Separating hydrogen (H₂) from carbon dioxide (CO₂) might sound like a niche lab exercise, but in reality it sits right at the heart of the “hydrogen economy” and low-carbon technologies.

Whenever hydrogen is produced from fossil fuels — for example, via the water–gas shift reaction — you don’t just get pure H₂. You get a mixture that contains a lot of CO₂, small amounts of CO, sometimes H₂S, and other impurities. If you want clean hydrogen for industry, fuel cells, or future energy systems, you must separate it efficiently and at a reasonable cost.

Traditional separation methods like cryogenic distillation or pressure swing adsorption (PSA) work, but they’re energy-intensive and expensive. That’s why membranes have become so attractive: thin selective barriers that allow some gases through faster than others. They can be simpler, cheaper, and easier to scale.

In the study we’re translating into blog form here, researchers designed a new kind of membrane based on metal–organic frameworks (MOFs), specifically a layered material called ZIF-L. By cleverly combining zinc (Zn) and cobalt (Co) in one structured membrane, and controlling how the crystals grow, they created a Co–Zn-ZIF-L membrane that can separate H₂ and CO₂ with both high speed (permeance) and high selectivity.

The result is a membrane that not only beats many existing ZIF-based membranes but even surpasses the famous “Robeson upper bound” for H₂/CO₂ gas separation — a benchmark curve that describes the performance limit of many polymer membranes.

Let’s walk through what they did and why it matters, in clear and simple language.

Why Is H₂/CO₂ Separation So Difficult?

On paper, separating hydrogen from carbon dioxide sounds easy. After all, they are different gases. But at the molecular level, the challenge becomes clear:

• Similar size:
  The “kinetic diameters” of the two molecules — a measure of how big they look to pores — are very close:

  • H₂ ≈ 2.9 Å

  • CO₂ ≈ 3.3 Å
    This means you can’t just use a “big vs. small” sieve in a crude way; the pores must be extremely precise.

• Different polarity:

  • CO₂ is a polarizable molecule with a quadrupole moment and interacts strongly with certain materials.

  • H₂ is non-polar and interacts more weakly.

Membranes can use both size differences and chemical affinity to separate gases. But getting both high selectivity (strong discrimination between H₂ and CO₂) and high permeance (fast transport of H₂) is difficult. Many polymer membranes suffer from a trade-off: as you increase selectivity, permeability drops, and vice versa.

This is where metal–organic frameworks come in.

What Are MOFs and ZIF-L?

Metal–organic frameworks (MOFs) are crystalline materials made from metal ions (or clusters) connected by organic linkers. Think of them as scaffolds with tunable pores. They can offer:

• Very high surface areas

• Tailored pore sizes

• Tunable chemical environments inside the pores

Within MOFs, zeolitic imidazolate frameworks (ZIFs) are a special family where metal ions (like Zn²⁺ or Co²⁺) are linked by imidazolate ligands. They resemble traditional zeolites structurally, but often with better chemical flexibility and stability.

ZIF-L is a particular layered ZIF:

• It is built from tetrahedral metal centers (Zn²⁺ or Co²⁺) coordinated to four imidazole ligands.

• Instead of forming a 3D cubic framework, these tetrahedra assemble into 2D layers that stack to form a lamellar structure.

• It has two relevant pore types:

  • Interlayer channels (~0.3 nm): narrow gaps between layers, close to the size of H₂ and CO₂ — ideal for molecular sieving.

  • Six-membered ring windows (~0.4 nm): in-plane openings that facilitate gas diffusion.

So ZIF-L is promising for gas separation, especially when you want to finely discriminate between very similar molecules.

However, single-metal ZIF-L (e.g., purely Zn-based) has some limitations:

• The pores are relatively rigid.

• The chemical environment is uniform (only one type of metal site), which can limit performance.

To push beyond these limits, the authors turned to heterostructures.

The Idea: A “Homogeneous Heterogeneity” Co–Zn-ZIF-L Membrane

The concept sounds contradictory at first: homogeneous heterogeneity. What they mean is:

• The membrane is structurally continuous and uniform (homogeneous at the device scale).

• At the microscopic level, it contains two types of metal sites (Zn and Co) arranged in a controlled way (heterogeneous at the local level).

By combining Zn and Co in ZIF-L and controlling how and where each metal appears in the layer, the researchers aimed to:

• Tailor pore size and flexibility

• Adjust electronic structure and adsorption behavior

• Improve mechanical stability and membrane adhesion

• Enhance H₂/CO₂ selectivity and H₂ permeance simultaneously

To achieve this, they needed:

1. A robust way to grow ZIF-L crystals as a continuous membrane.

2. A method to introduce cobalt in a controlled, oriented fashion onto an initial zinc-based ZIF-L layer.

Step 1: Creating a ZIF-L Seed Layer on the Support

Growing MOF membranes directly on polymer supports is not trivial. In this work, they used a polysulfone (PSf) support, which doesn’t naturally bond strongly to ZIF-L. So they first engineered a polymer–metal interface that helps ZIF-L form.

They used two polymers:

• Polyvinyl alcohol (PVA) – rich in –OH groups

• Polyallylamine (PAH) – rich in –NH₂ groups

These polymers were cross-linked and used to chelate Zn²⁺ ions, forming a PVA–PAH–Zn layer on top of the support. This layer does several things at once:

• Smooths over surface defects on the PSf support.

• Provides functional groups (–OH, –NH₂) that can:

  • Bind to Zn²⁺ ions

  • Interact with the ZIF-L framework during growth

• Restricts how quickly metal ions and ligands diffuse, promoting ordered growth at the interface.

From this tailored interface, a Zn-ZIF-L seed layer is grown via interfacial diffusion. The result:

• A ZIF-L layer about 1 μm thick.

• A preferred orientation where the (100) plane is parallel to the support, helping to expose the 2D pores in a useful direction for gas transport.

• Better adhesion between membrane and support.

This seed layer is the platform on which the final Co–Zn-ZIF-L membrane is constructed.

Step 2: Building the Co–Zn-ZIF-L Heterostructure by Secondary Growth

Once the Zn-ZIF-L seed layer is in place, the researchers perform a secondary growth step using a cobalt-containing solution to create the final Co–Zn-ZIF-L membrane by heteroepitaxial growth.

Key points about this step:

• Heteroepitaxial growth means a second crystalline material (here, Co-containing ZIF-L) grows in registry with the underlying Zn-ZIF-L structure.

• The interface between the seed layer and new crystals is coherent enough that the structure remains continuous, not patchy.

What emerges is:

• A membrane about 1.7 μm thick.

• A clear a-axis preferred orientation: X-ray diffraction shows strong intensity for specific planes (like (800)), which align with that direction.

• A layered distribution of metals:

  • Zn concentrated mainly in the lower half of the membrane (closer to the support).

  • Co enriched in the upper half (facing the feed gas).

This gradient in metal composition gives a “homogeneous heterogeneous” structure: one continuous membrane, but with internal compositional variation and dual metal environments.

How Do We Know the Structure Is Really There?

The paper uses various characterization tools; in blog form, we’ll just summarize the essential conclusions (not every technical detail):

• FTIR (infrared spectroscopy):

  • Shows the presence and growth of Zn–N and Co–N coordination bonds.

  • As the membrane grows from PVA–PAH–Zn to Zn-ZIF-L and then to Co–Zn-ZIF-L, the characteristic peaks associated with metal–ligand coordination become stronger and more defined.

• XRD (X-ray diffraction):

  • Confirms the crystals have the expected ZIF-L structure.

  • Demonstrates that the membrane is oriented, not randomly oriented: certain crystal planes dominate.

• Elemental mapping (EDS):

  • Visually confirms Zn and Co are distributed through the thickness, with Zn more at the bottom and Co more at the top.

Together, these results confirm that:

• The ZIF-L framework has formed properly.

• Both Zn and Co are incorporated.

• The membrane is continuous and oriented.

Adsorption Behavior: Why H₂ and CO₂ Behave Differently

To understand separation, you must understand how each gas interacts with the material.

For Co–Zn-ZIF-L powders, adsorption measurements at 303 K show:

• H₂ adsorption:

  • Roughly linear with pressure.

  • Indicates relatively weak interactions — H₂ can enter the micropores easily but does not bind strongly.

• CO₂ adsorption:

  • Follows typical Langmuir-type behavior, consistent with stronger, saturable adsorption.

  • CO₂ interacts more strongly due to:

    • Its larger size and quadrupole moment.

    • Possible interactions with exposed metal sites (like Co²⁺) and nitrogen-containing ligands.

This difference is key:

• The membrane adsorbs CO₂ more strongly, which helps discriminate between gases but can also risk blocking pores if not carefully designed.

• H₂ stays more mobile and diffuses faster.

In this system, the separation mechanism is dominated not by solubility differences, but by diffusion differences — H₂ diffuses much faster through the Co–Zn-ZIF-L framework than CO₂ does.

Gas Separation Performance: Numbers That Matter

The performance of the final Co–Zn-ZIF-L membrane was measured using a H₂/CO₂ gas mixture under standardized conditions. Here are the headline results at 298 K and 1.0 bar:

• H₂ permeance:
  ≈ 2156 GPU
  (GPU is a standard unit of gas permeance; higher numbers mean faster transport.)

• H₂/CO₂ selectivity:
  ≈ 31.6

These values are very significant because:

• The membrane exceeds the Robeson upper bound (2008) for H₂/CO₂ separation, which is a widely referenced benchmark that expresses the trade-off between permeability and selectivity for many polymer membranes.

• Compared to other reported ZIF-L-based membranes, this Co–Zn-ZIF-L design offers both high flux and strong selectivity.

The study also decomposed the performance into:

• Solubility coefficient: how much each gas dissolves in the membrane.

• Diffusion coefficient: how fast each gas moves through it.

Result:

• CO₂ has higher solubility (it interacts more with the framework),

• but H₂ has a much higher diffusion coefficient, and the diffusion contrast is what really drives separation.

How Operating Conditions Affect Performance

1. Temperature

As temperature increases:

• Both H₂ and CO₂ permeances increase.

• This behavior is consistent with activated diffusion: gas molecules need energy to hop through tiny pores, and higher temperature supplies that energy.

Despite the increase in permeance, the membrane maintains its separation ability, which is encouraging for practical operation across different temperatures.

2. Pressure

As pressure increases:

• Both H₂ and CO₂ permeances decrease.

• Selectivity may drop at higher pressures.

Why? Because:

• CO₂ is strongly adsorbed by ZIF-L.

• At higher pressures, more CO₂ molecules crowd into the pores, especially near active sites.

• This can block pathways that H₂ would otherwise use, reducing H₂ transport and disturbing the ideal sieving behavior.

So, in practice, there is an optimal pressure range where the membrane delivers the best combination of high flux and high selectivity.

3. Moisture (Wet vs. Dry Conditions)

Real gas streams often contain water vapor, so the team tested membrane stability in both dry and humid feeds.

• Under dry conditions, over 65 hours of continuous operation, the membrane maintained:

  • H₂ permeance around 2100 GPU

  • H₂/CO₂ selectivity around 32

• When water vapor was introduced:

  • The permeance of both gases decreased because water competes for transport pathways.

  • However, CO₂ (being larger and more strongly interacting) is more hindered than H₂.

  • As a result, H₂/CO₂ selectivity actually increased under humid conditions.

This is a valuable outcome: not only is the membrane stable over long operation, but it also responds to water vapor in a way that can be beneficial for purification.

Computational Insight: Molecular Dynamics Simulations

To complement the experiments, molecular dynamics simulations were used to estimate diffusion coefficients of H₂ and CO₂ in:

• Co–Zn-ZIF-L

• Zn-ZIF-L

• Co-ZIF-L

The key finding:

• In Co–Zn-ZIF-L, the diffusion coefficient of H₂ is significantly higher than that of CO₂.

This supports the idea that:

• The addition of Co modifies the pore environment and energy landscape inside the ZIF-L structure.

• The dual-metal system (Co + Zn) and oriented pore structure create a microenvironment where H₂ moves easily while CO₂ faces more resistance.

Zn-ZIF-L alone shows high flux but poor selectivity. Co-only systems help, but the synergistic Co–Zn combination gives the best balance.

Why the Co–Zn-ZIF-L Membrane Works So Well

Putting it all together, the excellent performance of this membrane comes from a combination of structural and chemical factors:

1. Oriented growth along the a-axis

   • The crystal alignment ensures that the sub-nanometer pores are aligned in a direction favorable for gas transport.

   • This reduces tortuosity and lets H₂ move quickly.

2. Homogeneous heterogeneous structure (Zn at bottom, Co enriched at top)

   • Zn-rich seed layer ensures good adhesion and basic framework integrity.

   • Co-rich upper region adjusts electronic structure and adsorption behavior in the active separation zone.

3. Dual role of the polymer network (PVA/PAH)

   • Creates a controlled interface for seed formation.

   • Enhances bonding and mechanical robustness.

   • Controls diffusion of metal ions and ligands during growth, avoiding defects and gaps.

4. Pore size and chemistry tuned for H₂/CO₂

   • Pore sizes around the borderline of H₂ and CO₂ kinetic diameters allow for molecular sieving.

   • The Co–Zn environment and imidazole ligands create differentiated adsorption and diffusion behaviors.

5. Good mechanical stability and interfacial strength

   • The membrane remains intact and performant for extended operation, even under wet conditions.

What This Means for Hydrogen and Carbon Management

From a practical point of view, this work shows a viable route to:

• High-performance H₂/CO₂ separation membranes that:

  • Are relatively thin and defect-free

  • Have high H₂ permeance

  • Provide strong H₂/CO₂ selectivity

  • Maintain stability over many hours and in the presence of moisture

Such membranes could be integrated into:

• Hydrogen purification units downstream of reformers or water–gas shift reactors

• Carbon capture processes that simultaneously purify H₂ and concentrate CO₂

• Future modular, membrane-based systems that reduce the energy penalty of gas separation compared to conventional PSA or cryogenic routes

Beyond this specific system, the work also demonstrates a general strategy:

• Use polymer-assisted interfacial diffusion to control MOF membrane orientation and defect formation.

• Build heterostructured MOF membranes by combining different metals within the same framework for finely tuned performance.

In the broader context of the hydrogen economy and carbon neutrality goals, materials like the Co–Zn-ZIF-L membrane provide a promising toolbox for making gas separations more energy-efficient, compact, and scalable.