The Catalyst — Mar 10, 2026

The theme this fortnight is translation — elegant lab results finally surviving contact with the real world. Roll-to-roll printed organic solar cells just got a proper mechanistic autopsy revealing the bottleneck was never the physics, iron catalysts are eating into palladium's territory on two fronts simultaneously, and a frozen electrolyte that shouldn't conduct lithium ions does anyway. Meanwhile, the chemical industry is splitting in two: commodity giants are slashing jobs and shuttering plants while the enabling tools for the next economy — 2 nm chip equipment, self-driving structural-materials labs, AI-designed catalysts — ship to factory floors.

There are roughly 80 palladium atoms per trillion in Earth's crust. Iron makes up about 5% of the planet. Every time a chemist swaps Pd for Fe and gets comparable performance, an industrial reaction becomes dramatically cheaper.

This fortnight delivered two such results. At Nagoya, a team built an iron photocatalyst that achieves asymmetric catalysis — controlling which mirror-image product forms — with far fewer expensive chiral ligands than previous designs. Getting enantioselectivity from iron is mechanistically hard: multiple accessible spin states tend to scramble stereochemistry. Nagoya's approach constrains the iron's coordination geometry to enforce the right electronic state during catalysis. Separately, a Peking University group reported in Nature Chemistry an iron catalyst for alkene hydroformylation — the cornerstone industrial reaction producing aldehydes for detergents and plastics — achieving turnover numbers comparable to rhodium systems.

These aren't incremental. Nagoya covers asymmetric fine chemistry; Peking covers bulk commodity catalysis. Together they suggest iron's catalytic moment is systemic, not anecdotal. The numbers to watch in follow-ups: ee values, substrate scope, and catalyst robustness under real feedstock impurities. Those will tell you whether these are methods or miracles.

Organic solar cells have promised roll-to-roll manufacturing for over a decade. The pitch — photovoltaics printed like newspapers on flexible film — keeps stumbling on the same problem: the nanoscale morphology that makes lab devices efficient falls apart under industrial gravure printing. Or so everyone assumed.

A new arXiv preprint from Taranenko et al. challenges that assumption directly. Using commercially available PM6:Y12 materials and non-halogenated solvents, they demonstrated that favorable bulk morphology and exciton harvesting survive gravure printing just fine. The real performance gap comes from optical interference within the printed layer stack and slow charge transport — engineering problems, not fundamental materials limits.

That's a reframe. If the active blend isn't the villain, the roadmap changes: optimize optical stacks and transport layers, stop recursing on blend chemistry. The team reports this as the highest efficiency for a fully roll-to-roll-compatible gravure-printed NFA device to date, with industrial-style demos hitting ~14% efficiency in the reported devices at 10 m/min printing speeds and surviving 1,000 flexing cycles with <5% degradation over the 1,000-cycle test. This is a preprint — treat the numbers as provisional — but the diagnostic framework is unusually rigorous. The non-halogenated solvent detail isn't a footnote: chlorinated solvents are industrial hazards, and eliminating them is a prerequisite for any real manufacturing process.

Solid-state batteries usually mean exotic ceramics or sulfide glasses. A UNIST–KAIST team basically asked "what if we just froze the stuff we already use?" and made it work. Their "organic ice" electrolyte — 0.2 m LiTFSI in ethylene carbonate, crystallized — conducts Li⁺ at ~0.64 mS·cm⁻¹ with a transference number of ~0.8, numbers in the same ballpark as some polymer electrolytes. Li⁺ hops along channels in the EC crystal lattice while the solvent stays immobilized, and the system forms a Li₂O-rich SEI that dramatically extends cycle life in LiFePO₄‖Li cells.

This isn't drop-in commercial tomorrow — temperature window, mechanical robustness, and scale-up are open questions. But it's a sharp reminder that "solid-state" doesn't require completely new supply chains. The broader solid-state picture is converging: ION Storage Systems just became the first U.S. developer to pass customer qualification for a ceramic electrolyte cell, McGill published a polymer-ceramic sponge hybrid that eliminates interfacial resistance, and QuantumScape qualified a sulfide electrolyte for pilot production. The metallurgy of production — coating, stack assembly, pressure control — is now the gating factor, not the chemistry.

If we're going to electrify petrochemicals, ethylene (200+ Mt/year) is the big boss fight. A Northwestern-led team in Nature Synthesis shows an electrocatalyst that makes ethylene directly from CO₂ dissolved in a capture liquid, not from purified gas. The problem is subtle: carbonate solutions hold very little free CO₂ near the electrode, so classical CO₂ reduction gives H₂ instead of C–C coupling. Their analysis shows low *CO coverage kills the usual symmetric CO–CO coupling pathway.

The solution: a dilute alloy catalyst that promotes asymmetric CO–CHO coupling, which has a lower kinetic barrier and still forms the C₂ bond even when *CO is scarce. That's a mechanistic design choice, not "let's try another metal." Energy efficiency roughly doubles versus previous reactive-capture ethylene demos. The big next step is nasty: proving long-term operation in genuine flue-gas capture liquids with real impurities. But this is the first result that makes "skip the CO₂ purification step entirely" look plausible for the most important petrochemical on Earth.

Self-driving labs have mostly been a story about battery cathodes and drug candidates — functional materials with fast feedback loops. Structural materials — alloys and ceramics for extreme heat, pressure, or radiation — have lagged because failure takes months to years, microstructure matters across nanometers to millimeters, and thermomechanical processing is brutally hard to automate.

A Johns Hopkins preprint describes AIMD-L, an automated laboratory built specifically for this problem. Two custom instruments — HELIX for shock studies and MAXIMA for XRD/XRF — collect data two to three orders of magnitude faster than conventional systems, processing 100+ samples per day. The key innovation isn't the robots; it's the feedback architecture. Rapid surrogate tests under harsher-than-service conditions serve as proxies for long-term performance, with ML models learning the mapping between surrogate results and actual service life. Amid Army Research Lab funding, the work sits amid defense-relevant applications.

Complementing AIMD-L: a Berkeley team demonstrated 41 of 58 AI-predicted materials synthesized in a single automated run, and a separate campaign used an MLIP to screen thousands of Cu-alloy surfaces, producing a Cu–Zn catalyst hitting 72% Faradaic efficiency in the reported experiment for CO₂-to-ethylene. Self-driving labs are no longer just producing datasets — they're delivering function-oriented hits that merit scale-up.

Applied Materials' 2 nm Chipmaking Suite — Applied unveiled production tools for Gate-All-Around transistor manufacturing: atomic-level smoothing, angstrom-precision etching, and a Spectral ALD system that swaps tungsten contacts for molybdenum, claiming ~15% reduction in critical contact resistance in internal tests versus previous-generation equipment. These aren't research prototypes; they're capital equipment shipping to fabs now.

Atinary's Self-Driving Labs-R — Atinary launched its Boston-based SDL platform targeting industrial R&D customers. The shift from academic demo to commercial product changes procurement dynamics for any materials team considering automation.

Seprify's Cellulose-Based TiO₂ Replacement — Swiss biomaterials startup Seprify raised €13.4M Series A to scale cellulose microstructures that scatter light like titanium dioxide. Funding targets hundreds-of-tonnes-per-year production — the classic inflection from lab demo to supply-chain-grade material.