

Thermodynamics
Since the electrolyzers operate in an opposite way with respect to batteries and fuel cell as they work with
an electricity supply, the cathode here represents the negative terminal, while the anode the positive one.
Reduction and oxidation reactions must respect the atomic and charge balances, while the global reaction
is a neutral stoichiometry. Accordingly, the number of produced/consumed electrons is balanced. The
connection between stoichiometry and charge is expressed by the Faraday’s constant 𝑭, which is defined
as the electric charge of one mole of electrons:

𝑒 − = 1.6022 ∙ 10−19 [𝐶]

𝑁𝐴𝑣 = 6.022 ∙ 1023 [𝑚𝑜𝑙 −1 ]

𝐶
𝐹 = 𝑒 − 𝑁𝐴𝑣 = 96485 [ ]
𝑚𝑜𝑙

5
 PROCESSES FOR HYDROGEN

Of course, a supply of electricity is necessary to carry out the reactions, and both the terminals are
connected to an external utility which provides the energy required.
Let’s now consider a generic system, in thermodynamic and mechanical equilibrium with the
surroundings and exchanging heat and work with it, subject to a reversible transformation that proceeds
through infinitesimal equilibrated steps. Moreover, it’s in steady state and the all the functions are fully
differentiated. It’s known that:
𝑑𝑆𝑡𝑜𝑡 ≥ 0
𝑑𝑆𝑠𝑦𝑠𝑡 + 𝑑𝑆𝑠𝑢𝑟𝑟 ≥ 0

The fact that the system proceeds through equilibrated infinitesimal steps implies:

𝑑𝑄 = −𝑑𝑄𝑠𝑢𝑟𝑟
And so:
𝑑𝑄
𝑑𝑆 =
𝑇
𝑑𝑄𝑠𝑢𝑟𝑟
𝑑𝑆𝑠𝑢𝑟𝑟 =
𝑇𝑠𝑢𝑟𝑟
Since the system is in equilibrium with the surroundings, their temperature is the same:
𝑇𝑠𝑢𝑟𝑟 = 𝑇
𝑑𝑄
𝑑𝑆𝑠𝑢𝑟𝑟 = −
𝑇
𝑑𝑆𝑠𝑢𝑟𝑟 = −𝑑𝑆
𝑑𝑄
𝑑𝑆 − ≥0
𝑇
From the first principle of thermodynamics:

𝑑𝑈 = −𝑝𝑑𝑉 + 𝑑𝑄 + 𝑑𝑊𝑒
𝑑𝑄 = 𝑑𝑈 + 𝑝𝑑𝑉 − 𝑑𝑊𝑒
𝑑𝑈 + 𝑝𝑑𝑉 − 𝑑𝑊𝑒
𝑑𝑆 − ≥0
𝑇
𝑑𝑈 + 𝑝𝑑𝑉 − 𝑇𝑑𝑆 − 𝑑𝑊𝑒 ≤ 0
Since pressure and temperature are constant, they can be brought inside the derivative:

𝑑(𝑈 + 𝑝𝑉 − 𝑇𝑆) − 𝑑𝑊𝑒 ≤ 0

𝑑(𝐻 − 𝑇𝑆) − 𝑑𝑊𝑒 ≤ 0
𝑑𝐺 − 𝑑𝑊𝑒 ≤ 0
This is the basic relation for a system in which a transformation spontaneously proceeds towards
equilibrium when a kind of work is supply. If the canonical parameters (temperature, pressure, and
composition, or extent of a reaction) are put inside the system:
𝜕𝐺 𝜕𝐺 𝜕𝐺
𝑑𝐺(𝑇, 𝑝, 𝜆) = ( ) 𝑑𝑇 + ( ) 𝑑𝑝 + ( ) 𝑑𝜆
𝜕𝑇 𝑝,𝜆 𝜕𝑝 𝑇,𝜆 𝜕𝜆 𝑇,𝑝

6
 PROCESSES FOR HYDROGEN

𝜕𝐺
𝑑𝐺(𝑇, 𝑝, 𝜆) = 𝑉𝑑𝑝 − 𝑆𝑑𝑇 + ( ) 𝑑𝜆
𝜕𝜆 𝑇,𝑝

Being pressure and temperature constant:

𝜕𝐺
(𝑑𝐺(𝑇, 𝑝, 𝜆))𝑇,𝑝 = ( ) 𝑑𝜆 = Δ𝐺𝑑𝜆
𝜕𝜆 𝑇,𝑝

Reminding our system:

𝑑𝐺 𝜕𝑊𝑒
( ) −( ) ≤0
𝑑𝜆 𝑇,𝑝 𝜕𝜆 𝑇,𝑝

Δ𝐺 − Δ𝑊𝑒 ≤ 0
The delta is not the difference between the initial and final state, but the amount of electrical work or free
energy exchanged associated to the extent. So, if the electrical work supplied to the system is bigger than
the Gibbs free energy, the transformation occurs. Actually, when dealing with electrolyzers, the
equilibrium is just the starting point, and the operating point is far from it, because otherwise the reaction
rate would be zero. The equilibrium is stated by:
Δ𝐺 = Δ𝑊𝑒
The electrolysis convention is such that:

Δ𝑊𝑒 = 𝑛𝐹𝐸𝑁
And so:
Δ𝐺 = 𝑛𝐹𝐸𝑁
And we can derive the Nernst equation:
𝑁𝑆
𝜈
Δ𝐺 = Δ𝐺𝑅0 (𝑇) + 𝑅𝑇𝑙𝑛 ∏ 𝑎𝑖 𝑖 (𝑇, 𝑝, 𝑦̅)

𝑁𝑆
𝑅𝑇0 𝜈
𝐸𝑁 = 𝐸 + 𝑙𝑛 ∏ 𝑎𝑖 𝑖 (𝑇, 𝑝, 𝑦̅)
𝑛𝑒 𝐹
Where 𝐸 0 is the standard potential of the cell and 𝐸𝑁 is the equilibrium potential, or Nernst potential.
The Nernst equation describes how the equilibrium potential varies with respect to thermodynamic
parameters. This equilibrium is never reached, but it’s a good approximation of what happened when the
current of the cell is stopped. Now, it’s of interest to assess how the Nernst potential changes with pressure
and temperature:
𝑑𝐸𝑁 1 𝑑Δ𝐺 Δ𝑆
( ) = ( ) =−
𝑑𝑇 𝑝,𝑛̅ 𝑛𝑒 𝐹 𝑑𝑇 𝑝,𝑛̅ 𝑛𝑒 𝐹

So, if the temperature increases, the Nernst potential decreases, generating a beneficial effect because the
lower the potential, the lower the energy consumption of the electrolyzer to perform the reaction.
According to the type of electrolyzer, the reactions, operating conditions, materials and configuration are
different. For the pressure, we need to distinguish between electrolyzers working with liquid water (PEM
and alkaline) or vapor water (solid oxide). For those working with liquid water, assuming ideal gases and
pure water, we have:

7
 PROCESSES FOR HYDROGEN

1
𝑑𝐸𝑁 1 𝑑 1 𝑑Δ𝐺𝑅0 𝑑 3
( ) = [Δ𝐺𝑅0 + 𝑅𝑇𝑙𝑛 (𝑝𝐻2 𝑝𝑂2 )] = [ + 𝑅𝑇 [ln (𝑝2 ) + ln(𝑦𝐻2 𝑦𝑂2 )]]
𝑑𝑝 𝑇,𝑛̅ 𝑛𝑒 𝐹 𝑑𝑝 2 2𝐹 𝑑𝑝 𝑑𝑝

Since the derivative of Δ𝐺𝑅0 and of the molar fraction on the pressure is zero:
𝑑𝐸𝑁 3𝑅𝑇
( ) =
𝑑𝑝 𝑇,𝑛̅ 4𝐹𝑝

This means that the higher the pressure, the higher the Nernst potential. However, this derivative presents
an asymptote:
𝑑𝐸𝑁
lim ( ) =0
𝑝→+∞ 𝑑𝑝 𝑇,𝑛̅

Instead, if we consider a solid oxide electrolyzer, the derivative is also affected by water partial pressure,
thus:
𝑑𝐸𝑁 𝑅𝑇
( ) =
𝑑𝑝 𝑇,𝑛̅ 4𝐹𝑝

In other words, the Nernst equilibrium reaches in a faster way the asymptote.
Another important feature of an electrolyzer is the so-called thermoneutral point, corresponding to an
operating point that is adiabatic and isothermal at the same time. For an electrolyzer, working in those
conditions is beneficial for several reasons. Let’s first begin with the assumption of isothermicity,
considering the electrolyzer as a black box, with one inlet and one outlet streams, and input power and
heat (indeed, apart from electrical power, there is a heat input corresponding to the enthalpy of the
reaction). So:
𝑇𝑖𝑛 = 𝑇𝑜𝑢𝑡 = 𝑇𝑡ℎ
Where 𝑇𝑡ℎ is the temperature corresponding to the thermoneutral point. The energy balance is:
̇ + 𝑄̇ = 0
𝐻̇𝑖𝑛 − 𝐻̇𝑜𝑢𝑡 + 𝑃𝑒𝑙
Where:
𝑁𝑆 𝑇
0,𝑟𝑒𝑓
𝐻̇ (𝑇, 𝑝, 𝑛̅) = ∑ 𝑛̇ 𝑖 [Δ𝐻𝑖 (𝑇𝑟𝑒𝑓 ) + ∫ 𝑐̃𝑝,𝑖 𝑑𝑇]
𝑇𝑟𝑒𝑓

And thus:
𝑁𝑆 𝑇𝑡ℎ 𝑁𝑆 𝑇𝑡ℎ
0,𝑟𝑒𝑓 0,𝑟𝑒𝑓 ̇ + 𝑄̇
∑ 𝑛̇ 𝑜𝑢𝑡,𝑖 [Δ𝐻𝑖 (𝑇𝑟𝑒𝑓 ) + ∫ 𝑐̃𝑝,𝑖 𝑑𝑇] − ∑ 𝑛̇ 𝑖𝑛,𝑖 [Δ𝐻𝑖 (𝑇𝑟𝑒𝑓 ) + ∫ 𝑐̃𝑝,𝑖 𝑑𝑇] = 𝑃𝑒𝑙
𝑇𝑟𝑒𝑓 𝑇𝑟𝑒𝑓

If we use the extent of reaction approach, it is:

𝑛̇ 𝑜𝑢𝑡,𝑖 = 𝑛̇ 𝑖𝑛,𝑖 + 𝜈𝑖 𝜆

For one reaction systems, as in this case. Now we have:

𝑁𝑆 𝑇𝑡ℎ 𝑁𝑆
0,𝑟𝑒𝑓 ̇ + 𝑄̇
𝜆 ∑ 𝜈𝑖 Δ𝐻𝑖 (𝑇𝑟𝑒𝑓 ) + 𝜆 ∫ ∑ 𝜈𝑖 𝑐̃𝑝,𝑖 𝑑𝑇 = 𝑃𝑒𝑙
𝑇𝑟𝑒𝑓

8
 PROCESSES FOR HYDROGEN

𝑇𝑡ℎ 𝑁𝑆

̇ + 𝑄̇
𝜆 [Δ𝐻𝑅0 (𝑇𝑟𝑒𝑓 ) + ∫ ∑ 𝜈𝑖 𝑐̃𝑝,𝑖 𝑑𝑇] = 𝑃𝑒𝑙
𝑇𝑟𝑒𝑓

̇ + 𝑄̇
𝜆Δ𝐻𝑅0 (𝑇𝑡ℎ ) = 𝑃𝑒𝑙
This is the basic relationship that most of the electrolyzers are based on. Now, we know that:
𝑖
𝜆=
𝑛𝑒 𝐹
̇ = 𝑖𝐸𝑐𝑒𝑙𝑙
𝑃𝑒𝑙

𝑖Δ𝐻𝑅0 (𝑇𝑡ℎ )
= 𝑄̇ + 𝑖𝐸𝑐𝑒𝑙𝑙
𝑛𝑒 𝐹
The thermoneutral voltage can be assessed minding that this point is defined by adiabatic conditions:

Δ𝐻𝑅0 (𝑇𝑡ℎ )
𝐸𝑡ℎ =
𝑛𝑒 𝐹
That can be better visualized as:

𝑄̇ = 𝑖(𝐸𝑡ℎ − 𝐸𝑐𝑒𝑙𝑙 )
In other words, the thermoneutral voltage defined the transition of the system between exothermicity and
endothermicity: if the actual voltage of the cell is lower the thermoneutral one, the electrolyzer works in
endothermic conditions, and a heating system is required (endothermic conduction); if the actual voltage
of the system is higher the thermoneutral one, the electrolyzer works in exothermic conditions and a
cooling system is required (exothermic conduction). Most of the electrolyzers work in exothermic
conduction, while those at high temperature, such as solid oxide, can work at the thermoneutral point.
The dependence of the thermoneutral voltage on the temperature is positive and it is defined by the
Kirchoff law:
𝑁𝑆
𝑑𝐸𝑡ℎ 1 𝑑Δ𝐻𝑅0 (𝑇𝑡ℎ )
( ) = ( ) = ∑ 𝜈𝑖 𝑐̃𝑝,𝑖
𝑑𝑇 𝑝,𝑛̅ 𝑛𝑒 𝐹 𝑑𝑇 𝑝,𝑛̅

While its dependence on pressure is null:

𝑑𝐸𝑡ℎ
( ) =0
𝑑𝑝 𝑇,𝑛̅

This is beneficial as we can pressurize the system and have high pressure hydrogen without affecting the
thermoneutral point, instead of pressurizing it after the production. One important outcome of this
feature regards the renewable energy sources. Indeed, they are inherently intermittent, and if the
temperature of the electrolyzers changes a lot, it might switch from an exothermic conduction to an
endothermic conduction, creating technical problems. This is why is important to have electrical energy
storage system that can efficiently work for the system of interest.
When the half reactions of the electrolyzer are analyzed, the electrochemical potential 𝝁
̃ 𝒊 is important
as it considers the potential of each phase involved in the half reaction:

𝜇̃𝑖 (𝑇, 𝑝, 𝑦̅, 𝜑) = 𝜇̂ 𝑖 (𝑇, 𝑝, 𝑦̅) + 𝑧𝑖 𝐹𝜑 = 𝜇𝑖0 (𝑇, 𝑝0 ) + 𝑅𝑇𝑙𝑛(𝑎̂𝑖 ) + 𝑧𝑖 𝐹𝜑

Where 𝜑 is the Galvani potential (electrical potential) of the phase of the i-th species, 𝜇̂ 𝑖 is the chemical
potential of the species, 𝑧𝑖 is the charge of the species, and 𝑎̂𝑖 is its activity in the solution or in the mixture.

9
 PROCESSES FOR HYDROGEN

A species moves from a region at high electrochemical potential to a region at low electrochemical
potential, according to temperature, pressure, and composition. The difference with the chemical
potential is that the energy of an electrically charged species is affected by the electric field 𝜑, whereas
the potential of a neutral species is unaffected. The delta of the electrochemical Gibbs free energy Δ𝐺̃
differs from the delta of chemical Gibbs free energy Δ𝐺 by the inclusion of the effects from the large-scale
electrical environment:
𝑁𝑆

Δ𝐺̃ = ∑ 𝜈𝑖 𝜇̃𝑖 (𝑇, 𝑝, 𝑦̅, 𝜑)

As for the chemical potential, the electrochemical potential is the partial molar derivative of the
electrochemical Gibbs free energy 𝐺̃ :

𝜕𝐺̃𝑖
𝜇̃𝑖 (𝑇, 𝑝, 𝑦̅, 𝜑) = ( )
𝜕𝑛𝑖 𝑇,𝑝,𝑛
𝑖≠𝑗 ,𝜑

For a half reaction, the equilibrium criterion is given by a null electrochemical Gibbs free energy:

Δ𝐺̃ = 0
Whereas it is spontaneously proceeding when:

Δ𝐺̃ < 0
Now, considering the oxygen evolution reaction (OER):
1
𝐻2 𝑂 → 𝑂2 + 2𝐻 + + 2𝑒 −
2
The electrochemical potentials are:
0
𝜇̃𝐻+ = 𝜇̃𝐻 ̂𝐻+ ) + 𝐹𝜑𝑒𝑙
+ + 𝑅𝑇𝑙𝑛(𝑎

𝜇̃𝑒 − = 𝜇̃𝑒0− + 𝑅𝑇𝑙𝑛(𝑎̂𝑒 − ) − 𝐹𝜑𝑚𝑒𝑡𝑎𝑙

𝜇̃𝑂2 = 𝜇̃𝑂0 2 + 𝑅𝑇𝑙𝑛(𝑎̂𝑂2 )

0
𝜇̃𝐻2 𝑂 = 𝜇̃𝐻 2𝑂
+ 𝑅𝑇𝑙𝑛(𝑎̂𝐻2 𝑂 )

1
Δ𝐺̃ = 2𝜇̃𝐻+ + 2𝜇̃𝑒 − + 𝜇̃𝑂2 − 𝜇̃𝐻2 𝑂
2
Where 𝜑𝑒𝑙 and 𝜑𝑚𝑒𝑡𝑎𝑙 are the potential of the electrolyte and of the metal, respectively. If we assume the
activity of the electrons in the metal as equal to the unit, it becomes:
2
0 1 𝑎̂𝐻 +𝑎̂𝑂0.5
2 𝑒
𝑎̂2−
2𝜇̃𝐻 ̃𝑒0− + 𝜇̃𝑂0 2 − 𝜇̃𝐻
+ + 2𝜇
0
2𝑂
+ 𝑅𝑇𝑙𝑛 ( ) − 2𝐹(𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 ) = 0
2 𝑎̂𝐻2 𝑂
2
0
Δ𝐺𝑅,𝑂𝐸𝑅 𝑅𝑇 𝑎̂𝐻 ̂𝑂0.5
+𝑎
2
𝜑𝑒𝑙 − 𝜑𝑚𝑒𝑡𝑎𝑙 = 𝐸𝑁,𝑂𝐸𝑅 = + ln ( )
2𝐹 2𝐹 𝑎̂𝐻2 𝑂
2
0
𝑅𝑇 𝑎̂𝐻 ̂𝑂0.5
+𝑎
2
𝐸𝑁,𝑂𝐸𝑅 = 𝐸𝑂𝐸𝑅 + ln ( )
2𝐹 𝑎̂𝐻2 𝑂

Considering this, we can also understand when a half reaction occurs, stated by:

Δ𝐺̃ < 0

10
 PROCESSES FOR HYDROGEN

𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 > 𝐸𝑁,𝑂𝐸𝑅

The anodic overpotential is defined as the additional potential that has to be applied in order to carry
out the operation:
𝜂𝑎𝑛 = 𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 − 𝐸𝑁,𝑂𝐸𝑅 > 0

Basically, when current is provided, the electrons leave the OER electrode and its potential increases,
becoming more positive.
By applying the same procedure to the HER:

2𝐻 + + 2𝑒 − → 𝐻2
We get:
0 𝑎̂𝐻
Δ𝐺𝑅,𝐻𝐸𝑅 𝑅𝑇
𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 = −𝐸𝑁,𝐻𝐸𝑅 = − − ln ( 2 22 )
2𝐹 2𝐹 𝑎̂𝐻+ 𝑎̂𝑒 −

0
𝑅𝑇 𝑎̂𝐻
𝐸𝑁,𝐻𝐸𝑅 = 𝐸𝐻𝐸𝑅 + ln ( 2 22 )
2𝐹 𝑎̂𝐻+ 𝑎̂𝑒 −

So, in order for the HER to occur, it must be:

𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 < −𝐸𝑁,𝐻𝐸𝑅

And the cathodic overpotential is defined in a different way:

𝜂𝑐𝑎𝑡 = 𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑒𝑙 + 𝐸𝑁,𝐻𝐸𝑅 < 0

11
 PROCESSES FOR HYDROGEN

At the end, the potential of the cell 𝐸𝑁 is the difference in electrochemical potential of electrons between
the leads connected to the positive and the negative electrodes, divided by the Faraday’s constant. The
cell potential according to the electrochemical reactions is defined as the electrostatic potential of the
electrode at the positive terminal with respect to the electrostatic potential of the electrode at the negative
terminal:
+ −
𝐸𝑁 = 𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑚𝑒𝑡𝑎𝑙 = 𝐸𝑁,𝑂𝐸𝑅 + 𝐸𝑁,𝐻𝐸𝑅

As a result, the equilibrium potential of the cell is associated to the variation of the Gibbs free energy of
the global reaction, not to the variation of the electrochemical Gibbs free energy. At the end, the potential
of the cell is the difference between the potential of the positive electrode and the negative:
𝑐𝑎𝑡 𝑎𝑛 𝑐𝑎𝑡,𝑒𝑞 𝑎𝑛,𝑒𝑞 𝑎𝑛 𝑐𝑎𝑡 )
𝐸𝑐𝑒𝑙𝑙 = 𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑚𝑒𝑡𝑎𝑙 = (𝜑𝑚𝑒𝑡𝑎𝑙 − 𝜑𝑚𝑒𝑡𝑎𝑙 ) − (𝜑𝑒𝑙 − 𝜑𝑒𝑙 + 𝜂𝑎𝑛 − 𝜂𝑐𝑎𝑡
12
 PROCESSES FOR HYDROGEN

𝐸𝑐𝑒𝑙𝑙 = 𝐸𝑁 + 𝜂𝑂ℎ𝑚 + 𝜂𝑎𝑛 − 𝜂𝑐𝑎𝑡

𝐸𝑐𝑒𝑙𝑙 = 𝐸𝑁 + ∑ 𝜂𝑖
𝑖

Where the ohmic overpotential is related to the ohmic resistances:

𝜂𝑂ℎ𝑚 = 𝑅𝑖
While the anodic and cathodic overpotentials can be assessed as result of two contributions:
𝜂𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑑𝑒 = 𝜂𝑎𝑐𝑡 + 𝜂𝑑𝑖𝑓𝑓

The activation overpotential, due to kinetics of the half reactions, and diffusion or concentration
overpotential, due to gradient of concentration and thus diffusion.

Kinetics
In order to have a thorough understanding of the kinetics of an electrolyzer, the polarization curve is
one of the most important tools that we can rely on. It defines the variation of the cell’s potential as a
function of the applied current density. At fixed operating conditions (T, p, composition), an
electrochemical process evolves faster the larger its distance from the equilibrium, that in turn depends
on:
- Half reactions at the electrodes, thus on activation polarization.
- Charge transfer across the electrolyte and across the electrode/electrolyte interface, thus on
ohmic polarization.
- Reactants’ mass transport towards the electrode active surface.
- Difference between open cell voltage and Nernst potential 𝐸𝑁 .

First of all, there is the open cell voltage at zero current, represented by 𝐸𝑁 . Then, the first amount of
energy put in the system upon electricity is devoted to activate the reactions occurring at the electrodes,
determined by the activation polarization region (non-linear). The next part, the ohmic polarization

13
 PROCESSES FOR HYDROGEN

region, is due ohmic resistances that are linear with respect to the current. At the end, the cell runs into a
limiting current because of diffusion limitations, in which any current can overcome those limitations.
At time zero, when the metal electrode is immersed in the solution, the charge transfer process begins.
𝐻 + ions move from the electrolyte to the electrode. An
energetic barrier must be overcome, which is still not
charged (Zero Charge Energy Barrier). According to the
transition state theory, the rate of this process is written
as follows:

𝛥𝐺⃗𝑅0 𝛥𝑆⃗𝑅0 ⃗⃗𝑅0

𝛥𝐻
𝑟⃗ = 𝐶𝐻2 + exp (− ) = 𝐶𝐻2 + exp ( ) exp (− )
𝑅𝑇 𝑅 𝑅𝑇

𝐸⃗⃗𝑎𝑐𝑡
𝑟⃗ = 𝑘⃗⃗0 𝐶𝐻2 + exp (− )
𝑅𝑇

And so for the reverse reaction:

𝐸⃖⃗𝑎𝑐𝑡
𝑟⃖ = 𝑘⃖⃗0 𝐶𝐻2 exp (− )
𝑅𝑇

As time proceeds, the spontaneous charge transfer process forms an excess of charge in the proximity of
the electrode, a region that is called double layer, which is limited by the Outer Helmholtz Plane (OHP).
As charge accumulates and 𝐻2 forms, the reverse reaction activates (HOR). Hydrogen adsorbs on the
metal, loses an electron and forms 𝐻 + ions. The
electron moves from the electrode to the electrolyte.
The charging process drives the metal/solution
interface to the equilibrium potential. Under this
dynamic equilibrium condition, the electrical potential
of the oxidized state increases by Δ𝐺 = 𝑛𝑒 𝐹𝐸. The
charge constantly crosses the interface, from the
electrolyte to the electrode and vice versa, but no
current is flowing. In this equilibrium state, the direct
rate equals the reverse rate, and the net rate is zero. The
corresponding current density is called exchange
current density 𝒊𝟎 , which is analogous to the rate used
in chemical kinetics, while 𝜷 is called apparent charge
transfer coefficient:

𝛥𝐺⃗𝑅0 + (𝛽 − 1)𝑛𝑒 𝐹𝐸 𝐸⃗⃗𝑎𝑐𝑡 2 (1 − 𝛽)𝑛𝑒 𝐹𝐸

𝑟⃗𝑒𝑞 = 𝐶𝐻2 + exp (− ) = 𝑘⃗⃗0 exp (− ) 𝐶𝐻+ exp ( )
𝑅𝑇 𝑅𝑇 𝑅𝑇
(1 − 𝛽)𝑛𝑒 𝐹𝐸
= 𝑘⃗⃗ 𝐶𝐻2 + exp ( )
𝑅𝑇

Δ𝐺⃖𝑅0 + 𝛽𝑛𝑒 𝐹𝐸 𝐸⃖⃗𝑎𝑐𝑡 𝛽𝑛𝑒 𝐹𝐸 𝛽𝑛𝑒 𝐹𝐸

𝑟⃖𝑒𝑞 = 𝐶𝐻2 exp (− ) = 𝑘⃖⃗0 exp (− ) 𝐶𝐻2 exp ( ) = 𝑘⃖⃗ 𝐶𝐻2 exp ( )
𝑅𝑇 𝑅𝑇 𝑅𝑇 𝑅𝑇

𝑖0 = 𝑛𝑒 𝐹𝑟⃗𝑒𝑞 = 𝑛𝑒 𝐹𝑟⃖𝑒𝑞

In other words, the exchange current density is the current at which the rates of oxidation and reduction
reactions are equal, and the overall rate is zero, namely at the thermodynamic equilibrium.

14
 PROCESSES FOR HYDROGEN

Then, to activate the reaction, a more positive potential must be applied compared to the equilibrium
one, such that the potential curve grows further, as shown in the graph. The overpotential 𝜂 is referred to
the potential of the electrode 𝑉 at a current different
than equilibrium, i.e., at the operative current 𝑖 of the
cell. The overpotential is intended as the measure of the
offset of the half reaction from its equilibrium potential:

𝜂 =𝑉−𝐸
And the final rate is:

𝛥𝐺⃗𝑅0 + (𝛽 − 1)𝑛𝑒 𝐹𝐸 + (𝛽 − 1)𝑛𝑒 𝐹𝜂

𝑟⃗𝜂 = 𝐶𝐻2 + exp (− )
𝑅𝑇
(1 − 𝛽)𝑛𝑒 𝐹𝜂
= 𝑟⃗𝑒𝑞 exp ( )
𝑅𝑇

𝛥𝐺⃖𝑅0 + 𝛽𝑛𝑒 𝐹𝐸 + 𝛽𝑛𝑒 𝐹𝜂

𝑟⃖𝜂 = 𝐶𝐻2 + exp (− )
𝑅𝑇
𝛽𝑛𝑒 𝐹𝜂
= 𝑟⃖𝑒𝑞 exp (− )
𝑅𝑇
The Butler-Volmer rate equation is derived as the offset of the electrochemical rate from the equilibrium
condition:
(1 − 𝛽)𝑛𝑒 𝐹𝜂 𝛽𝑛𝑒 𝐹𝜂
𝑟 = 𝑟⃗𝜂 − 𝑟⃖𝜂 = 𝑟⃗𝑒𝑞 (exp ( ) − exp (− ))
𝑅𝑇 𝑅𝑇

(1 − 𝛽)𝑛𝑒 𝐹𝜂 𝛽𝑛𝑒 𝐹𝜂 (1 − 𝛽)𝑛𝑒 𝐹𝜂 𝛽𝑛𝑒 𝐹𝜂

𝑖 = 𝑛𝑒 𝐹𝑟⃗𝑒𝑞 (exp ( ) − exp (− )) = 𝑖0 (exp ( ) − exp (− ))
𝑅𝑇 𝑅𝑇 𝑅𝑇 𝑅𝑇

Usually, the cathodic and anodic charge coefficients are grouped as 𝛽𝑐 and 𝛽𝑎 , respectively:
𝛽𝑐 𝑛𝑒 𝐹𝜂 𝛽𝑎 𝑛𝑒 𝐹𝜂
𝑖 = 𝑖0 (exp ( ) − exp (− ))
𝑅𝑇 𝑅𝑇
From a practical perspective, the shape of the current density presents a trend like that shown in the graph
below, where for negative currents the scope of the cell is reversed, switching from an electrolyzer to a
fuel cell, producing power through the oxidation of hydrogen.

15
 PROCESSES FOR HYDROGEN

If the overpotential is high enough, the reverse component of the rate becomes negligible, and the Tafel
law is found (high field approximation):
𝛽𝑐 𝑛𝑒 𝐹𝜂
𝑖 = 𝑖0 exp ( )
𝑅𝑇
From which the overpotential can be directly derived:
𝑅𝑇 𝑅𝑇
𝜂=− ln(𝑖0 ) + ln(𝑖) = 𝑎 + 𝑏𝑙𝑛(𝑖)
𝑛𝑒 𝐹𝛽𝑐 𝑛𝑒 𝐹𝛽𝑐
If the overpotential is small enough, the Butler-Volmer equation can be linearized (low field
approximation):
𝛽𝑐 𝑛𝑒 𝐹𝜂
𝑖 = 𝑖0
𝑅𝑇
𝑖 𝑅𝑇
𝜂=
𝑖0 𝛽𝑐 𝑛𝑒 𝐹

Regarding the efficiency of an electrolyzer, there are several definitions:

16
 PROCESSES FOR HYDROGEN

• The electrolysis efficiency accounts for the real electrical work delivered to the system, which
includes heat dissipations and electrical losses. Depending on the physical state of the water, it
can be referred to LHV or HHV:
𝑝𝑟𝑜𝑑 𝑝𝑟𝑜𝑑
𝑛̇ 𝐻2 Δ𝐻𝐶0 𝑛̇ 𝐻2 LHV
𝜂𝐸𝑙 = − =
𝑃𝑒𝑙 𝑖𝐸𝑐𝑒𝑙𝑙
• The voltage efficiency takes into account that the voltage applied to the cell 𝐸𝑐𝑒𝑙𝑙 is higher than
the reversible potential, due to irreversible phenomena associated with kinetics and electrical
contacts:
𝐸𝑁
𝜂𝑉 =
𝐸𝑐𝑒𝑙𝑙
• The faradaic efficiency takes into account that the applied current is larger than the theoretical
one to produce hydrogen, since complete water conversion is not reached due to thermodynamic
and kinetic reasons:
𝑝𝑟𝑜𝑑
𝑛𝑒 𝐹𝑛̇ 𝐻2
𝜂𝐹 =
𝑖
• A steam or water utilization factor can also be defined as the amount of moles converted over
the inlet amount of moles supplied, which corresponds to the conversion:
𝑝𝑟𝑜𝑑
𝑛̇ 𝐻2 𝑖𝜂𝐹
𝑈𝐹 = 𝑖𝑛 = 𝑖𝑛
𝑛̇ 𝐻2 𝑂 𝑛𝑒 𝐹𝑛̇ 𝐻2𝑂
• The specific amount of energy in [kWh] per mole of hydrogen produced is also a technical
parameter used for the evaluation of the electrolyzer:
𝑖𝐸
𝑊 ̂𝐸𝑙 = 𝑐𝑒𝑙𝑙
𝑝𝑟𝑜𝑑
𝑛̇ 𝐻2

Alkaline Electrolyzer
Alkaline electrolyzers use an electrolyte solution in which 𝑂𝐻 − ions are dissolved and pass from the
cathode to the anode, through a membrane. The hydrogen is produced at the cathode at high pressure,
and it can be extremely pure (up to 99.999%). This is the most established technology, where industrial
applications date back to 1900.
They require abundant and
inexpensive materials, such as
nickel, iron, steel, and usually
KOH for the electrolyte. Since
they work well with renewable
energy at the small scale, an off-
grid remote area with water,
wind or sun can be suitable for
the application. However, their
applications are limited to small
scales and special cases, and they work at moderate or low current density. They present an expensive
installation, reversible operations (fuel cells) are unfeasible, and improvements in safety, operability, and
durability are needed. The usual operating conditions are 1.8-2.2 [V], 200-400 [mA/cm2], 70-90 [°C], and
1-35 [atm]. Combining the two half reactions:
𝐸𝑂𝐸𝑅 = 0.4 [𝑉], 𝐸𝐻𝐸𝑅 = −0.83 [𝑉]
𝐸𝑁 = 1.23 [𝑉]
0
Δ𝐻𝑅,𝑙𝑖𝑞
Since the thermoneutral potential 𝐸𝑡ℎ (298𝐾, 1𝑏𝑎𝑟) = 2𝐹
= 1.48 [𝑉], these devices work in
exothermic conditions.
17
 PROCESSES FOR HYDROGEN

From the graphs above, we can see that the highest contribution in terms of overpotential is given by the
activation of the reactions, while resistance-wise, the alkaline electrolyzer is the worst above all. To the
right, the different overpotential are shown with respect to the current density applied. In this case, they
get higher quite quickly with higher currents, this is why they operate at low current densities.
Regarding the electrolyte, the most used is the potassium hydroxide, even though the sodium hydroxide
is cheaper. The main reason behind the choice of KOH is the relatively high vapor pressure of water at
80 [°C] in 20% NaOH solution, that
could lead to technical problems. The
chart below shows the variation of the
electrolyte conductivity on the salt
concentration in the aqueous solution at
different temperature. In most of the
concentration intervals, KOH shows a
higher conductivity than NaOH. The
amount of salt shall be adjusted to
balance the losses due to gaseous
product, that are usually stated to 1 [mg]
per [Nm3] of hydrogen. Moreover, the
aqueous solution must be regularly
changed due to the accumulation of impurities, mainly caused by corrosion of metallic parts, degradation
of the electrodes and of the separator. The ohmic resistances in the electrolyzer are calculated through
the Ohm law:
𝐿 𝐿
𝑅𝑂ℎ𝑚 = 𝜌 =
𝑆 𝑘𝑆
1
𝜌=
𝑘
𝑆
Where 𝑘 is the conductivity in [𝑚], 𝜌 the resistivity in [Ω ∙ 𝑚], 𝐿 the interelectrode gap, and 𝑆 the electrode
surface area. Unfortunately, is tough for the two electrolytic compartments to work at the optimal
concentration as one side of the cell is concentrating the aqueous solution (cathode, where 𝑂𝐻 − is
produced), while the other side is diluting the solution (anode, where water is produced). For this reason,
the electrolyte is withdrawn and recycled to try to maintain it around the optimal concentration.
The diaphragm shall selectively transport only 𝑂𝐻 − ions, prevents short circuits between the electrodes,
and minimize oxygen and hydrogen crossover. In order to fulfil such requirements, the membrane must

18
 PROCESSES FOR HYDROGEN

have micropores in the order of 0.1 − 50 [𝜇𝑚], must be stable against corrosion (in basic solutions), and
must be electrically insulating. Despite the first use of asbestos as membranes, that are made of dangerous
materials, the use zirfon perl (85 wt% of 𝑍𝑟𝑂2 and 15 wt% of polysulphone) has become predominant,
thanks to the higher conductivity of polysulphone. The problem with having a polymeric membrane is
their possible instability at high pressure, that could create technical problems. For this reason, ceramic
of zircon has been introduced inside. Moreover, since polysulphone is hydrophobic, some hydrophilic
materials are added in order to facilitate the transport of anions.
The electrodes materials shall have good corrosion resistance, high conductivity, and high catalytic
activity in both the half reactions. Indeed, usually the same materials are used for both the electrodes,
such as nickel doped with iron for the anode, and nickel-sulphur alloy o mild steel for cathode. However,
dealing with such cheap materials could lead to very easy corrosion and oxidation of the electrodes,
especially during start-up and shut-down phases.

Some adverse effects must be taken into account during the operation of the device:

• The cross-over leakages could cause dangerous explosive mixtures by mixing oxygen and
hydrogen as can partly dissolve in the aqueous electrolyte, independently on the amount
produced, according to the Henry’s law. Usually, the safety limit is set to 4% of hydrogen in
oxygen, to not reach the lower flammable limit at 25%. Also, there is a partial load limit that has
not to be exceeded in order to not overcome the cross-over limit, that is 40% of the maximum
current. The low load not only has effect on the cross-over, but also on the stability of the
electrodes, that could easily oxidize in those conditions.
• The champagne effect is featured by the expulsion of the electrolyte from the compartments with
very rapid rise of the current as a large amount of gas is produced.
• The impurities in the electrolyte can cause parasite reactions (𝑀𝑔2+ , 𝐶𝑎2+ , 𝐶𝑙 − ), or foams
(organic species). They shall be in sufficiently low amount to avoid the blockage on the electrode
surface or diaphragm. The chloride anions are oxidized when the current density exceeds the
hydroxyl ions limiting current. Moreover, chlorine 𝐶𝑙2 at the anode surface is highly corrosive to
the metal structures. Eventually, the parasite reactions can cause the formation of salts:
2𝐶𝑙 − → 𝐶𝑙2 + 2𝑒 − (𝑎𝑛𝑜𝑑𝑖𝑐)
𝑀𝑔2+ + 2𝑂𝐻 − → 𝑀𝑔(𝑂𝐻)2 (𝑐𝑎𝑡ℎ𝑜𝑑𝑖𝑐)
𝐶𝑎2+ + 2𝑂𝐻 − → 𝐶𝑎(𝑂𝐻)2 (𝑐𝑎𝑡ℎ𝑜𝑑𝑖𝑐)
One of the most concerning problems regarding alkaline electrolyzers is the bubble resistances creating
on the electrode surfaces due to gas formation. This can cause some diffusion issues to the water to get
from the electrolyte to the metallic surface. Both the conductivity of the electrolyte and the resistivity of
the electrode are influenced by the volumetric fraction of dissolved gas 𝑓 and by the bubble surface
coverage 𝜃, respectively:
𝑘𝑔 = 𝑘(1 − 1.5𝑓)

𝜌 = 𝜌0 (1 − 𝜃)1.5
Where 𝑘𝑔 is the electrolyte conductivity with dissolved gas. Remedies to these problems could be
electrodes more hydrophilic, additives in the electrolyte solution to reduce the surface tension, control
the flow pattern to force bubbles to leave the electrodes mechanically. Moreover, since in the catholyte
there is the production of 𝑂𝐻 − , as the solution gets more concentrated the viscosity increases, making
the pumping expenses higher.
Regarding the configuration, an alkaline electrolyzer can be built in two different ways:

• The monopolar configuration is featured by alternate electrodes connected to the DC power

supply, and individual cells are located in parallel, such that 𝑉𝑡𝑜𝑡 = 𝑉𝑐𝑒𝑙𝑙 . So, the same reactions

19
 PROCESSES FOR HYDROGEN

occur on both sides of the electrodes. It’s easy to fabricate, reliable and flexible, it can work with
high electrical currents at low voltages but presents high ohmic resistances.
• The bipolar configuration foresees that only the two end-electrodes are connected to the DC
current supply, and each couple of adjacent electrodes form a cell connected via the electrolyte.
HER and OER occur at both the sides of each electrode, according to the charge. The total
voltage, in this case, is given by the voltage of all the cells: 𝑉𝑡𝑜𝑡 = 𝑁𝑐𝑒𝑙𝑙 𝑉𝑐𝑒𝑙𝑙 . It presents low ohmic
losses thanks to the lower distance between the electrodes connected to the grid and can be
exploited for high-pressure applications. However, implementing this configuration involves
structural complexity, high precision, electrolyte recirculation system and gas/electrolytes
separators.

A usual process is represented by the picture below. Demineralized water is fed to the hydrogen/water
separator, where also the outlet of the cathode compartment is sent. The water/electrolytes solution, after
filtration, then circulates in the equalization line, that pumps it into the electrolyzer after being cooled
down. From the pressurized stack, that is composed by a pressure vessel to enclose the cells, the outlets
are sent to the separators, from which oxygen and hydrogen exits to go to the purification section, that
foresees also a demister. During start-up and shut-down, pressurization with nitrogen gas to purge the
system and avoid flammable mixtures is performed. The purificators for the gas products are featured by
gravity gas separators. The KOH concentration is adjusted to a constant value by mixing the separated
solutions in a mixing pipe, that is the equalization line. Indeed, the cathodic side will have a more
concentrated solution with respect to the anodic side, and this can be counterbalanced by mixing their
solutions.
Nowadays, large scale electrolyzers are the Oerlikon, Norsk Hydro, and Zdansky-Lonza. The simplest
to operate is the first one, but the Norsk presents a higher efficiency and reliability, even if operated at
atmospheric pressure and needs to be replaced after 8-10 years, while the last one can be used for high-
pressure applications, up to 30 [bar]. Generally, these large-scale electrolyzers can produce from 10 up to
100 [Nm3/h] of hydrogen, with a total current of 2.5 [kA], power consumption of 4.1-5 [kWh/Nm3],
𝜇𝑉
55-60% of energy efficiency, and a long lifetime (less than 3 [ ℎ ]).

Pressurized electrolyzers bring some benefits to the production of hydrogen as it simplifies the separation,
filtration, and drying, it reduces the stack voltage, the energy consumption, and the size of the cooling
system. However, a large amount of non-conductive bubble can be formed into the electrolyte (cross-
over) and their size is decreased. The ohmic resistances of the electrolyte are increased, as well as the
presence of parasitic reactions, reducing the faradaic efficiency.
To face the problem of cross-over and ion transport losses, the zero-gap configuration has been
developed. The two electrodes are placed very close to each other and separated by an anionic electrolyte
membrane (AEM), that allows to operate at high-pressure and to reduce ohmic resistances. Indeed,
reducing the distance the ions have to cover, the ohmic resistances are minimized, but also the amount

20
 PROCESSES FOR HYDROGEN

of gases dissolved in the electrolyte solution are reduced as they are proportional to its volume. Moreover,
the cell has some channels between the
electrodes, in the interelectrode gap, to facilitate
the circulation of the electrolyte through the
compartments. Heterogeneous membranes are
made of an anion 𝑂𝐻 − exchange material
embedded in an inert compound. Due to inert
part. These membranes have good mechanical
strength, but they are usually thick.
Interpenetrating polymer network (IPN) are a
combination of two polymer networks without
covalent bonds between. They are easy to
produce with a wide variety of polymers. The
conductive polymer is not bounded to the
polymer matrix, it can slowly diffuse out of the
membrane during operation, reducing the ionic
conductivity. Besides, this anionic membrane
it’s not still stable at high pH, that is normal in alkaline electrolyzers, leading to possible corrosion of the
interelectrode gap membrane.

Proton Exchange Membrane Electrolyzer

The proton exchange membrane (PEM) electrolyzers are a reasonably mature technology (1960s), with
a wide range of operational current density (much more hydrogen can be produced), excellent dynamic
response to variable electricity input and cycling, compact and small stacks for a desired production rate
(compared to alkaline), and differential pressure operations (up to 100 [bar]). However, they are
expensive, due to platinum catalysts and titanium plates, and with high investment costs, up to 1900-
2300 [€/kW] vs 1000-1200 [€/kW] of the alkaline electrolyzer. Moreover, their use at the large-scale is
still scarce, also due to their low durability. Eventually, the reversible operation as a fuel cell is very
complex.

Also in this case, water is present in the liquid form, and as such its activity is taken equal to the unit in
the assessment of the half potentials:
𝐸𝑂𝐸𝑅 = 1.23 [𝑉], 𝐸𝐻𝐸𝑅 = 0 [𝑉]
𝐸𝑁 = 1.23 [𝑉]
Being the thermoneutral voltage equal to that of the alkaline electrolyzer, even PEM electrollyzers work
in exothermic conditions. Usually, the operating conditions range between 1.8-2.2 [V], 400-2000
[A/cm2], 50-90 [°C], and 1-700 [bar]. One powerful advantage with respect to the alkaline electrolyzer is

21
 PROCESSES FOR HYDROGEN

the absence of water at the cathodic side, preventing the presence of a unit performing the drying of
hydrogen.

From the polarization curve, it’s clear the effect of temperature and pressure. As stated by the
thermodynamics, the low temperature has a beneficial effect on the cell voltage, while the pressure has
the opposite effect, increasing the voltage to
reach a certain current density. A compromise
can be found between a low voltage and a
reasonable pressure for the produced hydrogen.
Looking at the overpotential losses, instead, the
one related to the activation of the OER is the
highest, followed by the ohmic losses in the
electrolyte. The picture to the side shows a
sketch of a PEM electrolyzer components. In the
middle there is the proton conductive
membrane, that is not impermeable, but its
porous can let the water pass through it. Indeed,
the membrane shall be humidified in order to be
conductive to the protons. Coated on the
membrane surface are the catalysts layer. Next
to the MEA (membrane electrode assembly) there is the porous transport layer, through which water and
hydrogen diffuse up to the bipolar plate or separator plate, able to separate the gas from the liquid thanks
to special channels. Usually, the membrane is made of perfluorinated sulphonated acids (PFSA)
ionomeric materials, where the fluorination allows to ionize the sulphonic acid group and to provide
chemical stability. The acidity of the membrane is close to that of a 1 [M] aqueous sulphuric acid solution.
The goodness of the membrane is given by the ion exchange capacity (IEC), namely the concentration
of sulphonic acid per mass of dry membrane. The microstructure is a network of clustered hydrophilic
areas enhancing the protons conductivity. The sulphonated ending groups are organized as inverted
micelles and connected by pores. Here, the swelling becomes critical as it defines the upper current limit
(2.5 [A/cm2]), and in turn it depends on the type of reinforcement. When no reinforcement is used, the
swelling proceeds through all the directions. With microporous reinforcement, the membrane is protected
from microcrack formation. Also, above 90 [°C] the water swelling becomes too high, and the membrane
loses the mechanical properties.

22
 PROCESSES FOR HYDROGEN

The minimum critical current density is given by the cross-permeation effects, that for safety and
efficiency reasons must be avoided. The hydrogen diffusion crossover is due to pressure gradients,
concentration gradients, and electro-osmotic drag of water (through the pores). The crossover can cause
the recombination of oxygen and hydrogen in water, leading to a lower faradaic efficiency. Start-up and
shut-sown transients become crucial in terms of safety as the lower critical limit of current density must
not be crossed.
At the hydrogen evolution electrode, due to the highly acidic environment, platinum group metal is
required. A triple phase boundary is required to activate the HER. The protons come from the
membrane, the hydrogen gas desorbs from the platinum, and the electrons are transported by the carbon.
Carbon black or carbon nanotubes are used as support. On the other hand, the oxygen evolution
environment is highly oxidative, and only few electrocatalyst can be adopted (carbon can’t be used as it
will oxidize to carbon dioxide). Noble metals like Ru, Rh, Ir, Pt can be used, where 𝐼𝑟𝑂2 black is the best
one. Here, the porous transport layer plays a key role in diffusing the electrons between the bipolar plates
and the catalyst, water to the catalyst, ensuring efficient removal of the gases, and providing mechanical
support to the membrane. It shall be highly conductive and sufficiently porous, usually in titanium. The
bipolar plates are responsible for the separation of the two cells in the stack, the homogeneous
distribution of water, and reducing pressure drops for water recirculation. Below, a typical process
including the production of hydrogen with a PEM electrolyzer.

Demineralized water enters the system through the oxygen separator, then it is filtered to remove
impurities that can come from ions and metallic parts detaching from some parts of the electrolyzer.
Water is pumped, passes an ion exchanger to maintain the pH constant, cooled down and fed to the
anodic side, because it’s the part where water is consumed. The water entering the cathodic side has the
same fate. Around 50-300 times the water required is supplied at higher current densities. Working at
high pressure can minimize bubbles size and diffusive overpotential and facilitate the separation, but a
less pure hydrogen is obtained due to cross-over. The so-called balanced operations (𝑃𝑐𝑎𝑡 = 𝑃𝑎𝑛 )
guarantee a high MEA protection and can be run at atmospheric pressure (easier), but a hyperbaric

23
 PROCESSES FOR HYDROGEN

chamber is required for pressurization. The differential operations (𝑃𝑐𝑎𝑡 > 𝑃𝑎𝑛 , up to 100 [bar]) reduce
the intensity of the gas drying process, the work required for compression, but can experience a
mechanical failure due to the membrane.
PEM electrolyzers can range from small scale to the large scale, varying the surface are of the electrodes,
the power consumption and, consequently, the hydrogen production. For small scale, around 30 [cm2]
and 100 [W] are required to produce 1-10 [g/h] of hydrogen, while at the large scale above 700 [cm2] and
10 [MW] are required to produce more than 10 [kg/h] of hydrogen. In general, 100 [kWh/kg] is the
standard for bulk hydrogen production.

Solid Oxide Electrolyzer

This type of electrolyzer differs from the others for the fact that uses water steam to perform the synthesis,
instead of liquid water. As such, it works at high temperatures (700-900 [°C]), not only to maintain the
steam, but also to improve the conductivity and other properties of the material used, that are usually not
that much conductive. Up to 30% of the energy contribution for electrolysis comes from the heat, that is
industrial heat and costs less than electricity, lowering the expenses, but also it comes from heat wastes
(200-300 [°C] gas streams can vaporize water). The reversible operation of fuel cell is feasible. It can
operate on a total range of load (0-100%) thanks to a non-porous electrolyte, and a wide variety of
configurations can be adopted (tubular, planar, supported). Moreover, it is chemically flexible, meaning
that it can perform other kind of reactions such as 𝐶𝑂2 /𝐻2 𝑂 co-electrolysis, because the electrolyte is 𝑂2−
ions conductive.
Drawbacks of this technology regards the low TRL (not mature), mechanical failure due to fragility of
the components made of ceramics, expensive installation and manufacturing costs, fast degradation rates
due to high temperatures, and limitation for the pressurized operations.
24
 PROCESSES FOR HYDROGEN

The operating conditions vary from 500 to 2000 [mA/cm2], but atmospheric pressure is used to maintain
the gas phase. Here, contrarily to the others electrolyzers, the thermoneutral voltage is set to 𝐸𝑡ℎ =
1.29 [𝑉], while the Nernst potential is:

𝑅𝑇 𝑃𝐻2 𝑃𝑂0.5
𝐸𝑁 = 𝐸𝑁0 + ln ( 2
)
2𝐹 𝑃𝐻2 𝑂

That is 0.83 [V] if there is 90% of water and 10% of hydrogen, 0.93 [V] if there
is 50% oh water and 50% of hydrogen. Since the voltage is lower than the
thermoneutral one, solid oxide electrolyzers work in endothermic
conduction.
The typical structure of the SOE is shown in the microscope picture below.
The anodic current collecting layer (ACCL) is made of LSM and is 50 −
60 [𝜇𝑚]. Then, there is the functional layer (AFL, 5 − 20 [𝜇𝑚]), and the
electrolyte made of a highly oxide ions conductive ceramics (YSZ, 2 −
10 [𝜇𝑚]). In the other side, the cathode functional layer has similar properties to the AFL, but it’s made
of nickel-YSZ, as well as the substrate (1.5 [mm]).
The solid electrolyte must possess high oxide ions
conductivity, low or null electrical conductivity
(insulator), and gas-tight mechanically resistant. Its
conductivity follows a temperature dependence
based on an Arrhenius law:
𝜎0 𝐸𝑎𝑐𝑡
𝜎= exp (− )
𝑇 𝑅𝑇
The most common ones are Yttria stabilized
zirconia (YSZ, the most used because of the low
cost), gadolinium doped ceria (GDC), and
samarium doped ceria (SDC). Zirconia-based
electrolytes need some insights in the phase
diagram of 𝑍𝑟𝑂2 . Pure 𝑍𝑟𝑂2 has low oxide ions conductivity, monoclinic structure up to 1170 [°C],
tetragonal up to 2370 [°C], and cubic fluorite above 2370 [°C]. The most favourable structure for 𝑂2− is
the last one. The addition of an alkaline oxide (𝑌2 𝑂3 ) dopant is required to stabilize the fluorite phase
down to room temperature, otherwise the electrolyte would crack because of the phase change. The solid
solution of zirconia and YSZ performs between 800 and 1000 [°C] and introducing trivalent dopants
(𝑌 3+ ) allows to create oxygen vacancies in the lattice. Indeed,
the zirconia itself doesn’t have vacancies, but they are
introduced by doping zirconia with yttrium oxide. Basically,
by introducing the yttrium oxide in the lattice, some 4+ Zr are
replaced with 3+ Y, and the charge balance is satisfied by the
holes vacancies that have 2+ charge:

2(3+ − 4+ ) + 2+ = 0

𝑂2− + 𝑉𝑂∙∙ → 𝑂𝑂𝑋

This is the so-called vacancy hopping and diffusion
mechanism, that allows the electrolyte to efficiently transport
the oxide ions from the cathode to the anode. Moreover, yttria enhances the zirconia conductivity as it
peaks at a certain dopant level, depending on the nature of the dopant. Notably, YSZ performs better at
8-9 mol% of 𝑌2 𝑂3 , but the highest conductivity has been observed with 𝑆𝑐 3+ due to the close match of its
atomic radius with 𝑍𝑟. However, ScSZ is much more expansive than YSZ, because scandium is rare.
25
 PROCESSES FOR HYDROGEN

Usually, the standard target is 0.1 [Ω ∙ 𝑐𝑚2 ], achieved with 40 [𝜇𝑚] thickness with 8YSZ electrolytes, but
as the temperature decreases down to 700 [°C] other more conductive materials shall be used. According
to the thickness of the various parts of the cell, three different configurations can be adopted: the
electrolyte could the thickest part in the electrolyte-supported cells (ESC); the cathode, or hydrogen
electrode, is the thickest part in the cathode-supported cells (CSC), as it would be the part that needs to
be maximized to have a higher production of hydrogen; anode, cathode, and electrolyte are all thin, and
the cell is supported by a metallic foam next to the cathode in the metal-supported cells (MSC).
Regarding the anodic part, 𝐿𝑎𝑀𝑛𝑂3 -based perovskite oxides are used in SOE. The A site is a divalent or
trivalent alkaline or a rare earth, while the B site is a transition metal with 3+
or 4+ valence. This a key point, because the transition metals, being able to
change the valence, can vary the amount of oxygen that can be stored by the
lattice, and thus its adsorption. The doping of the A site with Sr is a common
practice as it can improve the electrical conductivity up to a certain limit, that
is defined by the thermal expansion coefficient: if the strontium is too much,
the ceramics can excessively expand upon temperature and break. Perovskites
are the preferential catalysts for the 𝑂2 adsorption. The composite anodes are
made of LSM/YSZ materials, with 50/50 vol%, where YSZ are part of the
electrolytes, able to transport oxide ions and maximize the surface between
the electrode and the electrolyte. In this case, the mechanism follows the so-
called three phase boundary:

2− −
1
𝑂𝑌𝑆𝑍 → 2𝑒𝐿𝑆𝑀 + 𝑂2,𝑔𝑎𝑠
2
In this mechanism, is important to understand the phenomenon of
percolation of electrons. The electrons in the LSM phase are active only
if they percolate, or, in other words, they form straight lines in the
electrode. Looking at the picture, the case C would not work for the OER
because the electrons inside the blue phase are trapped in that space,
while in the case B the line formed is too short to guarantee a good
activity. In alternative, the mixed ionic and electronic conductive
(MIEC) oxides can transport both ions and electrons, with good conductivity. Contrarily to the TPB
mechanism, here the transport runs through the bulk of the electrode and the oxide ion is oxidized at the
electrodes surface. This guarantees no percolation issues.
For the cathode, nickel is the most used metal, dispersed in YSZ. It is cheap, chemically active, and
𝐻
remains metallic in a wide range of 𝐻 2𝑂 ratios, but at least the amount of hydrogen shall be 10%. However,
2
pure nickel cannot be used as its TEC is too high, it has a high sintering tendency, and it can only transport
electrons, those are the reasons why YSZ is used, with an optimal wt% of 60 of nickel in YSZ. Also in
this case, a percolating structure is needed. The global HER is:
− 𝑋
𝐻2 𝑂 + 𝑉𝑂,𝑌𝑆𝑍 + 2𝑒𝑁𝑖 → 𝐻2 + 𝑂𝑂,𝑌𝑆𝑍

Where the RDS is the electronation of water, that is adsorbed on the YSZ surface.
The interconnects serve as electrical connectors between two adjacent cells and as anode/cathode
separators. They must be good electronic conductors and oxide-ions insulators (they shall transport only
electrons), chemically stable in both reducing and oxidizing environments, and thermally matching the
TEC of anode and cathode. The metallic connectors (𝐶𝑟2 𝑂3 alloyed steels) are good conductors and ions
insulators, cost-effective, but oxide scales with high contact resistance. Ceramic interconnectors (alkaline-
earth-doped), unfortunately, are mixed ionic and electronic conductors at low oxygen pressure,
mechanically weak, but don’t form scales.

26
 PROCESSES FOR HYDROGEN

The main deactivation phenomena featuring the SOE are anode delamination, chromium volatilization,
and nickel migration. Delamination of the anode is mainly due to the high pressure of oxygen at the
electrolyte/anode interface: the electrode detaches from the electrolyte surface, increasing the resistance.
For this reason, there is a maximum current density that we can input inside the electrolyzer to not
quickly deactivate the anode. A possible solution could be working with pure oxygen at the anodic side,
in order to not have partial pressure gradients that can detach the electrode, but pure oxygen at those high
temperatures would oxidize any metallic part of the compartment. It can also be attributed to the thermal
cycles experienced by the cell: increasing and decreasing the temperature can deteriorate the anode, this
is why is difficult to match SOEC with intermittent sources of electricity. Chromium volatilization, whose
22 wt% is needed to match the TEC of YSZ, is caused by the chromium species formed in the metallic
interconnectors, that in turn can form volatile gaseous compounds that precipitate at the TPB. Nickel has
a high mobility in the 650-950 [°C] range and can accumulate out of the cermet, leaving the
cermet/electrolyte interface.
As already mentioned, SOE can have different configurations. The planar SOEC stack uses electrode-
supported cells with extremely thin electrolyte (down to 2 [𝜇𝑚]), compact modular stack with high power
density and parallel and series assembly. The material used shall be high-temperature resistant, such as
refractory glass or glass-ceramic sealings. For the interconnectors, a planar standard heat exchanger is
designed. The tubular SOEC bundles are electrolyte-supported cells with thick electrolyte (> 200 [𝜇𝑚]),
that provide resistance to thermal cycling. Here, simple low temperature sealing and almost perfect gas
tightness are proved up to 15 [bar], and this the reason why is sometimes preferred to the planar one.
They present low power density, high internal resistance, and bulky stack arrangement. The methods for
manufacturing are established but they are expensive. Multiple shapes are possible, with different heat
and current heat exchange.

Above, two different block diagrams are shown. To the left, the simplest one, with a high temperature
heat source, that shouldn’t be based on combustion but on solar-thermal or nuclear sources, that must
sustain the operation with YSZ-based systems (700-900 [°C]), and two exchangers to recycle some heat
from the products. For the high temperature heat exchangers, special materials and efficient heat recovery
are required. To the right, the idea is to couple the SOEC with a Rankine cycle and the production of
high-pressure steam. The upstream of the cell is the same, but the air enriched in oxygen is then sent to a
radiation receiver (solar) and a part of it is going to the electrolyzer, while the remaining part is expanded
and used to produce high-pressure steam. A SOEC can also be coupled with nuclear energy to produce
27
 PROCESSES FOR HYDROGEN

pink hydrogen, where nuclear electricity compensates the load curves during low load periods. Indeed,
both nuclear plants and SOEC can’t be shut down, and the cell must be maintained at the operating
temperature, even with no current. So, base load from the nuclear plant can be used to produce heat.

Unitized Regenerative Fuel Cell Systems

These are systems that can work either in electrolysis mode or fuel cell mode, depending on the load
curve and on the way it is interpreted. In the peak shaving mode (left side), there is a fixed energy range
in which the device usually should work at the open cell voltage (OCV), namely at 𝐸𝑁 → 𝑃𝑒𝑙 = 0. Even
though the electrical power, either input or output, is zero, there is still the need of supply or subtract
some heat to maintain the system temperature at a certain level (recirculation system, pumping, etc). In
the load levelling mode (right side), the conduction of the device depends on the load without any energy
range, but the energy level is fixed at the OCV: below it, the device works as a fuel cell, otherwise as an
electrolyzer; it works at the OCV only in certain instants.

Above there is a representation of a usual system working in this way. The direct current busbar is the
part that connects all the components directly or via adapted power electronics. The generator is either a
photovoltaic unit, or a wind energy converter, or a hybrid system. A battery allows to handle short term
storage from hours to a few days, and start-up and shutdown. The electrolyzer/fuel cell system
compensate for long term storage and seasonal differences in electricity production, while it is connected
to compressors and high-pressure tanks to store hydrogen and oxygen at elevated energy density.
Eventually, there is a central energy management system monitoring all energy flows and states of charge
of the storage systems, and an inverter to convert the direct current produced or consumed by the
28
 PROCESSES FOR HYDROGEN

electrolyzer/fuel cell system because the grid needs alternate current. In this system, the state of the
battery charge is the main control parameter of the energy storage system, because it states whether the
device is working in electrolyzer or fuel cell mode. However, full charging or full discharging of the
battery can deteriorate it, reducing its service life. So, the central energy management system allows SBC
values between 30% and 80%, and the disconnecting points are 30% for the fuel cell and 80% for the
electrolyzer. However, the number of switching operations are minimized, and the electrolyzer works
either at constant load or in its permissible range, while the fuel cell usually works at a set point of power.
The regenerative fuel cell systems (RFC) are devices able to couple a water electrolysis cell and 𝐻2 /𝑂2
fuel cell. It stores electricity into chemicals by splitting the water molecules into gaseous hydrogen and
oxygen, and it generates electricity by the combustion of hydrogen with oxygen in the fuel cell. In these
systems, the power output and the storage capacity are decoupled. The discrete RFC (DRFC) is a tandem
device with separate electrolyzer and fuel cell units. The unitized RFC (URFC) is a hybridized and
integrated device which combines the electrolyzer and the fuel cell into a single unit that operates
reversibly. Of course, this second option requires two fully reversible electrodes. Generally, pressurized
operations are preferred because large capacitance and/or smaller volumes can be implemented.

The only cell that cannot be used as both electrolyzer and fuel cell is the alkaline one. So, the unitized
systems can be featured either by PEM or SOEC, while the discrete systems by an alkaline electrolyzer
and a PEM fuel cell, because is challenging to couple a ceramic device as the SOC with a non-ceramic
one due to temperature differences. The reversibility of a cell is of utmost importance in these operations.
It mainly depends on the shape of the
polarization curve, as shown to the side. since
the resistance of a cell is defined as:
𝑑𝐸𝑐𝑒𝑙𝑙
𝑅=
𝑑𝑖
We have to look at the behaviour of the curve
in the point where the switching occurs, so at
𝑖 = 0. The SOC has a low resistance at null
current, thus its reversibility is almost always
possible. The PEM, instead, presents a
resistance tending to infinite in that point,
meaning that the reversibility is hardly possible in those conditions. Moreover, the bottleneck of the
system is usually the oxygen electrode.
If a PEM is used as reversible electrolyzer/fuel cell, it can be symmetric in terms of catalyst used for the
half reactions: platinum for both the hydrogen and oxygen electrodes, because it’s active also in the OER
and the iridium can’t be used for ORR. Moreover, the PTL must be done in titanium oxide as happens

29
 PROCESSES FOR HYDROGEN

for a normal PEM electrolyzer, because carbon cannot be used due to its possible corrosion at the oxygen
electrode during electrolysis. With a reversible PEM, there are two operating modes: the same electrode
deals with same gas mixture (either hydrogen or oxygen
and water); the gas mixture is switched from fuel cell to
electrolysis. In the former case, the electrode changes the
charge, while the latter case the electrode maintains the
charge. The roundtrip efficiency is the ratio between the
energy in output from the fuel cell and the energy in input
for the electrolysis:
𝑄𝐹𝐶 𝑉𝐹𝐶 𝑉𝐹𝐶
𝜂𝑅𝑇 = ≅
𝑄𝐸𝐿 𝑉𝐸𝐿 𝑉𝐸𝐿
Where 𝑄 is the charge, namely the amount of hydrogen
produced. If we analyze the system at a certain amount of
hydrogen produced, the efficiency is given by the ratio of
the two voltages at fixed current. Looking at the chart to
the side, it’s clear that at a certain current, the voltage of the
electrolysis is higher than the voltage of the fuel cell,
resulting in a roundtrip efficiency of 44%. It can be interpreted therefore as a higher overpotential to get
a useful current density.
Comparing a PEM cell in fuel cell or electrolysis mode, there are some differences. First, the electrolyzer
would cost more due to noble metals even at the oxygen side (iridium). However, the fuel cell requires
thinner electrolytes due to worse performance in terms of efficiency. Moreover, the fuel cell in sensitive
to the so-called water flooding, according to which if there is too much water in the membrane, the
performance worsens due to pore obstruction. For this reason, the PTL is made of a hydrophobic
material. Instead, in the electrolysis the water is a reactant, and as such it needs to be brought inside the
PTL, that is made of a hydrophilic material. Eventually, the fuel cell is less efficient than an electrolyzer
because of the heat lost during the conversion between chemical energy in electrical energy. Indeed, in
the electrolysis, the heat that has to be removed because of the exothermic operation can be used for
second applications, such as trying to operate the device at the thermoneutral point. Moreover, the fuel
cell has a limiting point in terms of efficiency, because it is supposed to not burn all the fuel that is
entering, because in case of complete conversion the power output would be zero (it is a high current
corresponding to null voltage). Because of that, the amount of hydrogen entering the system (coming
from an electrolyzer) has more energy than the output one.
The heat exchange is important inside the stack, because
heat has to be removed both in fuel cell and electrolysis
mode (if dealing with a PEM, as in the picture to the side).
When the power installed is sufficiently low, an external
heat exchanger is implemented in the system, but when
the power is larger the heat exchange needs to be
integrated inside the stack, and notably inside the bipolar
plates. In that case, the bipolar plates shall be thicker to let
the coolant pass, increasing the resistances and the
possibility of leakage inside the cell. In fuel cells more than
𝐴
half of the energy is heat at 1 [𝑐𝑚2 ], and the sensible heat
absorbed by the product streams is negligible due to the
low specific heat. However, temperature level is important
as above 90 [°C] the membrane degrades, while below 60

30
 PROCESSES FOR HYDROGEN

[°C] the vapour pressure of water gets lower, causing electrode flooding.
The only cell that is fully reversible is the SOC, but only used in the unitized system, because is hard to
couple a ceramic cell with a non-ceramic one. So, the same materials are used in a SOEC and in a SOFC,
also called reversible SOC (ReSOC). They have a higher roundtrip efficiency than PEM-URFCS due to
higher individual ones in electrolysis and fuel cell mode (around 61% at 750 [°C]). The only issue could
be the degradation rate, also due to the redox cycling (polarization changing). Usually, SOFC are never
operated below 0.7 [V] because of high thermal gradients that can lead to thermomechanical stresses to
the materials. On the other hand, SOEC are operated at the thermoneutral voltage, and that’s the reason
why the excess of heat cannot be recovered in SOFC. Indeed, heat integration requires operations below
the thermoneutral point and above the maximum power density voltage 𝐸𝑚𝑎𝑥 . Hydrogen conversion in
SOFC mode and water conversion in SOEC mode are not complete.

Above there is a ReSOC process scheme. The preheaters are necessary both on air and fuel side to avoid
any thermal stresses. Indeed, the ceramics are at 700 [°C], and any low temperature stream could lead to
breakage. In electrolysis mode, 10% v/v hydrogen in steam is required to avoid nickel oxidation, and air
flow is minimal due to corrosion. In fuel cell mode, the air flow controls the stack temperature, and
anodic recirculation is required because of low hydrogen conversion. Here, the recirculation blower
requires a minimum amount of steam, otherwise the Nernst potential would tend to infinite. Evaporator
and the stack heaters are electrical.
The SOEC are the only ones that can run a 𝐶𝑂2 co-electrolysis, meaning that they can perform a
reduction of the carbon dioxide and of water simultaneously:

𝐻2 𝑂 + 2𝑒 − → 𝐻2 + 𝑂2−

𝐶𝑂2 + 2𝑒 − → 𝐶𝑂 + 𝑂2−
Producing syngas as a product. Considering that there is also the WGS reaction going on, by regulating
the temperature one can tune the CO/H2 ratio of the syngas, according to thermodynamics. Indeed, no
𝐶𝑂2 electrolysis would occur above 831 [°C] as the RWGS is present. However, since the enthalpy of
reaction is higher, in this case the thermoneutral voltage is 1.48 [V] vs 1.3 [V] of the normal electrolysis.
The ionic and electronic current density follows the trend reported in the graph below. A SOFC is more
than a simple electrochemical cell as most of the space inside can be exploited to have an active catalytic
surface. Usually, the hydrogen electrode is 400 − 500 [𝜇𝑚] thick, while the electrochemically active
31
 PROCESSES FOR HYDROGEN

thickness is only a few microns, as shown in the chart. This means that most of the part of the hydrogen
electrode is there for mechanical reasons, as already mentioned. Moreover, the catalytic layer is formed
by nickel supported on YSZ, which is active in most of the
oxidizing/reforming reactions. For this reason, external
reformers can be integrated. The external reformer presents
a first reforming with a hydrocarbon and water, producing
syngas, that then is fed to the fuel cell reacting to carbon
dioxide and water. Direct internal reformers, with all the
operations occurring inside the cell starting from
hydrocarbons and water, guarantee higher efficiency.
However, direct internal electro-oxidation with no water is
unfeasible on Ni/YSZ. The 𝐻2 𝑂/𝐶𝐻4 feed has been proved
with high excess of steam with nickel on YSZ, but H/C
ratios higher
than methane
are very
difficult to process because higher hydrocarbons would
crack before reaching the operating temperature of the cell.
These concepts can be merged in the so-called power-to-
methane electrochemical energy storage. Here, in the
discharge fuel cell mode, the endothermic methane steam
reforming is couple with exothermic hydrogen electro-
combustion, while in the charge electrolysis mode the
exothermic methanation is coupled with endothermic
𝐻2 𝑂/𝐶𝑂2 co-electrolysis. This operation can reduce the
operating temperature and pressure in favour of methane-
rich products. The roundtrip efficiency is around 70% up to
250 [MW] for the ReSOC. An extremely large storage
capacity is guaranteed when dealing with underground
cavern storage for 𝐶𝑂2 and 𝐶𝐻4 .

Solar Thermochemical Hydrogen Production

In this technology, the solar energy is thermally stored and used to split the water molecule. So, any
electrical input is used to perform an electrolysis, and any photovoltaic panel is adopted. Solar panels
are used to concentrate the solar incoming energy, and that’s why they are called parabolic solar
concentrators. Indeed, solar rays are not parallel, but they can be assumed to originate at a disk which
subtends an angle 𝜃 of 0.0093 radian. Given that the solar energy corresponds to 1 [kW/m2] (in absence
of any obstacle), the purpose of the solar concentrator is to increase this energy by concentrating the solar
rays on a smaller area. The reason behind adopting the concentrators is that there are lots of losses along
the path of solar rays, such as geometrical (not exact parabolic profile) and optical imperfections (poor
reflectivity of the mirror), and shading effects (clouds). The concentration ratio, telling how much the
radiation is increases, can be calculated according to the angle of the incoming rays (Φ):
4
𝐶≅ sin2 Φ
𝜃2
Below, four kinds of concentrators are shown. The parabolic through concentrators have a parabolic
shape that concentrates the solar rays on the focus of the parabola, where a pipe is built and wherein a
fluid is passing, that has to be heated by the solar energy. The linear Fresnel concentrators are flat and
concentrate the solar rays on a unique pipe passing where the energy is concentrated. The dish/engine
has a shape such that the solar rays are directed on a single point, it is mainly used for low-power

32
 PROCESSES FOR HYDROGEN

applications. Eventually, the central receiver is made of several flat concentrators that heat up a
cylindrical receiver in the centre of the plant.

Below, some typical values for the concentrators are shown. Parabolic and Fresnel types have similar
concentration factors and temperatures around 400 [°C], this is why they cannot be used for hydrogen
synthesis as the temperature is too low. However, they
can be used for other chemical reactions as the power
produced is sufficiently high. The parabolic dish has
huge concentration factor and temperature reached,
and that’s why they can be used for high-power
applications as such a high amount of energy would
melt any kind of material, and thus they are also small.
The central receivers are the most suitable
concentrators for hydrogen production.
On the other hand, the receiver has 4 different types of configurations. The direct internal receivers use
fluids or particle streams directly exposed to the concentrated solar radiation. Indirect irradiated receivers
use absorbing surfaces exposed to the concentrated solar radiation (the chemical species are directly
heated up). In cavity type receivers the radiation enters through an aperture into an insulated enclosure
containing the absorber. External type receivers are cylindrical indirect receivers where heat is conducted
across the walls to the thermal fluid.
At 𝑇 > 1000 [𝐾], the net power absorbed decreases due to radiative losses through the aperture (it starts
to emit radiation as well). For a perfectly insulated cavity receiver (neither convection nor conduction
heat losses), the adsorption efficiency is given by:

𝑄𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 − 𝜀𝐴𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 𝜎𝑇 4 𝑄𝑛𝑒𝑡,𝑎𝑏𝑠

𝜂𝑎𝑏𝑠 = =
𝑄𝑠𝑜𝑙𝑎𝑟 𝑄𝑠𝑜𝑙𝑎𝑟

The mean flux concentration ratio 𝑪 ̃ over an aperture is the capability of the collection system to
concentrate the solar energy. It is normalized with respect to the incident normal beam insolation and
averaged by the area:

33
 PROCESSES FOR HYDROGEN

𝑄𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒
𝐶̃ =
𝐼𝐴𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒

It can be assumed that the aperture captures all the incoming power (𝑄𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 = 𝑄𝑠𝑜𝑙𝑎𝑟 ), and that the
cavity receiver is a perfectly insulated blackbody (𝜀 = 1). The Carnot limit must be included to account
for the solar to fuel efficiency of the process, because the receiver will be included in a power cycle:

𝜎𝑇𝐻4 𝑇𝐶
𝜂𝑖𝑑𝑒𝑎𝑙 = (1 − ) (1 − ) = 𝜂𝑎𝑏𝑠,𝑖𝑑𝑒𝑎𝑙 𝜂𝐶𝑎𝑟𝑛𝑜𝑡
𝐼𝐶̃ 𝑇𝐻

The stagnation temperature is the highest temperature that the ideal solar receiver achieves. At this
temperature, all the solar energy is being re-radiated as fast as it is absorbed. Looking at the chart, the
solar to fuel efficiency presents a maximum. Indeed, the Carnot efficiency gets higher as the temperature
increases, but the absorption efficiency, at a certain point, starts to drop due to the radiation emitted by
the aperture, and becomes zero at the stagnation temperature. so, the optimal temperature is found
between these two efficiencies.

The idea behind solar hydrogen is to heat water as much as possible to split it into hydrogen and oxygen.
By looking at the reaction:
1
𝐻2 𝑂 → 𝐻2 + 𝑂2
2
We see that the Gibbs free energy of reaction Δ𝐺𝑅0 = 0 → 𝑇 ≈ 4330 [𝐾]. Of course, being an endothermic
reaction, only above that temperature the reaction is favoured. Also, being with an increase in the number
of moles, it is favoured at low pressure, and this is a problem because ideally the hydrogen would be
produced already pressurized for downstream operations. Moreover, materials that can handle those
temperatures are not easy to find.

34
 PROCESSES FOR HYDROGEN

One of the first attempts to carry out this synthesis was in 80s, where a reactor composed of a lens was
designed. It has a parabolic dish shape, and it can concentrate the solar energy in the reactive chamber,
where steam is injected. At least temperatures of 2500 [K] must be
achieved to have a non-null steam conversion. The reaction occurs,
and the products are separated immediately at high temperatures,
because upon quenching the mixture would explode. So, the
mixture is separated through a zirconia membrane, that is semi-
permeable and selective on hydrogen, while steam and oxygen are
just exiting the chamber.
However, since it is a semi-
permeable membrane, the
separation is not perfect, and there
would be some oxygen and
hydrogen in the two mixtures.
Moreover, the quenching is one of
the most important parts of the
downstream process, because a
enormously quick cooling is
important to prevent any
explosion. It can be carried out by
fresh gas injection (cold
hydrogen), or by nozzle expansion by pressurizing the system upstream,
to reach cooling rates in the order of millions of [K] per second. In terms
of performances, the quenching is dramatically worsening the efficiency
of the cycle, because it is removing something that can be used for other purposes as it has a great amount
of energy, but the cooling is strictly necessary. Besides, the faster the cooling, the more irreversible and
inefficient the exchange is.
Another idea stems from the so-called water splitting thermochemical cycle, where the solar energy is
used to heat up a metal oxide to be reduced and oxidized once again, producing oxygen and hydrogen.
The solar energy is concentrated in the solid oxide, that is reduced, producing the metal and pure oxygen.
Then, the metal is fed to the water splitting reactor,
where steam is injected to oxidize the metal and
producing hydrogen. Eventually, the metal oxide is
recycled to the solar reactor. It could be a good idea to
reverse the cycle during nighttime, where hydrogen
and oxygen are fed to produce power. This possible
configuration has some benefits with respect to the
previous one, because it avoids extremely high
temperatures and bypasses the complex separation of
oxygen and hydrogen. However, the two-step cycle is
almost ideal, because actually just few metal oxides can
perform reduction/oxidation in two steps, most of
them need more than one step to be reduced and
oxidized, lowering the efficiency of the process. some
attempts have been made to run the cycle with sulphuric acid.
One of the first attempts has been made on zinc oxide, because ideally it represents a two-step cycle, the
most favourable one, even from a thermodynamic viewpoint, through the following reactions:
1 𝑘𝐽
𝑍𝑛𝑂 → 𝑍𝑛 + 𝑂2 Δ𝐻𝑅0 (298𝐾) = 478 [ ]
2 𝑚𝑜𝑙

35
 PROCESSES FOR HYDROGEN

𝑘𝐽
𝑍𝑛 + 𝐻2 𝑂 → 𝑍𝑛𝑂 + 𝐻2 Δ𝐻𝑅0 = −104 [ ]
𝑚𝑜𝑙
The reduction of zinc is conducted at 2000 [°C], while its hydrolysis at 500 [°C], introducing in the cycle
an inefficiency in terms of temperature gap, because the stream needs to be quenched. Indeed, 33% of
the input is lost by quenching with cold nitrogen. Moreover, there are safety issues concerning the cooling
about forming solid particles in presence of oxygen, that can incur in explosions, while inside the receiver
zinc is liquid. Due to the high temperatures, refractory materials like alumina shall be used, and they are
covered by ZnO thick tiles. The second most favourable cycle is done with iron oxide:
1
𝐹𝑒3 𝑂4 → 3𝐹𝑒𝑂 + 𝑂2 𝑇 = 1400 [°𝐶]
2
3𝐹𝑒𝑂 + 𝐻2 𝑂 → 𝐹𝑒3 𝑂4 + 𝐻2 𝑇 = 800 − 1000 [°𝐶]
We can immediately see that the temperature gap is lower than zinc process (lower inefficiencies). Even
this process is carried out with multiple phases, as a thermodynamic based process implies (liquid and
solid). Pure 𝐹𝑒3 𝑂4 cannot be used due to sintering (loss of surface area), and for this reason it is doped
with zinc and nickel, while supported catalytic structures are required to lower the temperatures. Indeed,
lower hydrolysis temperatures reduces heat losses and allows for fixed bed monolith structures. The only
cycle that has been successfully run with iron and ferrites-based material is the Hydrosol project at 100
[kW], where refractory SiC, multichannel honeycomb coated with iron-based material mixed oxides has
been used. It was operated in continuous with alternate cycles of 𝐹𝑒3 𝑂4 dissociation and 𝐹𝑒𝑂 hydrolysis.

The carbothermal cycles are based on the fact that by introducing some carbon in the system, the
temperature at which the metal oxides are reduced, producing hydrogen, is dramatically decreased.
However, high quality additional energy inputs are required, chemical potential or electrochemical work.
Coal or coke are preferential reducing agents in blast-furnaces processes. Complications due to
intermediate carbide and oxycarbide formation are present.
Eventually, the sulphur cycles are the last kind of cycle through which hydrogen can be made, and they
are developed for very high temperature nuclear reactors (thanks to their perfect sealing). The baseline
step is the splitting of sulphuric acid:
𝐻2 𝑆𝑂4 → 𝑆𝑂3 + 𝐻2 𝑂 𝑇 = 500 − 800 [°𝐶]
36
 PROCESSES FOR HYDROGEN

1
𝑆𝑂3 → 𝑆𝑂2 + 𝑂2 𝑇 > 800 [°𝐶]
2
𝑆𝑂2 + 2𝐻2 𝑂 → 𝐻2 𝑆𝑂4 + 𝐻2 𝑇 = 80 − 90 [°𝐶]
The last one can be electrochemically performed with a PEMC, with a relatively low amount of energy
(𝐸 0 = 0.17 [𝑉]), as done in the Mark cycle. Of course, the cell must be built with special materials as the
environment is highly acidic. In general, in the whole process attention must be paid to sulphur emissions
as they are harmful to the environment. Instead, the Cristina cycle perform all the steps in one reactor:
1
𝐻2 𝑆𝑂4 ↔ 𝑆𝑂2 + 𝑂2 + 𝐻2 𝑂
2
There are two adiabatic volumetric receivers-reactors, where sulphuric
acid is fed, and the steps occur. In the 𝐻2 𝑆𝑂4 vaporization reactor, there
a SiC foam and the material used are stainless steel and quartz glass.
There is the direct injection of liquid sulphuric acid in air within the
foam, while in the 𝑆𝑂3 cracker there is a SiC honeycomb coated with
Cr-doped 𝐹𝑒2 𝑂3 catalyst. The cycle is supposed to produce hydrogen
during daytime, and sulphuric acid in nighttime.
The sulphur iodine cycle uses iodine to perform a second cycle in
addition to the sulphur one, forming an entirely thermal process. After
the sulphuric acid decomposition at 900 [°C], oxygen is separated and the mixture 𝐻2 𝑂/𝑆𝑂2 is mixed
with some iodine, producing iodide acid and sulphuric acid. Here, there is a large excess of water to form
acid solutions (iodide is hard to dissolve), and the physical state of iodine is critical as its fusion
temperature is 113 [°C] and the Bunsen reaction is carried out at 120 [°C]. Moreover, the separation
between the sulphuric acid and iodide acid is critical as it requires an azeotrope distillation. Then, the
iodide acid is decomposed at 300 [°C] upon heat and hydrogen is produced. If all the energy comes from
solar radiation, we have an entirely solar-based process.

HYDROGEN FROM COAL

Clean coal technologies refer to the group of various technologies intended for the energetic and
chemical utilization of coal with the smallest possible environmental impact. CCTs aim at eliminating
the emissions of particulates, ashes, 𝑁𝑂𝑥 , 𝑆𝑂𝑥 , chlorine, ammonia, and 𝐶𝑂2 . Carbon capture and storage
(CCS) refers to several technologies that capture 𝐶𝑂2 at some stage of processes as combustion for power
generation, or gasification. The captured 𝐶𝑂2 is then pressurized to around 100 [bar] or more and
transported to a storage site. Generally:

37
 PROCESSES FOR HYDROGEN

• Black hydrogen comes from black coal (anthracite) using gasification, producing 14 −
16 𝑡𝐶𝑂2 /𝑡𝐻2 .
• Brown hydrogen comes from brown coal (lignite) using gasification, producing 18 − 20 𝑡𝐶𝑂2 /𝑡𝐻2
• Blue hydrogen is black, brown, or grey hydrogen with CO2 capture and utilized. Blue hydrogen
and blue power require more than 95% of CO2 captured.

Carbon Capture Technolgies

During coalification, the content of humidity and organic volatiles decreases, and the amount of fixed
carbon grows. The ultimate analysis shows that the fractions of oxygen and hydrogen drop, while that of
carbon increases. With growing C and decreasing oxygen, the lower heating value of coal increases.
Most of the CO2 transportation currently occurs via pipeline. Operating pressures of the existing CO2
pipelines are in the range of 85 to 210 bar where CO2 is a dense‐phase fluid over a wide range of
temperatures (TC = 31°C PC = 74 bar). The compression and/or liquefaction stage represents a significant
use of energy, as high as 12% of the loss of efficiency of a power plant. Little difference is observed when
dealing with clean CO2 streams (> 95% v/v purity), but large penalties (increases in power requirements
of between 12‐30%) are incurred for less concentrated streams.

There are three main carbon capture configurations. The post-combustion capture represents the CO2
capture right after the combustion chamber. Steam is separated to produce power, while the flue gases
are mainly composed of nitrogen for 70%
and CO2 for 3% to 15%. For this reason,
since CO2 is more diluted, the separation
results more difficult than in other scenarios.
Moreover, the high temperature right after
the combustion doesn’t favour the
separation for an absorption. One of the
most adopted configurations is the pre-
combustion capture, where air is separated
to perform a gasification of a fuel in almost
pure oxygen and producing syngas and a
higher amount of CO2. Then, thanks to the
high temperature, a WGS can be performed
in the same chamber to increase the amount
of hydrogen and CO2, that is about 40%. In
this case, the separation is easier. The last
option would the oxycombustion,
performed with pure oxygen. In this case,
the exhaust gases recycle is necessary to lower the high temperature due to the stoichiometric combustion.
The integrated gasification combined cycle (IGCC) is a common process using coal to produce syngas
through gasification, combining it with steam generations and gas turbines. The 5% moisture coal is first
prepared and grinded to be suitable for the following entrained bed reactor, where the gasification occurs
at 28 [bar] and 1500 [°C]. Steam and 95% pure oxygen are used as fluidizing agents, and recycled syngas
is used to cool down the system. The products are withdrawn and cooled down by producing steam,
integrated in a separated cycle. The solids are removed, and the CO2 is separated through a scrubber.
Then, the acid gases are also removed, and the syngas passes through a usual gas turbine where it is
burned, sent to the heat recovery and steam generation (HRSG) and the exhausted released in
atmosphere. The benefits brought by this process are higher CO2 fraction in the exhaust as well as their
pressure, reduction of compression costs and loads, and lower volumetric flow rates, reducing the
volumes. However, the standard cycle doesn’t include the CO2 capture, and the clean up of cold and hot
gas reduces the efficiency. Moreover, an ASU is required, and the separation of solid particles is complex.

38
 PROCESSES FOR HYDROGEN

The calcium looping technology (CaL) exploits the spontaneous reaction between CO2 and CaO, that
is exothermic:

𝐶𝑂2 + 𝐶𝑎𝑂 → 𝐶𝑎𝐶𝑂3

The simplest way to implement it is to have some power production by burning a fuel in air, having some
flue gases entering a carbonator, where the above reaction occurs and store the CO2 inside the calcium
carbonate. Then, the calciner needs heat
to release the CO2, as well as a
limestone, the natural rock that is able
to produce CaO, that is recycle back to
the carbonator. The heat for the
calcination is produced by burning a
fuel with pure oxygen. One important
advantage of this process is the very
high selectivity the CO2 is absorbed by
the calcium oxide, so that all the other
flue gases can be separated in an
effective way at high temperatures. So, the energy penalty of a CaL can be much lower with respect to a
conventional amine scrubber. Moreover, limestone is widely available industrially, it’s not harmful and
it's very cheap. A CaL process can be integrated into a cement manufacture via utilization of the spent
calcium carbonate.
The calcination occurs between 1173K and 1273K, with 90% CO2 and 10-30% of steam. The sorbent
mass after calcination (baseline) remains constant, while the mass after carbonation reduces with the
number of cycles to 8-10% of its initial value. Sintering, grain growth of CaO, pore shrinkage and
embrittlement deactivate the sorbent. The CaCO3 formed on the surface of a CaO particle has larger
molar volume. This product layer grows over the mouths of pores, and seals them off. Some of the drop
in carrying capacity is due to closure of pores on carbonation, which do not reopen. Upon the first
calcination, a highly dispersed porous CaO is formed, which is highly reactive. The following
carbonation does not reach completion due to pore blocking. Some pores do not reopen upon the next
calcination, and this process continues until a rigid backbone of considerably less reactive sorbent is

39
 PROCESSES FOR HYDROGEN

formed after many cycles. This process is accelerated by sintering. Limestone (initial capacity 44/56 =
0.79 gCO2/g) and dolomite (CaCO3‐MgCO3 initial capacity 44/96 = 0.46 gCO2/g) are preferential
sorbents.
Hydration is the bets method to activate limestone sorbents for CO2 capture. It creates cracks in the CaO
particles by forming the hydroxide and forms larger pores. Weakening is a drawback of this utilization.

𝐶𝑎𝑂 + 𝐻2 𝑂 → 𝐶𝑎(𝑂𝐻)2 Δ𝐻𝑅0 < 0

𝐶𝑎(𝑂𝐻)2 + 𝐶𝑂2 → 𝐶𝑎𝐶𝑂3 + 𝐻2 𝑂 Δ𝐻𝑅0 < 0

Even though the thermodynamics would require temperatures below 500 [°C], the gasification needs
higher temperatures and pressure up to 80 [atm]. There is also the competition with the sulphonation
reaction:
1
𝑆𝑂2 + 𝐶𝑎𝑂 + 𝑂2 → 𝐶𝑎𝑆𝑂4
2
1
𝑆𝑂2 + 𝐶𝑎𝐶𝑂3 + 𝑂2 → 𝐶𝑎𝑆𝑂4
2
Where calcium sulphate is an irreversibly lost solvent, with very limited use.
The CaL can be used to integrate a further gasification to produce hydrogen, by using the heat released
by the sorption of CO2, in the HyPr-Ring process. The WGS reaction is prompted by mass removal
effect, namely by removing the calcium carbonate that precipitates, increasing the H2 production
efficiency up to 70%. The combustion of residual coal with pure oxygen from the ASU is also integrated.
Issues are the high operating pressure (70 bar), the excess of steam required and the H2 purity, because
some methane is also formed.

A second application of the CaL is the Alstom Power process, where the only difference from the
classical CaL is the use of calcium sulphate, that reacts with carbon to produce CO:

𝐶𝑎𝑆𝑂4 + 4𝐶 → 𝐶𝑎𝑆 + 4𝐶𝑂

𝐶𝑂 + 𝐻2 𝑂 → 𝐶𝑂2 + 𝐻2
𝐶𝑂2 + 𝐶𝑎𝑂 → 𝐶𝑎𝐶𝑂3
𝐶𝑎𝑆 + 2𝑂2 → 𝐶𝑎𝑆𝑂4
Three possible configurations are present: producing pure H2, producing syngas, producing power. The
standard limestone loop is added to a calcium sulphate loop exclusively to produce pure H2. Oxygen is
delivered via reaction of the calcium sulphate with carbon in coal.
40
 PROCESSES FOR HYDROGEN

The unimixed fuel process technology can use different feedstocks (coal, biomasses) to produce H2 and
electricity in adjustable ratio. Three fluidized beds connected in series are implemented, performing an
integrated gasification, a CaCO3 regeneration, and an oxidation. The idea is to not use pure oxygen, so
that an ASU is not necessary. It’s featured by high pressures and temperatures. Sorbent and metal oxide
particles elution and particle intermixing are main issues, decreasing the gas purity, but also the
complexity of the system. It’s near zero carbon emissions process, with 96% of carbon capture efficiency
and 60% hydrogen production efficiency.

HYDROGEN FROM BIOMASSES

Biomasses are material which contains fixed carbon and renovates with high enough frequency to allow
for large supply. Living organisms and recently died organisms are included in biomasses, such as plants,
wood, algae, and animal residues. Biomasses belong to three categories: lignocellulose feedstocks
(gasification products), starchy feedstocks (ethanol), and triglyceride feedstocks (biodiesel). The main
products obtainable form biomasses are gasification products, suitable for making syngas or hydrogen,
ethanol, and biodiesel.

41
 PROCESSES FOR HYDROGEN

Plants use solar energy to combine CO2 and water, forming oxygen and carbohydrate units, organized
in sugar molecules, according to the photosynthesis process:
𝑘𝐽
𝑛𝐶𝑂2 + 2𝐻2 𝑂 + 𝑙𝑖𝑔ℎ𝑡 → (𝐶𝐻2 𝑂)𝑛 + 𝑛𝑂2 Δ𝐻𝑅0 = 470 [ ]
𝑚𝑜𝑙𝐶
Of course, it’s a highly slow process as the reaction is very endothermic, and fixing CO2 with water
requires a great amount of energy, but research is focused on finding plants that are able to quickly
perform this process. Biomass energy with carbon capture and storage (BECCS) allows to achieve
negative carbon emissions.
There are generally three main methods for characterizing the biomass: the elemental analysis exploits
the Van Krevelen diagram; the proximate analysis gives the moisture (water is present in the biomass
due to hydrogen bonding), volatile matter VM (when heated to 950 [°C]), the remaining fixed carbon
FC, the ash (mineral) together with the high heating value
HHV. The proximate analysis is derived from
thermogravimetric analysis (TGA): some small
samples of biomass are heated up in the form of powder,
to avoid any mass and heat transfer limitation and make
the process quicker. The ultimate/elemental analysis
gives the composition of the biomass in weight
percentage of carbon, hydrogen, and oxygen as well as
sulphur and nitrogen. The proximate analysis gives
moisture, volatile matter VM, the fixed carbon
remaining at that point, the ash along with the HHV. By
crosslinking these two analyses, the relative amount of lignin, cellulose, and hemicellulose can be
assessed as the biomass can be considered as a linear combination of cellulose, hemicellulose, and
lignin. The atom balances of C/H/O allow to define biomass composition in terms of three reference
components.
Exactly as per coal gasification processes, also for biomasses an essential step is the grinding, drying,
control of the ashes, and appropriate calibration of the gasifying agents. Due to the heterogeneity of the
composition, elemental analysis and proximate ones are crucial. Tars and char are two typical products
of the biomass gasification. Tars are high-boiling and highly viscous organic compounds which condense
at 300-400 [°C], such as toluene, naphthalene, phenols, and other multiple-rings aromatics. they are liquid
or semisolid at room temperature, and they are also known as bio-oil or pyrolysis oil. Char is the solid
byproduct of pyrolysis and gasification, mostly composed of carbon (>90% after a 600 [°C] process).
Gasification and pyrolysis of biomasses are featured by many reactions thereby that is very difficult to
model all of them. The rates of gasification with steam and CO2 are orders of magnitude slower than
those of the pyrolysis. Below 900 [°C], it takes about 1 minute to gasify more than half of the char, while
the gasification of volatiles is much more rapid. Instead, when gasification occurs with oxygen and steam,
oxygen is preferentially leading to complete the combustion of volatiles, followed by reforming of char
and other products. The cold gas efficiency (CGE) is the ratio between the LHV of the produced syngas
and that of the biomass feedstock:
𝐿𝐻𝑉𝑠𝑦𝑛𝑔𝑎𝑠 𝑛̇ 𝑠𝑦𝑛𝑔𝑎𝑠
𝜂𝐶𝐺𝐸 =
𝐿𝐻𝑉𝑏𝑖𝑜𝑚𝑎𝑠𝑠 𝑛̇ 𝑏𝑖𝑜𝑚𝑎𝑠𝑠
High CGE can be obtained with low gasification temperatures and/or using oxygen instead of air. Of
course, the tar content must be minimized as it is a fatal byproduct, causing fouling of equipment and
poisoning the downstream catalysts. Also, the presence of methane in the products shall be minimized
as it is inert to fuel synthesis and WGS conversion processes.

42
 PROCESSES FOR HYDROGEN

Gasifiers
The reactor can be in co-current or counter current with the gases, and fixed bed, fluidized bed, or
entrained flow reactor. Usually, the heat is provided by the partial oxidation of the biomasses inside the
gasifier (directly heated) or from an external source (indirectly heated).

Updraft and downdraft gasifiers differ for the direction of the flowrates. The updraft or counter
current gasifiers are less sensitive to fuel size and moisture content as compared to downdraft. They
are usually directly heated as the outlet gas can heat up the inlet biomass, and the oxygen inside can
produce heat by performing partial oxidation, with high temperature gradients inside the bed.
Despite the amount of methane and tars is high in the syngas, thanks to the moderate temperature
the CGE is high. The downdraft or co-current gasifiers give relatively cleaner gas (low tar content)
and they are preferred for engine applications. They are also directly heated, but the CGE is lower.

Entrained flow gasifiers (left) are in co-current and directly heated, with harsh conditions (1200-1500
[°C]). They can quickly convert biomasses with low residence time, and thanks to the extreme conditions
inside they are tar- and methane-free due to cracking. However, the CGE is low, and a molten slag is
released from the bottom of the gasifier, both due to the high temperatures.
In bubbling fluidizing bed gasifiers (right), an O2/steam flow passes through a well-mixed bed of
biomass and other solids (sand and CaO) for the bed fluidization. Sand and CaO can help in absorbing
some heat from the environment, but also CaO can absorb some CO2 and the carbon capture can be
integrated with the biomass gasification. Even here, the system is directly heated, and it can keep a
uniform temperature, that is below the ashes softening limit to avoid agglomeration and de-fluidization
(800-900 [°C]). The velocity of the gases is not enough to entrain the solid particles, so that there is no
need of including a solid separation unit. Eventually, they can be pressurized and produce high-pressure
syngas.

In circulating fluidized bed (left), O2/steam flow through a well-mixed bed of biomass and other solids
at high velocity leading to bed fluidization and solid entrainment. It is in co-flow with direct heating, then
the solids and the syngas are separated in a cyclone. It can keep a uniform temperature, always below the
ashes softening point. Despite the possible pressurization, it has high methane and tar content.
43
 PROCESSES FOR HYDROGEN

Eventually, dual fluidized bed (right) is featured by two interconnected fluidized beds (1 BFB + 1 CFB)
indirectly heated. Unconverted char from the gasifier flows in the combustor, where it burns with air.
Heated solids from the combustor flow back to the gasifier providing the heat of reaction. Pressurizing is
not feasible, and the temperature must be kept below the softening point of ashes. The syngas is nitrogen
free even without using pure oxygen, because the combustion is separated from the gasification.

The capacity range for a gasifier is measured in odt/day (oven-dried ton/day), that corresponds to
𝑜𝑑𝑡
1 [𝑜𝑑𝑡] = 18 [𝐺𝐽], or 100 [ ] = 20 [𝑀𝑊]. Of course, the plant size affects the selection of the type of
𝑑𝑎𝑦
the oxidant. Oxygen-blown gasifiers are suitable only in large scale plants, due to the strong economy of
scale of the air separation unit. Co-gasification of coal and biomasses is an interesting option to increase
the size of the plant and improve the economic competitiveness.

44
 PROCESSES FOR HYDROGEN

The GoBiGas plant has a gasifier that is a DFB fluidized with steam, working below 900 [°C], with a 6-
12% of methane in raw gas. Actually, it’s not intended to produce hydrogen but methanol, whose
production syngas is produced in the 1, 2, and 3 pieces of plant. Olivine activated with sulphur,
potassium, and recycled ash (10-15%) is used to reduce the tar production during gasification (they can
catalyse its reforming), while limestone is included to integrated carbon capture. The convection path is
designed for a flow with uneven particle load, and the first downdraft chamber has walls cooled down by
water. The updraft chamber, instead, has integrated heat exchangers that preheat combustion air and
produce steam at 320-350 [°C]. The second convection has also integrated heat exchangers to preheat
combustion air, with flue gases exiting at 170 [°C].

To clean the raw gas coming from the gasifier, an oil gas scrubber can be employed. The OLGA plat is
an example of such separation for tar removal. The raw gas enters the plant at 350 [°C] and is cooled
down prior to a cyclone to separate coarse particles, ash, char, and soot. Then, it is fed to a stripping
column, where it gets in contact with the oil coming from the top, that is trapping all the affine species
and goes to the oil recovery system. The gas exits from the top and the remaining oil parts are removed
by an electrostatic precipitator working as a demister and recycled back to the stripper. Then, it enters a
second absorber, operated above the water dew point to avoid mixing with tar, where it’s additionally
cleaned by oil and then sent to a water condenser and scrubber to remove the residual humidity. The oil
used in the second absorber is then regenerated and stripped by air to remove volatiles. The targets of this
plant are a clean gas with a tar dewpoint less than 5 [°C], namely less than 10 [mg/Nm3] of phenol and
less than 30 [mg/Nm3] of naphthalene in the gas.

45
 PROCESSES FOR HYDROGEN

HYDROGEN FROM NATURAL GAS

Grey hydrogen refers to the H2 produced from natural gas via steam reforming, producing around 9 −
10 𝑡𝐶𝑂2 /𝑡𝐻2 , turquoise hydrogen is produced from natural gas via pyrolysis and thus without carbon
emissions, and blue hydrogen and blue power, as already mentioned, are grey hydrogen and power
produced with more than 95% of carbon capture.

Steam Methane Reforming

The steam reforming is the preferential way nowadays to produce hydrogen as it is a well-established
technology that produces less emissions than black and grey hydrogen:

𝐶𝐻4 + 𝐻2 𝑂 → 3𝐻2 + 𝐶𝑂 Δ𝐻𝑅0 > 0

The feed is composed of desulfurized natural gas and steam, entering in a fire tubular reformer/furnace,
where the ideal operating conditions are 700-850 [°C], 30-40 [atm], and 3-4 H2O/CH4 ratio. The reformer
consists of a radiation section with catalytic tubes and burners arrangement, and a convective section for
heat recovery. The heat is generated by
combustion of a fraction of the incoming
natural gas, and the sensible heat of
combustion is used to preheat the feed and
generate superheated steam at high
pressure. Other than water, CO2 can be
partially recycled to the supply to tune the
H2/CO ratio through the WGS. The
catalyst consists of 15-25 w% of 𝑁𝑖/𝛼 −
𝐴𝑙2 𝑂3 or 𝑁𝑖/𝑀𝑔𝐴𝑙2 𝑂3 loaded in hundreds
of 7-12 [m] long tubes of small diameter (7-
13 [cm]) to obtain a high surface to volume
ratio and maximize the heat transfer, that
has to be evenly distributed.

46
 PROCESSES FOR HYDROGEN

The chemical efficiency of the SMR process is the ratio of the LHV of H2 produced and the LHV of
methane consumed:
𝐿𝐻𝑉𝐻2 𝑛̇ 𝐻2
𝜂𝑆𝑀𝑅 =
𝐿𝐻𝑉𝐶𝐻4 𝑛̇ 𝐶𝐻4

Since methane is combusted to produce steam, the overall efficiency includes the heat required to
vaporize water:
𝑣𝑎𝑝
𝐿𝐻𝑉𝐻2 𝑛̇ 𝐻2 + Δ𝐻𝐻2 𝑂 𝑛̇ 𝑠𝑡𝑒𝑎𝑚
𝜂𝑜𝑣𝑒𝑟𝑎𝑙𝑙 =
𝐿𝐻𝑉𝐶𝐻4 𝑛̇ 𝐶𝐻4

A classical process produces up to 60000 [Nm3/h] of H2, with overall efficiencies around 84% and SMR
efficiencies up to 63%. The hydrogen produced must have at least 99.95% purity, while the CO2 and CO
shall be limited to 10 ppm as a maximum, and other impurities such as H2S, HCl, or ammonia must be
avoided.
The desulfurization is the first step of cleaning of the natural gas, where the H2S is adsorbed on a ZnO
catalyst operating at 300-400 [°C], but also a hydrodesulfurization can also be used over a Ni-Mo or Co-
Mo catalyst. The pre-reforming is an adiabatic low temperature reforming step that is applied upstream
of the conventional one. It is performed below 500 [°C], and its target is the complete conversion of higher
hydrocarbons fed to the main reformer. Indeed, its catalyst has a high nickel content and a very high
surface area, to maximize the performances. It allows to decrease the H2O/CH4 ratio and increase the
heat load on the reformer tubes, namely, less energy consumption and smaller equipment. After the
classical reformer, there is the high-temperature shift reactor. The gases, containing up to 5% of CO,
are cooled down to 350 [°C] in the process boiler, to convert the CO into H2 in the WGS reactor. An iron
oxide catalyst is used in a one-stage adiabatic reactor, and thus the gases temperature rises to 400-450
[°C], even with incomplete conversion. After the HT WGS, the gas is cooled down to ambient
temperature and water is condensed prior to the parallel PSA, composed of at least 3 up to 12 parallel
adsorbers. The purge and decompression gases of PSA are sent to a large buffer vessel to compensate for
variations in pressure and gas concentrations and are recycled to the reformer with a small amount of
fuel gas.
Steam reforming with CCS can be implemented, to produce blue hydrogen. The CO2 can be captured
at different steps along the process chain:

• Pre-combustion CO2 capture: it is trapped from the shifted syngas by chemical absorption with
aqueous MDEA solutions, thanks to the high partial pressure of CO2 in the syngas stream (16%
at 25 bar).
• CO2 capture from the tail gas is economically disadvantageous due to the low partial pressure of
CO2 (51% at 2 bar).
• The post-combustion CO2 capture from the flue gas (21% at 1 bar) uses MEA scrubbing
absorption. Here, we can capture the highest amount of CO2 as the water can be condensed upon
cooling. The solvent is regenerated by heating in a stripper unit, whereby the CO2 is released for
further utilization.
Usually, the tail gas capture is always discarded. For the option 1 there are two configurations, shown
below. The first can capture 50-55% of the CO2, the second 65% but with the implementation of a second
recycle before the PSA. This can lower the pressure at the PSA, making it less stressed. The products are
recycled and burned to supply the energy necessary to the reforming. The third option can capture 90%
of the CO2, and since the flue gases must be cooled down, the CCS can be integrated with steam
generation and power export.

47
 PROCESSES FOR HYDROGEN

A standard SMR is self-supporting, and it can export electrical power or steam. Instead, when CCS is
implemented, some of the heat or electricity produced is used to sustain an additional unit, that’s the
reason why the efficiency is decreasing. Moreover, a SMR emits 9 kg of CO2 per kg of produced
hydrogen, including CO2 formed through WGS. The CO2 solvent is regenerated by using low pressure
steam, because high pressure steam decreases efficiency. On the other hand, CCS SMR requires
additional electricity to capture, compress, and liquefy CO2 at the expense of internal electrical power.
So, SMR plant efficiencies are reduced by 5-14%, with carbon capture efficiencies ranging from 55 to
90%.

48
 PROCESSES FOR HYDROGEN

The other way of conducting the steam reforming is the autothermal reforming, where natural gas reacts
with water and oxygen to thermally support the process under adiabatic conditions. The ideal
stoichiometry is:

𝐶𝐻4 + 0.5𝑂2 → 𝐶𝑂 + 𝐻2
But the process is better described by a different sequence of reactions:

𝐶𝐻4 + 0.5 𝑂2 → 𝐶𝑂2 + 𝐻2 𝑂

Occurs first in a burner to achieve extremely high temperatures. The endothermic steam reforming and
WGS:

𝐶𝐻4 + 𝐻2 𝑂 → 𝐶𝑂 + 3𝐻2
𝐶𝑂 + 𝐻2 𝑂 → 𝐻2 + 𝐶𝑂2
Occur in a catalytic zone downstream, decreasing the temperature and producing syngas. Actually, the
O2/CH4 ratio is slightly higher than the stoichiometric one, to achieve complete combustion of methane
and to avoid the formation of coke that would foul the catalyst downstream. The oxygen must be pure
to not dilute too much the syngas. It can also be pressurized up to 100 [bar]. The process achieves high
energy efficiencies (88%) but requires pure oxygen with high production costs. It can also be designed
with integrated CCS.

Compared to the standard SMR, the autothermal one has a larger electricity consumption due to the
ASU, hence the decarbonization of its energy mix is required with the use of electrical energy with low
emission factor in order to decrease the total emissions. Similar overall emissions can be obtained when
the heat recovery and thermal integration between SMR and CCS is maximized, and HP steam is used
in a turbine, and LP steam is sent to the amine regeneration column.

49
 PROCESSES FOR HYDROGEN

Methane Pyrolysis
Methane pyrolysis is a bridging technology for climate-friendly hydrogen production:
𝑘𝐽
𝐶𝐻4 → 𝐶 + 2𝐻2 Δ𝐻𝑅0 = 75 [ ] , Δ𝐺𝑅0 (545°𝐶) = 0
𝑚𝑜𝑙
It’s also energetically favoured over SMR from liquid water:

0
𝑘𝐽 0
𝑘𝐽
Δ𝐻𝑅,𝑀𝑃 = 37.5 [ ] Δ𝐻𝑅,𝑆𝑀𝑅 = 63 [ ]
𝑚𝑜𝑙𝐻2 𝑚𝑜𝑙𝐻2
And all the existing pyrolysis processes are operated at atmospheric pressure. Undesired saturated and
unsaturated hydrocarbons and polycyclic aromatics are fatal byproducts, and any pyrolysis process
achieves complete methane conversion. The carbon product,
the carbon black, can be a valuable industrial product. From
the thermodynamics, we can see that methane and hydrogen
are not the only products. As methane is consumed, species like
ethylene, acetylene, and benzene form, depending on the
operating temperature. This could be a problem as acetylene is
also explosive. The methods to perform pyrolysis are
essentially three: thermal decomposition (>1000 [°C]), plasma
decomposition (2000 [°C]), and catalytic decomposition (600-
1000 [°C]). Each of them has different configurations and
operating conditions.
The thermal pyrolysis (TRL around 4) is featured by two
alternating furnaces (5 minutes each, until the temperature goes
down to 900 [°C]) filled by refractory bricks, where heat is
provided by burning methane and hydrogen with air. The
byproducts are condensed and removed by spraying some
water. Then, the solid particles are again removed by a cyclone,
The operating conditions are harsh (1400 [°C]). The BASF
thermal MP process adopts a single moving bed reactor with
carbon particles,
where heat is provided by electrical heating. Pyrolysis occurs
on the carbon granules surface, and condensation byproducts
are formed. The thermal pyrolysis can also be performed in
liquid metal media, such as Ti, Pb, Sn, metal alloys, and salts.
The thermal cleavage of methane occurs at the gas/liquid
interface of the bubble, where the liquid metal can also
provide catalytic activity. A carbon film coats the heated wall
after several days of operation. The produced carbon floats as powder on the liquid tin, and the separation
is easier thanks to the lower density.
Plasma pyrolysis (TRL around 8) is characterized by extremely high gas
temperatures (up to 10000-15000 [K]) and equally high electron temperatures.
Electrons, ions, atoms, radicals, and molecules therein are under thermal
equilibrium. The thermal energy and the ionization energy are transferred to
the gas via a direct current discharge between an anode and a cathode, which
ionizes the gas molecules and heats the gas by Joule effect. An issue is the
direct presence of the torch inside the reactor, that reacting can produce a non-
pure carbon. This can be limited by adopting graphite electrodes.

50
 PROCESSES FOR HYDROGEN

The catalytic methane pyrolysis (TRL around 3) is performed with either nickel, cobalt, iron-based
catalysts, or with activated carbon, carbon black, or graphite. In the former case, the catalysts experiences
carbon diffusion inside the pores and precipitation, leading to high cost for the materials and CO2
footprint for the regeneration of the catalyst. Indeed, the carbon deposition leads to deactivation, reactor
clogging, high pressure drops, decline in heat transfer, and interruption of the reaction. The idea is to
adopt a very cheap catalyst (iron), perform the process, collect the carbon, and throwing away the
catalyst. In the latter case, the costs are much lower, as well as the toxicity and a higher tolerance to
sulphur impurities. There is no need for regeneration, but the activities are usually lower. The heating of
the reactor can be done in two ways: the direct heating foresees the combustion of natural gas in two
internal heaters, while the indirect heating needs an additional combustor where the catalyst is in contact
with air that burn the residual carbon deposited. In the second case, either an additional separator for the
CO2 or an ASU is required.

Solar Thermal Reforming

The solar energy can be used to perform the thermal reforming essentially by concentrate the energy up
to 1000 [kW/m2] on a volumetric receiver, or by concentrate it to 200 [kW/m2] on a fluid passing through
a duct composed by metallic foams. The metallic foams can better distribute the heat, maximize it, and
evenly heat the fluid. However, not the whole fluid is in contact with the radiation, because heat is
exchange through conductivity and convection, thus this is less efficient than a direct concentration as in
the first case. In light of this, there are three ways of performing the reforming: a fluid (air) passes through
the receiver, where it’s heated up, and it transfers the heat to the actual reformer; the receiver consists of
the actual tubes where the reforming takes place, with high temperature gradients along the thickness;
the receiver is the actual reformer, that is transparent, and is filled with a catalytic foam inside (the most
efficient one).

The CSIRO solar reformer has an annular reactor composed of a tube disposed in a circular way. Inside,
it’s filled with nickel-based pellet catalyst and air is used as an indirect heating medium. It can be
pressurized up to 20 bar and 850 [°C]. The receiver has a 2.5 [m] diameter and the concentration results
in 800-900 [W/m2].
The WIS solar reformers are directly irradiated receiver, with rhodium-based catalyst doped with
alumina. It absorbs 200-300 [kW], at 750-860 [°C], 3.5 bar, and 85% methane conversion. It can’t be
pressurized because of the lens, that is very fragile.
BASF solar thermal pyrolysis is performed with 16000 heliostats and a 200 m high tower, in a 270 [MW]
plant. Fluidized bed reactor with external heat by molten salts as heat transfer/ heat storage fluid is
adopted (HMgCl2-KCl). A heater powered by hydrogen is used to balance the thermal storage tanks and
solar irradiation transients.
A membrane reformer reactor can be used to free the reaction of SMR and WGS from the limitation of
chemical equilibrium by extracting hydrogen, that is separated thank to a membrane made of palladium.
The temperature drops from conventional 700-800 [°C] to 500-550 [°C]. Up to 90% CO2 concentration
in the off gas enables capture by direct liquefaction. Fuel cell quality hydrogen is produced, with a purity
of 99.99%, avoiding the use of a PSA. Of course, the permeation side should be at lower pressure to drive
the process, meaning that hydrogen needs to be re-pressurized.
51
 PROCESSES FOR HYDROGEN

HYDROGEN SAFETY AND FLAME THEORY

Fire is a process that releases thermal energy as a result of a chemical reaction between a fuel and an
oxidant. The fuel can be gaseous, liquid, or solid. Regardless, the flame is a chemical phenomenon which
involves gaseous species. So, to start a fire, three elements must be simultaneously present; a fuel, an
oxidant, and a source of heat which provides ignition. Consequently, to avoid or extinguish a fire, one
needs to: avoid the presence of the fuel, avoid the presence of the oxidant, or remove the heat supply.
A flame is the visual manifestation of the region where combustion occurs. However, differently from
common fuels, a few fuels give rise to invisible flames which emit only in the UV region. Hydrogen is
one of these fuels, that when is burned in air, the flame is invisible in daylight, due to its low radiation
heat and to its high UV component.
Flammability range is the compositional range of the fuel/air mixture
wherein the fire keeps burning spontaneously, once combustion is ignited.
The molar fraction of the fuel which limit the flammability range are called
lower flammability limit (LFL, minimum % of the fuel in the mixture)
and upper flammability limit (UFL, maximum % of the fuel in the
mixture). The mixtures whose composition falls in these limits are
flammable. They are calculated at standard 25 [°C] and 1 [atm], to control
the behaviour of a mixture in environments that could be dangerous for
people, in case of leakages. These limits can be calculated according to
ignition theories. However, for practical scopes of safety, experimental
values are used, due to wide uncertainty associated with the estimates
provided by the models, and to experimental measurements that are
carried out under conditions close to the real ones. These experiments
measure also the minimum ignition energy (MIE), namely the smallest
amount of energy required to ignites a mixture. Upon increasing the
energy of the spark, the flammability range of the mixture grows until an
asymptotic value. The MIE corresponds to a created flame kernel of
critical size that is not extinguished by heat dissipation towards the cold
environment, and thus, to be conservative, the flammability limits are defined in the range of the value
of MIE that reaches the asymptote. We see that hydrogen is very sensitive to the ignition energy, reaching
an asymptote (with methane) at 10 [mJ], while its MIE is around 0.01 [mJ].
For hydrocarbons mixtures in air, at ambient pressure and temperature, given 𝐶 the % molar fraction:

• LFL is approximately half of the stoichiometric fraction 𝐶𝐿𝐹𝐿 ≅ 0.5𝐶𝑠𝑡𝑜𝑖𝑐 .

• UFL is almost double the stoichiometric fraction 𝐶𝑈𝐹𝐿 ≅ 2𝐶𝑠𝑡𝑜𝑖𝑐 .
When the mixture contains more than one fuel in air, they can be assessed with a weighted average,
according to the Le Chatelier’s law:
100
𝐿𝐹𝐿𝑚𝑖𝑥 =
𝐶
∑𝑖 𝑖
𝐿𝐹𝐿𝑖
100
𝑈𝐹𝐿𝑚𝑖𝑥 =
𝐶
∑𝑖 𝑖
𝑈𝐹𝐿𝑖
This law leads to large deviations when inert are present, and for the UFL of mixtures of fuels largely
different than each other. Also, flammability limits widely vary depending on the chemical nature of a
species, since they depend on kinetic and thermodynamic factors. Good correlations exist between the
LFL of common fuels and the adiabatic flame temperature. As an empirical rule, a mixture supports a

52
 PROCESSES FOR HYDROGEN

flame when the adiabatic flame temperature 𝑇𝐴𝐷 reaches 1200 [°C]. The existence of this correlation
highlights that ignition phenomena are mostly determined by thermal effects. Notable exceptions to this
criterion are given by hydrogen and ammonia. The former is very sensitive to variation of the
composition, while the latter is difficult to ignite.
The lower flammability limits keep almost constant passing from air to oxygen: the limiting reactant is
the fuel, and an excess of oxygen acts a thermal diluent, similarly to nitrogen in air. The UFL grows
significantly in pure oxygen, which now becomes the limiting reactant. The variations that occur when
passing from oxygen to air lead to a subsystem of the effect of adding an inert to the mixture. To visualize
the data associated with ternary mixtures which are obtained from this operation, ternary diagrams are
used. The pure species (H2, O2, N2) are located at the vertexes of the
triangle. The composition associated to the generic point M is obtained
by tracing segments parallel to the side directly facing the pure
component. However, to better visualize the effect of adding an inert in
a mixture, simplified plots are useful. The effect of the inert depends on
the nature of the inert itself. The different extensions of their
flammability limits are explained by considering that their specific heats
grow according to the following series:

𝐻𝑒 < 𝑁2 < 𝐻2 𝑂 < 𝐶𝑂2

Meaning that helium has the lowest specific heat, while CO2 the best.
Inert such as CCl4 and CH3Br have a strong influence due to their kinetic
effect as they favour the termination steps. When temperature increases,
the limits range become wider. This is mostly due to a thermal effect.
Indeed, the amount of reactant which leads to a certain adiabatic
temperature decrease. At the lower flammability limit, the mixture is
predominantly composed by air, and therefore the adiabatic temperature is almost constant and a linear
relation between the LFL and the LHV is established (if 𝑇𝑖𝑛 = 25 [°𝐶]):
𝐿𝐹𝐿𝐻2 (𝑇)
= 1 − 0.00157(𝑇 − 273)
𝐿𝐹𝐿(𝐻2 ) (25°𝐶)

A similar situation holds for the upper flammability limit. The

mixture is mostly composed of hydrogen, and a linear
relationship can be established. However, the chemical
mechanism is more complex due to the occurrence of parasitic
side reactions:
𝑈𝐹𝐿𝐻2 (𝑇)
= 1 + 0.00044(𝑇 − 273)
𝑈𝐹𝐿𝐻2 (25°𝐶)

The flash point is the lowest temperature at which a liquid fuel

originates vapours that form a flammable mixture in air, that in the graph above is shown as 𝑇𝐿 , while
the upper flash point is 𝑇𝑈 . It can be assessed as:
𝑠𝑎𝑡
𝑃𝑦𝐿𝐹𝐿 = 𝑃𝑓𝑢𝑒𝑙 (𝑇𝐿 )

It is experimentally determined with standard measurements. It correlates well with volatility and LFL
of the fuel examined. For safety considerations, it must be noted that liquids, if sprayed, can give rise to
flammable mixtures also below the flash point. The autoignition temperature is the temperature 𝑇𝑖𝑔 at
which a fuel spontaneously burns in air, without any external ignition. The AIT is scarcely useful in
safety. However, it is primarily important in engine design and in controlled combustion. It depends on
kinetic parameters, such as mixture composition, intrinsic chemical nature of the fuel, and pressure (it

53
 PROCESSES FOR HYDROGEN

decreases with pressure). Usually, when storing a fuel, for safety reasons it must be below the flash point,
to avoid any possible rise of flammable mixtures:

• A: at the surface we are below the flash point, and the concentration in proximity of the surface
is below the LFL, decreasing by stepping away from the surface. The liquid doesn’t start a fire
after an ignition.
• B: the surface is above the flash point and below the upper flash point, and the vapours in
proximity of the surface fall in the flammability range. The liquid is set on fire after an ignition
close to the surface.
• C: the surface is above the upper flash point, and the concentration at the surface is above the
UFL, but by decreasing away from the surface it falls in the flammability range. The liquid is not
set on fire after ignition close to the surface, but it starts a fire if the ignition is far from the surface.

The effect of the pressure on the LFL is almost negligible. In contrast, the UFL can grow significantly
with pressure (especially for light paraffins), because almost all the homogeneous combustions are very
promoted by pressure. The impact of the modification, and the extent of the growth, strongly depend on
the nature of the chemical species and on the kinetics of combustion. Hydrogen is an exception to this
behaviour. Indeed, its behaviour from a radical point of view is peculiar as it can generates very active
species, and the effect of pressure is totally kinetic in this framework.
The flames can be of various types: premixed, when fuel and oxidant are mixed prior to combustion;
diffusive, when fuel and oxidant are fed separately, and mixing occurs with a diffusive process towards
the flame front and occurs simultaneously to combustion; laminar/turbulent, depending on the fluid
dynamics regime that prevails during combustion, where mass and heat transport phenomena are
important; deflagrating/detonating, depending on the flame propagation mechanism, either thermal or
with shock waves.
The flame front that propagates within the flammable mixture is called combustion wave, and it can
generate a pressure wave with a mechanism analogous to that of an accelerating piston. For instance, in
a tube with two open sides, the combusted gases
(which are at a higher T and lower density than
fresh gases) expand and are discharged from the
opening close to the ignition point. The flame
propagates in the fresh stagnant gases. In tubes
with one closed side, where the ignition point is
located, the combusted expanding gases cannot
be discharged and therefore they move towards
the fresh gases, which start moving. Here, the
burning velocity is defined as a relative motion
with respect to an observer that is moving jointly with the gas mixture:

54
 PROCESSES FOR HYDROGEN

• 𝑢1∗ is the absolute velocity of the fresh unburned gases.

• 𝑠𝐿 is the relative velocity of the flame front.
• 𝑢𝐹∗ is the absolute velocity of the flame front.
It results in:
𝑠⃗𝐿 = 𝑢
⃗⃗𝐹 + 𝑢
⃗⃗1
Then, if 𝑢1∗ = 0 the fresh gases are stagnant, and the flame cannot be stabilized. If 𝑢𝐹∗ = 0, then the flame
is stationary, while if 𝑢𝐹∗ < 0 then there is a flashback.

The Mallard-Le Chatelier model assumes that the flame propagation occurs through heat exchange, and
it is stabilized when the heat diffusion from the flame front is sufficiently high to heat the fresh unburned
gases up to the autoignition temperature 𝑇𝑖𝑔 . Here, a flame shall be homogeneous, premixed, laminar,
deflagrant (the pressure gap is small, so it is isobaric), adiabatic, and stationary. In the graph below, a
typical laminar flame profile with this method is applied, where 𝑇𝐴𝐷 is the adiabatic flame temperature,
𝑇1 is the initial temperature of the fresh gases, and 𝛿 is the length of the reaction zone. The conductive
heat flux, that exchanges heat from the combusted gases to the non-combusted, specific of the surface in
the 0+ position where the flame front is equal to:
𝑑𝑇 𝑇𝐴𝐷 − 𝑇𝑖𝑔
𝜆( ) ≅
𝑑𝑥 0+ 𝛿
Where 𝜆 is the thermal conductivity of the reacted gases, and the derivative is linearly approximated. The
conductive heat equals the convective heat required to heat the reactants from 𝑇1 to 𝑇𝑖𝑔 :
𝑇𝐴𝐷 − 𝑇𝑖𝑔
= 𝜌𝑠𝐿 𝑐̂𝑝 (𝑇𝑖𝑔 − 𝑇1 )
𝛿
And so:
𝜆 𝑇𝐴𝐷 − 𝑇𝑖𝑔 1
𝑠𝐿 =
𝜌𝑐̂𝑝 𝑇𝑖𝑔 − 𝑇1 𝛿

The contact time of the gases within the flame

front is equal to the time of the reaction 𝑡𝑅 :
−1
𝑟 −1 𝑘𝐶 𝑛
𝛿 = 𝑠𝐿 𝑡𝑅 → 𝑡𝑅 = ( ) = ( )
𝐶 𝐶
𝐸𝑎𝑐𝑡 𝑛−1 −1
= (𝐴𝑒𝑥𝑝 (− )𝐶 )
𝑅𝑇
Where 𝑛 is the overall order of reaction. From
this relation, the burning velocity 𝑠𝐿 is derived
according to the Mallard-Le Chatelier model:

55
 PROCESSES FOR HYDROGEN

𝜆 𝑇𝐴𝐷 − 𝑇𝑖𝑔 𝐸𝑎𝑐𝑡 𝑛−1

𝑠𝐿 = √ 𝐴𝑒𝑥𝑝 (− )𝐶
𝜌𝑐̂𝑝 𝑇𝑖𝑔 − 𝑇1 𝑅𝑇

So, the burning velocity is nothing but the velocity with which the flame propagates within the flammable
mixture. It is a relative velocity, which is defined as the velocity of the flame front with respect to the
non-combusted mixture. Upon decreasing the amount of inert, the adiabatic flame temperature grows,
then the burning velocity increases if the quantity of inert decreases. The pressure influences the burning
velocity depending on the order of the reaction: if the order 𝑛 < 2 (like hydrogen), the burning velocity
decreases if the pressure increases; if 𝑛 = 2, it keeps constant; if 𝑛 > 2 increases if pressure increases.
Indeed, the dependence on pressure, that appears in the density and in the fuel concentration, is:
𝑛−2
𝑠𝐿 ~𝑃 2

Regarding the temperature, in the relation the part that has the strongest influence is without any doubt
the exponential part, and thus:
1
𝑠𝐿 ~ exp (− )
2𝑇
But its behaviour with temperature can be divided into two parts, as shown in the plots below.

The left part, approximated with a parabola, is the deflagrating part, while the right part is the detonating
part. Regarding the composition, the adiabatic flame temperature has a maximum that corresponds to
the stoichiometric amount of fuel, because there is the lowest amount of inert. In the lean combustion
part, the heat released is absorbed by the air in excess, while in the rich combustion there are also cracking
reactions occurring, that being endothermic absorb some heat. However, even if the adiabatic
temperature has the maximum in stoichiometric conditions, the burning velocity has the maximum in
slightly richer conditions, because more 𝐻 radicals are formed in those conditions, favouring the
combustion under a kinetic effect. Moreover, the flame velocity increases dramatically if detonation
occurs, passing from [cm/s] to [km/s].
As a tule of thumb, hydrocarbons have always a burning velocity around 0.5 [m/s], while hydrogen has
the highest one, around 3 [m/s]. Indeed, hydrogen is the smallest element and as such it can diffuse a lot
and at very high velocities.
A very used model to understand the difference between the deflagration and detonation, is the Rayleigh
model. We first account for continuity equation and momentum balance, where 2 refers to the combusted
part and 1 to the non-combusted part:
𝜌1 𝑣1 = 𝜌2 𝑣2
56
 PROCESSES FOR HYDROGEN

1 1
𝑃1 + 𝜌1 𝑣12 = 𝑃2 + 𝜌2 𝑣22
2 2
That can be rearranged as:
11 11
𝑃1 + (𝜌1 𝑣1 )2 = 𝑃2 + (𝜌 𝑣 )2
2 𝜌1 2 𝜌2 2 2
11 11
𝑃2 − 𝑃1 = (𝜌1 𝑣1 )2 − (𝜌 𝑣 )2
2 𝜌1 2 𝜌2 2 2
1 1 1 1 𝜌2 − 𝜌1
𝑃2 − 𝑃1 = (𝜌1 𝑣1 )2 ( − ) = (𝜌1 𝑣1 )2
2 𝜌1 𝜌2 2 𝜌1 𝜌2
𝑃2 − 𝑃1 1 (𝜌1 𝑣1 )2
= >0
𝜌2 − 𝜌1 2 𝜌1 𝜌2
This is the first Rayleigh equation. Bringing the velocity out of the brackets:
1 1
𝑃1 + 𝑣1 (𝜌1 𝑣1 ) = 𝑃2 + 𝑣2 (𝜌1 𝑣1 )
2 2
1
𝑃2 − 𝑃1 = 𝜌1 𝑣1 (𝑣1 − 𝑣2 )
2
𝑃2 − 𝑃1 1
= 𝜌 𝑣 >0
𝑣1 − 𝑣2 2 1 1
This is the second Rayleigh equation. In a deflagration, there is a very little pressure drop between the
pressure of the combusted gases and non-combusted ones, such that:

𝑃2 < 𝑃1
But if so, then:
𝑣2 > 𝑣1
𝜌2 < 𝜌1
Meaning that the flame front is accelerating, the combusted gases are trying to escape, and since the
density of the combusted gases is lower it’s a thermal concept. In the detonation, instead, the combusted
gases have much higher pressure:

𝑃2 ≫ 𝑃1
𝑣2 < 𝑣1
𝜌2 > 𝜌1
So, here there is a kind of deacceleration of the combusted gases, and the fresh gases are compressing the
combusted gases. Since the density of the fresh gases is lower, it’s a mechanical concept.
In a deflagration, the flame front propagates according to a thermal mechanism: the combustion heat is
transferred from the hot combusted gases to the fresh unreacted gases, igniting the combustion reaction.
The flame front moves with velocities which can reach a few meters per second at most. These velocities
are much smaller than the speed of sound (velocity of pressure waves in a fluid). In a deflagration, the
combustion wave generates the pressure wave: these two waves are separated, and the velocity of the
combustion wave is subsonic. The pressure gap is very small, almost negligible (isobaric process).

57
 PROCESSES FOR HYDROGEN

In a detonation, the energy is released with a much larger energy than in a deflagration. Indeed, the
combustion wave and the pressure wave are coupled and overlapped, and lead to the formation of an
explosive wave which propagates in the fluid with supersonic speed ([km/s]) compared to the conditions
of the unreacted gases ahead of the flame front. The flame front
propagates because of the ehaitng induced by the adiabatic
compression, which is due to the movement of the pressure
wave, and which, in turn, leads to the ignition of the fresh
gases. The explosive wave self-supports, thanks to the energy
released by the combustion reaction. Therefore, in a
detonation, the pressure wave generates the combustion wave.
Here, the pressure gap values across the pressure wave are high
(few tens of bar). Even though the detonation is not possible for
all the flammable mixtures, it must be avoided in any case.
However, a detonation can occur either abruptly due to a shock
wave or progressively starting from a deflagration. In the latter
case, the size of the vessel wherein the transition occurs
becomes crucial, because the walls can act as a third body. In
this framework, the ratio of critical turbulent to laminar
burning velocity is important. For instance, this ratio is the
lowest for hydrogen, meaning that it is the most prone among
fuels to explode, and the most sensitive. The higher the combustion speed, the higher the risk of reaching
the detonation point in closed rooms or pipelines. However, if the shock strength which makes the
transition to occur is calculated, hydrogen has the highest strength. Sensitivity to turbulence is also a
major issue.

PHYSICAL METHODS FOR HYDROGEN STORAGE

Physical-based storage technologies are state of the art as their TRL is around 8, being techniques still
under research. The main parameters for the choice of H2 storage and transport devices are:

• Gravimetric density, that refers to the gravimetric energy density of a fuel on a mass basis, based
on its LHV.
• Volumetric density, that is the LHV of a fuel expressed on a volumetric basis.
• Specific energy storage density, that refers to the number of service hours available in a vessel,
given in [kWh/kW].
• Range for mobile applications, expressed in [km] or in [km/kg H2].

58
 PROCESSES FOR HYDROGEN

The ways of storing hydrogen are essentially three:

• Cryogenic storage comprises both liquid hydrogen storage, located in large volume tanks for
annual storage or small volume tanks for special uses, and storage in active carbons at very low
temperature, with a large storage volume.
• Metal hydride storage, with either large or smaller volumes. Hydrogen can create metal
hydride through exothermic reactions and be released upon heating.
• Gas pressure storage, with either very large volumes at high pressure or underground storage.

The plot above shows the economy of the different storage methods. It’s clear that metal hydrides still
have some drawback in terms of cost-effectiveness as the economy of scale for them is almost null, being
expansive and with relatively small storage volumes. Then, there are high-pressure vessels and cryogenic
storage, while underground storage has the highest economy of scale as it is even out of the plot.

59
 PROCESSES FOR HYDROGEN

The plot above is the pressure-volume diagram for hydrogen, to understand which the best operating
conditions are to store it. One peculiar characteristic of hydrogen is the very small liquid phase are
compared to the others. Therefore, there are not many ways to operate to store hydrogen in the liquid
form, but the most popular one is at 1 [atm] and at 20 [K], so almost at the boiling point. Other major
problems are the extremely low density of hydrogen, around 70.9 [kg/m3] at 1 [atm], and its low critical
temperature, around 33 [K].

Compressed Gaseous Hydrogen Storage

These are one of the most common ways to store hydrogen, being above-ground conventional systems.
There are different types of vessels:

• Gas tanks (steel cylinders, bottles) up to 70 [L] at 200 [bar].

• Low pressure containers (spherical), above 15000 [m3] up to 16 [bar] or 30000 [m3] up to 11
[bar].
• High pressure containers up to 20 [m3] at 200 [bar].
• 35 [MPa] standard for automotive cylinders applications.
• 70 [MPa] standard for fuel-cell powered vehicles.
Usually, the pressure containers are spherical for two main reasons: to minimize the surface to volume
ratio and thus the heat transfer from the environment to the gas, and to minimize the thickness to be
applied to the tank.
The tank requirements differ from the application, depending on if it is stationary (type 1 or 2) or mobile
(type 3 or 4). The bursting pressure shall be at least 157.5 [MPa] for tanks used in mobile applications,
and the standard about no burst says that if damaged, the container can leak hydrogen, but it should
never burst, namely leak before break. The thermally controlled valve is necessary because hydrogen
must leak out in case of fire to avoid mechanically weaking the cylinder. The refilling standard says that
for tanks operated at 70 [MPa], complete refilling must take less than 3 minutes. Eventually, the weight
standard states that 9 [kg] tanks are state of the art for mobile applications.
One of the most critical parts of the storage regards the mechanical compression. The ideal isentropic
compression work is:
𝛾−1
𝛾 𝑃𝑜𝑢𝑡 𝛾
𝑊 = 𝑅𝑔𝑎𝑠 𝑇𝑖𝑛 [( ) − 1]
𝛾 − 1 𝑃𝑖𝑛

Where:
𝑐̃𝑝
𝛾=
𝑐̃𝑣
While:
𝛾−1
𝑃𝑜𝑢𝑡 𝛾
𝑇𝑜𝑢𝑡 = 𝑇𝑖𝑛 ( )
𝑃𝑖𝑛
The ideal isothermal work, instead is:
𝑃𝑜𝑢𝑡
𝑊 = 𝑅𝑔𝑎𝑠 𝑇𝑖𝑛 ln ( )
𝑃𝑖𝑛
Practical and real work includes non-ideal gas behaviour, friction losses, parasitic losses, and the
efficiency drops to 50%. The compressors used are several:

60
 PROCESSES FOR HYDROGEN

• Radial turbo compressors: standard only for extremely large amounts of hydrogen (>50000
[Nm3], ammonia synthesis).
• Oil-lubricated piston compressors: used for the compression of hydrogen for hydrogenation of
oil products.
• Water-lubricated piston compressors: for applications with small quantities of hydrogen and no
tolerance to oil.
• Dry-running piston compressors: standard for high-purity hydrogen, the pistons are sealed with
specially guided PTFE piston rings. Leaked hydrogen is fed back to the suction side via an active
carbon filter.
• Diaphragm compressor: standard for small high-purity hydrogen.
Below are shown the vessel types for gaseous hydrogen storage.

Type III and IV have a linear overwrapped by carbon-fibre composite material to provide strength. The
use of carbon-fibre composites results in significantly lower weight than all metal pressure vessels would
have. They differ in the kind of linear and weight (type III is light and has an aluminium liner, type IV is
superlight and has a polymeric liner) and in the composite with carbon fibres. Polymers liners are made
of high-density polyethylene or polyamide.

Hydrogen Liquefaction
At 25 [°C], normal H2 (n-H2) consists of 75% ortho H2, (o-H2, parallel spin), and 25% para H2 (p-H2,
antiparallel spin). The conversion from ortho to para is exothermic and spontaneous, but kinetically
𝑘𝐽
extremely low (day scale), because the temperature is very low. Below 70 [K], the Δ𝐻𝑅0 = −1.4 [𝑚𝑜𝑙],
𝑘𝐽
which exceeds the Δ𝐻 𝑣𝑎𝑝 = 0.9 [𝑚𝑜𝑙], namely the p-H2 evaporates, and thus the transition from ortho to
para must be performed before the liquefaction. 𝐹𝑒(𝑂𝐻)3 , 𝐹𝑒2 𝑂3, 𝑁𝑖(𝑂𝐻)2, 𝐶𝑟2 𝑂3 , gel, and activate
charcoal are effective catalysts, as well as magnetic fields.
The decompression in a throttle valve leads to a decrease in temperature if the Joule-Thomson 𝜇
coefficient is positive. The temperature at which a negative 𝜇 coefficient (heating by expansion) changes
to positive is the inversion point:

𝜕𝑇 1 𝜕𝑉̃ 𝑉̃
𝜇(𝑇, 𝑃) = ( ) = [𝑇 ( ) − 𝑉̃ ] = (𝑇𝛼 − 1)
𝜕𝑃 𝐻 𝑐̃𝑝 𝜕𝑇 𝑃 𝑐̃𝑝

61
 PROCESSES FOR HYDROGEN

Where 𝛼 is the thermal expansion coefficient, 𝑉̃ the molar volume, and 𝑐̃𝑝 the molar heat capacity.
However, hydrogen has a peculiar behaviour regarding this
coefficient, such that it becomes negative at a certain
temperature.
A small-scale liquefaction plant use throttle expansion, while
a large scale adds also a turbine expansion. The liquefaction
of H2 implies the complete conversion of o-H2 in p-H2, but if
the o/p conversion rate is large compared to the retention
time, major evaporation losses occur. All impurities, except
for helium, become solid and can block the refrigeration unit.
While an
electrolysis
hydrogen (>99.99%) undergoes catalytic conversion of
oxygen and adsorptive drying, a PSA hydrogen
(>99.5%) undergoes TSA adsorption. The theoretical
minimum demand for hydrogen liquefaction is 3.92
[kWh/kg], but a real plant requires 10-13 [kWh/kg],
which corresponds to a 30-40% efficiency.
The cold N2 vapours and the cold H2 flash gases are used
to cool the gaseous H2 supplied to the flash unit. liquid
N2 bath (77 [K]) is used with integrated heat exchanger.
Compression from 1-2 [bar] to 80-100 [bar] is applied,
the flash pressure is 1.2-1.4 [bar] and a cooling to 40-45 [K] prior the flash is required. The o/p conversion
reactor is included at different temperature levels or in the heat exchangers. The flash doesn’t liquefy
much hydrogen to allow the gaseous stream to be enough to cool down the feed.

The low-temperature Brayton cycle has typically two expansion stages driven by a single gas compressor.
The working fluid (N2, He or H2) is compressed up to 80 [bar], cooled to room temperature, and expanded
to 20 [K]. Compression from 1-2 [bar] to 25-40 [bar] is applied, as well as a flash pressure of 3-4 [bar] and
62
 PROCESSES FOR HYDROGEN

a cooling to 22-30 [K] prior to it. Helium Brayton cycles are preferred for small capacitance, while H2
cycle for large capacitance.

The boil-off occurs when the temperature rises in the vessel and H2 boils. If unchecked, the pressure
build-up pushes the vessel beyond its design limit. It depends on thermal insulation, tank size and shape,
and o/p ratio. The heat insulation between inner and outlet walls of the vessel consists of:

• Perlite (silica, alumina) for large containers. Typical losses are less than 0.1% per day.
• Superinsulation is usual for small size containers. It consists of several insulated layers wrapped
around the inner vessel. Aluminium foil (reflecting radiation) separated by glass fibres mat, or
layers of silver or gold coated plastic foil.
• Foam is used to insulate the space between the two walls of the double-walled tank, wherein
vacuum is applied.
• Vacuum is applied between the vessel walls to minimize the heat losses by conductivity.
The refueling standards refer to common procedure to be applied when hydrogen shall be fed into a tank
or a car, where 70% of the cases are represented by gaseous hydrogen. The tank filling time should be
less than 3 minutes, that is the standard for liquid fuel refueling. No level indicators can be used for
gaseous hydrogen, because the density is calculated through temperature and pressure. Filling requires
compression, so that the temperature in the tank increases, and a thermal compensation is necessary, as
well as the amount of fuel injected is lower because of the higher temperature. The standard filling
requires a temperature of 15 [°C] at 70 [MPa] and 10% underfilling. So, there is an upper temperature
limit of vehicle GH2 tank, that is 85 [°C], while the lower is -40 [°C]. This is why room temperature
hydrogen cannot be fed, otherwise it might overcome the upper limit. Below, some alternatives for the
refueling of a hydrogen vehicle are shown. System (a) is the simplest one, with a single stack, composed
of a GH2 tank, a compressor, a little can where hydrogen is fed and cooled down (the cooling is not
applied to the big tank). The system (b) is a multiple-stack system, composed of low-pressure, medium-
pressure, and high-pressure stack, to have time to cooled down and avoid the inversion of the Joule-

63
 PROCESSES FOR HYDROGEN

Thomson effect. The last system (c) takes into account also evaporation losses from a liquid H2 tank, it
makes use of a dry pump, where hydrogen is also recycled, and it eventually evaporates to be fed to the
vehicle.

The cryo-compressed hydrogen storage consists of storing H2 in a pressure vessel that can operate at
cryogenic temperatures (as low as 20 [K]) and high pressures (350 [atm]). This vessel can be fueled with
LH2, compressed GH2, or cryogenic H2 at elevated supercritical pressures, namely cryo-compressed H2.
By operating in a wide range of conditions, the concept
is to improve flexibility and merge the best features of
LH2 and compressed GH2 storage. The vessel plays the
most important role in this storage system as it can act as
heat sinks for the stack in FCEV. Indeed, CcH2 brings
higher volumetric and gravimetric storage capacity, fast
refueling, high dormancy (period of inactivity before the
tank releases H2), small sensitivity to heat transfer, better
control of T and P, and higher safety. The tank for this
configuration will be a merge of the GH2 and LH2 tanks.
It’s a type III tank with a carbon fibre reinforced
aluminium internal vessel in the inside (a), a vacuum super-insulation fibre (b), and the external metal
vessel for the vacuum enclosure (c). The only difference from the LH2 tank is the presence of two heat
exchangers, one external (h) and one internal (d). At the top, there is a thermally optimized suspension
to minimize losses due to conduction (e), and the vehicle receptacle for refueling (j).
The underground storage foresees geological storage sites that are forced options in case of large storage
capacity (>5 [GWh]), for longer-term grid stability, and seasonal balancing with discharge times of
several days. They have long-lasting scientific and technological experience (natural gas), hydrogen allow
a faster filling or draining compared to natural gas, and it permeates faster, levelling the pressure
gradients. However, they have lower density and lower volumetric LHV thank natural gas, poorly known
subsurface behaviour of H2 (diffusion in rocks, interactions with the microbial fauna), and too small
compression system available. The major interactions of hydrogen are expected with brine and microbial
inventories. The former is salty water at saturation point or with high salt concentration. In underground
water aquifers, common NaCl is found, but also other salts, and brine is mainly found in aquifers and
salt caverns. Hydrogen solubility in brine is smaller than in non-salty water up to 20%, which is why salt
cavern and former salty aquifers are preferential, because hydrogen permeate less in them compared to
non-salty water. The solubility of hydrogen in water is the key for tubing corrosion. The latter are
microorganisms such as methanogenic archea that can activate microbial metabolism processes with H2,
CO2, or sulphate compounds to form corrosive H2S. Reduction of CO and CO2 can be an issue when H2
storage occurs in depleted oil/gas reservoirs or in converted former natural gas or town gas storage.
Variations of pH and redox conditions can also be relevant to the microbial metabolism. The most
important cavern storage for hydrogen are salt caverns, where salt is gastight and inert to reactions with
H2, and it’s not degraded. Their depth is 1000-2000 m, with geothermal gradients that reduce the capacity.
Usually, a blanket with a lower density than water or brine is injected, like N2 or oil. Saline aquifers are
systems where an impermeable cap rock overlies the reservoir rock and prevents H 2 from buoyancy

64
 PROCESSES FOR HYDROGEN

upward migration. The reservoir rocks are sandstones or are sandstones or clastic sediments. A certain
volume of porosity is used for gas storage, and the pores are filled with fresh water or brines.
Electrochemical hydrogen compression is a non-mechanical method for pressurizing hydrogen. PEM
and PCC electrolyte membranes are used in the devices. Here, isothermal conditions are achieved much
more easily, H2 separation is reached together with compression, and issues with lubricants oils and
mechanical sealing are avoided. However, gas cross-over and permeation in protonic membranes, as well
as high ohmic resistance and fragility of ceramic membranes are present. The reversible potential 𝐸𝑂𝐶𝑉
(open cell voltage) is:

𝑅𝑇 𝑃𝐻𝑜𝑢𝑡
2
𝐸𝑂𝐶𝑉 = ln ( 𝑖𝑛 )
2𝐹 𝑃𝐻2

At 𝐸𝑂𝐶𝑉 no compression is possible. The EHC cell needs the total resistance 𝑅𝐸𝐻𝐶 to be overcome:

𝐸𝐸𝐻𝐶 = 𝐸𝑂𝐶𝑉 + 𝑅𝐸𝐻𝐶 𝑖

The resistance is based on the electrolyte membrane area 𝐴𝑚 . The specific work 𝑊𝐸𝐻𝐶 per unit of
compressed hydrogen moles is:

𝐸𝑂𝐶𝑉 𝑖 𝑅𝐸𝐻𝐶 𝑖 2
𝑊𝐸𝐻𝐶 = +
𝑛̇ 𝐻2 𝐴𝑚 𝑛̇ 𝐻2

The hydrogen embrittlement is a diffusion phenomenon caused by ionization, thermal dissociation, or

chemisorption and dissociation on active metal surfaces by hydrogen. The formation or dissolution of
atomic hydrogen in metals causes embrittlement, because it occupies interstitial sites, or it accumulates
in lattice vacancies and dislocations. Such defects are present especially if the material surface is activated
by plastic deformation. Hence, cyclic stress loading makes the material more prone to hydrogen attack.
If the concentration of H atoms overcomes a certain value, cracks form.
Decarburization, also known as high temperature hydrogen attacks, is typical of ferritic steel, it occurs
>200 [°C] and >3.25 [bar] of partial pressure. Increasing the number of deformation cycles and increasing
temperature rapidly promote the decarburization rate:

𝐹𝑒3 𝐶 + 4𝐻 → 3𝐹𝑒 + 𝐶𝐻4

Moreover, metal hydrides can be formed, especially with Ti, Mo, W, and Fe, and hydrogen
recombination causes blistering, flacking, and cracking. The hydrogen corrosion can be predicted by the
Nelson chart, that is the result of
experimental exposure to hydrogen
of several metals for 35 years. The
dashed lines represent surface
decarburization, the solid lines
represent hydrogen attack. At
constant temperature (<200 [°C])
and constant pressure up to 90
[MPa], no effects are found.
moreover, plastics are not subject to
H2 attack at ambient temperature,
and specific polymers are suitable at
higher temperature (PVC and PE up
to 60 [°C], PTFE up to 250 [°C]).
Copper and alloys resist up to 400 [°C], nickel and incoloy up to 540 [°C].
65
 PROCESSES FOR HYDROGEN

METAL HYDRIDES
Metal hydrides storage for hydrogen is a reversible chemical storage technique that can lead to significant
reduction of the storage volumes. The main advantages are safety, no spill of hydrogen and no explosive
or flame hazards can occur even after an accident, since hydrogen stays in the metal, and storage density
far above the compressed H2 and liquid H2. Hydrides formation is a batch dissociative adsorption process,
and it consists of a chemisorption followed by interstitial diffusion and solid solution:
𝑥
𝑀𝑒 + 𝐻2 ↔ 𝑀𝑒𝐻𝑥
2
Where the steps are similar to those of oxygen transport in SOEC:

𝐻2 + 2 ∗ ↔ 2𝐻 ∗
∗
𝐻𝑠𝑢𝑟𝑓𝑎𝑐𝑒 ↔ 𝐻𝑙𝑎𝑡𝑡𝑖𝑐𝑒 + ∗

𝐻𝑙𝑎𝑡𝑡𝑖𝑐𝑒 + 𝑀𝑒 ↔ 𝑀𝑒𝐻
Generally, three hydride categories are used:

• Ionic hydrides: salt-like hydrides (e.g., LiH and BaH2) are used as strong bases in organic
synthesis. They have too high decomposition temperature (>1000 [°C]), and dehydrogenation is
carried out with hydrolysis, and it is irreversible.
• Covalent hydrides: simple hydrides are stable if only highly polar (e.g., NH3, H2O, HF). Complex
covalent hydrides are alanates (Na[AlH4]), borohydrides (Li[BH4]), and amides (LiNH2).
• Interstitial metallic hydrides: hydrogen is dissolved as solid solution from the ppm range up to
20% fractions. H atoms are incorporated in the atomic form in the metal lattice structure and
occupy interstitial sites and vacancies. Ternary interstitial hydrides are AB2, AB3, AB, and A2B
compounds, where A is a lanthanide or alkaline (La, Ce, Mg, Y, Zr) and B is a transition element
(Ni, V, Fe).
All these types of hydrides are developed and studied to reduce the reaction enthalpy and the temperature
of the hydrogenation/dehydrogenation process. it’s worth it to look at the volumetric and gravimetric
densities for metal hydrides, compared to pressurized hydrogen storage. They both take into account the
tank volume and weight. For pressurized systems, there is a maximum for the volumetric density as by
increasing the pressure the thickness of the hydrogen tank gets higher to sustain the mechanical stress, as
well as the amount of material used to build the tank. Therefore, the hydrogen fraction decreases. For
liquid hydrogen storage, the volumetric density reaches an asymptote in correspondence of the covolume
𝐿
of hydrogen, namely the maximum density that can be achieved by compressing it (𝑏𝐻2 = 0.02661 [𝑚𝑜𝑙]
𝑘𝑔
and 𝜌𝐻2 ,𝑚𝑎𝑥 = 76.6 [𝑚3 ]). All the metal hydrides are above the usual pressurized tank for gaseous and
liquid hydrogen storage.

66
 PROCESSES FOR HYDROGEN

Thermodynamics and Kinetics

The 𝛼-phase is the solid solution, the 𝛽-phase is the metal hydride phase. The co-existence region is
characterized by a flat plateau. Van’t Hoff plots are derived for the desorption reaction as follows
(reference is made to ideal gas and 1 [atm]), assuming equilibrium conditions:
2𝑀𝑒𝐻 ↔ 2𝑀𝑒 + 𝐻2
By assuming ideal solid (activity equal to one), and ideal gas, the equilibrium conditions is stated by:

Δ𝐺𝑅0 = −𝑅𝑇𝑙𝑛(𝑃𝐻2 ) = Δ𝐻𝑅0 − 𝑇Δ𝑆𝑅0

Δ𝐻𝑅0 Δ𝑆𝑅0
𝑙𝑛(𝑃𝐻2 ) = − +
𝑅𝑇 𝑅
𝑃2 Δ𝐻𝑅0 1 1
ln ( ) = − ( − )
𝑃1 𝑅 𝑇2 𝑇1
In the bell-shape region, there is a mixture of the two phases, and the working area should be the 𝛽-phase.
Above the critical temperature, there is only one phase that is closer to the 𝛼-phase in term of
characteristics. The right-hand side plot describes the kinetics of the adsorption and desorption processes.
As it starts to be pressurized, the system switches from the 𝛼-phase to the 𝛽-phase, reaching a maximum
in the concentration of hydrogen in the metal. Then, desorption occurs at a different temperature and
pressure profile than adsorption (usually, at a higher temperature, even inf in the plot is the other way
around). The range between the maximum concentration and the final one after desorption is the
maximum capacity of the metal hydride. The reason below operating at different temperatures between
adsorption and desorption is related to the residual amount of hydrogen that remains in the hydride if
operating at the same temperature, that might be a problem.

67
 PROCESSES FOR HYDROGEN

The activation is the procedure required to use the hydride for the first time, remove the passivation
layer, and bring it to its maximum H-capacity and hydriding/dehydriding kinetics. Decrepitation is the
self-pulverization of large metal particles into powders, due to the brittle nature of the hydriding alloys.
It affects heat transfer properties and the tendency of migrating to undesirable parts of the equipment.
When the application is open-ended and uses new fresh H2 for each ads/des cycle, poisoning (loss of
capacity without a degradation of the kinetics), retardation (loss of kinetics but not of capacity), reaction
and corrosion can all occur. This is why the
resistance to gaseous impurities (e.g., S or
CO) is an important feature. Cyclic stability
involves disproportionation and formation of
pyrophoric compounds. Hydride powders tend
to burn spontaneously when suddenly exposed
to air due to an accidental tank rupture.
Toxicity may be an issue in case of ingestion or
inhalation of the powders. Usually, the tank
(right figure) is composed of several plates
inserted in a header tube, where the storage alloy is deposited. The hydrogen enters a filter prior to the
tank, and it forms the hydride once in contact with the alloy. Externally, there is a coolant (water) that is
cooling the environment. The drawback of the system is the filling time, 30 minutes.

Types of Hydrides
The interstitial hydrides of type AB5 have been discovered in1969. The A is La, Ca, or lanthanide mixture
mischmetal, while B side is Ni, Al, Mn, or Co. LaNi3 is the prototype and the best performing. They are
easy to activate as no passive layer is needed, and present a good tolerance to ppm impurities, but CO is
a poison. Ads/des kinetics is better than heat transport. In AB2 hydrides, A is Ti, while B is V, Cr, Mn,
or Fe. They have higher mass capacity than AB5 and they can be used below 25 [°C] for heat pumps and
refrigerators. AB hydrides are mainly represented by TiFe, with good average properties in between AB5
and AB2. Activation is complex, but they are economic, can achieve 2%wt of H2 in ambient conditions
and work well for stationary storage. A2B hydrides are based on Mg and Fe, Ni, Co, they have high H
content and form a non-metallic covalent bond. Mg2Fe is inexpensive and it’s a hydrogen stabilized
compound, but it requires high temperatures.
Complex hydrides (alanates, borohydrides, amides) are group I and II metal with complex anions of
[𝐴𝑙𝐻4 ]−, [𝐵𝐻4 ]−, and [𝑁𝐻2 ]− (respectively). They have slow kinetics, with high energy barriers, stepwise
68
 PROCESSES FOR HYDROGEN

mechanism and high temperatures required for desorbing H2 at 1 [atm]. The reversibility can be an issue,
with undesired decomposition products (due to stepwise mechanism). Alanates first decomposition step
occurs at 230 [°C] to give NaAlH6, while the final step gives Al and NaH at 250 [°C]. Mixing or doping
with Ti or Fe compounds enhances the dehydrogenation rates. Na[AlH4] is low cost, environmentally
friendly, but too unstable because of the generation of Al and NaH. Alanates can also react with water,
giving the hydroxide and hydrogen, and this form can be used in solution. Borohydrides are ionic
crystalline powders that release H2 via hydrolytic reaction with water and are stored in solution in excess
of water. They can also be used as boranes for H2 production via NH3 transportation.

Applications
An application of the hydride storage is the hydrogen compression, no mechanical parts are used as the
thermochemical compression is adopted, with compression ratios of 3-10 reached. Single or dual stage
systems are used, where in the latter case a step of adsorption (with cooling) and a step of desorption
(with heating) are performed. In the desorption step, since the partial pressure of hydrogen follows the
chemical equilibrium:

Δ𝐺𝑅0
𝑃𝐻2 = exp (− )
𝑅𝑇

It increases exponentially with the temperature. This is why the thermal conductivity of the metal hydride
plays a vital role. This is used when the efficiency is not critical, rather safety, reliability, and simplicity
of operation.
Another interesting application is for fuel cell powered vehicles. The only fuel cell that can be installed
in a vehicle is a PEMFC, that requires cooling as the heat is detrimental to the polymeric membrane. So,
as shown in the left-hand side picture, the PEMFC is coupled with a cycle in which an oil circulates,
needed both to heat up the hydride tanks to release hydrogen and to cool down the fuel cell. Then, it
passes in a pump and a filter, with a preheater for the start-up. This first option is generally used for big
vehicles, such as tracks. The right-hand side system is preferred for small systems, such as cars or scooters.
Even in this case there is the thermal coupling between the stack and the hydride tank: the tank is heated
up by electrical heat during start up, and by burning part of the hydrogen fed to the fuel cell during
operation. Indeed, in a fuel cell not all the hydrogen can be converted, otherwise it will reach zero voltage.

HYDROGEN PURIFICATION
Nowadays, the purification methods for hydrogen are several, differing from principles, feed gas, purity
and recovery achieved, and scale of use.

69
 PROCESSES FOR HYDROGEN

Membrane Gas Separation

The membrane gas separation for hydrogen is based on a selective barrier that allows the separation of
compounds based on molecular properties, such as molecular size, strength of adsorption, or solubility
in the membrane material. Membranes can be porous and non-porous, polymeric, ceramic, metal alloy,
and carbonaceous, but the most used purification types are cellulose acetate and polysulphone
membrane, palladium silver alloy membranes, and proton conductive ceramic oxide. The benefits that
this separation brings are low energy consumption, due to no phase transfer, mild conditions, low
pressure drops, no additional phase required, continuous and easy separation, and the absence of moving
parts, except for the recycle compressor in gas separation applications. However, it’s featured by fouling,
short lifetime, low selectivity, and no economic of scale.
As shown in the process scheme, the pressure difference is the driving force of the process, achieved with
compression and vacuum pump. The feed gas is compressed and cooled down below the dew point of
condensable species, that are removed and withdrawn. Droplets, particles, and dust must be removed by
a filter prior to the membrane. The separation is highly dependent on temperature as it affects diffusivity.
The outlets of the system are the retentate, namely the material retained on the high-pressure side of the
membrane, and the permeate, that is the material extracted through the membrane.

The membrane is a dense, non-porous, thin separation layer that is coupled with a thicker, porous support
layer. So, there are solution diffusion phenomena in the separation layer, while gas diffusion in the porous
layer. The thinnest polymeric layers achieved are 30-40 [nm], but 70-90 [nm] are the standard. The
pressure difference is between 1 and 20 [atm], so that a mechanical support is required to prevent the
membrane to collapse. The standard porosity of the
supporting layer is usually 50%, but weak points
and flaws must be avoided. Depending on the
application, the membrane can be symmetric, with
just a dense, porous film, asymmetric, with a dense
top layer and a porous substrate, or composite, with
a top selective skin, a porous support layer, and a
mechanical support layer.

The generalized Fick diffusion flow rate is:

𝐷𝑖 𝐶𝑖 𝑑𝜇̂ 𝑖
𝐽𝑖 = −
𝑅𝑇 𝑑𝑧
Where:

𝜇̂ 𝑖 = 𝜇𝑖0 (𝑇, 𝑃) + 𝑅𝑇𝑙𝑛𝑎̂𝑖

And so:
𝑑𝑙𝑛𝑎̂𝑖
𝐽𝑖 = −𝐷𝑖 𝐶𝑖
𝑑𝑧

70
 PROCESSES FOR HYDROGEN

𝑑𝑙𝑛𝑎̂𝑖 = 𝑑𝑙𝑛(𝐶𝑖 𝛾𝑖 ) = 𝑑𝑙𝑛𝐶𝑖 + 𝑑𝑙𝑛𝛾𝑖

Considering isothermal diffusion, being 𝛾𝑖 the activity coefficient of the species in the mixture:
𝑑(𝑙𝑛𝐶𝑖 ) 𝑑𝑙𝑛𝛾𝑖 𝑑𝐶𝑖
𝐽𝑖 = −𝐷𝑖 𝐶𝑖 (1 + ) = −𝐷𝑖 (1 + 𝐴)
𝑑𝑧 𝑑𝑙𝑛𝐶𝑖 𝑑𝑧
The thermodynamic factor 𝑨 accounts for the non-ideality of the solid mixture. The concentration 𝐶𝑖 in
the solid is found using the solubility coefficient 𝑆𝑖 in [kmol/m3/Pa], which is specific for the membrane
and for the permeant species:

𝐶𝑖 = 𝑆𝑖 𝑓̂𝑖𝑉

Considering thin separation layers, ideal gas behaviour, and ideal solid solution (𝐴 = 0), the diffusive
flow is obtained:
𝐷𝑖 𝑆𝑖 𝑉,𝑅 𝐷𝑖 𝑆𝑖 𝑅 𝑄𝑖
𝐽𝑖 = − (𝑓̂𝑖 − 𝑓̂𝑖𝑉,𝑃 ) = − (𝑃𝑖 − 𝑃𝑖𝑃 ) = − (𝑃𝑖𝑅 − 𝑃𝑖𝑃 )
𝑧 𝑧 𝑧
Where 𝑃𝑖𝑅 is the partial pressure of the species in the retentate, and 𝑃𝑖𝑃 in the permeate. The product
between solubility and diffusivity is called permeability coefficient 𝑸𝒊 , and it’s given in Barrer units
𝑚𝑜𝑙∙𝑚
(1 𝐵𝑎𝑟𝑟𝑒𝑟 = 3.35 ∙ 10−16 [𝑚2 ∙𝑠∙𝑃𝑎]). The membrane separation factor 𝜶𝒊𝒌 is defined as:

𝐷𝑖 𝑆𝑖 𝑄𝑖
𝛼𝑖𝑘 = =
𝐷𝑘 𝑆𝑘 𝑄𝑘
And it is a kind of selectivity. The hollow fibre modules are preferential for H2 purification, where the
feed enters the pressure vessel from the top, inside a fibre bundle. There, the permeate exits the unit from
the opposite part of the retentate. The flat sheet modules are preferred for CO2, and they are formed by a
spiral wound module over the permeate pipe with perforation. So, the permeate passes through the unit
in counter flow with respect to the feed and retentate.
The palladium alloy separation membranes allow for extremely high H2 purity (99.9999%) and much
higher flowrates of H2 purified with respect to polymeric membranes (3-4 orders of magnitude higher),
but they are featured by high pressure drops. The 23-25 wt% alloy with silver is required to avoid
distortions and rupture, and to increase H2 diffusion. The maximum H2 throughput and durability are
found at 350 [°C] (with 300 [°C] as minimum T, to avoid the formation of the hydride) and 20 [bar] H2
pressure. Unsaturated hydrocarbons, halogens, and sulphur compounds lead to the rupture of the
membrane. The surface roughness allows for dissociative adsorption of H2, and it acts as a catalyst.
Depending on the controlling step, the order 𝑛 varies, but commonly, 𝑛 = 0.5 indicates that dissociated
H diffuses in the Pd lattice bulk. Distortions and limit temperature are due to 𝛼 and 𝛽 phase transition.
𝐷𝑖 𝑆𝑖
𝐽𝑖 = − 0.5 (𝑃𝑅0.5 − 𝑃𝑃0.5 )
𝑧𝐾𝑒𝑞

Pressure Swing Adsorption

The pressure swing adsorption (PSA) is a gas purification process based on physical adsorption operated
at almost constant temperature and using partial pressure difference. The components to be removed are
adsorbed at high system pressure and desorbed at low system pressure. Adsorption occurs according to
the physical equilibrium isotherms of the various components, and the amount adsorbed ([mol/kg
adsorbent]) is called loading (𝑞𝑎𝑑𝑠 − 𝑞𝑑𝑒𝑠 ).
Commonly, different adsorbent materials compose the unit. a first layer of alumina is placed in the
bottom of the column to retain water, activated carbon removes almost all the contaminants from the

71
 PROCESSES FOR HYDROGEN

hydrogen, such as CO2, CH4, and CO. Then, zeolites decrease CO below 10 ppm, and possibly N2, O2,
and Ar, but it can’t receive either water or CO2.
The standard PSA cycle has four steps, and a minimum of two columns is required. During the
adsorption, all the contaminants are removed from the feed stream, and H2 is produced at the maximum
pressure. During the counter-current blowdown (depressurization or dump) and before the
breakthrough of the contaminants, the feed is interrupted and the column pressure is reduced, partially
removing the contaminants. After the initial
desorption of the contaminants during
blowdown, during the counter-current purge the
column is washed with recycle H2, reducing the
partial pressure of all the contaminants, and
promoting further desorption. The final step is
the pressurization, where the column is
reconditioned to start the next cycle. In PSA for
H2 purification, the pressurization is performed
counter-currently to the feed direction with
recycled H2. The extension to a larger number of columns (standard is 4) allows to minimize the amount
of H2 lost in the tail gas. Another strategy to improve H2 recovery is by sequentially reducing the column
pressure before the blowdown in a pressure equalization step, or introducing an additional step which is
called provide-purge step.
The temperature swing adsorption (TSA) is a gas purification process based on the adsorption operated
by using temperature variations (thermal regeneration). The impurities are adsorbed at low temperature
and desorbed by heating and/or depressurization. Two adsorbers switch alternatively, and cycle times
range from some hours up to several days. TSA is highly selective and reaches higher purity levels than
PSA.

HYDROGEN TRANSPORT
The most established technology to transport hydrogen is of course through the pipelines, with either
gaseous hydrogen that is mixed with NG in existing grid (possible now for 5-10% of H2), or repurposing
of the existing pipelines (easier in case of lean blending <10%), or building new H2 pipelines, with useful
local transportation, low operating costs, and long asset lifetime. Liquefied hydrogen is a relatively well-
established technique at small scale, but other technologies such as storing H2 in ammonia or in LOHC
are promising.

Hydrogen Pipelines
A pipeline network (20-30 [bar]) is preferential for the large-scale distribution of hydrogen, because they
require high levels of initial investment, which pay off only with large H2 volumes. One alternative is
local or regional networks (micro-networks) later combined into transregional networks. Another
possibility is to feed H2 into public natural gas networks (hydrogen enriched natural gas), like the H2-rich
town gas or coke-oven gas with H2 content >50 vol%. However, retrofitting of existing gas lines requires
replacement of fittings and sealings with polymeric unions and flanges, quality of the delivered H 2,
vicinity to high-voltage lines and electromagnetic fields.
The Wobbe index is used to establish the interchangeability of different gas mixtures. It’s based on the
volumetric thermal power output of a burner, calculated based on the HHV. Interchangeable gases have
a similar 𝑊𝑠 and burning velocities:
𝐻𝐻𝑉
𝑊𝑠 =
𝑑𝑅0.5

72
 PROCESSES FOR HYDROGEN

2Δ𝑃 0.5
𝑣 = 𝐾( )
𝜌
𝑄 = 𝐴𝑣𝐻𝐻𝑉
Where 𝑑𝑅 is the relative density of the gas mixture compared to air, 𝑣 the efflux velocity from the burner
nozzle, 𝐾 a certain structural constant, and 𝑄 the volumetric thermal power flux across a burner with
cross section 𝐴. In other words, 𝑊𝑠 sets the thermal power output of a burner nozzle, that should remain
constant even by changing the gas that is burned. The small quenching gap of H2 requires tighter
tolerances at the burner head sealing. The tighter flame speed creates a risk for flashback. A grid
conversion requires burner heads and seals to be replaced even if the current Wobbe indexes are suitable
for H2.
High voltage direct current (HVDC) transportation is not included because it is not efficient across very
long distances above 5000 [km] due to energy losses associated with cable resistances and high capital
costs. Additionally, HVDC cannot provide all the benefits of chemical fuels, including energy storage or
provision of high-grade heat.

Liquid Organic Hydrogen Carriers

LOHCs are organic molecules able to store hydrogen in reversible hydrogenation/dehydrogenation
cycles. They are liquid aromatic or heteroaromatic compounds, which are converted via catalytic
hydrogenation under H2 pressure into the corresponding alicyclic or heteroalicyclic compounds.
Hydrogen is released at the utilization site by dehydrogenation. They provide wide availability and low
price (LOHCs are common solvents) and easy distribution with the existing infrastructure and regulation
for mineral oil-based fuels. On the other hand, the hydrogen capacity is currently limited to 7.2% on mass
basis, and toxicological impact is not negligible. The main relevant features of a LOHC should be:

• Organic hydrogen gravimetric storage at least 5.5 wt% based on the hydrogenated form.
• H2 liquid storage in both hydrogenated and dehydrogenated stages as the two stages shall be
maintained as similar as possible, due to a network based mainly on pumping.
• Melting point of the LOHC lower than -30 [°C] to avoid formation of solids and addition of
external solvents.
• Boiling point as high as possible, so that the vapour pressure is as low as possible, to allow for an
easy isolation of pure H2.
• Viscosity as low as possible for easy heat transfer and transport.
• No by-products during H2 loading step.
• Thermal stability as high as possible to avoid formation of CO, CO2, or coke.
• The two steps shouldn’t be limited by thermodynamics, to allow for a once-through charge and
no recycle.
• Target reaction enthalpy for endothermic dehydrogenation should be 40 [kJ/mol H2].
Standard LOHC are toluene/MCH, naphthalene/decalin, benzene/cyclohexane, and DBT/PDBT. The
toluene hydrogenation is carried out at 200-300 [°C] and from 10 to 50 [bar]. Either liquid or gas phase
(preferential) toluene and MCH are reacted. The catalyst is composed of bimetallic noble metal-based
such as Pt-Sn or Pt-Mn supported on alumina. In the Chiyoda Spera process, it is carried out in a fixed
bed tubular reactor in gas phase, and product separation is achieved with vapour and liquid separation
drums. The feed is superheated and mixed with recycled H2. The heat generated by the reaction is
removed by cooling water, with production of steam. The dehydrogenation is performed at 350-400 [°C]
and 1-3 [bar]. Side products are dealkylated species, and the same catalyst is applied. The reactor is also
a fixed bed tubular type, and heating is provided with hot oil.
Alternative LOHCs are N-containing molecules, that have better biodegradability than aromatics and
lower enthalpy of dehydrogenation per mole of H2. Indole, carbazole, and their derivatives are favourite.
73
 PROCESSES FOR HYDROGEN

Liquid phase batch reactors are used because of the high melting point of these compounds, with noble
metal-based catalysts, such as Pt and Rh. They significantly reduce the environmental impact and
enthalpy of dehydrogenation, but the capacity of hydrogen on a mass basis is quite lower than the other
LOHCs.

Ammonia
Ammonia is a carbon-free and energy-dense possible liquid fuel, that can store hydrogen and release it
upon cracking. It can be synthesized with a carbon-free process from RES, and it can rely on a well-
established infrastructure worldwide. It is produced at high purity, and it presents one of the top
gravimetric (18wt%) and volumetric H2 densities. Moreover, its cracking is a simple way to produce
hydrogen on-site/on-board. On the other hand, to date its production is not green as it is produced mainly
by grey hydrogen, and its production shall be made carbon-free. The power generation from ammonia to
utility scale is complex and the economic viability for integrating green ammonia is an issue. Eventually,
it is toxic and flammable, and it is corrosive to copper, brass, and zinc alloys. Anhydrous ammonia is
delivered worldwide via pipelines, railroads, ships, in inexpensive steel tanks. Indeed, its boiling point at
atmospheric pressure is -33 [°C], and it can be easily condensed and transported in large quantities even
at that temperature.
Alkaline earth halides, such as MgCl2, CaCl2, SrCl2, act as adsorbents, and lead to solid storage of
ammonia. Amine complexes are formed, with up to 6-8 NH3 mol per mol of salt. Adsorption occurs at
low to medium temperatures and low pressures. High mass capacitance and molar concentration are
achieved. Solid state storage significantly downplays toxicity, dramatically lowering the vapour pressure
of ammonia. Thanks to the optimal balance between capacity, adsorption pressure, and enthalpy, MgCl2
is preferential.
Green ammonia production is still at very low TRL. The standard ammonia synthesis is performed with
the Haber-Bosch process, where H2 and N2 react at 150-250 [bar] and 400-450 [°C] over a Fe-based
catalyst. Due to the low equilibrium single-pass conversion (15%), a gas recycle is required. Ammonia is
removed by condensation, and the build-up of inert (methane and Ar) is purged and recycled to the
reformer furnace. Large compressors associated with the SMR process air, HB synthesis feed,
refrigeration cycle and synthesis loop recycle are driven by steam turbines that use the waste heat from
the reformer. Large energy-intensive conventional HB plants work with NG feedstocks, while small-
decentralized green ammonia plants in the vicinity of a consumer are the target of green ammonia. The
higher the H/C ratio in the fossil fuel, the more efficient the ammonia synthesis process, with lower
energy consumption, due to better quality of the raw syngas. Higher pressure reformers, better
methanation catalysts allow for saving purification energy and compression energy. Low energy concepts
include better heat integration with HP steam production and integration of membrane PSA in the recycle
loop.

An interesting way to produce green ammonia is the Haldor-Topsoe process, where the separation
between oxygen and nitrogen is made by using all the oxygen to burn additional hydrogen. Steam is
supplied with the recycle product of the system, to not have infinite voltage at the SOEC and to provide
74
 PROCESSES FOR HYDROGEN

enough hydrogen to burn. Before every cell, there is a H2/O2 burner to consume oxygen and produce
steam to be used in the SOEC. The steam produced in the combustion is used to produce hydrogen.
Then, the mixture of H2, N2, water and argon is cooled and the condensate is separated.

The catalytic ammonia cracking is preferable for decentralized and miniaturized distributed H2
generation. Ru-based catalysts are used. Ammonia slip is the essential factor as a 99.8% NH3 conversion
means 1000 ppm in the H2/N2 mixture. Heat supply, H2 pressure, and downstream ammonia slip
adsorbent introduction are issues. Portable microreactors have been proposed for automotive
applications, where heat transfer is the key feature. NH3 or NH3/H2/N2 recycle mixtures are used to
provide heat, but also propane is applied. The microreactor consists of an inner annular-type cylindrical
shield that supplies heat to the NH3 decomposition occurring in the outer annular micro-reformer. At the
optimum conditions defined by propane/air inlet velocity of 3.1 [m/s] and NH3 flowrate of 10
[Ncc/min], Ru-catalyzed ammonia decomposition produces hydrogen corresponding to a thermal power
equivalent of 5.4 [W] at 98% conversion.
Ammonia electrolysis is performed in aqueous phase in alkaline electrolysis cells at room temperature
and low voltage. The Nernstian voltage is 0.06 [V]. Slow electrode kinetics, and the need of Ir-Ru
electrodes to overcome high overpotentials are issues:

2𝑁𝐻3 + 6𝑂𝐻 − ↔ 𝑁2 + 𝐻2 𝑂 + 6𝑒 −
6𝐻2 𝑂 + 6𝑒 − ↔ 3𝐻2 + 6𝑂𝐻 −
Solid oxide fuel cells are favourite due to their high operating temperature, since NH 3 spontaneously
decomposes into nitrogen and hydrogen at the anode. With standard SOFCs, selectivity is an issue: NOx
formation occurs with YSZ-based cells. Indeed, cracking of ammonia gives a mixture of nitrogen and
hydrogen. In alternative, a proton conductive EC can be used, where:

2𝑁𝐻3 → 𝑁2 + 6𝐻 + + 6𝑒 −
6𝐻 + + 6𝑒 − → 3𝐻2
The drawback here is that complete conversion cannot be achieved.

75