Shifeng Jin’s research while affiliated with Chinese Academy of Sciences and other places

Accurate crystal structure determination is critical across all scientific disciplines involving crystalline materials. However, solving and refining crystal structures from powder X-ray diffraction (PXRD) data is traditionally a labor-intensive process that demands substantial expertise. Here we introduce PXRDGen, an end-to-end neural network that determines crystal structures by learning joint structural distributions from experimentally stable crystals and their PXRD, producing atomically accurate structures refined through PXRD data. PXRDGen integrates a pretrained XRD encoder, a diffusion/flow-based structure generator, and a Rietveld refinement module, solving structures with unparalleled accuracy in seconds. Evaluation on MP-20 dataset reveals a record high matching rate of 82% (1-sample) and 96% (20-samples) for valid compounds, with Root Mean Square Error (RMSE) approaching the precision limits of Rietveld refinement. PXRDGen effectively tackles key challenges in PXRD, such as the localization of light atoms, differentiation of neighboring elements, and resolution of overlapping peaks. Overall, PXRDGen marks a significant advancement in the automated determination of crystal structures from PXRD.

Accurate crystal structure determination is critical across all scientific disciplines involving crystalline materials. However, solving and refining inorganic crystal structures from powder X-ray diffraction (PXRD) data is traditionally a labor-intensive and time-consuming process that demands substantial expertise. In this work, we introduce PXRDGen, an end-to-end neural network that determines crystal structures by learning joint structural distributions from experimentally stable crystals and their PXRD, producing atomically accurate structures refined through PXRD data. PXRDGen integrates a pretrained XRD encoder, a diffusion/flow-based structure generator, and a Rietveld refinement module, enabling the solution of structures with unparalleled accuracy in a matter of seconds. Evaluation on MP-20 inorganic dataset reveals a remarkable matching rate of 82% (1 sample) and 96% (20 samples) for valid compounds, with Root Mean Square Error (RMSE) approaching the precision limits of Rietveld refinement. PXRDGen effectively tackles key challenges in XRD, such as the precise localization of light atoms, differentiation of neighboring elements, and resolution of overlapping peaks. Overall, PXRDGen marks a significant advancement in the automated determination of crystal structures from powder diffraction data.

C 6 H 18 O 2 S 2 NaInBr 6 , orthorhombic, Pmn 2 1 (no. 31), a = 10.5451(3) Å, b = 7.1169(2) Å, c = 13.7034(4) Å, V = 1028.42(5) Å ³ , Z = 2, R gt ( F ) = 0.0354, wR ref ( F ² ) = 0.0923, T = 293 K.

C 12 H 32 O 10 Cu 2 Cl 4 , monoclinic, P 2 1 / n (no. 14), a = 9.0202(4) Å, b = 9.5791(4) Å, c = 13.6994(6) Å, β = 92.555(1) ° , V = 1182.53(9) Å ³ , Z = 2, R gt ( F ) = 0.0308, wR ref ( F ² ) = 0.0800, T = 293(2) K.

All two-dimensional (2D) materials of group IV elements from Si to Pb are stabilized by carrier doping and interface bonding from substrates except graphene which can be free-standing. The involvement of strong hybrid of bonds, adsorption of exotic atomic species, and the high concentration of crystalline defects are often unavoidable, complicating the measurement of the intrinsic properties. In this work, we report the discovery of seven kinds of hitherto unreported bulk compounds (RO)nPb (R = rare earth metals, n = 1,2), which consist of quasi-2D Pb square nets that are spatially and electronically detached from the [RO]δ+ blocking layers. The band structures of these compounds near Fermi levels are relatively clean and dominantly contributed by Pb, resembling the electron-doped free-standing Pb monolayer. The R2O2Pb compounds are metallic at ambient pressure and become superconductors under high pressures with much enhanced critical fields. In particular, Gd2O2Pb (9.1 μB/Gd) exhibits an interesting bulk response of lattice distortion in conjunction with the emergence of superconductivity and magnetic anomalies at a critical pressure of 10 GPa. Our findings reveal the unexpected facets of 2D Pb sheets that are considerably different from their bulk counterparts and provide an alternative route for exploring 2D properties in bulk materials.

Chemical intercalation has great potential to realize exotic electrical properties in van der Waals materials. Here we report the Ge intercalation in layered Nb dichalcogenides NbX2 (X = Se and...

Due to the limited resources of lithium source and their high price, lithium-ion batteries (LIBs) cannot meet the demands of future large-scale energy storage. Sodium-ion batteries (SIBs) with advantages of the abundant sodium resources and low cost, are one of the most promising alternatives to LIBs. Here, we firstly synthesize a new manganese-based polyanionic compound (Na3Mn2P3O11) through a simple sol-gel method. The resolved crystal structure indicates that single-phase Na3Mn2P3O11 compound belongs to the orthorhombic structure. When tested as for cathodes, it displays a poor electrochemical performance with a potential window of 1.8–4.3 V (versus Na/Na⁺), caused by the Jahn-Teller effect of Mn³⁺ generated during charging process. In this context, a series of Fe-doped Na3Mn2-xFexP3O11 (0.1 ≤ x ≤ 0.5) are prepared to improve the electrochemical performance. Results indicate that Fe substitution of Mn in Na2Mn2-xFexP3O11 structure can not only enhance the electrochemical performance, but also increase conductivity and achieve fast reaction kinetics. Noticeably, when iron-doped amount is 0.4, Na3Mn1.6Fe0.4P3O11 exhibits the best cycling (62.7 mA h g⁻¹ at 0.1 C over 100 cycles) and rate performance (19.7 mA h g⁻¹ at 5 C).

Li4Na2CsB7O14 with unique edge-sharing [BO4]5- tetrahedra, unprecedented [B14O28]14- cluster and high anisotropic thermal expansion has been obtained under the atmospheric pressure condition, indicating the high-pressure-driven is not an essential prerequisite...

Constructing highly-efficient heterojunction for photogenerated charges separation is essential to photocatalysis for solar energy conversion. In this work, we prepare an efficient photocatalyst, SiC/QD-BiVO4 composite, through in-situ self-assemble method. This effective heterojunction is constructed by controlling QD-BiVO4 orientated deposition on certain facets of SiC. Efficient electron-hole separation is achieved by this heterojunction resulted from the following two effects. One is that the selective distribution of QD-BiVO4 on SiC facets, rather than random deposition, reduces the transfer path-length of photoexcited carriers. The other one is that the negative shift of conduction band of BiVO4 quantum dots increases the electric potential difference in built-in field, which accelerates the carriers transfer rate at the interface. Consequently, the O2 production is enhanced to 2096 μmol h-1 g-1 in SiC/QD-BiVO4 photocatalyst. Moreover, the degradation rate of RhB is doubled. Our work exhibits a rational route to prepare environmental-friendly photocatalytic material with potential applications.

We report a new layered FeS compound Kx(C2H8N2)yFe2−zS2 synthesized by intercalating K and C2H8N2 into tetragonal FeS via a simple sonochemical route. This new compound crystallizes in a body-centered tetragonal unit cell, with the [K(C2H8N2)] and [FeS] layers alternately stacking along the c direction. The nominal concentration of K, x, can be adjusted from 0.25 to 0.45, and the lattices a and c contract from 3.6971(9) and 20.667(5) Å to 3.691(1) and 20.566(7) Å, respectively. When x < 0.25, the parent FeS is residual and when x > 0.45, K reacts with FeS directly to form K2Fe4S5 impurity. It is found that the C2H8N2 molecule has been co-intercalated in between the [FeS] layers along with K, evidenced by its content, y, having a linear dependence with x. Measurements indicate that Kx(C2H8N2)yFe2−zS2 is a semiconductor and it shows a weak ferrimagnetism below 50 K. More importantly, Fe depletion resulting from the charged K⁺ intercalation was revealed by composition analysis, which leads to the formation of disordered Fe vacancies in the [FeS] layers and hence hinders the enhancement of original superconductivity in the FeS parent.