The Catalyst — Apr 21, 2026

This is a substantive week, and the throughline is control — over interfaces, over reaction pathways, over what chemistry is even allowed to reach the market. A Nature Materials paper puts atomically thin hard masks into the 3D chip etch stack, the EU's replaceable-battery rule is forcing a format rethink that lands squarely on solid-state chemistry, and a Nature Physics result on twisted bilayer graphene flips the sign of what electron interactions are supposed to do to superconductivity. Meanwhile the boring-but-binding infrastructure — EPA's PFAS reporting window, TSCA new-chemical queue depth, Dow's delayed Path2Zero cracker — is doing quiet work that determines which of this week's lab wins ever see a reactor.

• Ultrathin van der Waals hard masks (Nature Materials, April 16): metal oxyhalide sheets that survive plasma etch at thicknesses no conventional mask can match.
• SynAsk (RSC Chemical Science): a chemistry LLM constrained by curated reaction databases, designed to refuse thermodynamically implausible outputs.
• Argonne's autonomous "boss" agent (Argonne National Laboratory): a closed-loop AI manager running polymer discovery end-to-end with the human tether substantially cut.
• Cronin group chemputation + LLM stack (Communications Chemistry, April 3): literature-auditing layer that evaluates whether a published procedure is actually executable by a robot.
• EPA interim PFAS destruction and disposal guidance: updated pilot-data framework for pyrolysis, mechanochemical destruction, and supercritical water oxidation.

As the industry pushes toward 2 nm and gate-all-around architectures, the mask becomes the limit: too thick and you lose feature fidelity; too thin and plasma eats it. A Nature Materials paper published April 16 proposes a way out — ultrathin van der Waals metal oxyhalides used as hard masks for 3D semiconductor etching. Because van der Waals compounds are held together by weak interlayer forces, they can be exfoliated to near-atomic thickness while still behaving as a coherent solid, and the reported oxyhalides are unusually plasma-resistant with a measurable smoothing effect during pattern transfer.

If this clears fab qualification, the winners are whoever can deposit these materials at wafer scale reproducibly — and the losers are the process teams that have been brute-forcing plasma chemistries to compensate for inadequate masking. The signal to watch: whether specific oxyhalide compositions survive the harshest fluorine-based etch chemistries at TSMC or Samsung, and whether anyone demonstrates conformal deposition over 3D topography. If follow-ups from the same group remain lab-scale by autumn, this becomes a beautiful paper rather than a supply-chain event.

The EU's requirement that all phones and tablets sold in the bloc feature user-replaceable batteries by 2027 is being covered as a consumer-rights story. The materials story underneath it is larger. Modern pouch cells depend on the chassis for mechanical compression — delamination, dendrite growth, and cycle life are all managed by the rigid phone frame. Remove the glue and the potting, and you've broken the engineering assumption that keeps the pouch format viable.

What changes if this succeeds: the center of gravity in consumer electronics shifts toward prismatic or cylindrical formats, or toward solid-state and semi-solid polymer electrolytes that are inherently more robust to handling. A complementary Nature Materials thread this fortnight goes further, using reductive electrophiles to deliberately form dense, ion-conducting passivation layers at the solid-electrolyte interface — the chemistry self-assembles the shield you need. Watch CENELEC's technical standards work: the definition of "replaceable" will determine whether binder chemistry and electrolyte formulation get 18 months or 18 weeks to adapt. If the standards land loose, pouch cells survive in modified form. If they land tight, solid-state pulls forward by years.

Xueshi Gao and Chun Ning Lau at Ohio State, working with Francisco Guinea's group, published a Nature Physics result on April 7 that inverts the conventional BCS intuition. Placing twisted bilayer graphene on a strontium titanate substrate let them tune electron–electron interactions in real time. When they increased those interactions, superconductivity decreased — the same forces that help electrons pair also compete against the superconducting state. In a conventional superconductor, weakening repulsion helps pairing; in magic-angle graphene, the relationship is double-edged and no simple model predicts it.

If this mechanistic picture holds up in independent replications, the design rules for moiré superconductors get a real foundation rather than a phenomenological one — and electrostatic control becomes a clean experimental knob for separating pairing from screening. The OSU team frames this as an early step; the signal to watch is whether a second group reproduces the interaction-tuning result, and whether the triplet-pairing interpretation circulating elsewhere in the field gets direct evidence. Room-temperature superconductivity remains distant, but the field is finally converging on mechanism.

The C–F bond sits near 130 kcal/mol, which is why PFAS destruction has historically meant incineration above 1000°C or harsh electrochemistry. A Nature paper this month reports an organic photoredox catalyst that cleaves C–F bonds sequentially at 40–60°C under visible light, recovering fluorine as inorganic fluoride that can in principle be recycled into new fluorochemicals. The radical anions generated by the photocatalyst reduce perfluoroalkyl substrates across a broad scope, and the authors demonstrate both hydrodefluorination and cross-coupling from the resulting carbon-centered radicals.

The temperature is the story: this could run on waste heat or solar thermal. If it works in real contaminated-water matrices — where competing oxidants, metals, and dissolved organics typically poison photocatalysts — remediation economics shift. The signal to watch is pilot-scale or flow-reactor data within the next two quarters. Regulatory context matters here too: the EPA moved the TSCA PFAS reporting window while revising the rule architecture, a shift that reads less as leniency than as the agency demanding cleaner data before it acts. And the EPA's updated interim destruction guidance makes clear that >99% destruction efficiency is not sufficient on its own — non-target analysis and full mass balance are now the bar. PFAS is becoming a validation problem as much as a bond-breaking one.

A decade of perovskite solar cell research has treated defects as the enemy, with passivation chemistry as the primary lever for efficiency. New imaging work reported around April 10 maps how charge carriers actually move through real perovskite films and finds that certain defect networks act as charge highways rather than traps — funneling electrons and holes apart before they can recombine. Some of the efficiency gains previously attributed to passivation may have been working through a different mechanism entirely.

A separate Nature Materials paper on April 6 attacks the failure mode from the other side: a bulk nano-heterointerface that secures molecular contacts and stabilizes the device boundary. Together these point to two design axes — intentional defect networks as internal charge guides, and deliberate contact chemistry that survives humidity and bias stress. For Oxford PV, Saule, and anyone else moving perovskites toward pilot production, the question is whether engineered defect networks reproduce across large-area films and survive accelerated aging. Reproducibility under stress is now the binding constraint, not peak efficiency.

Radicals react faster than chiral environments can usually steer them — that's been the conventional wisdom. A Nature paper this month reports enantioselective three-component radical cross-coupling using a thiamine-dependent enzyme paired with a photoredox catalyst. The enzyme's active site acts as a stereochemical cage around a photogenerated radical intermediate, delivering ketone products with exceptional stereoselectivity. A related JACS paper pairs Rose Bengal with pyridoxal-dependent threonine aldolases to alkylate unprotected alanine and glycine in the enzyme pocket, producing alpha-tertiary amino acids in one step.

Three-component couplings build molecular complexity fast, but the selectivity problem has kept them out of medicinal chemistry workflows. Using an enzyme pocket to confine radical generation is the kind of conceptual move that tends to generate a decade of follow-on work — watch for this to become a named reaction family within two years. Substrate scope will decide whether it's a general method or a beautiful proof of concept.

A Penn State–led team used high-resolution imaging to watch what happens to a vanadium dioxide substrate when an electrical pulse hits a device built on top of it. VO₂ undergoes a metal–insulator transition in a trillionth of a second, and the researchers discovered the substrate itself was locally undergoing that transition in response to the field from the device above — the substrate is participating in the circuit, not hosting it. This explains years of anomalous data where VO₂ devices outperformed or underperformed their models.

If substrate-as-active-element becomes a design principle, chip architects gain a new logic layer for 3D integration — and lose an assumption that has simplified device modeling for decades. The signal to watch is whether other phase-change oxides (NbO₂, Ti₂O₃) show the same field-coupled substrate response under operando imaging.

A van der Waals sheet thin enough to be a rumor holds its shape under plasma; a cell phone battery politely asks to be taken out for the evening; a graphene stack superconducts by spite. The quiet news is that EPA's filing cabinet may be a better leading indicator for industrial chemistry than any preprint server — which should concern anyone whose thesis depends on the next big paper rather than the next 148 exemption applications. Back next week.

Forward this to the labmate who still thinks the substrate is just holding the device up.