Methane is the stubborn end of the hydrocarbon family: a C–H bond worth 105 kcal mol⁻¹, ionization potential around 12.5 eV, pKa about 50–51. It is burned because it is cheap and unreactive. A five-person team at the Centre for Research in Biological Chemistry and Molecular Materials (CiQUS), Universidade de Santiago de Compostela, has now used that carbon anyway. They hung an allyl handle on methane with iron, collidine and an LED lamp, then walked the product to dimestrol, a nonsteroidal estrogen used in hormone therapy, in 20% overall yield.

The paper is Álvarez-Constantino, Martínez-Balart, Barbeira-Arán, Velasco-Rubio and Fañanás-Mastral, “Attenuated LMCT photocatalysis enables C─H allylation of methane and other gaseous alkanes,” Science Advances 11, eaea0783, published 7 November 2025 (received 25 June, accepted 2 October). DOI 10.1126/sciadv.aea0783. The first three authors share equal credit; Martín Fañanás-Mastral is corresponding. CiQUS posted the lab note on 13 November 2025; Phys.org carried it on 14 November. The chemistry is November 2025, flask-scale — not a 2026 discovery and not a medicines factory.

What happened

The missing reaction was C–H allylation of gaseous alkanes — attaching a simple allyl group so the gas becomes a handle a chemist can actually use. Most photocatalytic work on methane and ethane has gone after highly activated Michael acceptors. Those products are limited. Allylic chlorides are less reactive, and that mismatch is fatal when the photocatalyst is an iron chloride: the alkyl radical is chlorinated, the catalyst dies as Fe(II), and you get almost nothing.

Table 1 is the whole problem in one line. Ethane plus allylic chloride 1, 8 bar, acetonitrile, room temperature, 43 W, 370-nm Kessil LED, 16 hours: tetrabutylammonium decatungstate gave no product; a copper/decatungstate system the same lab had used on liquid alkanes also gave none; cerium and titanium chlorides gave traces; FeCl₃·6H₂O alone gave 5% isolated product 2. Add 2 equivalents of 2,4,6-collidine to 10 mol % FeCl₃·6H₂O and the isolated yield is 70% (99% based on recovered starting material). On a 3-mmol run with two lamps, the same product still came in 45% isolated (56% brsm).

That collidine is not a garnish. The authors isolate a crystal (Cambridge deposition 2418394) of a distorted tetrahedral [FeCl₄]⁻ anion wrapped in collidinium cations. Hydrogen bonds from collidine methyl groups to chlorine sit at 2.360 Å and 2.398 Å. The LMCT event that makes a chlorine radical is concentrated on a short axial Fe–Cl bond (1.957 Å). Photolysis without collidine makes Cl• about an order of magnitude faster. In a cyclohexane control, chlorination without collidine hits 70% chlorocyclohexane in 15 minutes (quantum yield Φ = 2.16, a chain); with collidine the same window gives 9%. Allylation versus chlorination flips from 44:56 (8% allylation) to 93:7 (45% allylation). The allylation quantum yield is Φ = 0.026 — about 36 photons per turnover — a closed photoredox cycle, not a chain.

Figure 2 is the part a synthetic chemist will actually use. Ethane at 8 bar gives products 2–15; electron-poor aromatics 5 and 6 are 99%. Propane at 1 bar gives 16–31 in good yields with low regioselectivity. Isobutane at 1 bar gives 32–35 in 64–91%. Methane at 50 bar, in a Parr vessel with a sapphire window, gives 36–43 at 0.1 mmol in CD₃CN. The best isolated methane yield in that set is product 38 at 75% (99% brsm); the boronic ester 43 is 19%. Scaled to 0.5 mmol in ordinary acetonitrile under a 70 W 370-nm LED, 38 still came in 48%, with solvent-activation product <5%. In a 100-ml reactor, ordinary acetonitrile works.

Then they used the handle. Figure 3 is the claim that leaked into headlines:

  1. Methane → photocatalytic allylation → photoexcited-nitroarene oxidative cleavage → propiophenone 44, 60% overall. That ketone is a documented intermediate toward d-propoxyphene, phenylpropanolamine and phenmetrazine. It is not those drugs.
  2. The same logic on a methoxy-substituted allylic chloride, then a McMurry coupling: allylation 31%, oxidation 76%, McMurry 84% → dimestrol 45, 20% overall from methane. The authors write, verbatim, that this is, “to the best of our knowledge, the first synthesis of a bioactive compound using methane as starting material.” Dimestrol itself is an old estrogen, not a new medicine.
  3. Ethane, 3 mmol, 8 bar → hydroboration → oxidation → 2-phenylpentanoic acid 46, 31% overall (step yields 45 / 68 / 45%). That acid is a listed intermediate for α-amino acids, BET-protein inhibitors and STING inhibitors — an intermediate, not a clinical candidate from this flask.

Funding is named: ERC CoG 863914-BECAME, Spain’s Agencia Estatal de Investigación, Xunta de Galicia, and ERDF.

Fañanás, in the CiQUS note: the core is a tetrachloroferrate anion stabilized by collidinium cations, “which effectively modulates the reactivity of the radical species generated in the reaction medium.” That sentence matches the crystal and the kinetics. The press line that the method “operates under mild temperature and pressure” is only half true. Temperature is room temperature. Ethane is 8 bar and propane 1 bar. Methane is 50 bar. Collidine is used at 2 equivalents, not catalytic loading. Those are the numbers; they are still a long way from a steam cracker, and they are not ambient methane-to-pills.

Why it matters

The interesting object is not “medicine from natural gas.” It is a kinetically compatible iron photocatalyst that lets an unactivated allylic chloride outrun chlorination on the lightest alkanes. That is a design rule — hydrogen-bond attenuation of Fe–Cl LMCT and radical ligand transfer — that other hydrogen-atom-transfer reactions on light alkanes can steal. The paper says so in the last paragraph.

If the rule holds, natural-gas carbon becomes a legal starting material for molecules that today start from petroleum-derived olefins. Propiophenone and dimestrol are demonstrations, not a pipeline. The iron is cheap and the lamp is an LED; the methane still has to be compressed to 50 bar and processed at 0.1–0.5 mmol. Climate copy will overreach: a lab’s worth of methane at 20% yield of dimestrol does not move the atmosphere. It does show that a gas people already extract can, in principle, be a carbon source instead of a fuel. “First bioactive from methane” is the authors’ literature claim, hedged with “to the best of our knowledge.” Repeat the hedge.

What to watch next

  1. Scale of the methane step. Product 38 at 0.5 mmol and 48% is the current ceiling in the paper. A flow reactor, or a larger batch that keeps ordinary acetonitrile, is the first industrial signal. Until then, 50 bar and 0.1 mmol is the fact.
  2. Catalytic collidine. Two equivalents of a pyridine is a stoichiometric organic additive sitting on top of 10 mol % iron. If a later paper keeps the 93:7 allylation/chlorination ratio with catalytic collidine plus an inorganic base (they already saw K₂HPO₄ plus 20 mol % collidine help ethane, at slightly lower yield), the method gets cheaper.
  3. Something besides dimestrol. The McMurry finish is a classic, harsh, titanium–zinc coupling. A bioactive made from methane and finished under the same mild photocatalytic conditions would be the real second paper. The acylation work in Cell Reports Physical Science is a different C–C bond; watch whether those two methods start sharing intermediates.
  4. Do not write “methane-to-medicine plant.” Dimestrol at 20% overall from 0.1 mmol of allylic chloride is a landmark for methane functionalization. It is not a supply-chain event. Treat aggregator headlines that skip the yield, the 50 bar, and the flask scale as a miss.

Sources

  1. Andrés M. Álvarez-Constantino, Pol Martínez-Balart, Sergio Barbeira-Arán, Álvaro Velasco-Rubio, Martín Fañanás-Mastral, “Attenuated LMCT photocatalysis enables C─H allylation of methane and other gaseous alkanes,” Science Advances 11, eaea0783 (7 November 2025). DOI: 10.1126/sciadv.aea0783 — https://www.science.org/doi/10.1126/sciadv.aea0783
  2. Open-access PMC copy (PMC12594199) — https://pmc.ncbi.nlm.nih.gov/articles/PMC12594199/
  3. CiQUS / USC, “Breakthrough Catalyst Turns Methane into Bioactive Compounds for the First Time,” 13 November 2025 — https://ciqus.usc.gal/en/news/breakthrough-catalyst-turns-methane-bioactive-compounds-first-time
  4. Phys.org, “Catalyst turns methane into bioactive compounds for the first time,” 14 November 2025 — https://phys.org/news/2025-11-catalyst-methane-bioactive-compounds.html
  5. Related, same group: Nair, Barbeira-Arán, Malga, Fañanás-Mastral, Cell Reports Physical Science 6, 102912 (2025). DOI: 10.1016/j.xcrp.2025.102912