Hexagonal germanium polytypes have emerged as promising direct-gap semiconductors for silicon-integrated optoelectronics, yet their optical properties remain largely unexplored beyond the well-studied 2H phase. We present a comprehensive theoretical study of optical properties of hexagonal 2H-, 4H-, and 6H-Ge polytypes through ab initio calculations of quasiparticle band structures, dipole transition matrix elements, and solution of the Bethe-Salpeter equation. While all three polytypes exhibit direct band gaps of increasing size from 2H to 6H, we reveal that the fundamental optical transition in 4H-Ge is parity-forbidden due to matching band parities at the valence and conduction band edges. This selection rule results in a radiative lifetime seven orders of magnitude longer than in 2H- and 6H-Ge, severely limiting light emission capabilities. To demonstrate that the selection rule can be lifted, we introduce controlled symmetry perturbations by substituting single Ge atoms with Si in each unit cell, breaking the crystal symmetry. This perturbation increases the optical matrix elements by up to two orders of magnitude and reduces radiative lifetimes for all perturbed polytypes. We also compute absorption coefficients and frequency-dependent dielectric tensors for both light polarizations, including excitonic effects up to 5 eV, providing complete optical characterization of ideal and symmetry-perturbed hexagonal Ge systems relevant for optoelectronic applications.

Graph neural networks have become the dominant machine-learning architecture for predicting materials properties from crystal structures. Yet the initialization of atomic node features has received comparatively little attention, and conventional approaches rely on static elemental descriptors that carry no information about the quantum-mechanical electronic environment of each atom in its crystalline host. Here we show that augmenting atomic node representations with site-projected orbital density of states (pDOS) fingerprints, computed directly from density functional theory calculations, yields systematic and substantial improvements in predictive performance.These representations are fused with Pettifor elemental embeddings at each atomic site before message passing. For the superconducting critical temperature T_c and the optical dielectric constant ε_∞,the pDOS augmentation reduces prediction errors by 22.9

We present a computational framework that integrates machine learning with high-throughput ab initio calculations to screen over 2.8 million compounds for metallic transport. We identify several intermetallic candidates with predicted high conductivities comparable to that of aluminum (36.59 × 10^6 S/m). We perform full electron–phonon coupling calculations for the top-performing materials, yielding results in excellent agreement with available experimental data. Our analysis reveals that while the noble metals (Ag, Au, Cu) define the practical ceiling for conductivity due to their unique electronic structure and low scattering, compounds like LiBePt_2 can achieve comparable performance by utilizing valence electrons from light elements to shift high-scattering d-states beneath the Fermi level. This study not only identifies novel high-performance conductors but also demonstrates the predictive power of combining statistical learning with detailed ab initio calculations.

Bismuth oxyhalides (BiOX, X = Cl, Br, and I) are promising photocatalytic materials whose electronic and optical properties can be systematically tuned through halogen alloying. This study presents a comprehensive computational and experimental investigation of halogen anion alloying effects on the electronic structure and optical properties of BiOCl1-xBrx, BiOCl1-xIx, and BiOBr1-xIx alloy systems. Using density functional theory calculations combined with the generalized quasichemical approximation, we systematically investigated band gaps and density of states for all symmetrically non-equivalent configurations in 24-atom supercells. Our calculations reveal significant band gap bowing behavior with bowing parameters of 0.98, 2.23, and 2.70 eV for BiOCl1-xBrx, BiOBr1-xIx, and BiOCl1-xIx alloys, respectively, representing substantial deviations from Vegard's law. Point defect formation energy calculations demonstrate that although halogen substitution is endothermic, the formation energies remain sufficiently low (2-143 meV per dopant atom) to enable experimental synthesis. The mixing enthalpies remain below 10 meV per formula unit across the entire composition range for all three systems. At typical synthesis temperatures, the configurational entropy contribution easily overcomes the enthalpy penalty, stabilizing the random solid solution. We successfully synthesized BiOCl1-xIx nanoparticles across the complete composition range (x = 0-1) with yields exceeding 90%, experimentally validating our predictions. UV-visible spectroscopy of the synthesized alloys confirms the predicted red-shift in absorption onset with increasing iodine content. While the electronic band structures exhibit strong bowing effects, the absorption spectra are reasonably captured by a linear interpolation between pure end-members, providing a practical approximation for targeted optical design applications. The minimum band gaps occur at approximately 75-78% heavier halogen content, offering optimal visible light absorption. These findings provide fundamental insights into halogen alloying mechanisms in bismuth oxyhalides and establish clear design principles for enhanced photo-absorption or photo-catalytic applications.

We present a novel approach to generate a fingerprint for crystalline materials that balances efficiency for machine processing and human interpretability, allowing its application in both machine learning inference and understanding of structure-property relationships. Our proposed material encoding has two components: one representing the crystal structure and the other characterizing the chemical composition, which we call Pettifor embedding. For the latter, we construct a non-orthogonal space where each axis represents a chemical element and where the angle between the axes quantifies a measure of the similarity between them. The chemical composition is then defined by the point on the unit sphere in this non-orthogonal space. We show that the Pettifor embeddings systematically outperform other commonly used elemental embeddings in compositional machine learning models. Using the Pettifor embeddings to define a distance metric and applying dimension reduction techniques, we construct a two-dimensional global map of the space of thermodynamically stable crystalline compounds. Despite their simplicity, such maps succeed in providing a physical separation of material classes according to basic physical properties.

Hydrogen-rich materials are among the most promising candidates for achieving high-temperature superconductivity due to their light atomic mass and strong phonon-mediated Cooper pairing. While many high-temperature superconducting hydrides have been reported experimentally, their practical applicability remains limited due to the required extreme conditions, motivating the search for stable or metastable superconducting phases at ambient pressure. Here, we combine ab initio electron–phonon coupling calculations with machine learning methods to explore over two million hydride structures, identifying more than 600 compounds with predicted critical temperatures (Tc) above 20 K. This dataset reveals a large chemical and structural diversity among potential superconductors, including perovskites, kagome lattices, and compounds with isolated hydrogen octahedra, among others. Despite this diversity, high-Tc materials consistently exhibit substantial hydrogen contributions to the electronic density of states at the Fermi level. Symbolic regression analysis quantitatively confirms this correlation. Thermodynamic analysis shows that all identified superconductors lie above the convex hull of stability, with energies typically exceeding 100 meV/atom above the hull. This implies that experimental synthesis can only be obtained through non-equilibrium approaches, including high-pressure or high-temperature methods followed by quenching, or by thin-film deposition techniques. Moreover, as many promising candidates resemble degenerate semiconductors, a possible synthesis route may involve controlled doping strategies of parent semiconducting compounds. This work substantially expands the known landscape of hydride superconductors and establishes design principles to guide future experimental realization.

The rapid rise of generative artificial intelligence is reshaping materials discovery by offering new ways to propose crystal structures and, in some cases, even predict desired properties. This review provides a comprehensive survey of recent advancements in generative models specifically for inorganic crystalline materials. We outline architectures, representations, conditioning mechanisms, data sources, metrics, and applications, and organize existing models into a unified taxonomy.

In this contribution we report on the synthesis, structure and optical characterization of Ruddlesden-Popper oxyfluorides La2Ni1-xCuxO2.5F3 (0 <= x <= 1) obtained by topochemical low-temperature fluorination of La2Ni1-xCuxO4 with polyvinylidene fluoride (PVDF). Our study reveals that the anionic ordering in the tetragonal unit cell of La2NiO2.5F3 persists even at high Cu substitution levels (x = 0.9), with minimal change in unit cell volume. This observation is contrary to expectations based on Jahn-Teller induced unit cell distortions, which were previously reported for the oxides La2Ni1-xCuxO4, as well as for the closely related oxyfluorides La2Ni1-xCuxO3F2. The pure copper-containing compound La2CuO2.5F3 crystallizes in a triclinic version of the same structure, and the symmetry lowering is attributed to the enhanced space requirements of the Jahn-Teller elongated CuO4F2 octahedra. The structural investigations based on XRD and ND Rietveld refinements are supported by low-field 19F MAS NMR experiments. We also report the results of diffuse reflectance UV-Vis measurements, which are complemented by DFT calculations. Here, we demonstrate a strong impact of the Cu substitution on the electronic structure of the oxyfluorides, resulting in band gap energies in the range of 3.4 eV to 1.3 eV, spanning the whole visible spectrum. Notably, first photocatalytic water splitting tests reveal a considerable hydrogen evolution activity for x = 0.2, highlighting the potential of Ruddlesden-Popper oxyfluorides for solar energy applications.

We perform a large scale search for two-dimensional (2D) superconductors, by using electron-phonon calculations with density-functional perturbation theory combined with machine learning models. In total, we screened over 140 000 2D compounds from the Alexandria database. Our high-throughput approach revealed a multitude of 2D superconductors with diverse chemistries and crystal structures. Moreover, we find that 2D materials generally exhibit stronger electron-phonon coupling than their 3D counterparts, although their average phonon frequencies are lower, leading to an overall lower T-c. In spite of this, we discovered several out-of-distribution materials with relatively high-T-c. In total, 105 2D systems were found with T-c> 5 K. Some interesting compounds, such as CuH2, NbN, and V2NS2, demonstrate high T(c )values and good thermodynamic stability, making them strong candidates for experimental synthesis and practical applications. Our findings highlight the critical role of computational databases and machine learning in accelerating the discovery of novel superconductors.

Tunability of the band gap energy is achieved for the highly fluorinated system La 2 Ni 1− x Cu x O 2.5 F 3 while retaining the overall structural distortion.

The theoretical maximum critical temperature (Tc) for conventional superconductors at ambient pressure remains a fundamental question in condensed matter physics. Through analysis of electron-phonon calculations for over 20,000 metals, we critically examine this question. We find that while hydride metals can exhibit maximum phonon frequencies of more than 5000 K, the crucial logarithmic average frequency ω log rarely exceeds 1800 K. Our data reveals an inherent trade-off between ω log and the electron-phonon coupling constant λ, suggesting that the optimal Eliashberg function that maximizes Tc is unphysical. Based on our calculations, we identify Li2AgH6 and its sibling Li2AuH6 as theoretical materials that likely approach the practical limit for conventional superconductivity at ambient pressure. Analysis of thermodynamic stability indicates that compounds with higher predicted Tc values are increasingly unstable, making their synthesis challenging. While fundamental physical laws do not strictly limit Tc to low-temperatures, our analysis suggests that achieving room-temperature conventional superconductivity at ambient pressure is extremely unlikely.

The kagome lattice has emerged as a fertile ground for exotic quantum phenomena, including superconductivity, charge density wave, and topologically nontrivial states. While AV3Sb5 (A = K, Rb, Cs) compounds have been extensively studied in this context, the broader AB3C5 family remains largely unexplored. In this work, we employ machine-learning accelerated, high-throughput density functional theory calculations to systematically investigate the stability and electronic properties of kagome materials derived from atomic substitutions in the AV3Sb5 structure. We identify 36 promising candidates that are thermodynamically stable, with many more close to the convex hull. Stable compounds are not only found with a pnictogen (Sb or Bi) as the C atom, but also with Au, Hg, Tl, and Ce. This diverse chemistry opens the way to tune the electronic properties of the compounds. In fact, many of these compounds exhibit Dirac points, Van Hove singularities, or flat bands close to the Fermi level. Our findings provide an array of compounds for experimental synthesis and further theoretical exploration of kagome superconductors beyond the already known systems.

This study presents a computational investigation of X4H15 compounds (where X represents a metal) as potential superconductors at ambient conditions or under pressure. Through systematic density functional theory calculations and electron-phonon coupling analysis, we demonstrate that electronic structure engineering via hole doping dramatically enhances the superconducting properties of these materials. While electron-doped compounds with X4+ cations (Ti, Zr, Hf, Th) exhibit modest transition temperatures of 1-9 K, hole-doped systems with X3+cations (Y, Tb, Dy, Ho,Er, Tm, Lu) show remarkably higher values of approximately 50 K at ambient pressure. Superconductivity in hole-doped compounds originates from stronger coupling between electrons and both cation and hydrogen phonon modes. Although pristine X3+4H15compounds are thermodynamically unstable, we propose a viable synthesis route via controlled hole doping of the charge-compensated YZr3H15 compound. Our calculations predict that even minimal concentrations of excess Y could induce high-temperature superconductivity while preserving structural integrity. This work reveals how strategic electronic structure modulation can optimize superconducting properties in hydride systems, establishing a promising pathway toward practical high-temperature conventional super-conductors at ambient pressure

There has been an ongoing race for the past several years to develop the best universal machinelearning interatomic potential. This progress has led to increasingly accurate models for predictingenergy, forces, and stresses, combining innovative architectures with big data. Here, we benchmarkthese models on their ability to predict harmonic phonon properties, which are critical for under-standing the vibrational and thermal behavior of materials. Using around 10 000 ab initio phononcalculations, we evaluate model performance across various phonon-related parameters to test theuniversal applicability of these models. The results reveal that some models achieve high accuracyin predicting harmonic phonon properties. However, others still exhibit substantial inaccuracies,even if they excel in the prediction of the energy and the forces for materials close to dynamicalequilibrium. These findings highlight the importance of considering phonon-related properties inthe development of universal machine learning interatomic potentials.

We present a computational study of the M3QX7 family of two-dimensional compounds, focusing specifically on their flat-band properties. We use a high-throughput search methodology, accelerated by machine learning, to explore the vast chemical space spawned by this family. In this way, we identify numerous stable 2D compounds within the M3QX7 family. We investigate how the chemical composition can be manipulated to modulate the position and dispersion of the flat bands. By employing a tight-binding model we explain the formation of flat bands as a result of a relatively loose connection between triangular M3QX3 clusters via bridges of X atoms. The model provides an understanding of the residual interactions that can impact the band dispersion. The same loose connection between clusters not only leads to strongly localized electronic states and thus flat electronic bands but also leads to localized phonon modes and flat bands in the phonon dispersion.

We present a range of inverse perovskite nitrides with an elpasolite-type superstructure. (Ca3N0.682(9))Sn and (Ca3N0.559(7))Pb are variants of the previously described (Ca3N)Sn and (Ca3N)Pb which contain less nitrogen and crystallize in Fm3̄m. (Ba3N0.5)Sn and (Ba3N0.5)Pb resemble the previously reported perovskites (Ba3Nx)Sn and (Ba3Nx)Pb, but with both the superstructure and octahedral tilting, resulting in space group R3̄. (Ca3N0.77(2))Si, (Ca3N0.669(6))Ge, (Sr3N0.5)Ge and (Ba3N0.5)Ge all crystallize in P21/n. Among these, only (Ca3Nx)Ge has been previously described as (Ca3N)Ge. (Ca3N0.77(2))Si is notably the first compound in which mutually isolated N3− and Si4− ions coexist. There also exists a version with composition (Ca3N0.86(6))Si, which crystallizes in the cubic perovskite aristotype structure with space group Pm3̄m. Similarly, there are versions of (Sr3N0.5)Ge, (Ba3N0.5)Sn and (Ba3N0.5)Pb with elevated nitrogen contents, less strongly tilted octahedra and no apparent superstructure. Electronic structure calculations indicate a metallic nature of the title compounds, with rather narrow improper band gaps for the strontium and barium compounds.

Ruddlesden-Popper oxyfluorides of the substitution series La2Ni1-xCuxO3F2 (0 <= x <= 1) were obtained by topochemical fluorination with polyvinylidene fluoride (PVDF) of oxide precursors La2Ni1-xCuxO4. The thermal stability and the temperature-dependent unit cell evolution of the oxyfluorides were investigated by high-temperature XRD measurements. The oxyfluoride with x = 0.6 shows the highest decomposition temperature of theta(dec) similar to 520 degrees C, which is significantly higher than the ones found for the end members La2NiO3F2 (x = 0) theta(dec) similar to 460 degrees C and La2CuO3F2 (x = 1) theta(dec) similar to 430 degrees C. The magnetic properties of all La2Ni1-xCuxO3F2 oxyfluorides were characterized by field- and temperature-dependent measurements as well as DFT calculations of the magnetic ground state. An antiferromagnetic ordering was derived for all substitution levels. For the Neel temperature (T-N), a nonlinear dependence on the copper content was found, and comparably high values of T-N in the region of 200-250 K were observed in the broad composition range of 0.3 <= x <= 0.8.

We present a novel approach to generate a fingerprint for crystalline materials that balances efficiency for machine processing and human interpretability, allowing its application in both machine learning inference and understanding of structure-property relationships. Our proposed material encoding has two components: one representing the crystal structure and the other characterizing the chemical composition, that we call Pettifor embedding. For the latter we construct a non-orthogonal space where each axis represents a chemical element and where the angle between the axes quantifies a measure of the similarity between them. The chemical composition is then defined by the point on the unit sphere in this non-orthogonal space. We show that the Pettifor embeddings systematically outperform other commonly used elemental embeddings in compositional machine learning models. Using the Pettifor embeddings to define a distance metric and applying dimension reduction techniques, we construct a two-dimensional global map of the space of thermodynamically stable crystalline compounds. Despite their simplicity, such maps succeed in providing a physical separation of material classes according to basic physical properties.

One of the major challenges in the development of universal machine learning interatomic potentials is accurately reproducing phonon properties. This issue appears to arise from the limitations of available datasets rather than the models themselves. To address this, we develop an extensive dataset of phonon calculations using density-functional perturbation theory (DFPT). We then show how this dataset can be used to train neural-network force fields, by implementing the training and the prediction of force constants in periodic crystals. This approach improves the quality of phonon properties prediction while reducing the number of structures needed for neural network training. We demonstrate the efficiency of this method using two examples of ternary phase diagrams: Ti-Nb-Ta and Li-B-C. In both cases, neural network predictions for the energy and forces show a considerable improvement, while phonon properties are predicted with high precision for all structures across the entire phase diagrams.