L'EPFL a décroché 50 millions d'euros de «grants» malgré les écueils
La course aux «ERC Grants» a souri à l’EPFL ces deux dernières années. Entre 2014 et 2015, onze chercheurs ont reçu un «Advanced grant» dans le cadre du programme «Horizon 2020». Ils obtiendront chacun un financement de l’ordre de 2,5 millions d’euros sur cinq ans.
C’est peut-être une forme de baroud d’honneur, tant l’avenir de la participation de la Suisse aux programmes de recherche et de financement européens est incertain. Les deux premières rondes d’attribution de bourses de l’European Research Council (ERC grants) dans le cadre du programme Horizon 2020 ont été très favorable au pays, qui décroche pas moins de 42 Advanced grants, d’un montant moyen de 2,5 millions d’euros sur cinq ans.
A elle seule, l’EPFL a obtenu 11 de ces bourses prestigieuses (voir la liste ci-dessous). Quatre ont été attribuées durant l’exercice 2014, sept à la suite des candidatures déposées en 2015.
Quatre «Starting grants», six «Consolidator grants» et trois «Proof of concept grants» viennent compléter le tableau des bourses ERC reçues par l’EPFL sous Horizon 2020, totalisant plus de 49 millions d’euros.
En raison de l’acceptation par le peuple suisse de l’initiative « contre l’immigration de masse » le 9 février 2014, la possibilité pour les chercheurs établis en Suisse de concourir pour ces bourses avait été suspendue, avant d’être réintroduite à titre provisoire le 15 septembre 2014.
La situation provisoire actuelle n’aura cours que jusqu’au 9 février 2017. Nos chercheurs restent toutefois éligibles pour les prochains appels à projets, dont les échéances s’échelonnent jusqu’à cette date.
EPFL H2020 Advanced Grants
Smit, Berend – MaGic
The Materials Genome in Action
It is now possible to make an enormous spectrum of different, novel nanoporous materials simply by changing the building blocks in the synthesis of Metal Organic Frameworks (MOF) or related materials. This unique chemical tunability allows us to tailor-make materials that are optimal for a given application. The promise of finding just the right material seems remote however: because of practical limitations we can only ever synthesize, characterize, and test a tiny fraction of all possible materials. To take full advantage of this development, therefore, we need to develop alternative techniques, collectively referred to as Materials Genomics, to rapidly screen large numbers of materials and obtain fundamental insights into the chemical nature of the ideal material for a given application. The PI will tackle the challenge and promise posed by this unprecedented chemical tunability through the development of a multi-scale computational approach, which aims to reliably predict the performance of novel materials before synthesis. We will develop methodologies to generate libraries of representative sets of synthesizable hypothetical materials and perform large-scale screening of these libraries. These studies should give us fundamental insights into the common molecular features of the top-performing materials. The methods developed will be combined into an open access infrastructure in which our hypothetical materials are publicly accessible for data mining and big-data analysis. The project is organized in three Work Packages, each centered around finding better materials for carbon capture: (1) screen materials for gas separations and develop the tools to predict the best materials for carbon capture; (2) gain insights into and develop a computational methodology for screening the mechanical properties of nanoporous materials; (3) achieve an understanding of the amine-CO2 chemistry in diamine-appended MOFs and use this to predict their performance.
De Micheli, Giovanni – CyberCare
Integrated Sensing Architectures and Tools for Health Care
This proposal addresses high-risk, high-reward research of integrated sensing and computing architectures, as well as of models, methods and tools for their design and operation. Such architectures provide the bridge between bio-systems and information processing systems, where a bio-system is an abstraction of a human in terms of biophysical parameters. Breakthroughs in data acquisition, processing and decision making support will enable new smart-health applications.
The essential research goals of this proposal are: biophysical data acquisition by novel programmable integrated sensor arrays and their design and test using a modular and structured architecture; data processing in situ and/or remotely using application-specific hardware and/or embedded software; a new robust synthesis methodology for data processing units based on a new logic structure; models, abstractions and software tools for reasoning about the acquired data, to validate health conditions and/or to provide remedies (i.e., therapy). The results of this research will be embodied in a demonstrator showing the effectiveness of these combined technologies in first-aid medical care.
The outcome of this research will have a deep and broad impact on health care, because it will improve diagnosis and therapy in a variety of cases. Namely, it will boost the quality and quantity of the acquired biophysical data, possibly in real time, by leveraging multiple sensing modalities and dedicated computing architectures. The use of formal methods for design, data evaluation and decision making support will enhance the quality of the diagnostic platforms and will ease their qualification and adoption. Moreover, the integration of sensing and electronics and their in-field programmability will reduce production cost and lower the barrier of adoption, thus providing for better and more affordable health care means.
Forro, Laszlo – PICOPROP
Photo Induced Collective Properties of Hybrid Halide Perovskites
The recent discovery of the organo-inorganic perovskite CH3NH3PbI3 as very efficient material in photoelectric conversion is multifaceted: it turns out that this compound is promising not only in photovoltaics, but it is lasing, it gives bright light emitting diodes, promising in water splitting and we are persuaded that it can play an important role in basic sciences, as well.
We have recently realized that under white light illumination the photoelectrons, due to their very long recombination time, stay in the conduction band and the resistivity of a single crystal shows a metallic behavior. If the lifetime is sufficiently long and the density of these excited carrier is high enough they could condense into a Fermi sea. The project’s goal is to realize this highly unusual state and to document its properties by magneto-transport and spectroscopic techniques. We will check in our model compound the long-sought superconductivity of photo-excited carriers, extensively searched for in cuprates, if we could stabilize it by fine tuning the interactions by hydrostatic pressure under constant illumination.
The availability of high quality samples is primordial for this program. It turns out that CH3NH3PbI3 is ideal compound, it seems to be almost free of charged defects (its room temperature resistance is 5 orders of magnitude higher than that of Phosphorus doped Silicon at 1013 cm-3 doping concentration) and we can grow excellent single crystals of it. Furthermore, it has a flexibility in material design: one can vary all the constituents, and even the dimensionality by making layered materials with the main chemical motifs. A special effort will be devoted to tune the spin-orbit coupling by different elements, since this could be at the origin of the long recombination time of the photo-electrons.
We suspect that the highly tunable, clean and disorder-free doping obtained by shining light on these ionic crystals opens a new era in material discovery.
Duboule, Denis – RegulHox
Topological organization of vertebrate regulatory landscapes: The Hox genes paradigm
The aim of this grant is to understand how mammalian developmental genes, which are usually pleiotropic, are controlled via long-range regulations and how chromatin partitions into large and discrete regulatory domains, generally matching Topologically Associating Domains (TADs). We want to understand how such domains emerged in evolution, how they are built during development and how they help implement enhancer functions. We will use a large genomic interval in mouse chromosome 2 containing the HoxD cluster as a paradigm, as it is covered by a large allelic series re-organizing the topology of this interval. Since the syntenic human locus (2q31) is affected in numerous genetic syndromes involving CNVs or large DNA re-arrangements, we believe this work will also help understand the mechanistic bases of human regulatory mutations. The approach will capitalize on our knowledge of mouse embryos, the implementation of cutting-edge genomic technologies and the unique collection of engineered mammalian chromosomes kept into living mice, which represent as many targeted re-organizations of both chromatin and regulatory topologies. It will require important technological development, in order to apply to mammalian embryos, methods (HiC) currently used for cell cultures or adult tissues. We think that the feasibility of this novel program is high, due the portfolio of experimental tools recently developed in our laboratory. Also, pilot experiments have been initiated to identify problems and preliminary results including the use of HiC on embryonic tissues suggest that the proposed experiments can be realized within delay. The novelty and originality of this program are in the interdisciplinary and system approach of genomic re-arrangements, as analyzed in vivo using recently developed methodologies, allowing to associate topological variations with regulatory modalities in a physiological context, during normal or genetically impaired embryonic development.
Trono, Didier – TRANSPOS-X
Transposable elements, their controllers and the genesis of human-specific transcriptional networks
Transposable elements (TEs) account for more than two thirds of the human genome. They can inactivate genes, provide novel coding functions, sprinkle chromosomes with recombination-prone repetitive sequences, and modulate cellular gene expression through a wide variety of transcriptional andposttranscriptional influences. As a consequence, TEs are considered as essential motors of evolution yetthey are occasionally associated with disease, causing about one hundred Mendelian disorders and possibly contributing to several human cancers. As expected for such genomic threats, TEs are subjected to tight epigenetic control imposed from the very first days of embryogenesis, in part owing to their recognition by sequence-specific RNA- and protein-based repressors. It is generally considered that the evolutionary selection of these TE controllers reflects a simple host-pathogen arms race, and that their action results in the early and permanent silencing of their targets. We have recently uncovered new evolutionary evidence and obtained genomic and functional data that invalidate this dual assumption, and suggest instead that transposable elements and their epigenetic controllers establish species-specific transcriptional networks that play critical roles in human development and physiology. The general objective of the present proposal is to explore the breadth of this phenomenon, to decipher its mechanisms, to unveil its functional implications, and to probe how this knowledge could be exploited for basic research, biotechnology and clinical medicine.
Chergui, Majed – DYNAMOX
Charge carrier dynamics in metal oxides
Transition metal (TM) oxides (TiO2, ZnO, NiO) are large gap insulators that have emerged as highly attractive materials over the past two decades for applications in photocatalysis, solar energy conversion, etc., all of which rely on the generation of charge carriers, their evolution and their eventual trapping at defects or a self-trapped excitons. Despite the huge interest for such materials, the very nature of the elementary electronic excitations (Frenkel, Wannier or charge transfer exciton) is still not established, nor is the way these excitations evolve after being created: excitonic polaron or charged polaron. Finally, the electron and hole recombine is also not clearly established because of issue of defects and trapping.
In order to tackle these issues, here we implement novel experimental tools that would provide us with hitherto inaccessible information about the charge carrier dynamics in TM oxides. Of importance is the ability to detect both the electrons and the holes. Some of these tools have been developed in the PI’s group: i) Ultrafast X-ray absorption spectroscopy (XAS) will provide information about the final metal d-orbitals and about the structural changes around it; ii) Ultrafast X-ray emission (XES) will provide information about hole states. While these two approaches are ideal element-selective ones, the localization of the electron at metal atoms represents a small proportion of the electron population. Therefore, ultrafast Angle-resolved photoemission spectroscopy (ARPES) will be used to map out the band structure changes in the system and the evolution of the conduction band electrons. Ultrafast 2-dimensional (2D) UV (<400nm) transient absorption spectroscopy allows the mapping of the time evolution of both the valence and the conduction bands by its ability to pump and probe above the band gap. Last, Fourier Transform visible 2D spectroscopy will allow the probing of gap state dynamics at high time resolution.
Shaposhnikov, Mikhail – NuBSM
From Fermi to Planck: a bottom up approach
The Standard Model of particle physics is a hugely successful theory that has been tested in experiments at ever increasing energies, culminating in the recent discovery of the Higgs boson. Nevertheless, some major riddles cannot be addressed by the Standard Model, such as neutrino oscillations, the existence of Dark Matter, the absence of antimatter in the Universe. New fundamental principles, interactions and unknown yet particles are required to address these questions. Much of the research done during the last three decades on physics ‘beyond the Standard Model’ (BSM) has been driven by attempts to find a ‘natural’ solution of the hierarchy problem: why the Planck and the electroweak scales are so different. The most popular approaches to this problem predict new particles with the masses right above the electroweak scale.
This project explores an alternative idea that the absence of new particles with masses between the electroweak and Planck scales, supplemented by extra symmetries (such as scale invariance) may itself explain why the mass of the Higgs boson is much smaller than the Planck mass. This calls for a solution of the BSM problems by extremely feebly interacting particles with masses below the electroweak scale. Along the same lines we also explore the possibility that cosmological inflation does not require a new field, but is driven by the Higgs field of the Standard Model.
The proposed model offers solutions for BSM puzzles and is among a few ones that can be tested with existing experimental technologies and are valid even if no evidence for new physics is found at the LHC.
Constructing such a theory requires consolidated efforts in domains of high-energy theory, particle physics phenomenology, physics of the early Universe, cosmology and astrophysics as well as analyses of the available data from previous experiments and from cosmology. We will make predictions and establish the sensitivity goals for future high intensity experiments.
Unser, Michael – GlobalBioIm
Global integrative framework for Computational Bio-Imaging
A powerful strategy for increasing the quality and resolution of medical and biological images is to acquire larger quantities of data (Fourier samples for MRI, projections for X-ray imaging) and to jointly reconstruct the complete signal by correctly reallocating the measurements in 3D space/time and integrating all the information available. The underlying image sequence is reconstructed globally as the result of a very large-scale optimization that exploits the redundancy of the signal (spatio-temporal correlation, sparsity) to improve the solution. Due to recent advances in the field, we are arguing that such a “bigger data” integration is now within reach and that our team is ideally qualified to lead the way. A successful outcome will profoundly impact the design of future bioimaging systems.
We are proposing a unifying framework for the development of such next-generation reconstruction algorithms with a clear separation between the physical (forward model) and signal-related (regularization, incorporation of prior constraints) aspects of the problem. The pillars of our formulation are: an operator algebra with a corresponding set of fast linear solvers; an advanced statistical framework for the principled derivation of reconstruction methods; and learning schemes for parameter optimization and self-tuning. These core technologies will be incorporated into a modular software library featuring the key components for the implementation and testing of iterative reconstruction algorithms. We shall apply our framework to improve upon the state of the art in the following modalities: 1) phase-contrast X-ray tomography in full 3D; 2) structured illumination microscopy; 3) single-particle analysis in cryo-electron tomography; 4) a novel multipose fluorescence microscopy; 5) real-time MRI, and 6) a new multimodal digital microscope. In all instances, we shall work in close collaboration with the imaging scientists who are in charge of the instrumentation.
Ionescu, Adrian – Milli-Tech
Milli-Volt Switch Technologies for Energy Efficient Computation and Sensing
The Milli-Tech proposal aims at a novel technology platform serving both computation and sensing: electronic switch architectures, called steep slope switches, exploiting new device physics and concepts in emerging 2D materials to achieve operation at voltages below 100 millivolts. Such switches will have a subthreshold slope below 10mV/decade, significantly more abrupt than MOSFET thermal limit of 60mV/decade at room temperature and in great advance to any beyond CMOS switches. Such characteristics will dramatically improve both the energy efficiency of logic circuits and the transduction sensitivity for many classes of sensors.
The project will develop a technological platform called ‘millivolt technology’ focusing on low power digital and sensing/analog electronic functions exploiting steep slopes, with the goal of lowering the energy per useful function (computed and sensed bit of information) by a factor of 100x.
Such ultra-low operation voltage will contribute to solving major challenges of nanoelectronics such as power issues and it will enable energy efficient super-sensitive sensors for Internet-of-Everything (IoE).
Milli-Tech includes fundamental research on new solid-state steep slope device concepts: heterostructure tunnel FETs in 2D Transition-Metal-Dichalcogenides (TMD), 2D Van der Waals super-lattice energy filter switch and hybrid architectures combining two switching principles: band-to-band-tunneling and metal-insulator-transition or negative capacitance in VO2, used as additive technology boosters.
Milli-Tech plans breakthroughs by precise demonstrators: (i) energy efficient computation blocks for Von-Neumann ICs at sub-100mV, (ii) Active Pixel Sensors based on 2D TMD/GeSn tunnel FETs for IR imagers, (iii) Terahertz detectors based on hybrid 2D VO2/TMD switches (iii) ultra-sensitive 2D steep slope charge detectors for biosensing. The high-risk ‘millivolt technology’ will be highly rewarding by enabling the energy efficient revolution for IoE.
Martin, Olivier – NANOFACTORY
Building tomorrow’s nanofactory
The aim of this project is to translate the concept of production line to the nanoworld to develop what could become tomorrow’s nanofactory. So far, nanostructures are either chemically synthesized or produced using top-down approaches such as nanolithography, but no processes exist to take a few nanostructures and perform the basic operations required to assemble them into a more complex system. This proposal aims at addressing this need by realizing at the nanoscale the different functions that are required for a production line: receiving and moving raw nanomaterial in position, where it can be immobilized and worked on or transformed; combining different elements into more complex systems that support new functionalities. The project uses optical forces generated by plasmonic traps as enabling mechanism to act on raw material and the entire production line will be integrated into microfluidics, which will perform as an advanced conveyor belt. Local electrophoresis and photo-curable polymerization are used to locally modify and assemble raw nanoparticles. In addition to implementing challenging nanotechnologies, such as nanoscale electric contacts and perforated membranes, this project will also explore a fair amount of completely new physics, including the van der Waals interaction – which will be studied numerically and experimentally – the competition between optical and chemical forces or electrostatic attraction, and the detailed determination of the trapping potential produced by plasmonic nanostructures. The foreseen research is very comprehensive, including modelling, nanofabrication and explorations at the nanoscale. This ground-braking proposal will demonstrate how additive manufacturing can be implemented at the nanoscale.
Vetterli, Martin – SENSIT*
Sense and Sensitivity: Inverse Problems between Sparsity and Data Deluge
Around 90% of today’s exponentially growing Internet traffic is signal-based—a striking consequence of the proliferation of low-cost sensing. Beyond sheer quantity, these data span extremes of structure, modality, and quality. To make sense of this hodgepodge, new signal processing techniques are needed, and the paradigm of inverse problems is central to their design.
In SENSIT, we address this need by developing theoretical tools and algorithms to push the boundaries in key areas of inverse problems coming from acoustics and imaging:
1. We investigate original sampling modalities that balance sparsity models and abundance of data. A particular emphasis is on sampling at unknown locations, which we identify as a common abstraction for numerous problems.
2. We study regularization of inverse problems based on powerful geometrical objects such as Euclidean Distance Matrices (EDMs). We pursue extensions of EDMs and new algorithms for combinatorial cases, e.g. for ab initio crystallography.
3. We attack problems in computational acoustics using geometrical tools and EDMs. We aim at a complete characterization of indoor channels with mobile nodes, with immediate implications in source localization, source separation, cocktail party processing and wave-based “simultaneous localization and mapping” (SLAM).
4. We apply our tools to computational imaging, where our results on sampling at unknown locations, as well as insights from SLAM, can be used to simultaneously acquire higher resolution images, depth maps and reflectance functions.
In sum, we aim at providing theoretical foundations and tools for modern inverse problems in signal processing, where data deluge is prominent. We will investigate new models and algorithms, and a theory of their complexity, performance, and sensitivity. These advances will impact a wide range of applications including crystallography, wave-based signal processing, SLAM and computational imaging.
* en raison de sa nomination à la présidence de l’EPFL, Martin Vetterli a renoncé à son grant.