Five years ago, direct seawater electrolysis looked like a neat demo with a pending bug report: "works in clean water, fails when the ocean shows up." Today, Saj and colleagues are submitting a more serious patch: a fluorine-doped cobalt-iron layered hydroxide catalyst that makes hydrogen from alkaline seawater while telling chloride ions, politely but firmly, to go ruin someone else's electrode.

LGTM? Mostly. Approved with reservations.

The Bug: Salt Is Not a Passive User

Water electrolysis sounds simple in the same way "just deploy it" sounds simple. You pass electricity through water, hydrogen bubbles off one side, oxygen off the other, and everyone goes home feeling renewable.

Then seawater enters the repo.

Seawater contains chloride ions, and chloride ions are the kind of dependency that breaks your build only after the demo. At the anode, the system is supposed to run the oxygen evolution reaction, or OER. That reaction is slow because it has to move four electrons around, which is basically chemistry's version of making a committee agree on lunch. Worse, chloride can compete and cause chlorine-related reactions and corrosion. Recent reviews keep pointing to the same blocker: direct seawater electrolysis needs catalysts and systems that stay active, selective, and alive in salty, chemically messy conditions Liu et al., 2024, Corbin et al., 2024.

The Patch: Fluorine Doping, But Make It Useful

The new paper reports F-CoFe LMH-8, a fluorine-doped cobalt-iron layered metal hydroxide catalyst Saj et al., 2026. Layered hydroxides are thin, stacked materials with tunable metal sites. Think of them as molecular shelving units where the useful stuff is not the shelf, but all the weirdly specific things you can put on it.

Here, the authors use fluorine doping to tune the electronic structure of CoFe hydroxide. Translation: they sneak fluorine into the material so the iron sites behave differently. The paper argues this pushes Fe toward a high-spin configuration, improves binding energies for reaction intermediates, and creates a chlorophobic surface.

Nit: "chlorophobic" sounds like a villain in a beach episode, but the idea is clean. The catalyst likes water and oxygen-related species enough to do useful electrochemistry, while discouraging chloride from camping on the surface and corroding the active sites.

That matters because a catalyst that performs well in purified KOH but folds in seawater is not production-ready. It is a lab diva.

The Numbers: Surprisingly Decent Commit Message

For the half-reactions, the catalyst hit 81.23 mV at 10 mA cm^-2 for hydrogen evolution and 265.5 mV at 10 mA cm^-2 for oxygen evolution. Lower overpotential means less extra voltage wasted pushing the reaction uphill. Good.

The more interesting test is the full anion exchange membrane water electrolyzer, or AEMWE. AEM systems conduct hydroxide ions through a membrane, separating products while using alkaline chemistry and potentially cheaper catalysts than precious-metal-heavy alternatives. The authors report current densities of:

• 1.2 A cm^-2 in 1 M KOH at 2.3 V
• 1.02 A cm^-2 in 1 M KOH plus 0.5 M NaCl at 2.3 V
• 1.0 A cm^-2 in 1 M KOH seawater at 2.3 V

They also report 500 hours of continuous operation with a degradation rate of 0.15 uV h^-1. Blocking issue avoided, at least at this scale.

For context, other recent work is also pushing seawater systems toward practical current densities, including corrosion-resistant NiFe anodes at kilowatt scale Sun et al., 2024, molybdate-modulated NiFe oxide electrodes Shao et al., 2024, and chlorine-suppression strategies reviewed across catalysts, membranes, and full systems Wang et al., 2025. This paper fits the pattern: stop treating seawater like pure water with vibes. Engineer for the actual mess.

The AI Cameo: LSTM Enters, Wearing a Lab Coat

The authors also used a long short-term memory model, or LSTM, to forecast catalyst stability. LSTMs are older neural network architectures built for sequence data. They remember patterns over time better than a plain neural net, which has the memory discipline of a browser tab left open since 2019.

Here, the LSTM is not discovering chemistry from first principles. It is forecasting stability trends from experimental time-series data. Useful? Potentially. Needs careful validation? Absolutely. Prediction models in materials papers can be helpful, but only if they do not become decorative dashboards with statistical confetti.

Reviewer note: show me external validation across different electrolyzer builds, real seawater sources, temperatures, and long runtimes. Then we can talk.

Why This Is Worth Merging

The practical prize is obvious: green hydrogen systems that can use abundant seawater without expensive pretreatment and without producing a corrosion crime scene. If results like these reproduce and scale, they could help coastal hydrogen production, offshore renewable integration, and industrial electrolysis setups where freshwater demand is a real constraint.

But no hype merge. A 500-hour lab test is encouraging, not a warranty. Real seawater brings biofouling, magnesium and calcium deposits, changing salinity, membrane degradation, gas crossover, and maintenance costs. The ocean is not a reagent bottle. It is a chaotic soup with excellent lawyers.

Still, the core design choice is strong: use fluorine doping to tune active sites and repel chloride, then test the material in an AEM device instead of stopping at half-cell bragging rights. That is the kind of scope discipline reviewers like to see.

Final review: clever, focused, experimentally relevant. Needs scale-up, independent replication, and harsher field conditions. Approved with reservations.

References

1. Anandhan Ayyappan Saj et al. "High Polarity Doping of CoFe Layered Hydroxides: Bifunctional and Corrosion-Resistant Anion Exchange Membrane Seawater Electrolyzers." Nano-Micro Letters 18, 393, 2026. DOI: 10.1007/s40820-026-02230-8

2. Y. Liu et al. "Long-term durability of seawater electrolysis for hydrogen: from catalysts to systems." Angewandte Chemie International Edition 63, e202412087, 2024. DOI: 10.1002/anie.202412087

3. J. Corbin et al. "Challenges and progress in oxygen evolution reaction catalyst development for seawater electrolysis for hydrogen production." RSC Advances 14, 6416-6442, 2024. DOI: 10.1039/D3RA08648H

4. X. Sun et al. "Corrosion-resistant NiFe anode towards kilowatt-scale alkaline seawater electrolysis." Nature Communications 15, 10351, 2024. DOI: 10.1038/s41467-024-54754-5

5. L. Shao et al. "In situ generation of molybdate-modulated nickel-iron oxide electrodes with high corrosion resistance for efficient seawater electrolysis." Advanced Energy Materials 14, 2303261, 2024. DOI: 10.1002/aenm.202303261

6. "Comprehensive Chlorine Suppression: Advances in Materials and System Technologies for Direct Seawater Electrolysis." PMC, 2025. PMCID: PMC11754585

Disclaimer: This blog post is a simplified summary of published research for educational purposes. The accompanying illustration is artistic and does not depict actual model architectures, data, or experimental results. Always refer to the original paper for technical details.