Weekly Advanced Technologies〔91〕丨Breaking the Symmetry Barrier: Generative AI for On-Demand Crystal Design; Phosphorus Mobilization Triggered by Permafrost Thaw Alters Biogeochemical Cycling in Alpine Ecosystems

Crystals hold nature's symmetrical codes—now AI has learned to decipher this language, using "chemical intuition" to generate entirely new crystal structures, ushering in an intelligent new paradigm for materials discovery. Meanwhile, permafrost thawing not only drives carbon release but also profoundly impacts phosphorus cycling; new research reveals that thermal collapse is accelerating soil phosphorus activation, uncovering a previously overlooked ecological effect of permafrost degradation on nutrient balance.

Based on the weekly diary of technology provided by the daily list of the NCSTI online service platform, we launch the column "Weekly Advanced Technologies" at the hotlist of sci-tech innovation. Today, let's check out No.91.

1. Science Bulletin丨Breaking the Symmetry Barrier: Generative AI for On-Demand Crystal Design

Symmetry-Aware Generative Modeling of Crystalline Structures

A research team from the Institute of Physics, Chinese Academy of Sciences / Beijing National Laboratory for Condensed Matter Physics, in collaboration with Jilin University, has developed a novel generative AI model named CrystalFormer specifically for crystal material design and discovery. Unlike traditional simulation methods relying on energy optimization, CrystalFormer learns generative patterns of crystal structures from data, directly "guessing" plausible crystal configurations under symmetry constraints.

The model's core innovation lies in transforming crystal structures into sequential representations, constructing generation logic using Wyckoff position tables of space groups and the periodic table. Crystals are treated as "orderly filling of atoms in symmetric positions," a process modeled by an autoregressive neural network that progressively generates key information including atomic species, occupancies, and lattice parameters. By learning from vast datasets of known crystals, CrystalFormer internalizes fundamental solid-state chemistry principles and encodes this "chemical intuition" as neural network parameters.

CrystalFormer not only explores uncharted material spaces without constraints but can also generate crystals with specified structural features on demand. When integrated with property prediction models, it enables "inverse design"—first defining target properties, then inversely generating novel structures likely to meet these conditions. The entire process employs probabilistic modeling and Bayesian inference to efficiently identify high-potential candidate materials.

This work represents the first deep integration of mathematical symmetry in crystals with chemical heuristics within a generative modeling framework, establishing an efficient, intelligent, and scalable new paradigm for materials discovery.

2. Nature Climate Change丨Phosphorus Mobilization Triggered by Permafrost Thaw Alters Biogeochemical Cycling in Alpine Ecosystems

Thermokarst formation accelerates ecosystem phosphorus cycling by enhancing soil phosphorus transformation and plant phosphorus uptake.

Permafrost regions store approximately one-third of the global soil organic carbon, and their thawing has profound impacts on the climate system. However, current research on post-thaw nutrient cycling has largely focused on nitrogen, with phosphorus cycling remaining critically understudied. Leveraging the Thermokarst Monitoring Network Platform in the Qinghai-Tibet Plateau permafrost region, researchers from the Institute of Botany, Chinese Academy of Sciences (CAS) conducted a systematic investigation into the response mechanisms of ecosystem phosphorus cycling to permafrost thaw.

By integrating ³¹P nuclear magnetic resonance (³¹P-NMR), ³³P isotopic labeling, and metagenomic sequencing, the team discovered that thermokarst formation significantly increases the activation rate of total phosphorus in surface soils while enhancing plant phosphorus uptake capacity, collectively accelerating ecosystem phosphorus cycling.

Mechanistic analyses revealed that the observed phosphorus activation primarily stems from upregulated microbial functional genes involved in phosphorus cycling, highlighting the pivotal role of microbial activity in driving phosphorus transformation. Concurrently, the enhancement in plant phosphorus uptake correlates closely with adaptive root morphological changes, increased root exudation, and improved competitive capacity for soil phosphorus resources.

This study has systematically revealed for the first time the response patterns of key phosphorus cycling processes in ecosystems under permafrost thaw, demonstrating the synergistic effects of microbial functionality and plant adaptation in accelerating phosphorus cycling. The findings have filled a critical knowledge gap in permafrost phosphorus cycling research and provided essential scientific basis for understanding carbon-phosphorus interactions and their climate feedbacks, holding significant implications for predicting ecological trajectories in permafrost regions.

3. Communications Biology丨The Thalamus's Hidden Role: A Color Gatekeeper in Our Brain

Global feature-based attention exhibits layer-specific modulation and functional connectivity in the lateral geniculate nucleus (LGN)

Human beings can rapidly recognize targets based on features like color and shape, even when such information lies outside the focus of attention—an ability termed "feature-based attention". While existing studies have revealed its mechanisms within the cerebral cortex, whether and how subcortical structures participate remains unclear. A research team from the Institute of Biophysics, Chinese Academy of Sciences (IBP-CAS), collaborating with other institutions, has for the first time uncovered the critical role of subcortical nuclei in global feature-based attention using high-resolution functional magnetic resonance imaging (fMRI) at 7 T.

The study designed a red-green contrast visual stimulation task to investigate the color sensitivity properties of the parvocellular layers in the lateral geniculate nucleus (LGN). Results demonstrated that when stimuli in the unattended visual field shared the same color as the attended target, the color responses in the LGN parvocellular layers were significantly enhanced. Concurrently, both feedforward and feedback connections between the LGN and primary visual cortex were notably strengthened, indicating that the LGN plays an early "gating" role in feature-based attention.

Further analysis revealed that the deep layers of the superior colliculus participate in regulating global feature-based attention, while the ventrolateral subdivision of the pulvinar coordinates attentional information transfer across different brain regions. Notably, the visual thalamus exhibited increased responses to unattended stimuli that matched the target color, while responses to attended stimuli were conversely suppressed. This "seesaw" phenomenon was not observed in cortical areas, suggesting a thalamus-dominated, spatially global attentional sampling mechanism.

This study has for the first time demonstrated the critical role of subcortical structures in feature-based attention. The findings provide key evidence for a unified attention theory incorporating both cortical and subcortical mechanisms, advancing our understanding of the brain's information filtering system.

4. Nature Energy丨Operational Stability Realized: Organic Photovoltaics Withstand Extreme Damp Heat and Thermal Cycling

Stability Strategies for Organic Solar Cells

Organic solar cells (OPVs), with their lightweight, flexible, and solution-processable advantages, show great potential for near-space applications, building integration, and IoT. However, their stability under extreme environments like damp heat and thermal cycling has remained an application bottleneck.

A team from the Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences, has proposed a systematic solution. Through coordinated optimization of materials, interfaces, and encapsulation, they achieved for the first time high-performance OPVs with exceptional stability under harsh conditions (85°C/85%RH damp heat and -40°C to 85°C thermal cycling).

The research team innovatively developed a UV-Vis absorption spectral differentiation analysis method, defining the absorption onset temperature (Tonset) to characterize the thermal stability of active layer microstructures. Using this approach, they discovered that crosslinked fullerene doping could significantly increase the Tonset of blend films, establishing a universal strategy for designing intrinsically thermally stable active layer materials. By combining this material strategy with interface optimization, the team fabricated inverted ternary OPV devices that maintained excellent performance even at 150°C.

Addressing moisture permeation issues, the team has achieved the first quantitative analysis of 2D vs. 1D diffusion rates, establishing a kinetic model for edgewise moisture penetration along encapsulation adhesives and adhesive/substrate interfaces. This enabled precise identification of dominant permeation pathways. The resulting aluminum foil-butyl tape encapsulation structure maintained: 94% efficiency retention after 1,032 hours at 85°C/85%RH and 94% retention through 200 thermal cycles (-40°C to 85°C). The study demonstrates that high-efficiency OPVs possess exceptional environmental robustness, providing both scientific foundations and technical blueprints for commercialization.