The traditional approach to designing better catalysts focuses on tweaking the active sites where chemical reactions happen, like adjusting the electronic properties of metal atoms or changing what atoms surround them. But researchers have now demonstrated that an often overlooked feature, the physical arrangement of layers in the catalyst's structure, can be just as important as the chemistry of the active sites themselves.

Their discovery centers on a subtle architectural feature: hydrogen bonds that form between layers of the catalyst, reaching across the nanoscale gap to stabilize reaction intermediates. This structural cooperation boosts both the activity and selectivity of a catalyst designed to produce hydrogen peroxide, one of the most versatile and environmentally friendly chemicals in modern industry.

The Hydrogen Peroxide Challenge

Hydrogen peroxide serves as a powerful oxidant and disinfectant across industries from medicine to manufacturing. Global demand reaches approximately 6 million metric tons annually, with needs projected to grow. Yet over 95% of production still relies on the anthraquinone process, an energy intensive industrial method that carries serious safety risks and generates significant pollution.

An alternative exists: electrochemical synthesis, which produces hydrogen peroxide directly from oxygen and water using electricity. The process, particularly through what chemists call the two electron oxygen reduction reaction, offers intrinsic advantages. It's safer, generates no toxic byproducts, and can be deployed on site wherever needed. The challenge has been developing catalysts efficient enough to make the approach economically viable, especially across the wide pH range needed for practical applications like water treatment and disinfection.

Most research has concentrated on tailoring the electronic structure of single atom catalysts, manipulating properties like the d band center of metal atoms to optimize how strongly they bind reaction intermediates. These efforts typically model the catalyst as a single flat layer containing specific metal sites, a simplification that provides useful insights but misses important physics.

The problem is that real catalysts aren't single layers. Many have stacked structures where layers sit just 0.3 to 0.4 nanometers apart, close enough that the layer above an active site might significantly influence its behavior. This steric effect from the layered architecture has been largely ignored.

When Layers Cooperate

The research team focused on a one dimensional metal organic framework containing nickel atoms coordinated by four nitrogen atoms from organic ligands. This material, with nickel centers surrounded by amine groups, has a distinctive layered structure where chains stack with precise spacing.

Using density functional theory calculations, the researchers first examined what would happen at a nickel site in a single isolated chain. The results showed weak binding of the key reaction intermediate, a species called hydroperoxyl (written as *OOH in catalyst notation). The binding energy was too high, at 4.82 electron volts, meaning the catalyst would have low activity but high selectivity for the desired two electron pathway that produces hydrogen peroxide rather than fully reducing oxygen to water.

Simply changing the metal center or adding functional groups like hydroxyl to the coordination sphere could shift this binding energy, but not enough to reach the optimal value around 4.23 eV where both high activity and high selectivity converge.

The picture changed dramatically when the calculations included a second layer of the framework positioned above the active site, mimicking the real structure where chains stack 0.35 nanometers apart. Now, with both layers present and hydroxyl groups coordinating the nickel atoms, the binding energy dropped to 4.21 eV, nearly ideal for efficient hydrogen peroxide production.

The Hydrogen Bond Advantage

What causes this dramatic improvement? The answer lies in an elegant piece of molecular cooperation. When the *OOH intermediate binds to a nickel site on the bottom layer, the oxygen atoms sit just 1.9 to 2.0 angstroms from hydrogen atoms on nitrogen groups in the layer above. This distance is characteristic of a hydrogen bond, a relatively weak but important interaction where a hydrogen atom acts as a bridge between two electronegative atoms.

The calculations revealed telltale signs of hydrogen bonding. The N-H bonds in the upper layer stretched slightly when OOH bound below. The hydrogen atoms lost electron density, becoming more positive. Most tellingly, the binding energy of OOH strengthened substantially compared to configurations without the upper layer present.

This internal hydrogen bond acts similarly to surface ligands that other researchers have used to tune catalytic activity, but with crucial advantages. The layered structure provides a stable, precisely positioned hydrogen bond donor that can repeatedly form and break during catalysis without dissolving or degrading, unlike surface coatings that might wash away.

The effect works synergistically with the hydroxyl groups coordinated to the nickel atoms. The intrinsic oxygen affinity of nickel provides the baseline, the hydroxyl coordination shifts the binding energy in the right direction, and the hydrogen bond from the layer above fine tunes it to the optimal value.

Experimental Validation

The theoretical predictions needed experimental confirmation. The team synthesized the nickel based framework along with versions containing manganese, iron, cobalt, and copper for comparison. X ray characterization confirmed the expected layered structure, with the distinctive fishbone like stacking pattern predicted by theory.

When tested for oxygen reduction activity, the nickel catalyst showed the highest selectivity for hydrogen peroxide production, maintaining over 90% selectivity across a wide potential range in alkaline solution and over 85% in neutral conditions. The performance surpassed the other metal variants, exactly as the theoretical volcano plot based on binding energies had predicted.

Particularly revealing was the comparison with a deliberately disordered version of the nickel catalyst, prepared by accelerating the synthesis to prevent proper layer formation. This amorphous material showed lower selectivity and activity, providing direct evidence that the ordered layered structure indeed enhances performance.

To confirm the hydrogen bonding mechanism, the researchers used sophisticated spectroscopy techniques that could observe the catalyst during operation. Infrared spectroscopy showed the characteristic signatures of *OOH intermediates on the nickel sites. Crucially, the N-H stretching vibration from the organic ligands shifted to lower frequency during oxygen reduction, exactly what would happen if those N-H groups were forming hydrogen bonds with the reaction intermediates.

Beyond Theory: Practical Performance

In practical flow cell tests simulating real world conditions, the catalyst demonstrated impressive capabilities. In neutral sodium chloride solution (simulating seawater conditions), it produced hydrogen peroxide at rates exceeding 13.5 moles per gram of catalyst per hour while maintaining over 80% efficiency. The material operated stably for 24 hours.

In alkaline solution, the production rate more than doubled to 27 moles per gram per hour. These numbers place the catalyst among the best reported performers, whether in alkaline or neutral conditions.

The practical utility extends beyond raw production numbers. The team demonstrated that hydrogen peroxide generated in physiological saline could completely sterilize both E. coli and S. aureus bacteria within 30 minutes. In simulated wastewater, the system rapidly degraded organic dyes when coupled with UV light in a Fenton type oxidation process.

Even at the challenging industrial scale current densities of 200 milliamps per square centimeter, the catalyst maintained over 80% selectivity, showing potential for practical implementation.

The Non-Coordination Revolution

Perhaps the most important insight from this work extends beyond hydrogen peroxide production. For decades, catalyst design has focused almost exclusively on the coordination environment, what atoms directly bond to the metal center and how those bonds affect electronic structure.

This research demonstrates that non-coordination features, the physical architecture surrounding active sites, can be equally important. The layers above active sites aren't just spectators; they're active participants in the catalytic mechanism.

The implications ripple outward. If internal hydrogen bonds can enhance oxygen reduction, what other reaction types might benefit from similar architectural engineering? The concept could apply to any catalyst where reaction intermediates contain hydrogen bond acceptor groups, a broad category spanning everything from fuel cells to chemical synthesis.

The findings also highlight a limitation in common computational approaches. Modeling catalysts as single isolated layers, while computationally convenient, misses crucial physics. More sophisticated models that include the full three dimensional structure, even when computationally expensive, may be necessary to capture the real behavior of layered materials.

For the specific goal of hydrogen peroxide production, the work addresses a pressing need. Decentralized, on demand production of this essential chemical could transform applications from point-of-use water treatment to medical disinfection, particularly in resource limited settings. A catalyst that works efficiently across pH ranges, from alkaline to neutral conditions, removes a major barrier to widespread deployment.

Design Principles for the Future

The research establishes clear design principles that could guide the next generation of catalysts. First, consider the full three dimensional structure, not just the two dimensional coordination environment. Second, look for opportunities to position functional groups precisely where they can interact with reaction intermediates through non covalent interactions like hydrogen bonds. Third, use rigid frameworks rather than flexible surface ligands to ensure those interactions remain stable over many catalytic cycles.

Metal organic frameworks, with their tunable structures and precise atomic arrangements, emerge as ideal platforms for implementing these principles. The same design logic could extend to other layered materials, from graphene based systems to layered double hydroxides, anywhere that interlayer spacing can be controlled to enable productive interactions between layers.

The work also suggests that catalyst screening and optimization might benefit from explicitly considering steric and architectural effects alongside electronic ones. A metal atom might show mediocre performance in isolation but excel when placed in the right structural context, with supportive interactions from the surrounding framework.

As chemistry moves toward more sustainable processes, such insights become increasingly valuable. The most efficient catalysts will likely be those that exploit every available design dimension, from electronic structure to geometric arrangement to architectural cooperation. This research shows that some of the most powerful design levers have been hiding in plain sight, in the spaces between layers we rarely thought to examine closely.

Publication Details

Published online: April 30, 2025
Journal: Nature Communications
Publisher: Springer Nature
DOI: https://doi.org/10.1038/s41467-025-58628-2

Credit and Disclaimer

This article is based on original research published in Nature Communications by scientists from Tianjin University, Peking University, and the Chinese Academy of Science. The content has been adapted for general audiences while maintaining complete scientific accuracy. Readers are strongly encouraged to consult the full research article for comprehensive technical details, complete datasets, detailed methodologies, experimental protocols, and supplementary information via the DOI link provided above. All scientific findings, data, and conclusions presented here are derived directly from the original publication, and full credit belongs to the research team and their institutions.