Hydrogen Permeation through Proton-Conducting Ceramic Membranes: Transport, Selectivity, and Electrochemical Pumping
Hydrogen Permeation through Proton-Conducting Ceramic Membranes: Transport, Selectivity, and Electrochemical Pumping
Dense proton-conducting ceramic membranes can separate hydrogen from gas mixtures at elevated temperatures. Their selectivity arises from a charge-transfer sequence: hydrogen is dissociated at one surface, transported through the ceramic as protonic defects, and recombined as molecular hydrogen at the opposite surface. The electronic charge required to complete this process may be transported through the membrane itself or through an external electrical circuit.
These two operating modes must be distinguished. A mixed protonic–electronic conducting (MPEC) membrane can support spontaneous hydrogen permeation under a hydrogen chemical-potential gradient. A predominantly proton-conducting electrolyte can instead be used in an electrochemical hydrogen pump, where an external voltage drives proton transport and electrons travel through an external circuit.
This guide develops both mechanisms, derives useful flux relations, and discusses material selection, membrane architecture, experimental measurement, selectivity and practical limitations.
1. Two different ceramic hydrogen-separation modes
1.1 Spontaneous permeation through a mixed conductor
A dense MPEC membrane transports protons and electronic charge internally. On the feed side, hydrogen dissociates:
\[ \mathrm{H_2(g)\rightleftharpoons 2H^+ + 2e^-} \]
Protons move through the ceramic under the hydrogen chemical-potential gradient. The electronic carriers provide the complementary charge transport. At the permeate side, protons and electrons recombine:
\[ \mathrm{2H^+ + 2e^-\rightleftharpoons H_2(g)} \]
No external electrical power is intrinsically required to sustain this spontaneous permeation mode; the driving force is the difference in hydrogen chemical potential between the two sides. In practice, an electronic short circuit or a deliberately designed electronic pathway may be needed in some membrane architectures to reduce electronic resistance.
1.2 Electrochemical hydrogen pumping
An electrochemical hydrogen pump uses a proton-conducting electrolyte and electronically conducting electrodes. Hydrogen is oxidized at the feed electrode, protons cross the electrolyte, and electrons travel through an external circuit. At the permeate electrode, hydrogen is regenerated.
For the forward pumping direction:
\[ \mathrm{H_2\rightarrow 2H^+ + 2e^-} \]
\[ \mathrm{2H^+ + 2e^-\rightarrow H_2} \]
An applied electrical potential controls the pumping rate. The external circuit is not a minor detail: it is the electronic path that closes the electrochemical cell.
1.3 Key distinction
| Feature | MPEC membrane | Electrochemical pump |
|---|---|---|
| Proton transport | Through ceramic | Through ceramic electrolyte |
| Electronic transport | Through membrane or internal connected phases | Through electrodes and external circuit |
| Main driving force | Hydrogen chemical-potential gradient | Applied electrical potential |
| External power | Not intrinsically required for spontaneous permeation | Required for pumping against or beyond the equilibrium potential |
| Main material requirement | Mixed protonic and electronic transport | High proton conductivity and suitable electrodes |
A proton-conducting ceramic does not automatically behave as a spontaneous hydrogen-permeation membrane. If electronic transport is insufficient, sustained internal ambipolar permeation is suppressed.
2. Proton transport in oxide ceramics
In many perovskite oxides, protons are incorporated by hydration of oxygen vacancies. A simplified hydration reaction is
\[ \mathrm{H_2O(g)+V_O^{\bullet\bullet}+O_O^x \rightleftharpoons 2OH_O^{\bullet}} \]
Here, \(OH_O^{\bullet}\) represents a proton bound to a lattice oxygen ion. The proton is not generally a free interstitial proton; it is associated with oxygen and migrates through local rearrangement of O–H bonds and transfer between neighbouring oxygen sites.
The proton concentration depends on temperature, water-vapour partial pressure, oxygen partial pressure, vacancy concentration and acceptor-dopant chemistry. Hydration is often favoured at lower temperature and higher water activity, but the actual equilibrium depends on the material and gas environment.
2.1 Proton migration
A useful microscopic picture involves two coupled motions:
- Reorientation of the O–H bond around a lattice oxygen.
- Transfer of the proton to a neighbouring oxygen site.
The measured proton conductivity reflects carrier concentration and mobility. A material may have a high concentration of incorporated protons but modest conductivity if proton migration is slow.
2.2 Defect chemistry is atmosphere-dependent
Hydrogen separation tests are not always performed in dry hydrogen alone. Water vapour, oxygen, carbon dioxide and other gas components can change the defect populations and surface chemistry. Therefore, proton conductivity measured in humidified gas cannot automatically be used to predict performance in a dry feed.
3. Ambipolar transport and the MPEC membrane
For spontaneous hydrogen permeation, protonic and electronic charge transport are coupled. A Wagner-type expression uses the ambipolar conductivity:
\[ \sigma_{\mathrm{amb}}= \frac{\sigma_{\mathrm{H^+}}\sigma_{\mathrm{e}}} {\sigma_{\mathrm{H^+}}+\sigma_{\mathrm{e}}} \]
where \(\sigma_{\mathrm{H^+}}\) is proton conductivity and \(\sigma_{\mathrm{e}}\) is total electronic conductivity.
If electronic conductivity is much larger than proton conductivity,
\[ \sigma_{\mathrm{amb}}\approx\sigma_{\mathrm{H^+}} \]
If proton conductivity is much larger than electronic conductivity,
\[ \sigma_{\mathrm{amb}}\approx\sigma_{\mathrm{e}} \]
Thus, the lower of the two conductivities tends to control the ambipolar response. Increasing proton conductivity alone may have little effect if electronic transport remains the bottleneck.
4. Wagner-type equation for spontaneous hydrogen permeation
For a dense, homogeneous MPEC membrane under steady-state conditions, a common idealized expression for hydrogen flux is
\[ J_{\mathrm{H_2}}= \frac{RT}{4F^2L} \int_{\ln p_{\mathrm{H_2}}^{\mathrm{perm}}}^{\ln p_{\mathrm{H_2}}^{\mathrm{feed}}} \frac{\sigma_{\mathrm{H^+}}\sigma_{\mathrm{e}}} {\sigma_{\mathrm{H^+}}+\sigma_{\mathrm{e}}} \,d\ln p_{\mathrm{H_2}} \]
where:
- \(J_{\mathrm{H_2}}\) is hydrogen molar flux in \(\mathrm{mol\,m^{-2}\,s^{-1}}\);
- \(R\) is the gas constant;
- \(T\) is absolute temperature;
- \(F\) is the Faraday constant;
- \(L\) is membrane thickness;
- \(p_{\mathrm{H_2}}^{\mathrm{feed}}\) and \(p_{\mathrm{H_2}}^{\mathrm{perm}}\) are hydrogen partial pressures at the feed and permeate sides;
- \(\sigma_{\mathrm{H^+}}\) and \(\sigma_{\mathrm{e}}\) are local protonic and electronic conductivities.
The prefactor follows from the transfer of two electronic charges per hydrogen molecule. The expression is written for a positive feed-to-permeate flux when the feed-side hydrogen partial pressure is higher.
If the ambipolar conductivity is approximately constant across the membrane, the expression becomes
\[ J_{\mathrm{H_2}}\approx \frac{RT\sigma_{\mathrm{amb}}}{4F^2L} \ln\left( \frac{p_{\mathrm{H_2}}^{\mathrm{feed}}} {p_{\mathrm{H_2}}^{\mathrm{perm}}} \right) \]
This approximation is useful for screening and interpreting trends. It should not be used blindly if conductivity changes substantially with hydrogen partial pressure, temperature, water activity or local defect chemistry.
The equation describes ideal bulk ambipolar transport. It does not automatically account for finite surface reaction rates, gas-film resistance, concentration polarization, grain-boundary effects, cracks, seal leakage or chemical degradation.
5. Worked example: estimating an ideal MPEC flux
Consider an illustrative membrane with:
- \(T=873\ \mathrm{K}\);
- \(L=1.0\ \mathrm{mm}=1.0\times10^{-3}\ \mathrm{m}\);
- \(\sigma_{\mathrm{H^+}}=5\ \mathrm{S\,m^{-1}}\);
- \(\sigma_{\mathrm{e}}=50\ \mathrm{S\,m^{-1}}\);
- \(p_{\mathrm{H_2}}^{\mathrm{feed}}=0.50\ \mathrm{bar}\);
- \(p_{\mathrm{H_2}}^{\mathrm{perm}}=0.05\ \mathrm{bar}\).
First, calculate the ambipolar conductivity:
\[ \sigma_{\mathrm{amb}}= \frac{5\times50}{5+50} =4.545\ \mathrm{S\,m^{-1}} \]
Then apply the constant-conductivity approximation:
\[ J_{\mathrm{H_2}}\approx \frac{(8.314)(873)(4.545)} {4(96485)^2(1.0\times10^{-3})} \ln\left(\frac{0.50}{0.05}\right) \]
\[ \boxed{ J_{\mathrm{H_2}}\approx 2.08\times10^{-4}\ \mathrm{mol\,m^{-2}\,s^{-1}} } \]
This value is an ideal bulk-controlled estimate, not a measured membrane performance. Surface exchange, gas-side transport and microstructural resistance can lower the experimental flux.
6. Electrochemical hydrogen pumping: Nernst potential and Faraday’s law
For an electrochemical hydrogen pump, the equilibrium potential associated with a hydrogen partial-pressure ratio is described by the Nernst relation:
\[ E_{\mathrm{rev}}= \frac{RT}{2F} \ln\left( \frac{p_{\mathrm{H_2}}^{\mathrm{perm}}} {p_{\mathrm{H_2}}^{\mathrm{feed}}} \right) \]
The sign depends on the chosen cell and voltage convention. The reversible voltage magnitude is
\[ |E_{\mathrm{rev}}|= \frac{RT}{2F} \left| \ln\left( \frac{p_{\mathrm{H_2}}^{\mathrm{perm}}} {p_{\mathrm{H_2}}^{\mathrm{feed}}} \right) \right| \]
The actual applied voltage must also overcome ohmic losses and electrode overpotentials. Consequently, the cell voltage under load is generally greater in magnitude than the reversible value for the same pumping direction.
For an ideal Faradaic process, two electrons are transferred per hydrogen molecule. The hydrogen production or pumping rate is
\[ \dot n_{\mathrm{H_2}}=\frac{I}{2F} \]
and the area-normalized flux is
\[ J_{\mathrm{H_2}}=\frac{j}{2F} \]
where \(I\) is current and \(j=I/A\) is current density.
The Faradaic efficiency is
\[ \mathrm{FE}_{\mathrm{H_2}}= \frac{2F\dot n_{\mathrm{H_2,measured}}}{I} \times100\% \]
A low Faradaic efficiency can indicate gas leakage, parasitic electrode reactions, electronic leakage through the electrolyte, measurement errors or transient hydrogen storage.
7. Material families and selection criteria
Proton-conducting ceramics for hydrogen separation include acceptor-doped barium cerates, barium zirconates, barium zirconate–cerates and other perovskite-related oxides. Some compositions are primarily proton conductors and are therefore suited to electrochemical pumps. Others are modified or combined with an electronic conducting phase to create MPEC membranes.
A material-selection assessment should include:
- Proton conductivity under the intended temperature and gas composition;
- Electronic conductivity for spontaneous MPEC permeation;
- Chemical stability against \(\mathrm{H_2O}\), \(\mathrm{CO_2}\), hydrogen and impurities;
- Thermal expansion and mechanical strength;
- Sinterability and achievable gas tightness;
- Compatibility with electrode and support materials;
- Surface catalytic activity for hydrogen dissociation and recombination.
7.1 Single-phase mixed conductors
A single-phase MPEC material transports protons and electronic charge through the same crystallographic phase. This can simplify interfaces and reduce the need for a separate electronic network. However, obtaining sufficiently high protonic and electronic conductivities together with chemical stability is challenging.
7.2 Dual-phase membranes
A dual-phase membrane combines a proton-conducting phase with an electron-conducting phase. The two phases must form continuous pathways through the membrane thickness. A disconnected electronic phase cannot provide a low-resistance path, even if its intrinsic conductivity is high.
The percolation threshold, phase distribution, interfacial reactions and thermal expansion mismatch are therefore central design variables.
7.3 Metal–ceramic composites
A metal phase can provide electronic conductivity while a ceramic phase transports protons. These composites may improve electronic transport, but introduce concerns about metal coarsening, oxidation, chemical compatibility, thermal cycling and long-term connectivity.
8. Membrane thickness, surface exchange and rate limitation
For ideal bulk-controlled permeation,
\[ J_{\mathrm{H_2}}\propto\frac{1}{L} \]
Reducing membrane thickness lowers bulk transport resistance. However, hydrogen dissociation and recombination occur at the surfaces. When the membrane becomes sufficiently thin, surface exchange or gas-side transport can dominate, and further thinning gives diminishing returns.
A useful qualitative resistance picture is
\[ R_{\mathrm{overall}}\sim R_{\mathrm{surface,feed}}+ R_{\mathrm{bulk}}+ R_{\mathrm{surface,perm}}+ R_{\mathrm{gas}} \]
This is a conceptual series-resistance model; quantitative treatment requires a coupled transport and surface-kinetics model.
Thickness-series experiments can help identify the limiting regime. An approximately inverse-thickness dependence supports bulk control. A near-constant flux over a range of thicknesses suggests that surface exchange or another thickness-independent resistance may be important. Support layers and microstructural changes must also be considered.
9. Hydrogen selectivity and competing gases
A dense, defect-free proton-conducting ceramic transports hydrogen through protonic defects rather than by ordinary molecular diffusion through open pores. This provides the basis for high hydrogen selectivity.
However, practical selectivity can be compromised by:
- Cracks, pinholes and seal defects;
- Electronic or ionic leakage pathways that do not produce the intended separation;
- Chemical reactions with carbon-containing or sulfur-containing gases;
- Surface poisoning and catalyst deactivation;
- Unwanted transport of other ionic species in mixed-conducting materials.
A membrane that shows a high hydrogen flux is not necessarily highly selective. Feed and permeate compositions should be measured for all relevant gas species, and gas-tightness should be verified independently.
10. Fabrication and membrane architecture
10.1 Self-supported dense membranes
Self-supported disks or tubes are straightforward for laboratory evaluation. Their thickness is often limited by mechanical strength and the need to withstand pressure differences and thermal stress.
10.2 Supported thin membranes
A thin dense selective layer can be fabricated on a porous support to reduce bulk transport resistance while retaining mechanical strength. The support must be sufficiently porous and permeable, and must not react with the membrane during sintering or operation.
10.3 Processing considerations
A typical ceramic route includes powder synthesis, calcination, milling, forming, sintering and final machining. For dual-phase materials, control of phase connectivity and reaction between phases is especially important.
The processing schedule should be selected to achieve:
- Phase purity and controlled stoichiometry;
- High density in the selective layer;
- Appropriate grain size and grain-boundary chemistry;
- Continuous protonic and electronic pathways where required;
- Chemical and thermomechanical compatibility with supports and seals.
A dense membrane should be checked for cracks and pinholes. High apparent hydrogen flux caused by non-selective leakage must not be interpreted as intrinsic proton transport.
11. Experimental measurement of hydrogen permeation
A typical spontaneous permeation experiment supplies a hydrogen-containing gas to one side of the membrane and a sweep gas or lower-hydrogen-pressure atmosphere to the other side. The permeate composition and total flow are measured after reaching steady state.
The hydrogen flux is
\[ J_{\mathrm{H_2}}= \frac{\dot n_{\mathrm{H_2}}}{A} \]
where \(A\) is the active membrane area and \(\dot n_{\mathrm{H_2}}\) is the hydrogen molar flow rate through the membrane.
For a measured dry-gas mole fraction \(y_{\mathrm{H_2}}\) and known total dry outlet flow,
\[ \dot n_{\mathrm{H_2}}= y_{\mathrm{H_2}}\dot n_{\mathrm{total,dry}} \]
The total outlet flow must be determined consistently. It should not automatically be assumed equal to the inlet sweep flow because permeated hydrogen adds to the outlet stream.
For an electrochemical pump, measure current, voltage, active area and permeate hydrogen flow. Compare the measured hydrogen rate with the Faradaic prediction \(I/(2F)\) and calculate Faradaic efficiency.
12. Separating bulk transport, electrode kinetics and gas-side effects
A reliable experiment should include control measurements.
Gas tightness: Perform leak tests before and after the experiment. Inspect seals and the membrane for cracks.
Flow-rate dependence: Vary feed and sweep flow rates. A strong dependence may indicate gas-film limitation or concentration polarization.
Thickness dependence: Compare membranes with different thicknesses while keeping composition, microstructure and operating conditions as similar as possible.
Electrode dependence: For electrochemical pumps, vary electrode composition, loading or active area to assess electrode polarization.
Electrical response: Use impedance spectroscopy where appropriate to separate ohmic and electrode contributions. Interpret equivalent-circuit elements cautiously and support assignments with controlled experiments.
Atmosphere control: Record hydrogen, water vapour, oxygen and other relevant gas partial pressures. Proton concentration and surface chemistry can depend strongly on humidity.
Steady-state verification: Record the time required for permeate composition, current and voltage to stabilize.
13. Common failure mechanisms
- Insufficient electronic conductivity in an MPEC membrane: limits ambipolar hydrogen transport.
- Low proton conductivity: limits both spontaneous permeation and electrochemical pumping.
- Poor surface kinetics: slows hydrogen dissociation or recombination.
- Chemical instability: reactions with \(\mathrm{CO_2}\) or water vapour can change phase composition and conductivity.
- Electronic leakage in a pump electrolyte: reduces the fraction of current used for proton transport.
- Cracking and seal failure: create non-selective gas crossover.
- Dual-phase disconnection: interrupts either the protonic or electronic transport network.
- Electrode degradation: increases polarization resistance and lowers pumping rate.
- Thermal expansion mismatch: causes delamination, cracking or seal failure during thermal cycling.
- Uncontrolled humidity: changes proton concentration and makes comparisons between tests unreliable.
14. Practical workflow for membrane evaluation
- Define the target hydrogen purity, recovery, feed composition, pressure and operating temperature.
- Decide whether the intended process is spontaneous MPEC permeation or electrochemical pumping.
- Measure protonic and electronic conductivity under relevant gas atmospheres.
- Assess chemical stability in hydrogen, steam, carbon dioxide and relevant contaminants.
- Estimate ideal MPEC flux using the Wagner-type relation, or estimate pump rate using Faraday’s law.
- Select membrane thickness, support and electrode architecture.
- Fabricate dense membranes and verify phase purity, microstructure and gas tightness.
- Measure permeate composition, total flow, temperature and pressure at steady state.
- For pumps, record current–voltage curves, impedance and Faradaic efficiency.
- Vary flow rate and membrane thickness to identify transport limitations.
- Conduct long-term and thermal-cycling tests.
- Characterize the membrane and electrodes after testing to identify degradation.
15. Reporting checklist
A reproducible hydrogen-separation report should state:
- Full membrane composition and synthesis route;
- Phase composition and density;
- Membrane thickness and active area;
- Single-phase, dual-phase or metal–ceramic architecture;
- Feed and permeate gas compositions;
- Hydrogen and water-vapour partial pressures;
- Total pressure and pressure difference;
- Operating temperature and temperature profile;
- Feed and sweep flow rates;
- Electrode composition and geometry, if applicable;
- Current and voltage for electrochemical pumping;
- Hydrogen flux and calculation method;
- Hydrogen purity, recovery and Faradaic efficiency where relevant;
- Leak-test method and results;
- Duration of stability testing and post-test characterization.
Further reading
- S. Cheng, X. Li, X. Huang, Y. Ling, S. Liu and T. Li, “Hydrogen separation via proton conducting ceramic membranes: A review,” International Journal of Hydrogen Energy 70 (2024), 654–665. https://doi.org/10.1016/j.ijhydene.2024.05.181.
- Z. Tao, L. Yan, J. Qiao, B. Wang, L. Zhang and J. Zhang, “A review of advanced proton-conducting materials for hydrogen separation,” Progress in Materials Science 74 (2015), 1–50. https://doi.org/10.1016/j.pmatsci.2015.04.002.
- H. Cheng, “Dual-Phase Mixed Protonic-Electronic Conducting Hydrogen Separation Membranes: A Review,” Membranes 12 (2022), 647. https://doi.org/10.3390/membranes12070647.
- J. Tong, P. Zhang, F. Zhuang, Y. Zheng, B. Liu, X. Qiao and X. Zhu, “Mixed-conducting ceramic membrane reactors for hydrogen production,” Reaction Chemistry & Engineering 9 (2024), 3072–3099. https://doi.org/10.1039/D4RE00372A.
- G. Zhang, X. Bao, G. Zheng et al., “Research Progress in Proton Conducting Ceramic Membranes for Hydrogen Separation,” Journal of Ceramics 44 (2023), 259–271. https://doi.org/10.13957/j.cnki.tcxb.2023.02.006.