arXiv subjects

M. V. Fernandez-Serra

Publications and source records attributed to M. V. Fernandez-Serra.

7 recordsLinked to original sources

arXiv2016

Continuous melting through a hexatic phase in confined bilayer water

Liquid water is not only of obvious importance but also extremely intriguing, displaying many anomalies that still challenge our understanding of such an a priori simple system. The same is true when looking at nanoconfined water: The liquid between constituents in a cell is confined to such dimensions, and there is already evidence that such water can behave very differently from its bulk counterpart. A striking finding has been reported from computer simulations for two-dimensionally confined water: The liquid displays continuous or discontinuous melting depending on its density. In order to understand this behavior, we have analyzed the melting exhibited by a bilayer of nanoconfined water by means of molecular dynamics simulations. At high density we observe the continuous melting to be related to the phase change of the oxygens only, with the hydrogens remaining liquid-like throughout. Moreover, we find an intermediate hexatic phase for the oxygens between the liquid and a triangular solid ice phase, following the Kosterlitz-Thouless-Halperin-Nelson-Young theory for two-dimensional melting. The liquid itself tends to maintain the local structure of the triangular ice, with its two layers being strongly correlated, yet with very slow exchange of matter. The decoupling in the behavior of the oxygens and hydrogens gives rise to a regime in which the complexity of water seems to disappear, resulting in what resembles a simple monoatomic liquid. This intrinsic tendency of our simulated water may be useful for understanding novel behaviors in other confined and interfacial water systems.

cond-mat.soft↗

arXiv2012

Ferroelectric PbTiO$_{3}$/SrRuO$_{3}$ superlattices with broken inversion symmetry

We have fabricated PbTiO$_{3}$/SrRuO$_{3}$ superlattices with ultra-thin SrRuO$_{3}$ layers. Due to the superlattice geometry, the samples show a large anisotropy in their electrical resistivity, which can be controlled by changing the thickness of the PbTiO$_{3}$ layers. Therefore, along the ferroelectric direction, SrRuO$_{3}$ layers can act as dielectric, rather than metallic, elements. We show that, by reducing the concentration of PbTiO$_{3}$, an increasingly important effect of polarization asymmetry due to compositional inversion symmetry breaking occurs. The results are significant as they represent a new class of ferroelectric superlattices, with a rich and complex phase diagram. By expanding our set of materials we are able to introduce new behaviors that can only occur when one of the materials is not a perovskite titanate. Here, compositional inversion symmetry breaking in bi-color superlattices, due to the combined variation of A and B site ions within the superlattice, is demonstrated using a combination of experimental measurements and first principles density functional theory.

cond-mat.mtrl-sci↗

arXiv2012

Anomalous Nuclear Quantum Effects in Ice

One striking anomaly of water ice has been largely neglected and never explained. Replacing hydrogen ($^1$H) by deuterium ($^2$H) causes ice to expand, whereas the "normal" isotope effect is volume contraction with increased mass. Furthermore, the anomaly increases with temperature $T$, even though a normal isotope shift should decrease with $T$ and vanish when $T$ is high enough to use classical nuclear motions. In this study, we show that these effects are very well described by {\it ab initio} density functional theory. Our theoretical modeling explains these anomalies, and allows us to predict and to experimentally confirm a counter effect, namely that replacement of $^{16}$O by $^{18}$O causes a normal lattice contraction.

cond-mat.mtrl-sci↗

arXiv2005

On electrons and hydrogen-bond connectivity in liquid water

The network connectivity in liquid water is revised in terms of electronic signatures of hydrogen bonds (HBs) instead of geometric criteria, in view of recent X-ray absorption studies. The analysis is based on ab initio molecular-dynamics simulations at ambient conditions. Even if instantaneous thread-like structures are observed in the electronic network, they continuously reshape in oscillations reminiscent of the r and t modes in ice (tau~170 fs). However, two water molecules initially joint by a HB remain effectively bound over many periods regardless of its electronic signature.

cond-mat.other↗

arXiv2005

Two exchange-correlation functionals compared for first-principles liquid water

The first-principles description of liquid water using ab initio molecular dynamics (AIMD) based on Density Functional theory (DFT) has recently been found to require long equilibration times, giving too low diffusivities and a clear over-structuring of the liquid. In the light of these findings we compare here the room-temperature description offered by two different exchange correlation functionals: BLYP, the most popular for liquid water so far, and RPBE, a revision of the widely used PBE. We find for RPBE a less structured liquid with radial distribution functions closer to the experimental ones than the ones of BLYP. The diffusivity obtained with RPBE for heavy water is still 20% lower than the corresponding experimental value, but it represents a substantial improvement on the BLYP value, one order of magnitude lower than experiment. These characteristics and the hydrogen-bond (HB) network imperfection point to an effective temperature ~3% lower than the actual simulation temperature for the RPBE liquid, as compared with BLYP's ~17% deviation. The too long O--O average nearest-neighbor distance observed points to an excessively weak HB, possibly compensating more fundamental errors in the DFT description.

cond-mat.soft↗

arXiv2005

Network equilibration and first-principles liquid water

Motivated by the very low diffusivity recently found in ab initio simulations of liquid water, we have studied its dependence with temperature, system size, and duration of the simulations. We use ab initio molecular dynamics (AIMD), following the Born-Oppenheimer forces obtained from density-functional theory (DFT). The linear-scaling capability of our method allows the consideration of larger system sizes (up to 128 molecules in this study), even if the main emphasis of this work is in the time scale. We obtain diffusivities that are substantially lower than the experimental values, in agreement with recent findings using similar methods. A fairly good agreement with D(T) experiments is obtained if the simulation temperature is scaled down by ~20%. It is still an open question whether the deviation is due to the limited accuracy of present density functionals or to quantum fluctuations, but neither technical approximations (basis set, localization for linear scaling) nor the system size (down to 32 molecules) deteriorate the DFT description in an appreciable way. We find that the need for long equilibration times is consequence of the slow process of rearranging the H-bond network (at least 20 ps at AIMD's room temperature). The diffusivity is observed to be very directly linked to network imperfection. This link does not appear an artefact of the simulations, but a genuine property of liquid water.

cond-mat.soft↗

arXiv2002

Model Hessian for accelerating first-principles structure optimizations

We present two methods to accelerate first-principles structural relaxations, both based on the dynamical matrix obtained from a universal model of springs for bond stretching and bending. Despite its simplicity, the normal modes of this model Hessian represent excellent internal coordinates for molecules and solids irrespective of coordination, capturing not only the long-wavelength acoustic modes of large systems, but also the short-wavelength low-frequency modes that appear in complex systems. In the first method, the model Hessian is used to precondition a conjugate gradients minimization, thereby drastically reducing the effective spectral width and thus obtaining a substantial improvement of convergence. The same Hessian is used in the second method as a starting point of a quasi-Newton algorithm (Broyden's method and modifications thereof), reducing the number of steps needed to find the correct Hessian. Results for both methods are presented for geometry optimizations of clusters, slabs, and biomolecules, with speed-up factors between 2 and 8.

cond-mat.mtrl-sci↗