Making a MXene in 30 Seconds Does Not Take 30 Seconds

The number is almost too good: under 30 seconds.

A team led by Shichen Xu and James Tour at Rice University has published a method that converts nine different MAX phases into MXenes in tens of seconds.1 It couples extremely rapid Joule heating to gas-phase chlorination and fluorination. A liquid etch that may occupy material for hours or days becomes short enough to fit inside a vertical video.14

The most useful number in the paper may nevertheless be 45.2 hours.

We opened the raw spreadsheet behind Figure 4. In the 10,000 Monte Carlo simulations used for the authors' technoeconomic and life-cycle analysis, the FJH-ClF process has a mean process time of 45.2 hours. Hydrofluoric-acid processing averages 569.1 hours in the same data, while the molten-salt route averages 221.5 hours.2

The two numbers measure different things: a reaction has one clock, while the manufacturing process around it has another, and keeping both visible makes the research much more useful.

Two clocks

A factory does not begin when a reaction starts or stop when the current is switched off.

Feedstock has to be prepared and loaded, atmosphere and reagents controlled, and the material then cooled, removed, separated or washed where required, handled safely, checked and moved into the next operation before the equipment is ready for another run. The Rice paper does not claim that this entire chain takes 30 seconds; its abstract says the nine MAX phases were converted through the selective etching step within 30 seconds each.1

Rice's own release makes the origin of the headline number unusually clear. Xu says a lab mate sent him a paper describing a 12–24 hour MXene etch. He wondered whether his lab's flash Joule heating could compress that waiting time, tried it and, in his account, got a result on the first attempt.4

The achievement is therefore attached to one bottleneck: selective etching of the MAX precursor on the way to a MXene.

A 2024 Nature Protocols paper shows how much sits around that step in a conventional workflow. Its reference procedure for Ti3C2Tx, the best-studied MXene, allocates roughly one week to making the MAX precursor, followed by two days for etching and another two days for delamination.5 The new Rice method leaves most of those surrounding operations in place and attacks the long chemical treatment in the middle.

The second clock is less impressive in a headline and much closer to the industrial question: how long is production capacity occupied for the job?

To be sure, the modeled improvement remains large: 45 hours is about 12.6 times shorter than 569 hours, but taking one operation from a day to a few seconds leaves the rest of the workflow on its own clock.

That distinction matters whenever a scientific result gets translated into a manufacturing claim.

Remove the aluminium

MXenes are a family of two-dimensional transition-metal carbides and nitrides. Their electrical, mechanical and surface properties have made them candidates for energy storage, electromagnetic shielding, electronics, catalysis and functional coatings.1 Many are made by starting from a layered MAX phase and removing the element that separates the useful layers.

Rice explains the structure with a sandwich. In a common example, titanium and carbon form the parts one wants to keep while aluminium sits between them. Turning that MAX phase into a MXene requires removing the aluminium selectively, without wrecking the neighbouring layers, and producing useful surface terminations.4

Selective is the important word.

Excess heat can destroy the structure, an etchant can attack more than the intended element, and insufficient treatment leaves the conversion incomplete. A useful route must therefore balance speed with selectivity, defect control, surface chemistry and repeatability; the stopwatch tells only one part of the result.

The established route often uses hydrofluoric acid, sometimes with HCl or with fluoride chemistry that generates HF in situ. The 2024 protocol explicitly says operators should have experience handling HF.5 Lewis-acid and molten-salt approaches avoid some parts of the conventional wet route but bring their own temperatures, durations, washing steps and chemistry.1

Rice moves the problem into a different workshop. Instead of waiting for a liquid solution to etch the precursor, it couples a very short thermal pulse to controlled gas chemistry.

The change reaches beyond elapsed time.

The flash

Flash Joule heating is deliberately violent. A large electrical pulse passes through a resistive target and turns electrical energy into heat directly inside or around the material, allowing temperature to rise and fall much faster than in a conventional furnace.

A 2025 review of the technique describes temperatures above 3,000 °C, heating rates above 100 °C per second and treatment durations ranging from milliseconds to seconds.6 Direct heating is the useful trick: instead of keeping an entire furnace hot for hours, much more of the input can be concentrated in the material and the brief event that matters.

For the MXene process, Rice describes sequential chlorination and fluorination. The MAX precursor is rapidly heated, then chlorine and tetrafluoromethane, CF4, are used in the gas-phase treatment. The reaction selectively removes the interlayer species, including aluminium in the systems described, while the fluorination step contributes surface termination chemistry.14

The team demonstrates the strategy on nine different MAX phases.1 That breadth is important. A reaction that worked once on one composition would remain an intriguing special case; the paper is trying to establish a more general recipe whose thermodynamic and kinetic conditions can be tuned across a family of materials.

Xu tells Rice that gas composition and exposure time can be adjusted to control how far etching proceeds.4

In production, that control may eventually matter more than the 30-second headline.

Less waiting is useful in any plant. More control over what is removed can also mean fewer rejected batches, easier changes between recipes and a better chance of reproducing the surface chemistry that gave a laboratory sample its properties in the first place.

The evidence currently stops at nine demonstrated MAX phases; an on-demand factory able to make arbitrary MXenes belongs to a later engineering problem.

An abstract already posted for an August 26, 2026 ACS presentation adds one useful scale marker. It specifies 10 seconds of chlorination followed by 20 seconds of fluorination, reports more than 97% aluminium removal from Ti3AlC2, and says the reaction system enabled gram-scale laboratory synthesis within 80 seconds.11 That is stronger evidence than a milligram-scale demonstration, while still describing a laboratory batch rather than continuous throughput, industrial equipment or thousands of qualified production cycles.

The raw numbers

The paper's strongest industrial argument is hidden in a spreadsheet instead of the microscope images.

Figure 4 compares FJH-ClF with HF and molten-salt routes through life-cycle assessment and technoeconomic analysis. The authors release the figure's source workbook and four underlying OpenLCA datasets on Zenodo.23

We extracted the 10,000 Monte Carlo draws for process time, energy and cost and checked them against the consolidated values in Panel 4i.

They turn the story from “a fast reaction” into “a process that still looks advantageous when more of the factory is counted.”

By contrast with the HF case at 652.95 MJ, mean process energy for FJH-ClF is 40.26 MJ, a factor of roughly 16.2 in the authors' model.2

Relative to the HF value of 12.59 dollars, the consolidated FJH-ClF cost measure is 2.61 dollars in the functional unit used by their analysis, roughly 4.8 times lower.2 We are deliberately not turning that into “$2.61 per kilogram”: the public source workbook exposes the figures, but the accessible article text does not give enough of the full system definition for us to restate the exact functional unit without qualification. The ratio itself is explicitly part of the authors' comparison.

Waste falls from 4.6124 to 0.9011 kg, about 5.1 times lower, while the aggregate toxicity indicator reaches about 7% of the HF value, a factor of roughly 14.3.2

And process time, our second clock, falls to 7.94% of the HF case.2

None of those numbers guarantees a future production line will hit the same ratios. They establish a narrower and more useful point: the reported advantage is not created solely by measuring the few seconds during which the sample flashes. It remains when the authors widen their model to additional material, energy and processing flows.

Passing that broader model is a useful hurdle.

A hundredfold faster reaction can be industrially irrelevant if purification then becomes ten times slower, separation consumes the saved energy or the new equipment costs more than all the reactors it replaces. Here, the authors' model suggests that the benefit survives beyond the photogenic moment.

The next test is whether it survives contact with a real production line.

Chemistry remains

Materials announcements have a bad habit of translating “no acid bath” into “clean.”

The source data refuse to cooperate with that simplification.

FJH-ClF does remove much of the long liquid-acid etch, but its published inventory includes chlorine, CF4 and argon.2 In the Figure 4 comparison, the FJH-ClF input list contains about 0.621 kg of Cl2, 2.626 kg of CF4 and 0.956 kg of argon.2

The CDC describes chlorine as a toxic, strongly irritating gas capable of causing serious harm after exposure.8 Rice therefore phrases the safety benefit carefully: moving to gas could reduce risk, and only if the manufacturing process is properly controlled.4

“Properly controlled” is doing useful engineering work there.

A production system needs containment, flow control, leak detection, appropriate scrubbing or capture, handling of reaction products and maintenance procedures that keep a time-saving process from becoming an occupational-safety experiment.

CF4 adds a different concern. The US EPA lists tetrafluoromethane as an extraordinarily persistent perfluorocarbon, with an atmospheric lifetime around 50,000 years and a 100-year global warming potential measured in the thousands of times that of CO2.7

That fact does not prove FJH-ClF has a worse climate footprint. The paper's own life-cycle analysis reaches the opposite overall conclusion for the routes it compares.12 It does mean that industrial scale-up has to keep CF4 emissions tightly controlled. “Gas instead of acid” is a change in process architecture, not a synonym for harmless chemistry.

The waste sheet makes the same point from another direction. FJH-ClF totals 0.9011 kg of waste in the comparison and lists roughly 0.374 kg HCl and 0.206 kg HF among the outputs.2 The route substantially lowers total waste and the modeled toxicity indicator, yet fluorine and acidic species have not vanished from the universe.

Publishing the LCA source workbook makes this accounting inspectable and makes the result much harder to compress into a single green adjective.

The scale-up trap

“Scalable” may be the most dangerous word in materials science after “simple.”

Depending on the paper, it can mean that physics permits a larger machine, that a lab has repeated the process on larger samples, or that an equipment architecture exists for continuous operation; occasionally it means little more than an optimistic adjective in the abstract.

The MXene paper describes FJH-ClF as scalable.1 By contrast, a 2025 review of flash Joule heating places much of FJH at bench scale and says that, although industrial scale-up is beginning for selected applications, equipment and engineering development are still required.6

Rapid heating creates its own scale problem.

In a small sample, current and temperature can be controlled across a limited volume. Increase the mass and electrical paths, thermal gradients, electrode contacts and gas flow all become harder to keep uniform. That matters enormously for a process whose selling point is selective chemistry controlled through temperature and exposure time. A reactor cannot be enlarged like a picture in a slide deck and retain the same gradients out of politeness.

The gas phase creates another layer. A small cell can be given a controlled pulse of chlorine and CF4. A production line must supply those gases, distribute them consistently, recover or destroy effluents, prove leak-tight operation and show that Monday's MXene resembles Friday's MXene after thousands of cycles.

Then there is everything before and after etching.

A 2024 Scientific Reports study applying a Safe and Sustainable by Design framework to Ti3C2Tx notes that long-term hazard data for MXenes remain incomplete and identifies precursor production, including titanium-related inputs, as an important part of the life-cycle burden.9 The result is a reminder that faster etching does not solve titanium extraction, MAX synthesis, later delamination, dispersion into a product or eventual end-of-life.

Thirty seconds already has enough work to do.

Flash Joule heating also has a commercial context: the 2025 review says several FJH processes and apparatus are Rice intellectual property, lists authors including Tour on patents or patent applications across the wider portfolio, and notes that some FJH technologies have been licensed to companies.6 The 2026 MXene paper declares no competing interests.1; those statements describe different scopes and can both be true. It simply means that industrialization may eventually involve both chemistry and licensing, as industrialization has an irritating tendency to do.

What actually changes

To be sure, the 30-second result still deserves attention.

Taking an etch from 12–24 hours to under half a minute is a profound change in residence time.4 Even when the entire process remains measured in tens of hours, removing a long chemical hold can shrink equipment requirements, free reactor capacity, reduce energy tied up in waiting and make recipe changes much less expensive.

At that point a process innovation starts changing the shape of the factory, not merely its stopwatch reading.

When etching occupies a reactor for two days, production volume is tied directly to reactor count and vessel size. Long liquid treatments also mean more material sitting in intermediate states, more chemical inventory and slower feedback when a recipe is wrong. A reaction that lasts seconds moves the bottleneck somewhere else: loading, gas handling, cooling, characterization, delamination or quality assurance.

The total process is not instant. The industrial problem has moved.

That can matter more than the raw acceleration factor.

Slow machinery encourages parallel machinery. Fast chemistry can expose that handling or inspection is now the limiting operation. Long residence times encourage large batches and intermediate inventory; short active treatment may make smaller batches and more recipe variation plausible, provided the surrounding system can keep up.

Nature's August 20 Research Briefing summarizes the work as converting MAX phases into MXenes in tens of seconds while reducing acid use, energy input and chemical waste.10 The summary fits the paper; adding the second clock turns it from a speed record into a manufacturing story.

Rice has not built a “30-second MXene factory.”

It has demonstrated a process in which the chemical step that used to justify hours of waiting abruptly stops being the slow piece. The authors' own source data suggest that this change still produces major reductions in time, energy, cost, waste and modeled toxicity when a wider process boundary is considered.2

Now comes the less cinematic work: turning a laboratory cell that can flash nine precursors into repeatable equipment that handles its gases safely, controls temperature across useful throughput and produces the same material often enough for somebody else to design a product around it.

If that works, the 30 seconds will have done exactly what a good process breakthrough should.

Not make the factory instantaneous.

Make the rest of the factory become the new problem.