Corrosion-resistant super steel for high-potential electrolysis
Researchers at the University of Hong Kong have developed SS-H2, a stainless steel that resists corrosion at electrical potentials up to 1700 mV in chloride-rich environments, matching titanium performance in saltwater electrolyzers while costing far less. Published findings and patent progress highlight its sequential dual-passivation mechanism, which forms a manganese-based protective layer atop the conventional chromium oxide film. This stands out as a materials advance that could cut structural component costs in electrolysis systems by roughly 40 times.
Conventional stainless steels fail via transpassive corrosion around 1000 mV because chromium oxide converts to soluble species, well below the ~1600 mV needed for water oxidation. Manganese was long viewed as detrimental to corrosion resistance, yet atomic-level data confirmed its unexpected role in the second passivation layer starting near 720 mV. The work builds on the team’s prior super-steel projects and has advanced to ton-scale wire production.
Key uncertainties include scaling laboratory alloys into practical meshes and foams that maintain performance under industrial electrolyzer conditions, plus long-term durability in real seawater systems. Current corrosion science cannot fully explain the manganese mechanism, leaving open questions about broader alloy design rules at high potentials.
Sources: ScienceDaily (University of Hong Kong), Materials Today.
Pre-Big Bang black holes as dark matter candidates
A University of Portsmouth study proposes that black holes formed in a prior contracting cosmic phase could have survived a bounce into our expanding Universe, potentially accounting for dark matter and early massive objects seen by JWST. The model treats the Big Bang not as a singularity but as a quantum-pressure-driven reversal from contraction, allowing compact relics larger than about 90 meters to persist as cosmic fossils. This offers a unified angle on inflation-like expansion, dark matter, and unexpectedly early structure formation.
Standard cosmology traces the Universe to a hot dense state 13.8 billion years ago but leaves the initial singularity, inflation trigger, and dark matter identity unresolved. Bounce scenarios replace the singularity with finite-density quantum effects analogous to those stabilizing neutron stars, generating density fluctuations that seed both post-bounce black holes and surviving pre-bounce ones. Relic gravitational waves or CMB imprints could serve as observational tests.
Tensions remain around whether enough such black holes form to constitute all dark matter and how precisely the bounce reproduces observed uniformity and acceleration. The framework is theoretical; distinguishing bounce relics from standard primordial black holes requires future gravitational-wave and CMB data that may not yet exist at sufficient sensitivity.
Sources: ScienceDaily (University of Portsmouth), Physical Review D.
Brain reasoning operates independently of language
MIT neuroscientists showed that people with severe aphasia from stroke damage solve complex logic puzzles—rule induction on number lists and visual matrices—as accurately as controls, while fMRI in healthy adults reveals language regions stay silent during both inductive and deductive reasoning. The multiple-demand network activates for rule discovery but not syllogisms, indicating distinct neural systems. The result, published in PNAS, directly tests long-standing claims that language is required for abstract thought.
Philosophers and cognitive scientists have long linked hierarchical structure in language to logical propositions. Prior work already separated language from object categorization and social reasoning; this study isolates formal logic using non-verbal tasks and patients whose language networks are compromised. Participants conveyed discovered rules via gesture or drawing, confirming preserved competence.
Uncertainties center on the precise circuits for pure deduction, which did not recruit the expected multiple-demand network, and on how these findings scale to more naturalistic or multi-step reasoning. Implications for AI architectures that entangle language and logic remain exploratory rather than prescriptive.
Sources: ScienceDaily (MIT), PNAS.