• ChemAlive - MARVEL collaboration wins grant to boost machine learning approaches to real-time chemical and reaction modelling

    Founded in December 2014, ChemAlive (www.chemalive.com) has already gained contract business with customers such as Biosynth in Switzerland and internationally with Saudi Aramco (Aramco Services Center) to validate quantum chemistry in an industrial setting and guide the design of specific on-line tools to be launched as a software as a service (SaaS) in the third quarter of 2018. ChemAlive has also won the support of influential start-up accelerators, taking away the Gold Award, for example, from the MassChallenge Accelerator.

  • 2D or not 2D? MARVEL algorithm answers the question

    Two-dimensional materials, such as graphene, are a new and exciting class of materials. No more than a few atomic layers thick, they have the most extraordinary properties, making them attractive for all kinds of applications. However, despite high expectations, progress in identifying new 2D materials has been slow: to date, only a few dozen have been identified experimentally. In their just-out Nature Nanotechnology paper, which made the cover page, "Two-dimensional materials from high-throughput computational exfoliation of experimentally know compounds", a team led by MARVEL director Nicola Marzari takes a computational route towards improving that count. 

  • AiiDA manages, preserves and disseminates the simulations, data and workflows of modern computational science

    Advances in high-performance computing, improvements in the computer codes behind quantum–mechanical simulations and the emergence of curated materials databases have turned high-throughput computing into an essential tool in materials discovery and design.  

  • MARVEL team shows how properties of amorphous aluminum oxide can be tuned by electrochemical anodizing

    Research groups supported by NCCR MARVEL used a combination of experimental and theoretical methods to show that structural and dielectric properties of amorphous aluminum oxide can be tuned through electrochemical anodizing, an industrially relevant surface treatment technique.

  • Machine learning accelerates discovery of new 2D structure of titanium dioxide

    A new combination of structure prediction and machine learning methods has led to the discovery of a new 2D titanium dioxide structure that could eventually be used in hydrogen generation and energy storage.

  • New machine learning approach could accelerate materials optimization and drug discovery

    Researchers have developed a machine-learning model that may greatly accelerate drug discovery by accurately predicting the interactions between a protein and a drug molecule using only a handful of reference experiments or simulations. The algorithm, which can also tackle materials science problems such as modelling the structure of silicon surfaces, promises to revolutionise materials and chemical modelling, and gives insight into the nature of intermolecular forces.

  • Machine learning presents a huge opportunity to identify new materials at a reduced cost

    A new machine learning-based approach to predicting the properties of materials cut the computing power needed to analyze two million crystals from more than 20 million hours on a supercomputer to an afternoon on a laptop, the cost from as much as CHF 2 million to, essentially, nothing, and identified 90 previously unknown, thermodynamically stable crystals in the process. 

  • Type II Dirac fermions, first predicted theoretically by MARVEL researchers, now experimentally verified

    The usual approach to scientific discovery is "experiments first, theory second". But sometimes things work the other way around. In a 2015 Nature article, a MARVEL team led by Alexey Soluyanov predicted that a novel type of particle, type II Weyl fermions, can be hosted as quasiparticles in topological semimetals. In a remarkable triumph for their computational approach, the existence of such quasiparticles has now been confirmed experimentally by five independent studies.

  • A new tool for discovering nanoporous materials

    In their paper Quantifying similarity of pore-geometry in nanoporous materials, Prof. Berend Smit and his MARVEL team present a mathematical method for describing and comparing pore shape in nanoporous materials. For nanoporous materials, pore shape is as important a determinant of performance as chemical composition. However, an effective method to quantify similarity of pore geometry had been lacking so far. 

  • A breakthrough on Weyl semimetals

    It is not often that materials science ventures into the territory of particle physics — but with the study of Weyl semimetals, it does. In Weyl semimetals, energy bands cross at pairs of points at the Fermi level, and the low-energy excitations near these crossing points fulfil the Weyl equation. In other words, thanks to their unusual band topology, Weyl semimetals realize Weyl fermions as quasiparticles. This is exciting news for fundamental physics: Weyl fermions have been predicted in 1929 but remain elusive in high energy experiments. Weyl semi-metals allow us to study these fundamental particles and their exotic physics in condensed matter.

  • Artificial intelligence helps in the discovery of new materials

    With the help of artificial intelligence, MARVEL chemists from the group of Prof. Anatole von Lilienfeld at the University of Basel have computed the characteristics of about two million crystals made up of four chemical elements. The researchers were able to identify 90 previously unknown thermodynamically stable crystals that can be regarded as new materials. 

  • Elusive quasiparticle discovered in diphosphide materials

    A team of MARVEL researchers of both the group of Prof. Oleg Yazyev at EPFL and the group of Prof. Matthias Troyer at ETHZ have predicted the existence of Weyl fermions in two Weyl semimetals, molybdenum diphosphide and tungsten diphosphide. The scientists used a high-throughput computational method screening a large database of existing materials, performing electronic-structure computations for each candidate.