Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: chemistry – Page 12

Mars dust storms are sparking electricity and rewriting the planet’s chemistry

Mars may look like a quiet, dusty world, but it’s actually buzzing with hidden electrical activity. Powerful dust storms and swirling dust devils generate static electricity strong enough to spark faint glowing discharges across the planet, triggering chemical reactions that reshape its surface and atmosphere. Scientists have now shown that these tiny lightning-like events can create a surprising mix of chemicals—including chlorine compounds and carbonates—and even leave behind distinct isotopic “fingerprints.”

Mars is often portrayed as a dry, lifeless desert, but it is far more active than it appears. Its thin atmosphere and dusty terrain create an environment where constant motion generates electrical energy. Dust storms and spinning dust devils sweep across the surface, continually reshaping the landscape and driving processes that scientists are only beginning to fully understand.

Planetary scientist Alian Wang has been studying this phenomenon in depth. In a series of studies, including recent work published in Earth and Planetary Science Letters, she has examined how these electrically charged dust activities influence the chemistry of Mars, particularly through their impact on isotopes.

Solar reactor uses old battery acid to turn plastic waste into clean hydrogen

Researchers have developed a solar-powered reactor to break down hard-to-recycle forms of plastic waste—such as drink bottles, nylon textiles and polyurethane foams—using acid recovered from old car batteries, and converting it into clean hydrogen fuel and valuable industrial chemicals. The results are reported in the journal Joule.

The reactor, developed by researchers from the University of Cambridge, is powered by the energy from the sun, and could be a cheaper, more sustainable alternative to current chemical-based recycling methods. The team says their method could create a circular system where one waste stream solves another.

Global plastic production is more than 400 million tons per year, yet only 18% is recycled. The rest is burned, landfilled, or leaks into ecosystems. The researchers believe that their method, known as solar-powered acid photoreforming, could become part of the solution to the global mountain of plastic waste.

Quantum-informed machine learning for predicting spatiotemporal chaos with practical quantum advantage

Ultimately, QIML proves that we don’t need a fully fault-tolerant quantum computer to see results. By using quantum processors to learn the complex “rules” of chaos, we can give classical computers the boost they need to make reliable, long-term predictions about the most turbulent environments in the natural world.

Modeling high-dimensional dynamical systems remains one of the most persistent challenges in computational science. Partial differential equations (PDEs) provide the mathematical backbone for describing a wide range of nonlinear, spatiotemporal processes across scientific and engineering domains (1–3). However, high-dimensional systems are notoriously sensitive to initial conditions and the floating-point numbers used to compute them (4–7), making it highly challenging to extract stable, predictive models from data. Modern machine learning (ML) techniques often struggle in this regime: While they may fit short-term trajectories, they fail to learn the invariant statistical properties that govern long-term system behavior. These challenges are compounded in high-dimensional settings, where data are highly nonlinear and contain complex multiscale spatiotemporal correlations.

ML has seen transformative success in domains such as large language models (8, 9), computer vision (10, 11), and weather forecasting (12–15), and it is increasingly being adopted in scientific disciplines under the umbrella of scientific ML (16). In fluid mechanics, in particular, ML has been used to model complex flow phenomena, including wall modeling (17, 18), subgrid-scale turbulence (19, 20), and direct flow field generation (21, 22). Physics-informed neural networks (23, 24) attempt to inject domain knowledge into the learning process, yet even these models struggle with the long-term stability and generalization issues that high-dimensional dynamical systems demand. To address this, generative models such as generative adversarial networks (25) and operator-learning architectures such as DeepONet (26) and Fourier neural operators (FNO) (27) have been proposed. While neural operators offer discretization invariance and strong representational power for PDE-based systems, they still suffer from error accumulation and prediction divergence over long horizons, particularly in turbulent and other chaotic regimes (28, 29). Recent work, such as DySLIM (30), enhances stability by leveraging invariant statistical measures. However, these methods depend on estimating such measures from trajectory samples, which can be computationally intensive and inaccurate in all forms of chaotic systems, especially in high-dimensional cases. These limitations have prompted exploration into alternative computational paradigms. Quantum machine learning (QML) has emerged as a possible candidate due to its ability to represent and manipulate high-dimensional probability distributions in Hilbert space (31). Quantum circuits can exploit entanglement and interference to express rich, nonlocal statistical dependencies using fewer parameters than their promising counterparts, which makes them well suited for capturing invariant measures in high-dimensional dynamical systems, where long-range correlations and multimodal distributions frequently arise (32). QML and quantum-inspired ML have already demonstrated potential in fields such as quantum chemistry (33, 34), combinatorial optimization (35, 36), and generative modeling (37, 38). However, the field is constrained on two fronts: Fully quantum approaches are limited by noisy intermediate-scale quantum (NISQ) hardware noise and scalability (39), while quantum-inspired algorithms, being classical simulations, cannot natively leverage crucial quantum effects such as entanglement to efficiently represent the complex, nonlocal correlations found in such systems. These challenges limit the standalone utility of QML in scientific applications today. Instead, hybrid quantum-classical models provide a promising compromise, where quantum submodules work together with classical learning pipelines to improve expressivity, data efficiency, and physical fidelity. In quantum chemistry, this hybrid paradigm has proven feasible, notably through quantum mechanical/molecular mechanical coupling (40, 41), where classical force fields are augmented with quantum corrections. Within such frameworks, techniques such as quantum-selected configuration interaction (42) have been used to enhance accuracy while keeping the quantum resource requirements tractable. In the broader landscape of quantum computational fluid dynamics, progress has been made toward developing full quantum solvers for nonlinear PDEs. Recent works by Liu et al. (43) and Sanavio et al. (44, 45) have successfully applied Carleman linearization to the lattice Boltzmann equation, offering a promising pathway for simulating fluid flows at moderate Reynolds numbers. These approaches, typically using algorithms such as Harrow-Hassidim-Lloyd (HHL) (46), promise exponential speedups but generally necessitate deep circuits and fault-tolerant hardware.

Quantum-enhanced machine learning (QEML) combines the representational richness of quantum models with the scalability of classical learning. By leveraging uniquely quantum properties such as superposition and entanglement, QEML can explore richer feature spaces and capture complex correlations that are challenging for purely classical models. Recent successes in quantum-enhanced drug discovery (37), where hybrid quantum-classical generative models have produced experimentally validated candidates rivaling state-of-the-art classical methods, demonstrate the practical potential of QEML even before full quantum advantage is achieved. Despite these strengths, practical barriers remain. QEML pipelines require repeated quantum-classical communication during training and rely on costly quantum data-embedding and measurement steps, which slow computation and limit accessibility across research institutions.

How a chemical reaction triggers brain inflammation in Alzheimer’s disease

The brain has its own immune system, which detects threats and mounts a defense. A growing body of evidence has shown that in Alzheimer’s disease, those immune cells are chronically overactivated, causing inflammation that damages the connections between brain cells.

Now, in a preclinical study using human Alzheimer’s brain cells, scientists at Scripps Research have identified a molecular switch—and potential drug target—responsible for driving that chronic inflammation.

The research, published in Cell Chemical Biology on April 23, 2026, centers on a protein called STING, which normally functions as part of the immune system’s early-warning system. In the brains of people with Alzheimer’s, the team discovered that STING undergoes a chemical modification known as S-nitrosylation (or SNO, a reaction involving sulfur, oxygen and nitrogen) that promotes its overactivation. Blocking this chemical change to STING in a mouse model of the disease decreased neuroinflammation.

Carbon nanotubes are closing the gap on copper conductivity

Carbon nanotubes are one technology that many observers believe hasn’t quite lived up to the extreme hype that surrounded them when they first appeared on the scene in the late 1990s. At that time, much was made of their extraordinary electrical, thermal, and mechanical properties, with predictions that they would revolutionize materials science, electronics, and daily life. But could we be closer to realizing some of that promise?

In a paper published in the journal Science, researchers describe a method for adding a chemical to carbon nanotube bundles that brings them closer to copper’s ability to conduct electricity.

Carbon nanotubes are nanoscale hollow cylinders of carbon atoms, a structure that allows electricity to flow through them with very low resistance. However, when you bundle millions of them together, as you would need for practical applications like power lines and electrical wiring, they lose some of their exceptional conductivity. Electrons move easily along individual nanotubes, but transferring charge between neighboring tubes in a bundle is much less efficient.

Natural-language AI helps chemists design molecules step by step

Designing molecules is one of chemistry’s most complex challenges. From life-saving drugs to advanced materials, each compound requires a precise sequence of reactions. Planning these steps demands both technical knowledge and strategic insight, making it a task that often relies on years of experience.

Two problems plague much of modern chemistry. The first is retrosynthesis: Chemists start from a target molecule and work backward to identify simpler building blocks and viable reaction pathways. Retrosynthesis involves countless decisions, from choosing starting materials to determining when to form rings or protect sensitive functional groups. While computers can explore vast “chemical spaces,” they often struggle to capture the strategic reasoning used by human experts.

The second problem is reaction mechanisms. These describe how chemical reactions unfold step by step through the movements of electrons. Mechanistic insight helps scientists predict new reactions, improve efficiency, and reduce costly trial and error. Existing computational methods can generate many possible pathways, but often lack the chemical intuition needed to identify the most plausible ones.

Chromosomes condense in three timed chemical waves during cell division, study shows

DNA does not float freely in the cell. Instead, it is wrapped around histone proteins to form structures called nucleosomes. These histones carry numerous chemical modifications that act as molecular signals, controlling how tightly the DNA is packaged and which genes are active. During cell division, this DNA-histone complex—known as chromatin—must be further condensed into compact, rod-shaped chromosomes. Histone modifications play a key role in this process: They change significantly during condensation and regulate the conversion of chromatin.

For the first time, researchers have precisely tracked how molecular marks on DNA proteins change during cell division—and disproved a long-held assumption in the process.

An international research team led by Professor Axel Imhof at LMU’s Biomedical Center and Professor William Earnshaw (University of Edinburgh) has analyzed these changes during cell division with unprecedented precision. To this end, the researchers developed an innovative method that synchronizes the division of cell populations. They then employed high-resolution mass spectrometry to precisely record the changes in histone modifications during cell division. The findings are published in Molecular Cell.

How electron structure affects light responses in moiré materials

In materials science, if you can understand the “texture” of a material—how its internal patterns form and shift—you can begin to design how it behaves. That’s the focus of the work of Zhenglu Li, assistant professor in the Mork Family Department of Chemical Engineering and Materials Science at USC Viterbi School of Engineering. Li’s recently published paper in PNAS, titled “Moiré excitons in generalized Wigner crystals,” demonstrates that the way electrons organize themselves inside a material determines how that material responds to light—and how this organization can be engineered.

“Moiré” is a word that will be familiar to anyone who follows fashion. In the context of textiles, it refers to a larger-scale interference pattern that appears when two repeating patterns are slightly misaligned. Imagine brushing a swatch of velvet in different directions; the material reveals different properties depending on how it is ruffled.

Likewise, in the context of nanoscale materials science, an independent, shimmering or wavelike pattern is formed when two overlapping atomically thin layers are overlaid at an acute angle. The new pattern, moiré superlattice, changes how electrons move, which can give the material unusual properties.

Boosting good gut bacteria population through targeted interventions may slow cognitive decline

The origin of neurodegenerative diseases like Alzheimer’s or dementia isn’t limited to the brain. The state of your gut can quietly set off a cycle of chronic, system-wide inflammation that nudges the brain toward cognitive decline. But how does the pathogenesis of a disease that seems purely brain-based begin in the gut—an organ that is mostly busy producing chemicals for digesting food?

It turns out these two entities are linked by the gut-brain axis, a two-way communication superhighway that constantly sends signals between the digestive tract and the central nervous system. It runs on chemical messengers like neurotransmitters and fatty acids, sharing information that shapes our memory, mood, and inflammation triggers.

An analysis of 15 studies involving more than 4,200 participants found that the gut-brain highway can be put to work as a drug-free route to support cognitive health. Tuning the gut microbiota through diet, supplements, or medical interventions such as fecal microbiota transplantation (FMT) can help improve memory, executive function, and overall cognitive performance, particularly in early or mild cases of cognitive impairment.

Frontiers: Year 2020 this gene therapy in mice shows promise for als gene therapy in humans

Gene therapy is an emerging and powerful therapeutic tool to deliver functional genetic material to cells in order to correct a defective gene. During the past decades, several studies have demonstrated the potential of AAV-based gene therapies for the treatment of neurodegenerative diseases. While some clinical studies have failed to demonstrate therapeutic efficacy, the use of AAV as a delivery tool has demonstrated to be safe. Here, we discuss the past, current and future perspectives of gene therapies for neurodegenerative diseases. We also discuss the current advances on the newly emerging RNAi-based gene therapies which has been widely studied in preclinical model and recently also made it to the clinic.

Gene therapy is an emerging therapeutic tool used to deliver functional genetic material to cells in order to correct a defective gene. By delivering a copy of a therapeutic gene to affected cells, the product encoded by that gene [i.e., its messenger RNA (mRNA) and/or proteins] will be continuously synthesized within the cell, utilizing the cell’s own transcriptional and translational machinery (Porada et al., 2013). The main advantage of this technology is that it offers a potentially life-long therapeutic effect without the need for repeated administration. Gene therapy can be used to correct defective genes by introducing a functional copy of the gene, by silencing a mutant allele using RNA interference (RNAi), by introducing a disease-modifying gene, or by using gene-editing technology (Grimm and Kay, 2007; Dow et al., 2015; Saraiva et al., 2016).

Gene therapy vectors can be either viral or non-viral. Different physical and chemical systems can be applied to deliver therapeutic genes to cells without the need of a viral vector. Non-viral vectors have no size limitation for the therapeutic gene, generally have a low immunogenicity risk, and can be produced at relatively low costs (Nayerossadat et al., 2012). However, due to the fact that high therapeutic doses are required when using non-viral technologies, and the resulting gene expression is generally transient, most gene therapies now rely on viral vectors. Numerous viral vector types have been tested in clinic, including vaccinia, measles, vesicular stomatitis virus (VSV), polio, reovirus, adenovirus, lentivirus, γ-retrovirus, herpes simplex virus (HSV) and adeno-associated virus (AAV) (Lundstrom, 2018).

/* */