People

The Executive Committee is in charge of the coordination and implementation of MARVEL's  activities, in collaboration with the Management Team.

The Strategic Committee is in charge of planning MARVEL's strategic activities and is composed of the Executive Committee and one representative each from CSCS, Empa, and PSI.

A team of administrative and scientific collaborators carries out MARVEL's management and financial duties and takes care of responsibilities such as communication, knowledge and technology transfer, education and equal opportunities.

The MARVEL data team, with one representative per project, is at disposal to address questions regarding open research data policy within MARVEL, with the role of being the point of contact in the group, project, institution; of raising awareness on data management; and of checking the minimal requirements presented in the MARVEL research data strategy. The data manager is Lidia Favre-Quattropani, MARVEL scientific manager. 

The Scientific Advisory Board has the formal role of suggesting research lines that need to be strengthened within the NCCR, or that have run their course, and act as an independent reviewer of the activities of the NCCR in addition to the official Review Panel of the SNSF. The Scientific Advisory Board is chaired by Giulia Galli.

The Industrial Advisory Board has the formal role of giving feedback to the NCCR regarding the industrial interest of research orientation and to suggest new directions. They act as independent reviewers of the activities of the NCCR in addition to the official Review Panel of the SNSF. The Industrial Advisory Board is chaired by Erich Wimmer.

Pillar 1 — Design and Discovery of Novel Materials. This pillar is dedicated to two phase II projects that have shown significant scientific and economic impact — metal-organic frameworks and alloys for additive manufacturing.

Recent developments in materials science and chemistry have created a problem of luxury: we can conceive of many more materials than we can ever imagine synthesizing and testing. Finding the optimal material for any particular application has therefore become increasingly difficult.

We think that identifying the highest-performing materials for a specific application will likely be achieved only by the adroit combination of high-throughput first-principles computations, theory and simulation, and machine learning, guiding experimental teams towards the most promising materials. To this end, we are building on the experience gained in MARVEL's phases I and II to develop a general framework for guiding experimental efforts in material discovery and creation. We are developing this framework in the areas of nanoporous materials (e.g., metal-organic frameworks (MOFs), covalent organic frameworks (COFs), zeolites, etc.) and multicomponent metal alloys (traditional dilute and, especially, new high-entropy alloys). Both topics have immense promise for applications but pose the challenge of a vast composition space. The computational methods developed in phase III will therefore have a significant impact.

The project is led by William Curtin and Berend Smit.

Pillar 2 — Machine Learning Platform for Molecules and Materials. This pillar aims to further strengthen the leadership of the Swiss research community in the use of machine learning to computational materials design.

Since the start of NCCR MARVEL, machine-learning (ML) techniques have become an integral part of the atomistic simulation toolbox, extending the reach and improving the accuracy of chemical modeling and driving materials discovery to new frontiers. MARVEL has made pioneering contributions to this field, ranging from the most fundamental understanding of the mathematical structure of descriptors used in atomistic ML, to challenging applications to basic chemistry and to materials of great technological relevance.

In the third and last phase of the NCCR, our ambitious research plan further bolsters our leadership in the application of ML to computational materials design.  It is built around a core infrastructural effort that both harmonizes the constellation of software tools developed during MARVEL's phase I and II and ensures a long-lasting legacy for the NCCR, keeping a forward-looking perspective. The final phase will allow us to continue advancing the state-of-the-art in atomistic machine learning, and integrate new PIs who will help deepen our understanding of the fundamental nature of atomistic machine learning and draw bridges with the study of quantum materials.

The project is led by Clémence Corminboeuf and Michele Ceriotti.

Pillar 3 — Digital Infrastructure of Open Simulations and Data. This pillar aims to provide the critical digital infrastructure needed to enable materials design and discovery in the 21st century. It focuses on the three main objectives of: 1) preparing for the exascale challenge; 2) making all codes, algorithms and workflows widely usable in a simple and optimized way, also for non-experts; and 3) delivering a self-sustaining long-term simulations infrastructure for the whole Swiss materials-science research ecosystem and beyond.

Preparing for the exascale challenge means tackling issues around code performance and scalability on (pre-)exascale machines via domain specific libraries (SIRIUS) and custom algorithms (Ensemble DFT, MaZe, ...), and developing the high-throughput performance to drive millions of individual HPC simulations via AiiDA. The services provided by the AiiDAlab infrastructure and the Quantum Mobile virtual machine will help us simplify access to codes, algorithms and workflows, ensure reproducibility of all research and facilitate FAIR access and sharing not only of the resulting data via the Materials Cloud portal, but also directly of the simulation capabilities.

By including an ability to deliver automated simulations as microservices, this pillar will establish a self-sustaining infrastructure for simulations that reaches well beyond the end of the NCCR. MARVEL's Digital Infrastructure will thus become a key enabler for future research in the field, extending to novel approaches such as autonomous laboratories with a closed feedback loop between automated simulations and robotic experiments, and thus further accelerating materials discovery and characterization.

The project is led by Giovanni Pizzi and Joost VandeVondele.

Pillar 4 — Long-term Integration in the Swiss Scientific Landscape. This pillar is dedicated to integrating MARVEL into the Swiss scientific landscape.

This translates into a strategic, long-term effort nucleating at PSI and Empa, where MARVEL's simulation and data capabilities will be developed further and become broadly available to the thousands of researchers within these labs. We will continue our strategic partnership with CSCS and ignite a new, lasting collaboration with Swiss Data Science Center (SDSC).

The efforts will be bolstered by PSI's new division dedicated to Scientific Computing, Theory and Data. Directed by Prof. Christian Rüegg, it comprises a number of scientific and technical laboratories, including a Laboratory for Materials Simulations developed in close partnership with MARVEL. This lab features three groups: Materials Software and Data, dedicated to expanding simulation capabilities for the prediction and characterization of materials properties; Multiscale Materials Modelling; and Light-Matter Interactions.  

Empa is also expanding its focus on accelerating materials design, all while targeting the reproducibility of experimental and computational research. Here, Dr. Carlo Pignedoli will take a leadership role. Resources are being invested in the creation of tools that boost cooperation between experiments and simulation. This enables high levels of standardization in how calculations for a specific experiment are performed and how results are analyzed and processed. Tools such as AiiDAlab are, in fact, specifically designed to enable experimentalists to independently access the submission of calculations and the analysis of the automatically processed data. 

The project is led by Nicola Marzari, Christian Rüegg and Carlo Pignedoli.

Continued Support — Advanced Simulation Methods. This project focuses on the integration of methods developed around correlated materials and on devices already started in phase II. The efforts are spread across three sub-projects: Accurate open source simulation framework for correlated materials; Correlations in realistic systems; and Electrical, thermal, and optical effects in  2D nanodevices.

In the first two phases of MARVEL, substantial resources were invested into the development of an accurate and parameter-free ab initio simulation method for correlated electron material. This class of materials exhibits some of the most remarkable phenomenon found in condensed matter systems, but their strongly correlated nature requires a theoretical treatment that goes beyond density functional theory (DFT) or weak-coupling perturbation theory. The new approach is based on the combination of dynamical meanfield theory (DMFT) with the GW method and has been successfully tested. The goal for phase III is to turn this complex simulation framework into an efficient, easily usable and well-documented open source code that can handle O(10) correlated orbitals per site, treat several sites in a unit cell, and read ab initio input from different widely used codes such as FLEUR or Quantum ESPRESSO. 

Simulations of atomistic models as close as possible to reality are key to success in meeting the challenge of realizing atomically precise bottom-up synthesis of graphene-derived structures, themselves opening up the way to the design of carbon-based nanomaterials and devices. A specific challenges in this regard is the lack of efficient and reliable methods capable of describing electron transport in reduced-size systems where screening effects play a fundamental role. The main objective in phase III is to develop a computational tool for calculating the conductance in these regimes.

Scaling down the dimensions of electronic devices to the sub-nanometer range not only influences their electronic transport properties, but also their electron-phonon and light-matter interactions. Being able to accurately model them has become critical to the design of ultra-scaled field-effect transistors (FET), thermoelectric nano-generators (TEG), and highly efficient photovoltaic (PV) cells. The objective of this project is therefore to set up an environment to automatically generate the required parameters and apply them to the modeling of nanodevices, for example ultra-scaled transistors with various 2D channel materials. The work will involve collaborations with experimental groups at ETHZ, EPFL, and Empa.

The project is led by Philipp Werner, Daniele Passerone and Mathieu Luisier.

Bonus — Leveraging Quantum Computers and Algorithms for Materials Discovery. This bonus project is driven by an interdisciplinary team that aims to develop and deploy advanced quantum algorithms for electronic structure simulations and material science.

The development of unprecedented capabilities in quantum computing hardware has marked the beginning of a potentially revolutionary new era for scientific discovery of quantum phenomena. The technology is perceived as having remarkable potential for scientific discovery, suggesting access to entirely unexplored computational realms that were unattainable in the classical setting.

While quantum hardware development is growing at an unprecedented pace, successful near- and mid-term applications of quantum computing to key scientific and technological discoveries will undoubtedly happen within the context of hybrid classical-quantum computing approaches. This bonus project allows us to establish new key synergies between the classical and the quantum world, aiming at closing the gap with more traditional approaches and set the path towards unexplored electronic structure calculations beyond the accuracy currently attainable with classical computing only. 

We are focusing on two main aspects. On one hand, we are pursuing the development of core quantum algorithms for the simulation of electronic problems, leveraging the most advanced hybrid classical-quantum algorithms based on variational methods. On the other hand, we are tackling the important task of interfacing well established classical software for electronic structure calculations with quantum computing accelerators through IBM's Qiskit platform. These approaches are of strategic scientific interest for Switzerland and instrumental in devising the next generation of applied quantum computing software and complementing the strong development of experimental platforms.

The project is led by Giuseppe Carleo.

Molecular materials, e.g. those formed by pharmaceutical compounds, pose formidable modeling challenges.  This D&D project has developed, demonstrated, and distributed breakthrough simulation techniques, powered by accurate quantum calculations, advanced statistical sampling and machine learning, to address these challenges, increasing the predictive power of simulations and bringing together computation and experiments.

The project D&D1 — Understanding Complex Molecular Crystals: Structures and Properties — unites the expertise of several groups within the MARVEL collaboration, and leverages some of the outcomes of the development work of phase I, to bring about a new level of understanding of molecular crystals. By combining physics-based and data-driven modeling this project is demonstrating how larger and more flexible compounds, such as those that make up a large fraction of last-generation drugs, can be modeled with high accuracy, describing their stability and their properties in a way that complements and augments experimental investigations.

The project is led by Michele Ceriotti.

The ultimate objective of the project D&D2 —  From Atoms to Additive Manufacturing: Computational Design of Complex Metal Alloys — is to develop a computational framework for the creation and control of precipitate-strengthened metal alloys, with a major application goal of guiding emerging additive manufacturing methodologies.  The technical objectives that underly this ultimate objective are to predict a priori, and validate experimentally, the structure and energetics of complex precipitate phases that form in advanced engineering metal alloys; the resulting mechanical strength (plastic flow stress) as a function of precipitate morphology and alloy chemistry; and the early-stage evolution of the alloy as a function of time-temperature history.

The project is led by William Curtin.

The D&D3 project — MARLON: MARVEL Design, Discovery and Engineering of Low-Dimensional Materials and Nanostructures — is dedicated to the design or discovery of novel low-dimensional materials and nanostructures and in the engineering and integration of these into novel architectures, devices, and processes. Key assets are the computational identification of a large portfolio of  about 1'800 novel two-dimensional (2D) materials, outcome of a phase I VP2 project, exquisite control of devices based on bottom-up nanostructures in the Empa laboratory of [email protected], driven by electronic-structure simulations, and in the involvement of a new PI at ETHZ that ensures a tight integration between first-principles calculations and large-scale device modeling (Mathieu Luisier, Nano TCAD).

The project is led by Nicola Marzari.

Recent developments in chemistry and materials science have made it possible to make millions of novel nanoporous materials. We can generate a spectrum of distinct materials simply by changing the building blocks in the synthesis route of metal-organic frameworks (MOFs) and related advanced nanoporous materials. Crucially, these materials have important applications related to energy, such as in carbon capture, and other gas separations, gas storage, and catalysis. Indeed, their unique chemical tunability offers the potential to decisively change the field: instead of relying on traditional methods of trial and error, we now have the chemistry at hand to tailor-make materials.

The promise of finding just the right material among millions of hypothetical structures, however, seems remote because of practical limitations. The number of materials we can synthesize, characterize, and fully test is only a tiny fraction of all possibilities. These experimental limitations make it difficult, if not impossible, to systematically search the entire chemical space of these materials. To take full advantage of these materials, project D&D4 aims to develop computational techniques to rapidly screen large number of possible materials and to obtain fundamental molecular insights into what the ideal material looks like for applications of these materials ranging from gas separation, (photo)catalysis, and sensing.

The project is led by Berend Smit.

Transition metal oxides are of fundamental interest because of their emerging properties -- exotic magnetism and superconductivity, divergent susceptibilities, metal-insulator transitions, etc. -- which are caused by the strong correlations between the electrons. Recent developments in experimental thin-film growth techniques offer the possibility to create oxide heterostructures, allowing materials with different chemistries to be combined to optimize existing behaviors and engineer entirely new functionalities. The primary objective of project D&D5 — Correlated Transition Metal Oxides and Heterostructures — is to design correlated oxides and heterostructures with novel functionalities that can enable disruptive technologies for electronics and energy applications. We develop the methodologies required for predicting the properties of correlated oxides, incorporate them into publicly available codes, and apply them to materials design, focusing in particular on the interplay between chemistry, dimensionality and defect chemistry, all of which are key to a complete description of oxide heterostructure physics and function.

The project is led by Nicola Spaldin.

Project D&D6 searches for novel topological materials by combining conceptual development in the classification of topological phases with computational materials design and discovery. The scope of materials includes topological insulators, semimetals, but also materials realizing quantum spin liquid phases and twisted moiré superlattices. The predictions are supported by a broad range of experimental techniques, notably by angle-resolved photoemission and optical spectroscopies.  The grand challenge of the project is realizing technological applications of materials belonging to this diverse, emerging class.

The project is led by Oleg Yazyev.

Towards next-generation electrochemical energy storage solutions, inorganic crystalline solid-state ionic conductors can play a major role in developing a safer and more energy dense battery.

Incubator 1 — Design and Discovery of Novel Solid-State Ionic Conductors — is a synergistic project based on continuous feedback between theory and experiment to accelerate the discovery of novel materials with improved ionic conductivity, wide electrochemical stability window, and favorable mechanical properties. Key contributions include the development of a mixed DFT method for computational screening with large structural databases (EPFL), expertise in first-principles methods and the development of neural-network techniques (IBM) for the theoretical characterization of diffusion and static properties as well as expertise in materials synthesis and characterization techniques including testing protocols for solid-state battery cells (PSI). Most of the computational contributions are available to the scientific community as AiiDA workflows.

The project is led by Teodoro Laino.

Supervised learning of materials properties of molecules and solid materials has been demonstrated to afford machine learning (ML) models which enable property predictions of new materials with low approximation error. Equally important, their computational cost is negligible when compared to conventional quantum methods, such as density functional theory (DFT). Over the first phase of MARVEL, von Lilienfeld and Ceriotti have made substantial progress on the development and application of unsupervised and supervised ML models. The findings have enabled the systematic and rigorous classification of materials, the identification and ranking of low-dimensional governing collective variables, as well as reaching an unprecedented combination of computational speed and predictive power across compositional and configurational compound space. The latter aspect subsequently enabled the discovery of ~90 crystals of the Elpasolite structure, predicted to be thermodynamically stable.

The project is led by Analole von Lilienfeld.

The Open Science Platform represents the core structural effort of MARVEL that would not have been possible and would not exist without this NCCR. At the same time, it aims to be a world-class effort — enabled by our expertise in automatic, robust high-throughput calculations — to develop a computational facility: an infrastructure, based on AiiDA, comparable to that of a synchrotron or another major brick-and-mortar installation where multiple services are offered to the scientific community. These services start with first-principles calculations: from the capability of obtaining equilibrium geometries and energetics with fail-safe “turnkey” solutions, to develop new methods and workflows for computational spectroscopies and microscopies, where vibrational, magnetic, and optical properties, and STM or TEM images can be obtained for a target material or molecule. The Materials Cloud portal provides access to these capabilities, in an Open Science model where calculations can be deployed on the cloud.

This project — HPC and Future Architectures —supports the deployment of the hardware resource for MARVEL, located at CSCS, and hence is mostly an infrastructure effort, with a significant element of co-design, to ensure that there is convergence on an optimal hardware and software roadmap. In addition, it sustains the development of the SIRIUS domain-specific library, which provides performance portability to pre-Exascale hardware for plane waves electronic-structure codes and guarantees their suitability for high throughput calculations. As matching services, CSCS keeps providing an optimal environment for the deployment of the Materials Cloud, and for provisioning the hardware resources.