Tungsten lets iron keep 93% of hydrogen capacity, not 11%

ENERGY AND GRID

Tungsten lets iron keep 93% of hydrogen capacity, not 11%

Researchers report an iron-tungsten powder that kept 93% of its hydrogen storage capacity over 30 cycles, against 11% for plain iron cycled the same way. The authors point to hydrogen users that need bulk storage, such as ammonia and methanol production, fuel refining, green steelmaking and sustainable aviation fuel. Iron stores hydrogen chemically: hydrogen strips the oxygen from iron oxide, leaving iron, and steam later turns the iron back into oxide, releasing the hydrogen. The obstacle has been sintering – hot particles fusing into a dense mass that gas cannot get into. The authors say earlier fixes, such as adding inert oxides between particles, slowed the loss without stopping it, and added weight. The authors are from the Hong Kong University of Science and Technology (Guangzhou) and Northwestern University.

The team blended iron oxide with tungsten oxide and reduced it with hydrogen at 800 °C, giving an alloy with 19 atomic percent tungsten. A 1.5 kg bed in a custom reactor, cycled between hydrogen and steam at 650–800 °C, stored and released 43.8 grams of hydrogen per cycle and stayed porous after 30 cycles. A plain iron bed cycled the same way packed into a dense block. Partial cycles, filling the bed to 25%, 50% and 75% of capacity 30 times each, left it able to recover nearly full capacity afterward.

The authors credit two effects. During cycling, tungsten vapor settles on the iron surfaces and opens up pores, so the powder swells rather than packs down – the "self-foaming" of the title. During idle periods at high temperature, tungsten-rich particles sit between the iron grains and keep them from merging.

The limits. The tests cover only 30 full cycles and up to 90 partial ones, short of the thousands a multi-year installation would need. The authors' estimate that tungsten content would fall from 19 to 10 atomic percent after about 3,000 cycles is a calculation, not a measurement. Scale-up to the 100 kg or tonne-scale modules they describe is proposed, not tested. Hydrogen stored per kilogram remains below that of liquid ammonia. After cycle 30, the powder smoldered when exposed to air at room temperature, a handling risk the authors note but do not address.

Read the paper: "Self-foaming sintering-resistant iron-tungsten powders enable high-cycle thermochemical hydrogen storage," Nature Communications (2026). DOI: 10.1038/s41467-026-78337-8 · Free PDF

MATERIALS AND MANUFACTURING

3D-printed copper composite survives over 100,000 fatigue cycles at 95% of its breaking stress

Researchers report a 3D-printed copper composite that survived more than 100,000 loading cycles at 95% of its breaking strength without failing. Fatigue – damage that builds up under repeated loading until a part cracks – limits how hard structural metals can be worked. The composite's ratcheting, the slow stretching a metal accumulates cycle by cycle, grew 10 to 1,000 times more slowly than in any known metal or alloy, the authors say. They report the highest fatigue ratio (peak cyclic stress as a share of breaking strength) yet for copper-based materials, ahead of copper strengthened by finer or graded grain structures. The authors are from Shanghai Jiao Tong University, the University of California, Los Angeles, and Xi'an Jiaotong University.

The team printed copper powder mixed with 4% boron carbide by volume, using a laser that melts the powder layer by layer (laser powder bed fusion). The fast melting and cooling broke the ceramic into amorphous nanoparticles – glassy, with no crystal structure. The printed composite had a tensile strength of 552 MPa, about 80% above printed pure copper at 310 MPa. Fatigue tests pulled the samples at 70–95% of breaking strength and stopped at fracture or at 10 million cycles.

A comparison batch, heated until the boron carbide particles crystallized, lasted 10 to 100 times fewer cycles at the same loads. The authors' explanation turns on dislocations, the line defects that pile up in a metal under load and eventually start cracks. Here the glassy particles absorb dislocations instead of letting them accumulate. X-ray measurements showed almost no rise in dislocation density during cycling, and simulations showed dislocations entering the particles and largely undoing themselves when the load came off.

The limits. All testing was at room temperature, with no data at high temperature or in corrosive conditions. Only one composition was tested, as small lab samples rather than structural parts. The paper gives no cost or scale-up figures, and no data on crack growth or fracture toughness. The simulations used simplified models at small scales.

Read the paper: "Decoupling Stable Cyclic Deformation From Defect Accumulation in Metals," Advanced Science (2026). DOI: 10.1002/advs.78331 · Free PDF

ENERGY AND GRID

Electrolyzer membrane holds voltage decay to 9 µV/h over 5,400 hours of operation

Researchers report a membrane for hydrogen electrolyzers whose operating voltage rose by only 9 microvolts per hour (µV/h) over 5,400 hours at 80 °C. A slowly rising voltage is the usual sign of a membrane wearing out, and membrane life is a limit on anion exchange membrane (AEM) electrolyzers, a lower-cost route to green hydrogen. In a harder test, at twice the current for 2,500 hours, the decay rate was 25 µV/h, against 43 µV/h for the commercial PiperION A60 membrane under the same conditions. The authors are from Southern University of Science and Technology, Shenzhen Polytechnic University and BriHyNergy (Shenzhen) Co., Ltd., in China.

The membrane separates the two halves of the cell while letting hydroxide ions through. The authors' membranes, a series they call QAI-x, attach the charged imidazolium groups to the plastic backbone through a benzene-ring tether. Earlier designs used a link that hydroxide attacks. The 9 µV/h figure came from the best of the series, in 1 M potassium hydroxide at 0.5 A/cm². In stronger alkali, 5 M, the voltage rose 16 µV/h over 2,000 hours. In 10 M, it rose 180 µV/h for 700 hours before speeding up, against 3,300 µV/h for PiperION A60.

A separate test soaked samples for 120 hours at 80 °C in hydrogen peroxide and iron (Fenton's reagent), a standard check for the chemical attack membranes face near the catalyst. The new membrane lost no measurable mass and kept over 90% of its tensile strength. PiperION A60 broke apart.

The limits. The tests ran in a single lab cell, with no multi-cell stack or field trial. In 10 M alkali, degradation kept accelerating after about 700 hours, so that figure is not a stable rate. The explanation of why the tether resists attack rests on simulation alone. A brief voltage drop at about 3,817 hours in the 5,400-hour test is attributed only tentatively to a change in contact resistance. Several authors work for BriHyNergy (Shenzhen) Co., Ltd., a company involved in the work.

Read the paper: "Aryl‐Tethered Pendant Sterically Shielded Imidazoliums for Durable Anion Exchange Membranes," Advanced Science (2026). DOI: 10.1002/advs.77767 · Free PDF

ENERGY AND GRID

Methanol rinse lifts organic solar cells to 20.65% certified efficiency, from about 19.5%

Researchers report a methanol rinse that raised the certified efficiency of a lab organic solar cell to 20.65%. The standard version of the same cell typically reaches 19.0–19.5%, the authors say. Organic solar cells use carbon-based light absorbers that can be laid on light, flexible film, and the authors present them as a lower-cost alternative to silicon. The rinse works on the cathode interlayer, the thin layer beneath the metal electrode that controls how much of the generated charge is collected as current. The authors are from Beijing University of Chemical Technology, the Chinese Academy of Sciences, Soochow University, Linköping University and Jilin University.

The method rinses the light-absorbing layer with methanol, then lays down a new interlayer material from a second solvent. The light absorber was a binary blend, meaning two materials, the simplest recipe. The best cell reached 20.72%, and 20.65% in independent certification. Its fill factor – the share of an ideal cell's power it actually delivered – was 82.7%. These were tiny cells, 0.0256 cm² in area. A 1 cm² cell reached 19.20%.

The authors say leftover methanol lines up the interlayer molecules to face the electrode. That strengthens the electric field that pulls charge out of the cell (0.82 V against 0.65 V untreated) and cuts the cell's internal resistance about five-fold. The rinse also helped two other material combinations.

The limits. The certified figure comes from the smallest lab cells, with no module-scale demonstration. In indoor lifetime testing, treated cells took 984 hours to fall to 80% of their starting output, against 330 hours untreated. No outdoor test was run. The account of how the molecules line up rests on spectroscopy and computer modeling. Only a handful of material combinations were tried.

Read the paper: "Directed assembly of cathode interlayers enables efficient binary organic solar cells," Nature Energy (2026). DOI: 10.1038/s41560-026-02148-w

ENERGY AND GRID

Recycled lithium metal electrode outlasts commercial foil, retaining 91.4% capacity over 400 cycles

Researchers report turning lithium from spent lithium iron phosphate (LFP) batteries directly into thin lithium-metal electrodes, recovering 98% of the lithium. In test cells, the recycled electrode kept 91.4% of its capacity over 400 cycles. Cells with commercial lithium foil faded after about 120. The authors point to the growing number of electric-vehicle LFP batteries reaching end of life, whose cathode material holds more lithium than mined ore. Most recycling today melts or dissolves the material and yields lithium salts that need further refining before reuse, they say. The authors are from the Ningbo Institute of Materials Technology and Engineering of the Chinese Academy of Sciences, Ningbo University of Technology, the University of Chinese Academy of Sciences, Fudan University and Hefei Gotion High-Tech Power Energy.

The process uses a redox mediator, a dissolved chemical shuttle the authors call TMAPO. It pulls lithium ions out of the spent powder and is regenerated at an electrode, while the lithium plates onto copper foil as a layer about 50 µm thick. Tests used simulated spent powder at 10–40 grams per liter.

The test cells paired the recycled lithium with a high-nickel cathode (NCM811) and little electrolyte, as in high-energy designs. Larger pouch cells kept 88.8% of capacity after 400 cycles, against about 110 cycles for commercial foil. The authors credit a protective surface layer formed during plating, which blocks moisture and limits reaction with the electrolyte, and a dense deposit that copes with swelling better than foil.

The limits. The headline results come from simulated spent powder. Real spent material was tried only in side experiments, though those recovered 97.42% of the lithium. Real material also needs pretreatment, since the process cannot fully reject binders and other metals. The mediator was shown stable over 8 days, not industrial durations. Testing stopped at coin and pouch cells. The authors call their cost analysis preliminary. One co-author works for Hefei Gotion High-Tech Power Energy, a battery maker.

Read the paper: "Closed-loop upcycling of residual lithium in LiFePO4 into tunable thin lithium metal electrodes," Nature Communications (2026). DOI: 10.1038/s41467-026-78696-2

INSTRUMENTATION AND MEASUREMENT

Lookup-table fix cuts phase error in shiny-surface inspection by up to 255x

Researchers report a one-time calibration that cut phase errors by up to 255-fold in deflectometry, a method for measuring the shape of shiny surfaces. The authors point to inspection of mobile phone cover glass, display panels and automotive paint, where blur, a screen's uneven brightness response and overexposure corrupt the measurement. Errors in the reconstructed 3D surface fell by up to 18-fold, to about a hundred-thousandth of a millimeter. The authors say most existing corrections handle one source of error at a time and break down when several occur together. The authors are from the University of Electronic Science and Technology of China.

Deflectometry shows striped patterns on a screen and films their reflection in the part. The way the stripes bend reveals the shape of the surface. Each camera reading is converted to a phase, its position within the stripe pattern, so an error in the phase becomes an error in the measured shape.

The authors found that blur, gamma distortion (the screen's uneven brightness response) and overexposure all bend the link between the computed phase and the screen's pixel positions. That link depends only on the screen. So it can be measured once at calibration and stored as a lookup table, instead of modeling each distortion separately.

They tested it on a lab concave mirror under four conditions: blur alone, and blur combined with gamma distortion, with overexposure, and with both. Phase errors fell by 99.1–99.6% and 3D surface errors by 84.5–94.5%, judged against a ZYGO interferometer measurement of the same mirror. Even after correction, the measured curvature of the mirror differed from the reference by 3.1–6.5%.

The limits. The method was tested on one lab mirror, not on manufactured parts such as phone glass, display panels or car paint. All gains are measured against the authors' own uncorrected setup, not against existing correction methods or a commercial deflectometry system. The four distortions were induced in the lab on that one object. Screens made of LED arrays, which lack a pixel structure, are left for future work.

Read the paper: "Generic phase correction method based on phase recoding for deflectometry," Optics Express (2026). DOI: 10.1364/oe.608523

Reply with corrections, tips or leads. If this was useful, forward it to someone who’d want it. Subscribe at clevernewstuff.com.