arXiv subjects

Roberto Bergamaschini

Publications and source records attributed to Roberto Bergamaschini.

13 recordsLinked to original sources

arXiv2026

Neural surrogates for crystal growth dynamics with variable supersaturation: explicit vs. implicit conditioning

Simulations of crystal growth are performed by using Convolutional Recurrent Neural Network surrogate models, trained on a dataset of time sequences computed by numerical integration of Allen-Cahn dynamics including faceting via kinetic anisotropy. Two network architectures are developed to take into account the effects of a variable supersaturation value. The first infers it implicitly by processing an input mini-sequence of a few evolution frames and then returns a consistent continuation of the evolution. The second takes the supersaturation parameter as an explicit input along with a single initial frame and predicts the entire sequence. The two models are systematically tested to establish strengths and weaknesses, comparing the prediction performance for models trained on datasets of different size and, in the first architecture, different lengths of input mini-sequence. The analysis of point-wise and mean absolute errors shows how the explicit parameter conditioning guarantees the best results, reproducing with high-fidelity the ground-truth profiles. Comparable results are achievable by the mini-sequence approach only when using larger training datasets. The trained models show strong conditioning by the supersaturation parameter, consistently reproducing its overall impact on growth rates as well as its local effect on the faceted morphology. Moreover, they are perfectly scalable even on 256 times larger domains and can be successfully extended to more than 10 times longer sequences with limited error accumulation. The analysis highlights the potential and limits of these approaches in view of their general exploitation for crystal growth simulations.

cond-mat.mtrl-sci↗

arXiv2025

Unified machine-learning framework for property prediction and time-evolution simulation of strained alloy microstructure

We introduce a unified machine-learning framework designed to conveniently tackle the temporal evolution of alloy microstructures under the influence of an elastic field. This approach allows for the simultaneous extraction of elastic parameters from a short trajectory and for the prediction of further microstructure evolution under their influence. This is demonstrated by focusing on spinodal decomposition in the presence of a lattice mismatch eta, and by carrying out an extensive comparison between the ground-truth evolution supplied by phase field simulations and the predictions of suitable convolutional recurrent neural network architectures. The two tasks may then be performed subsequently into a cascade framework. Under a wide spectrum of misfit conditions, the here-presented cascade model accurately predicts eta and the full corresponding microstructure evolution, also when approaching critical conditions for spinodal decomposition. Scalability to larger computational domain sizes and mild extrapolation errors in time (for time sequences five times longer than the sampled ones during training) are demonstrated. The proposed framework is general and can be applied beyond the specific, prototypical system considered here as an example. Intriguingly, experimental videos could be used to infer unknown external parameters, prior to simulating further temporal evolution.

cond-mat.mtrl-sci↗

arXiv2025

Learning Kinetic Monte Carlo stochastic dynamics with Deep Generative Adversarial Networks

We show that Generative Adversarial Networks (GANs) may be fruitfully exploited to learn stochastic dynamics, surrogating traditional models while capturing thermal fluctuations. Specifically, we showcase the application to a two-dimensional, many-particle system, focusing on surface-step fluctuations and on the related time-dependent roughness. After the construction of a dataset based on Kinetic Monte Carlo simulations, a conditional GAN is trained to propagate stochastically the state of the system in time, allowing the generation of new sequences with a reduced computational cost. Modifications with respect to standard GANs, which facilitate convergence and increase accuracy, are discussed. The trained network is demonstrated to quantitatively reproduce equilibrium and kinetic properties, including scaling laws, with deviations of a few percent from the exact value. Extrapolation limits and future perspectives are critically discussed.

cond-mat.stat-mech↗

arXiv2024

Phase-field Modelling of Anisotropic Solid-State Dewetting on Patterned Substrates

We present a phase-field model for simulating the solid-state dewetting of anisotropic crystalline films on non-planar substrates. This model exploits two order parameters to trace implicitly the crystal free surface and the substrate profile in both two and three dimensions. First, we validate the model by comparing numerical simulation results for planar substrates with those obtained by a conventional phase-field approach and by assessing the convergence toward the equilibrium shape predicted by the Winterbottom construction. We then explore non-planar geometries, examining the combined effects of surface-energy anisotropies and parameters controlling the contact angle. Our findings reveal that crystalline particles on curved supports lose self-similarity and exhibit a volume-dependent apparent contact angle, with opposite trends for convex versus concave profiles. Additionally, we investigate the migration of faceted particles on substrates with variable curvature. Applying this model to experimentally relevant cases like spheroidal and pit-patterned substrates demonstrates various behaviors that could be leveraged to direct self-assembly of nanostructures, from ordered nanoparticles to interconnected networks with complex topology.

cond-mat.mes-hall↗

arXiv2024

Interface energies of Ga2O3 phases with the sapphire substrate and the phase-locked epitaxy of metastable structures explained

Despite the extensive work carried out on the epitaxial growth of Ga2O3, a fundamental understanding of the nucleation of its different metastable phases is still lacking. Here we address the role of interface energies by Density Functional Theory calculations of alpha, beta and kappa Ga2O3 on (0001) Al2O3 substrates, and different Ga2O3 interlayers. In conjunction to surface energies and misfit strain contribution, we demonstrate that alpha Ga2O3 is the preferred phase in 2D islands, when the low growth temperatures and the high growth rates hinder 3D island nucleation. This quantitatively explains the phase-locking in Mist-CVD experiments

cond-mat.mtrl-sci↗

arXiv2024

Extreme time extrapolation capabilities and thermodynamic consistency of physics-inspired Neural Networks for the 3D microstructure evolution of materials via Cahn-Hilliard flow

A Convolutional Recurrent Neural Network (CRNN) is trained to reproduce the evolution of the spinodal decomposition process in three dimensions as described by the Cahn-Hilliard equation. A specialized, physics-inspired architecture is proven to provide close accordance between the predicted evolutions and the ground truth ones obtained via conventional integration schemes. The method can accurately reproduce the evolution of microstructures not represented in the training set at a fraction of the computational costs. Extremely long-time extrapolation capabilities are achieved, up to reaching the theoretically expected equilibrium state of the system, consisting of a layered, phase-separated morphology, despite the training set containing only relatively-short, initial phases of the evolution. Quantitative accordance with the decay rate of the Free energy is also demonstrated up to the late coarsening stages, proving that this class of Machine Learning approaches can become a new and powerful tool for the long timescale and high throughput simulation of materials, while retaining thermodynamic consistency and high-accuracy.

cond-mat.mes-hall↗

arXiv2024

Quantitative analysis of the prediction performance of a Convolutional Neural Network evaluating the surface elastic energy of a strained film

A Deep Learning approach is devised to estimate the elastic energy density $\rho$ at the free surface of an undulated stressed film. About 190000 arbitrary surface profiles h(x) are randomly generated by Perlin noise and paired with the corresponding elastic energy density profiles $\rho(x)$, computed by a semi-analytical Green's function approximation, suitable for small-slope morphologies. The resulting dataset and smaller subsets of it are used for the training of a Fully Convolutional Neural Network. The trained models are shown to return quantitative predictions of $\rho$, not only in terms of convergence of the loss function during training, but also in validation and testing, with better results in the case of the larger dataset. Extensive tests are performed to assess the generalization capability of the Neural Network model when applied to profiles with localized features or assigned geometries not included in the original dataset. Moreover, its possible exploitation on domain sizes beyond the one used in the training is also analyzed in-depth. The conditions providing a one-to-one reproduction of the ground-truth $\rho(x)$ profiles computed by the Green's approximation are highlighted along with critical cases. The accuracy and robustness of the deep-learned $\rho(x)$ are further demonstrated in the time-integration of surface evolution problems described by simple partial differential equations of evaporation/condensation and surface diffusion.

physics.comp-ph↗

arXiv2022

Morphological evolution via surface diffusion learned by convolutional, recurrent neural networks: extrapolation and prediction uncertainty

We use a Convolutional Recurrent Neural Network approach to learn morphological evolution driven by surface diffusion. To this aim we first produce a training set using phase field simulations. Intentionally, we insert in such a set only relatively simple, isolated shapes. After proper data augmentation, training and validation, the model is shown to correctly predict also the evolution of previously unobserved morphologies and to have learned the correct scaling of the evolution time with size. Importantly, we quantify prediction uncertainties based on a bootstrap-aggregation procedure. The latter proved to be fundamental in pointing out high uncertainties when applying the model to more complex initial conditions (e.g. leading to splitting of high aspect-ratio individual structures). Automatic smart-augmentation of the training set and design of a hybrid simulation method are discussed.

physics.comp-ph↗

arXiv2022

Doubling the mobility of InAs/InGaAs selective area grown nanowires

Selective area growth (SAG) of nanowires and networks promise a route toward scalable electronics, photonics and quantum devices based on III-V semiconductor materials. The potential of high-mobility SAG nanowires however is not yet fully realized, since interfacial roughness, misfit dislocations at the nanowire/substrate interface and non-uniform composition due to material intermixing all scatter electrons. Here, we explore SAG of highly lattice-mismatched InAs nanowires on insulating GaAs(001) substrates and address these key challenges. Atomically smooth nanowire/substrate interfaces are achieved with the use of atomic hydrogen (a-H) as an alternative to conventional thermal annealing for the native oxide removal. The problem of high lattice mismatch is addressed through an In$_x$Ga$_{1-x}$As buffer layer introduced between the InAs transport channel and the GaAs substrate. The Ga-In material intermixing observed in both the buffer layer and the channel is inhibited via careful tuning of the growth temperature. Performing scanning transmission electron microscopy and x-ray diffraction analysis along with low-temperature transport measurements we show that optimized In-rich buffer layers promote high quality InAs transport channels with the field-effect electron mobility over~10000~cm$^2$V$^{-1}$s$^{-1}$. This is twice as high as for non-optimized samples and among the highest reported for InAs selective area grown nanostructures.

cond-mat.mtrl-sci↗

arXiv2020

Reentrant behavior of the density vs temperature of indium islands on GaAs(111)A

We show that the density of indium islands on GaAs(111)A substrates have a non-monotonic, reentrant behavior as a function of the indium deposition temperature. The expected increase in the density with decreasing temperature, indeed, is observed only down to 160 °C, where the indium islands undertake the expected liquid-to-solid phase transition. Further decreasing the temperature causes a sizeable reduction of the island density. An additional, reentrant increasing behavior is observed below 80 °C. We attribute the above complex behavior to the liquid-solid phase transition and to the complex island-island interaction which takes place between crystalline islands in the presence of strain. Indium solid islands grown at temperatures below 160 °C have a face-centered cubic crystal structure.

cond-mat.mtrl-sci↗

arXiv2020

Kinetic Control of Morphology and Composition in Ge/GeSn Core/Shell Nanowires

The growth of Sn-rich group-IV semiconductors at the nanoscale provides new paths for understanding the fundamental properties of metastable GeSn alloys. Here, we demonstrate the effect of the growth conditions on the morphology and composition of Ge/GeSn core/shell nanowires by correlating the experimental observations with a theoretical interpretation based on a multi-scale approach. We show that the cross-sectional morphology of Ge/GeSn core/shell nanowires changes from hexagonal to dodecagonal upon increasing the supply of the Sn precursor. This transformation strongly influences the Sn distribution as a higher Sn content is measured under the {112} growth front. Ab-initio DFT calculations provide an atomic-scale explanation by showing that Sn incorporation is favored at the {112} surfaces, where the Ge bonds are tensile-strained. A phase-field continuum model was developed to reproduce the morphological transformation and the Sn distribution within the wire, shedding light on the complex growth mechanism and unveiling the relation between segregation and faceting. The tunability of the photoluminescence emission with the change in composition and morphology of the GeSn shell highlights the potential of the core/shell nanowire system for opto-electronic devices operating at mid-infrared wavelengths.

cond-mat.mtrl-sci↗

arXiv2019

Strain engineering in Ge/GeSn core/shell nanowires

Strain engineering in Sn-rich group IV semiconductors is a key enabling factor to exploit the direct band gap at mid-infrared wavelengths. Here, we investigate the effect of strain on the growth of GeSn alloys in a Ge/GeSn core/shell nanowire geometry. Incorporation of Sn content in the 10-20 at.% range is achieved with Ge core diameters ranging from 50nm to 100nm. While the smaller cores lead to the formation of a regular and homogeneous GeSn shell, larger cores lead to the formation of multi-faceted sidewalls and broadened segregation domains, inducing the nucleation of defects. This behavior is rationalized in terms of the different residual strain, as obtained by realistic finite element method simulations. The extended analysis of the strain relaxation as a function of core and shell sizes, in comparison with the conventional planar geometry, provides a deeper understanding of the role of strain in the epitaxy of metastable GeSn semiconductors.

physics.app-ph↗

arXiv2017

Optically reconfigurable polarized emission in Germanium

Light polarization can conveniently encode information. Yet, the ability to steer polarized optical fields is notably demanding but crucial to develop practical methods for data encryption and to gather fundamental insights into light-matter interactions. Here we demonstrate the dynamic manipulation of the chirality of light in the telecom wavelength regime. This unique possibility is enrooted in the multivalley nature of the conduction band of a conventional semiconductor, namely Ge. In particular, we demonstrate that optical pumping suffices to govern the kinetics of spin-polarized carriers and eventually the chirality of the radiative recombination. We found that the polarization of the optical transitions can be remarkably swept through orthogonal eigenstates without magnetic field control or phase shifter coupling. Our results anticipate spin-enhanced optoelectronic functionalities in materials backed by the mainstream microelectronics industry and provide guiding information for the selection and design of functionalized emitters based on novel group IV compounds.

cond-mat.mtrl-sci↗