Search

The discovery of new functional and stable materials is a big challenge due to its complexity. This work aims at the generation of new crystal structures with desired properties, such as chemical stability and specified chemical composition, by using machine learning generative models. Compared with the generation of molecules, crystal structures pose new difficulties arising from the periodic nature of the crystal and from the specific symmetry constraints related to the space group. In this work, score-based probabilistic models based on annealed Langevin dynamics, which have shown excellent performance in various applications, are adapted to the task of crystal generation. The novelty of the presented approach resides in the fact that the lattice of the crystal cell is not fixed. During the training of the model, the lattice is learned from the available data, whereas during the sampling of a new chemical structure, two denoising processes are used in parallel to generate the lattice along with the generation of the atomic positions. A multigraph crystal representation is introduced that respects symmetry constraints, yielding computational advantages and a better quality of the sampled structures. We show that our model is capable of generating new candidate structures in any chosen chemical system and crystal group without any additional training. To illustrate the functionality of the proposed method, a comparison of our model to other recent generative models based on descriptor-based metrics is provided.

Lamellar lithiated nitridonickelates have been investigated from both experimental and theoretical points of view in a wide range of compositions. In this study, we show that the nickel ion in lamellar lithiated nitridonickelates adopts an intermediate valence close to +1.5. This solid solution can therefore be written Li 3-1.5 x Ni x N with 0 ≤ x ≤ 0.68. Attempts to introduce more nickel into these phases systematically lead to the presence of the endmember of the solid solution, Li 1.97 Ni 0.68 N, with metallic nickel as an impurity. The LiNiN phase has never been observed, and first-principles calculations suggested that all the structural configurations tested were mechanically unstable.

The new ternary compounds La 15 Ni 13 Bi 5 and La 9 Ni 8 Sn 5 were obtained by arc melting under argon from appropriate amounts of the elements and subsequent annealing at 800 °C for 2 weeks. Single-crystal X-ray diffraction reveals that they represent two new structure types: La 15 Ni 13 Bi 5 crystallizes in the hexagonal space group P 62 m [ hP 33, a = 14.995(3), c = 4.3421(10) Å, V = 845.5(4) Å 3 , Z = 1] and La 9 Ni 8 Sn 5 in P 6 3 / m [ hP 88, a = 23.870(15), c = 4.433(3) Å, V = 2187(3) Å 3 , Z = 4]. The crystal structures of both compounds are characterized by hexagonal honeycomb-based motifs formed by Ni and Sn that extend along the c axis. The building motif with its three-blade wind turbine shape is reminiscent of the organic molecule triptycene and is unprecedented in extended solids. First-principles calculations have been performed in order to analyze the electronic structure and provide insight into chemical bonding. They reveal significant electron transfer from La to Ni and the respective p -element, which supports the formation of the polyanionic Ni- p -element network. DFT calculations suggest paramagnetic-like behavior for both compounds, which was confirmed by magnetic measurements.

High-throughput calculations can be applied to a large number of compounds, in order to discover new useful materials. In the present work, ternary intermetallic compounds are investigated, to find new potentially interesting materials for thermoelectric applications. The screening of stable nonmetallic compounds required for such applications is performed by calculating their electronic structure, using DFT methods. In the first section, the study of the density of states at the Fermi level, of pure elements, binary and ternary compounds, leads to empirically chose the selection criterion to distinguish metals from nonmetals. In the second section, the TiNiSi structure-type is used as a case-study application, through the investigation of 570 possible compositions. The screening leads to the selection of 12 possible semiconductors. The Seebeck coefficient and the lattice thermal conductivity of the selected compounds are calculated in order to identify the most promising ones. Among them, TiNiSi, TaNiP, or HfCoP are shown to be worth a detailed experimental investigation.

A very large thermoelectric figure of merit ZT = 6 at 380 K has recently been reported for Fe2V0.8W0.2Al in the thin-film form (B. Hinterleitner et al., Nature, 2019, 576, 85-90). In this form, Fe2V0.8W0.2Al experimentally crystallizes in a disordered A2 crystal structure, different from its bulk-form structure (L21). The first-principles calculations of the electronic structure performed in A2-Fe2V0.8W0.2Al supercells generated by the special quasirandom structure (SQS) method are thus reported here. These calculations unambiguously indicate that A2-Fe2V0.8W0.2Al is a ferromagnetic metal at 0 K, displaying a small Seebeck coefficient at 400 K (<30 μV K-1). The present results contradict the scenario of the occurrence of a deep pseudo-gap at the Fermi level, previously invoked to justify ZT = 6 in Fe2V0.8W0.2Al thin films.

Hexagonal La 2 Ni 7 and rhombohedral Y 2 Ni 7 are weak itinerant antiferromagnet (wAFM) and ferromagnet (wFM), respectively. To follow the evolution between these two compounds, the crystal structure and magnetic properties of A 2 B 7 intermetallic compounds ( A = La, Y, B = Ni) have been investigated combining x-ray powder diffraction and magnetic measurements. The La 2- x Y x Ni 7 intermetallic compounds with 0 ⩽ x ⩽ 1 crystallize in the hexagonal Ce 2 Ni 7 -type structure with Y preferentially located in the [ A 2 B 4 ] units. The compounds with larger Y content (1.2 ⩽ x < 2) crystallize in both hexagonal and rhombohedral (Gd 2 Co 7 -type) structures with a substitution of Y for La in both [ A 2 B 4 ] and [ AB 5 ] units. Y 2 Ni 7 crystallizes in the rhombohedral structure only. The average cell volume decreases linearly versus Y content, whereas the c / a ratio presents a minimum at x = 1 due to geometric constrains. The magnetic properties are strongly dependent on the structure type and the Y content. La 2 Ni 7 displays a complex metamagnetic behavior with split AFM peaks. Compounds with x = 0.25 and 0.5 display a wAFM ground state and two metamagnetic transitions, the first one toward an intermediate wAFM state and the second one toward a FM state. T N and the second critical field μ 0 H c2 increase with the Y content, indicating a stabilization of the AFM state. LaYNi 7 , which is as the boundary between the two structure types, presents a very wFM state at low field and an AFM state as the applied field increases. All the compounds with x > 1, and which contains a rhombohedral phase are wFM with T C = 53(2) K. In addition to the experimental studies, first principles calculations using spin polarization have been performed to interpret the evolution of structural phase stability for 0 ⩽ x ⩽ 2., (© 2020 IOP Publishing Ltd.)

The weak itinerant magnetic properties of A 2 Ni 7 compounds with A  =  {Y, La} have been investigated using electronic band structure calculations in the relation with their polymorphic crystal structures. These compounds crystallizes in two structures resulting from the stacking of two and three blocks of [A 2 Ni 4   +  2 ANi 5 ] units for hexagonal 2H-La 2 Ni 7 (Ce 2 Ni 7 type) and rhombohedral 3R-Y 2 Ni 7 (Gd 2 Co 7 type) respectively. Experimentally, 2H-La 2 Ni 7 is a weak itinerant antiferromagnet whereas 3R-Y 2 Ni 7 is a weak itinerant ferromagnet. From the present first principles calculation within non-spin polarized state, both compounds present an electronic density of state with a sharp and narrow peak centered at the Fermi level corresponding to flat bands from 3d-Ni. This induces a magnetic instability and both compounds are more stable in a ferromagnetic (FM) order compared to a paramagnetic state (ΔE  ≈  -35 meV/f.u.). The magnetic moment of each of the five Ni sites varies with their positions relative to the [A 2 Ni 4 ] and [ANi 5 ] units: they are minimum in the [A 2 Ni 4 ] unit and maximum at the interface between two [ANi 5 ] units. For 2H-La 2 Ni 7 , an antiferromagnetic (AFM) structure has been proposed and found with an energy comparable to that of the FM state. This AFM structure is described by two FM unit blocks of opposite Ni spin sign separated by a non-magnetic layer at z  =  0 and ½. The Ni (2a) atoms belonging to this intermediate layer are located in the [La 2 Ni 4 ] unit and are at a center of symmetry of the hexagonal cell (P6 3 /mmc) where the resultant molecular field is cancelled. Further non-collinear spin calculations have been performed to determine the Ni moment orientations which are found preferentially parallel to the c axis for both FM and AFM structures.

While the overwhelming number of papers on multi-principal-element alloys (MPEAs) focus on the mechanical and microstructural properties, there has been growing interest in these alloys as solid-state hydrogen stores. We report here the synthesis optimization, the physicochemical and the hydrogen sorption properties of Ti 0.325 V 0.275 Zr 0.125 Nb 0.275 . This alloy was prepared by two methods, high temperature arc melting and ball milling under Ar, and crystallizes into a single-phase bcc structure. This MPEA shows a single transition from the initial bcc phase to a final bct dihydride and a maximum uptake of 1.7 H/ M (2.5 wt%). Interestingly, the bct dihydride phase can be directly obtained by reactive ball milling under hydrogen pressure. The hydrogen desorption properties of the hydrides obtained by hydrogenation of the alloy prepared by arc melting or ball milling and by reactive ball milling have been compared. The best hydrogen sorption properties are shown by the material prepared by reactive ball milling. Despite a fading of the capacity for the first cycles, the reversible capacity of the latter material stabilizes around 2 wt%. To complement the experimental approach, a theoretical investigation combining a random distribution technique and first principle calculation was done to estimate the stability of the hydride.

The present study focuses on the impact of the vibrational frequencies on the thermodynamic behavior of hydrides, deuterides and tritides, using high scale harmonic phonon calculations based on first-principle calculations. 115 MH y hydrides were considered, for [Formula: see text] with M among 30 metallic elements. The results were found to be in good agreement with the available experimental data and pointed out trends on the evolution of the hydride zero point energy as a function of the crystal structure and the host metal nature. Based on this information, the vibration contribution to the formation enthalpy was deduced. This contribution is responsible for the differences between the enthalpies and therefore pressures of formation of the hydride, deuteride and tritide compounds. This so-called 'isotope effect' is experimentally observed but has never been studied by large scale calculations. A straightforward method has been developed allowing to quantify the isotope effect at non zero temperature. It explains the experimentally observed relative stability of hydride, deuteride and tritide compounds. As a major achievement, a new phenomenon was highlighted, which has never been anticipated, consisting in an inversion of the isotope effect when the temperature increases.

First-principles calculations were systematically performed for 31 binary metal-hydrogen (M-H) systems on a set of 30 potential crystal structures selected on the basis of experimental data and possible interstitial sites. For each M-H system, the calculated enthalpies of formation were represented as functions of H composition. The zero-point energy correction was considered for the most stable hydrides via additional harmonic phonon calculations. The sequence of stable hydrides (ground-state) given by the convex hull was found in satisfactory agreement with the experimental data. In addition, new high pressure dihydrides and trihydrides were predicted, providing orientations for new materials synthesis. The overall results provide a global overview of hydride relative stabilities and relevant input data for thermodynamic modeling methods.

Metastable β-As2Te3 (R3̅m, a = 4.047 Å and c = 29.492 Å at 300 K) is isostructural to layered Bi2Te3 and is known for similarly displaying good thermoelectric properties around 400 K. Crystallizing glassy-As2Te3 leads to multiphase samples, while β-As2Te3 could indeed be synthesized with good phase purity (97%) by melt quenching. As expected, β-As2Te3 reconstructively transforms into stable α-As2Te3 (C2/m, a = 14.337 Å, b = 4.015 Å, c = 9.887 Å, and β = 95.06°) at 480 K. This β → α transformation can be seen as the displacement of part of the As atoms from their As2Te3 layers into the van der Waals bonding interspace. Upon cooling, β-As2Te3 displacively transforms in two steps below T(S1) = 205-210 K and T(S2) = 193-197 K into a new β'-As2Te3 allotrope. These reversible and first-order phase transitions give rise to anomalies in the resistance and in the calorimetry measurements. The new monoclinic β'-As2Te3 crystal structure (P2(1)/m, a = 6.982 Å, b = 16.187 Å, c = 10.232 Å, β = 103.46° at 20 K) was solved from Rietveld refinements of X-ray and neutron powder patterns collected at low temperatures. These analyses showed that the distortion undergone by β-As2Te3 is accompanied by a 4-fold modulation along its b axis. In agreement with our experimental results, electronic structure calculations indicate that all three structures are semiconducting with the α-phase being the most stable one and the β'-phase being more stable than the β-phase. These calculations also confirm the occurrence of a van der Waals interspace between covalently bonded As2Te3 layers in all three structures.

Pd90Co10 and Pd75Co25 nanoalloys embedded into mesoporous carbon hosts have been prepared by two synthetic methods: direct and indirect. The average nanoparticles size can be tuned by both the temperature during thermal treatment and the chemical composition: the higher the treatment temperature and the richer the Pd composition, the larger the nanoparticle size. Twofold size- and composition-dependence of the hydrogen sorption properties at room temperature are evidenced. The Co substitution in Pd nanoalloys increases the equilibrium pressure at room temperature relative to nanosized Pd. The hydrogen sorption capacity decreases by Co substitution in Pd, as also demonstrated by SQS + DFT calculations.

In order to clarify controversial reports on the Fe-Re phase diagram, a new experimental investigation has been carried out. Three intermetallic phases have been evidenced, including the new report of the P phase found for the first time in a binary system. The phase relations involving the σ phase were established. In parallel, a first-principles study has been performed which provided the heat of formation of every ordered configuration for four intermetallic phases (D8(b), A12, A13 and P). The mixing energy of solid solutions (fcc, bcc, hcp) was calculated using the special quasi-random structure method. Calculations were performed with the help of the density functional theory, with and without spin polarization. From these results, in the frame of the Compound Energy Formalism using the Bragg-Williams approximation, the Fe-Re phase diagram has been computed without the use of adjustable parameters. Different thermodynamic parameters obtained experimentally and theoretically, as the site occupancies, are compared. The computed phase diagram presents several differences with the experimental one. To understand these differences, the influence of several parameters on the phase stability, such as the magnetic contribution has been evaluated.

The Frank-Kasper phases, known as topologically close-packed (tcp) phases, are interesting examples of intermetallic compounds able to accommodate large homogeneity ranges by atom mixing on different sites. Among them, the χ and σ phases present two competing complex crystallographic structures, the stability of which is driven by both geometric and electronic factors. Rhenium (Re) is the element forming the largest number of binary χ and σ phases. Its central position among the transition metals in the periodic table plays an important role in the element ordering in tcp phases. Indeed, it has been shown that Re shows an opposite site preference depending on which elements it is alloyed with. In the present work, χ- and σ-phase stability in binary Re-X systems is systematically studied by a first-principles investigation. The heats of formation of the complete set of ordered configurations (16 for χ and 32 for σ) have been calculated in 16 well-chosen systems to identify stability criteria. They include not only the systems in which χ-Re-X (X = Ti, Mn, Zr, Nb, Mo, Hf, Ta, W) or σ-Re-X (X = V, Cr, Mn, Fe, Nb, Mo, Ta, W) exist but also the systems in which both phases are not stable, including systems in which X is a 3d element from Ti to Ni, a 4d element from Zr to Ru, and a 5d element from Hf to Os. Careful analysis is done of the energetic tendencies as a function of recomposition, size effect, and electron concentration. Moreover, the site preference and other crystallographic properties are discussed. Conclusions are drawn concerning the relative stability of the two phases in comparison with the available experimental knowledge on the systems.

Nanocrystalline PrCo(3) powder has been synthesized by high-energy milling and was subsequently annealed from 873 to 1273 K for 30 min to optimize the extrinsic properties. The structure and magnetic properties of the nanocrystalline PrCo(3) have been investigated by means of x-ray and neutron diffraction as well as magnetization measurements. All compounds crystallize in the same PuNi(3) type structure, with grain sizes between 28 and 47 nm. As the annealing temperature increases, a maximum coercive field of 12 kOe at 300 K (55 kOe at 10 K) was obtained by annealing at 1023 K for a grain size of 35 nm. The refinement of the neutron powder diffraction patterns (NPD) of PrCo(3) from 1.8 to 300 K shows an expansion of the parameter a and a contraction of the parameter c, leading to a decrease of the ratio c/a. The evolution of the Co and Pr magnetic sublattices measured by NPD indicates that this compound is a highly anisotropic uniaxial ferromagnet with the easy magnetization axis parallel to c(-->). This experimental study has been completed by a theoretical investigation of the electronic structure of the PrCo(x) (x = 2, 3 and 5) compounds. Band structure calculations with collinear spin polarization were performed by using the local approximation of the density functional theory scheme implemented in the projector-augmented wave method. The electronic structure of PrCo(3) compound in both directions of spin shows that the majority of occupied states are dominated by the 3d states of Co, with a strong electronic charge transfer from Pr to Co. The PrCo(3) electronic structure can be explained by a superimposition of those of PrCo(2) and PrCo(5), as expected from its crystal structure. The magnetic anisotropy has been confirmed for PrCo(3), as a non-collinear spin calculation with the polarization along the c axis is shown to be more stable than with the polarization in the (a(-->),b(-->)) plane.

The catalytic hydrogenation of CO(2) at the surface of a metal hydride and the corresponding surface segregation were investigated. The surface processes on Mg(2)NiH(4) were analyzed by in situ X-ray photoelectron spectroscopy (XPS) combined with thermal desorption spectroscopy (TDS) and mass spectrometry (MS), and time-of-flight secondary ion mass spectrometry (ToF-SIMS). CO(2) hydrogenation on the hydride surface during hydrogen desorption was analyzed by catalytic activity measurement with a flow reactor, a gas chromatograph (GC) and MS. We conclude that for the CO(2) methanation reaction, the dissociation of H(2) molecules at the surface is not the rate controlling step but the dissociative adsorption of CO(2) molecules on the hydride surface., (This journal is © the Owner Societies 2012)

The site occupancies of the Mo-Ni-Re σ phase have been studied as a function of the composition in the ternary homogeneity domain by both experimental measurements and calculations. Because of the possible simultaneous occupancy of three elements on the five sites of the crystal structure, the experimental determination of the site occupancies was achieved by using combined Rietveld refinement of X-ray and neutron diffraction data, whereas calculation of the site occupancies was carried out by using the density functional theory results of every ordered (i.e., 3(5) = 243) configuration appearing in the ternary system. A comparison of the experimental and calculation results showed good agreement, which suggests that the topologically close-packed phases, such as the σ phase, could be described by the Bragg-Williams approximation (i.e., ignoring the short-range-order contributions). On the other hand, the atomic distribution on different crystallographic sites of the Mo-Ni-Re σ phase was found to be governed by the atomic sizes. Ni, having the smallest atomic size, showed a preference for low-coordination-number (CN) sites, whereas Mo, being the largest in atomic size, preferred occupying high-CN sites. However, the preference of Re, having intermediate atomic size, varied depending on the composition, and a clear reversal in the preference of Re as a function of the composition was evidenced in both the calculated and experimental site-occupancy results.

The total energies of all the ordered configurations of the σ and χ phases have been calculated by using first principles methodology in both Mo-Re and W-Re systems. These two complex structures possess 5 and 4 inequivalent sites generating 32 and 16 different ordered configurations, respectively, for a binary A-B system. The converged total energies of all the fully relaxed structures have been used to compute the occupancy of the inequivalent sites as a function of composition and temperature by using the Bragg-Williams approximation in the complete composition range. It is shown that the configurational entropy stabilizes the σ and χ phases in the Mo-Re and W-Re systems. The results evidence the preference of Re for lower coordination number site occupancy and are in very good agreement with the experimental measurements. Tentative ab initio phase diagrams have also been drawn.