Materials and Devices for Electronics
and Magnetism

The physical properties of the building blocks of conventional and spin electronics depend on the nature and atomic structure of the nanomaterials that constitute them. These properties are tuned for applications in MOS, photonics, magnonics, memories or sensors. In the MEM group (Materials and Devices for Electronics and Magnetism), we study the link between the atomic structure and the electronic, optical and magnetic properties of these nanomaterial assemblies, in order to understand the physical processes governing these properties. The group relies on expertise in elaboration, structural and chemical characterisation down to the nanoscale, micro- and nanomagnetism, spin dynamics, atomic scale modelling of properties.

Research topics

1. Spintronics, magnonics and nanomagnetism

Magnetic materials offer a wide variety of applications in the fields of magnetic sensors, information storage or electronic logic components. They also make it possible to design low-energy devices at the nanometric scale, thanks to the manipulation of spin current instead of charge current. In this context, one of the objectives of the MEM team is to understand the link between structural properties and static and dynamic magnetic properties (spin waves) in nano-materials in the form of thin films, nanowires or micro/nano structured objects. For this, the team relies on a wide range of experimental skills developed in the laboratory for material growth (sputtering, ultra-high vacuum evaporation), structural characterisation (electron microscopy, X-ray diffraction) and characterisation of magnetic properties (ferromagnetic resonance, PPMS magnetometry, Kerr effect magnetometry and imaging). Experimental studies are also based on micromagnetic finite element simulations and ab-initio calculations.

contact :

Heusler alloys of the X2YZ type are particularly interesting for the realization of radio frequency components (> 1 GHz) thanks to their low damping coefficient and their magnetization value higher than 1T. Moreover, these materials are potentially semi-metallic with a spin polarisation of the order of 100%, which is important for the design of giant magnetoresistance or tunnel sensors. Our team has long studied the influence of different types of crystal disorder on the values of its magnetic parameters via broadband ferromagnetic resonance measurements (30 GHz). One of our current activities is to take advantage of these properties to build reconfigurable radio frequency devices. For example, the absorption spectrum of a square antidot magnonic crystal can be strongly modified in the remanent state due to the cubic anisotropy of the material [1]. Indeed, under the effect of a weak ultra-short field pulse (qq mT, 1 ns), it is possible to obtain remanent states which may or may not couple with a radio frequency field (Fig 1.1b,c). Our future developments in this field consist in integrating these magnonic crystals into spin waveguides in order to realise frequency filter functions operating at zero field. To this end, we are studying the impact of nanofabrication methods (FIB, optical and electronic lithography, Ar+ ion beam etching, etc.) [2] on the number of dynamic modes accessible in the system (Fig 1.2).

Figure 1.1 a) SEM image of a magnonic crystal composed of a square antidot lattice of 300 nm side length and 900 nm spacing. The crystal is fabricated by electron lithography followed by Ar+ ion beam etching at 3 KeV. b) Micromagnetic simulations of the A1 and A4 remanent states corresponding to two directions of the saturation magnetic field pulse. c) RF absorption spectra measured in the crystal for the A1 and A4 configurations and micromagnetic simulations of the frequencies of the dynamic modes (blue) in the A4 configuration.

Figure 1.2: Evolution of absorption frequencies as a function of the external magnetic field in a magnonic crystal composed of a lattice of square antidots of 100 nm side length, separated by 300 nm. The crystal is fabricated by Ga+ ion etching at 30 KeV in a FIB. On the right, example of spatial profiles of the dynamic modes obtained by micromagnetic simulations (from [2]).  

[1] Cubic Anisotropy for a Reconfigurable Magnonic Crystal Based on Co2MnSi Heusler Alloy, S. Mantion and N. Biziere, Phys. Rev. Applied 17, 044054, 2022.

[2] Influence of Ga+ milling on the spin waves modes in a Co2MnSi Heusler magnonic crystal, S. Mantion and N. Biziere, J. Appl. Phys. 131, 113905, 2022.

contact :

From a fundamental point of view, the performance and functionality of spintronic components require a detailed understanding of the formation of remanent (= zero-field) magnetic states in nano objects. Our team is working in collaboration with the I3EM team of CEMES to understand how inhomogeneous states appear in cylindrically symmetric systems such as nanowires in relation to possible local chemical and geometrical variations of the object. To this end, we observe the magnetic states by electron holography and compare the magnetic phase images obtained with those reconstructed from micromagnetic simulations. For the latter, we try to include in the numerical models the geometry observed by electron microscopy as well as local variations of the magnetic parameters induced by compositional variations. Thanks to this quantitative approach, we have been able to observe transitions between uniform and vortex states in cylindrical multilayers [1,2] (Fig. 1.3) or transverse wall formation in solid nano-cylinders [3]. Our future developments concern the observation of dynamic states in nano-objects under radio frequency excitation by electron holography.

Figure 1.3: Transmission (a) and EFTEM (b) images of a Co25 nm/Cu15 nm stack. Experimental (c,d) and simulated (e,f) magnetic phase images of the remanent magnetic state for a HSAT saturation field perpendicular (c,e) or parallel (d,f) to the wire axis. The corresponding magnetic patterns are shown in (i) and (j). (g) and (h) Magnetic phase intensity profile plotted along the white arrows in (c) and (e) and integrated over the white areas in (d) and (f). From [2].

[1] Multi magnetic states in Co/Cu multilayered cylindrical nanowires studied by combination of off-axis electron holography imaging and micromagnetic simulations. N. Biziere, D. Reyes, T. L. Wade, B. Warot-Fonrose and C. Gatel,, J. Appl. Phys. 126, 163906, 2019.

[2] Magnetic Configurations in Co/Cu Multilayered Nanowires: Evidence of Structural and Magnetic Interplay. D. Reyes, N. Biziere, B. Warot-Fonrose, T. Wade, and C. Gatel,, Nanoletters 16, 1230, 2016.

[2] Imaging the Fine Structure of a Magnetic Domain Wall in a Ni Nanocylinder. N. Biziere, C. Gatel, R. Lassalle-Balier, M-C. Clochard,  J.E. Wegrowe, E. Snoeck, NanoLetters, 13, 2053, 2013.

2. Semiconductors for electronics

The aim of the “Semiconductor Materials for Microelectronics” theme is to observe, quantify, understand and model new processes, new structures and new materials used to manufacture the basic building blocks of ultimate microelectronics. This theme focuses on materials science issues. We have a long history of collaboration with major players in the field (applied research centres, foundries and equipment manufacturers) and we coordinate or participate in technological adventures, supported by national (ANR), European and especially industrial contracts. This theme corresponds to ∼3/8 of the research time of the group’s permanent staff.

contact :

Our objective is to understand the physics, i.e. the ingredients and driving forces, involved in the formation (nucleation), growth and transformation of heterostructures of nanocrystals, defects or layers (1D, 2D, 3D) in crystalline matrices under stress and thermal annealing. More precisely, we are interested in the characteristics, which we have measured experimentally, and in the modelling of the structure of nano and micrometric objects, but also in the simulation of the evolution kinetics (growth laws) of populations of precipitates in the presence of stresses, voluntarily or involuntarily applied.

A telling example of the application of this approach concerns the ionic implantation of H+ and He+ in Si. When followed by molecular bonding, it can be used to fracture and transfer Si thin films from a donor substrate to a host substrate, providing the most efficient method (Smart-CutTM technology) for the fabrication of SOI (silicon-on-insulator) substrates, a field in which France, with the company Soitec, is a world leader. The fracture of Si results from the growth of platelets, flat cavities filled with molecular hydrogen, and their elastic interactions which lead to the formation and growth of nano and micro cracks or pressurised blisters (Fig. 2.1).

Figure 2.1 Defects, precipitation, ripening, coalescence, transformation, fracture after H/He implantation in Si: a multi-scale study.

Indeed, all these systems (matrix/precipitates) originate from the precipitation, during thermal annealing, of strong supersaturations of point defects (chemical impurities, interstitials and vacancies) initially introduced in the matrix by ion implantation. These defects changed the mesh volume of the host matrix and were therefore expected to cause biaxial stress in the matrix plane [1] (Fig. 2.2).

Figure 2.2 (a) TEM image in cross-section; Strain (b), (c) out-of-plane εzz; (d), (e) in-plane εxx obtained in Si implanted with H by (b), (d) dark field holography, (c), (e) FEM using sub-nanometer defect model (f). The image on the right shows the elastic reaction of the material to the presence of the complexes and the stress σxx.

Consequently, this stress strongly impacts the type (structure, families, variants), growth rate and possible transformations of the complexes [2] (Fig. 2.3), extended defects and precipitates that form during thermal annealing of these supersaturated matrices.

Figure 2.3 Profils des différents complexes (a) et de la contribution des différents complexes à la déformation macroscopique (b), (c) déduites pour deux doses d’H implanté.

Ces caractéristiques, mesurables expérimentalement, reflétant directement l’énergie de formation des précipités, notre idée fut de générer et utiliser des champs anisotropes afin d’affecter ces populations et d’ainsi accéder à toutes les composantes de l’énergie de Gibbs de précipités 2D. Cette approche nous a permis de remonter aux paramètres fondamentaux de la précipitation, enthalpies de formation et énergies de liaisons des espèces, énergies élastiques, surfaciques et volumiques mises en jeu, diffusivités des espèces, et finalement de simuler le phénomène [3].

[1] N. Cherkashin, S. Reboh, A. Lubk, M. J. Hÿtch, A. Claverie, “Strain in Hydrogen-Implanted Si Investigated Using Dark-Field Electron Holography”, Applied Physics Express 6, 091301 (2013).

[2] N. Cherkashin, F.-X. Darras, P. Pochet, S. Reboh, N. Ratel-Ramond, A. Claverie,  “Modelling of point defect complex formation and its application to H+ ion implanted silicon”, Acta Materialia 99 (15),  187-195  (2015).

[3] N. Cherkashin, F. X. Darras, A. Claverie, “Determination of the Free Gibbs Energy of Plate-Like Precipitates of Hydrogen Molecules and Silicon Vacancies Formed after H+ Ion Implantation into Silicon and Annealing”, Solid State Phenomena 242, 190-195 (2015).

contact :,

GeSbTe-based alloys (GST) are phase change materials, i.e. they have extremely different physical properties depending on whether they are amorphous or crystalline. In DVDs and Blu-ray discs, binary information is stored thanks to the high contrast in optical reflectivity that these two phases present.

In electronic phase change memories (e-PCM), the information is contained in the pronounced difference in electrical conductivity between the crystalline and amorphous phases of these alloys. These states are then switched locally and reversibly between crystalline and amorphous phases using thermal pulses generated from electrical pulses. Heating above the melting temperature followed by rapid quenching results in the formation of an amorphous region of high resistivity, while more limited heating over a longer period of time results in the recrystallisation of the same region.

Recent work, including ours, has shown that beyond digital memories (2 bits), these alloys can also be used to fabricate multi-level memories (several bits) as well as memristors, capable of reproducing synaptic activity, and thus offer devices for artificial intelligence.

Despite their enormous potential, the development and industrialisation of e-PCMs requires a thorough understanding of the physical phenomena involved in the switching and storage mechanisms, and this within the framework of extremely small dimensions. Currently, most IC manufacturers are exploring the potential of these materials in collaboration with academics, and this project is no exception.

A simplified schematic of the architecture of an e-PCM cell and a MET image are shown below. It includes a heating filament (TiSiN), the phase change material (GST) and the top electrode (TiN). The TEM image shows the characteristic amorphous dome of the cell in the RESET (0) state. The dimensions illustrate the need for nanoscale studies using advanced microscopy techniques.

On the left, a simplified diagram of the “mushroom” structure of PCM cells. On the right is a TEM image of a real cell in the RESET state. Note the amorphous dome above the heating filament.


The “Ô-GST” project:

Little is known about the physical and chemical changes that affect the material during phase changes and are responsible for the electrical characteristics of the cell (resistive, conductive or intermediate) and the degradation mechanisms over time and use that affect it. For this reason, fundamental work is needed to understand the mechanisms by which the material transitions from amorphous to crystalline phase (and vice versa), the impact of the geometry, size and surrounding media of the cell on the final characteristics of the material and the associated device. Furthermore, the desired cell characteristics are obtained using materials of clearly non-stoichiometric compositions, which further increases the need for a thorough understanding of the atomic mechanisms involved and thus for characterisation at the nanoscale. In the last few years, we have already explored crystallisation and transport in Ge-rich GST alloys (GGST) as well as structure/property relationships in 28 nm technology cells and obtained some notable results defining the state of the art in the field today (1-10).

Phase separation during crystallisation of a Ge-rich GST alloy. Elemental mapping (TEM-EDX) in the 500°C annealed layers and corresponding ASTAR image (bottom). Large pure grains of GST 225 are embedded in a matrix of small Ge grains.

In this context, CEMES is collaborating with STMicroelectronics in the framework of a large project “Ô-GST” whose objectives are to

1) Identify the mechanisms and parameters governing the thermal crystallisation of GGST materials and the changes resulting from doping with N, C and H, in deposited layers but also within nanometric cells (28 and 18 nm technologies).

2) To understand the influence of the morphology of the GST domains (phases, grain sizes…) on the electrical characteristics of the material and on the performance and reliability (drift, retention / cycling) of PCMs based on these materials.

3) To explore the possibility of accessing intermediate resistivity states (IRS) and mimicking synaptic activity (analogue storage, cumulative storage and plasticity) using Ge-GST cells.

To achieve these objectives, we have formed a group of three permanent researchers from CEMES and LPCNO with complementary skills (theory, materials science, transport and electrical properties), an expert engineer from STMicroelectronics, 4 postdocs and a PhD student.

Typical example of chemical analysis by EELS of the active region of a PCM cell in the RESET state (resistive and amorphous). The forming step leads to the fabrication of a typical nanostructure, composed of Ge sidewalls and a GeTe roof, within which a homogeneous amorphous dome, rich in Ge and Sb, is formed.  The total thickness of the layer is 50 nm.


The activity is strongly supported by direct contracts with STMicroelectronics (Task Force GST SOW1 and SOW2, Cifre), by European IPCEI projects on microelectronics via the nano2022 and nano2025 programmes, for a total amount of more than 2 million euros. An exploratory project on IRS (intermediate resistivity states) that we are coordinating will be funded by the ANR from 2023 to 2027.


On the left, evolution of the resistivity of cells as a function of the heat treatment received. Initially, the amorphous material relaxes and becomes even more resistive (drift). Above 270°C, it becomes conductive.

Below, dark field images showing the stability of Ge grains at the edge of the dome (in white) while GST-225 grains (in black) recrystallise and invade the dome from 260°C.


Publications de l’équipe :

1) M. Agati et al, MRS Communications (2018),

In situ TEM shows that thermal crystallisation of Nitrogen-doped GGST alloys occurs around 380°C. It is homogeneous and leads to the formation of a two-phase Ge-GST material.

2) M. Agati et al, J. Mat. Chem. (2019),

Various TEM techniques, including EDX and ASTAR, as well as XRD, are used to show that the crystallisation of GGST proceeds through the nucleation of the Ge phase followed by the crystallisation of the GST-225 phase, which is formally identified.

3) R. Sinha Roy et al, Phys. Rev. B, (2019),

The electronic structure and electrical conductivity of non-stoichiometric and Si-doped Ge2Sb2Te5 cubic crystals have been calculated (first principles). The incorporation of excess Ge or Si on the normally lacunar sites of the Ge-Sb sublattice increases the conductivity by a factor of 4-5.

4) M. Agati et al, Applied Surface Science, (2019),

It is shown that air storage of amorphous GGST layers leads to the formation of an oxidised layer on the surface. When such layers are annealed, the selective oxidation of Ge leads to a redistribution of elements and heterogeneous crystallisation of the layer from the surface at a lower temperature than observed when the layers are encapsulated.

5) A. Bourgine et al, Solid State Electronics, (2020),

We have studied the conduction characteristics of GGST cells by I(V) between 100 and 500 K and by impedance spectroscopy. While it is generally accepted that conduction is of the Poole-Frenkel type, i.e. results from charge jumping between defects in the band gap of a homogeneous material, we show that this conduction can be described by transport between (nano)-grains of different conductivities and thus compositions. In the crystalline state, the conduction resembles that of Sb-doped Germanium.

6) Luong et al, Physica status solidi PSS-RRL, 2021, Vol.15(3),

A review article highlighting the specificities of Ge-rich GeSbTe (GGST) alloys, from a structural, chemical and electrical point of view.

7) Luong et al, Physica status solidi PSS-RRL, 2021, Vol.15(3),

We show by TEM and XRD that the incorporation of N in GGST tends to slow down the phase separation, crystallisation and growth processes during thermal annealing. The technological advantages of N doping are also discussed, considering the increased stability of the amorphous phase compared to its parent crystalline phase and its more divided microstructure.

8) Luong et al, Nanomaterials, (2021),

Ion implantation allows the fabrication of GST-225 samples in which the nitrogen concentration varies with depth. We were able to study the effect of nitrogen on the crystallisation of GST-225 by direct comparison in a TEM. The increase in the viscosity of the glass and the crystallization totally dominated by nucleation were explained by the reduction of the diffusivity of Ge in the presence of nitrogen. In nitrogen-rich regions, GST-225 grains are deformed in tension, suggesting the incorporation of N within the grains and not at their joints, as often assumed in the literature.

9) E. Rahier et al, ACS Appl. Electron. Mater. (2022),

The fine characterization by synchrotron XRD during isothermal annealing (in situ) of the crystallization of GGST alloys, combined with ex situ characterizations by STEM-HAADF, has allowed to elucidate the crystallization scenario of GGST alloys, thus putting an end to a long controversy.

10) Luong et al, Material Science in Semiconductor Processing, in press.

Specific samples with a solid source embedded in a GST matrix have been used to study the thermal diffusion of Ge through amorphous and crystalline GST-225 layers by in situ TEM and STEM-EDX. The diffusion becomes strong above 220°C and takes place via the grain boundaries. Ge segregates in the amorphous phase between the GST grains.

11) L. Laurin et al, IEDM Proc. 2022, in press.

We present a comprehensive study of the retention of SET and RESET states in PCM cells of GGST alloys by coupling electrical and physical characterizations. STEM-HHADF images suggest that Ge-rich amorphous residues present in the active region of the cells could be at the origin of the drift phenomena noted in these cells.

contact :

In general, the epitaxy of heterostructures imposes an accommodation of materials with different crystal lattices. The substrate being volumetric and rigid with respect to the deposited layer, the mesh of the deposited layer deforms and generates a biaxial stress in the plane of the substrate. As a result, this layer stores elastic energy anisotropically, which strongly affects the evolution of the system during growth or when a second phase is precipitated within it. An example that is widely studied by the scientific community concerns the growth of pseudomorphic layers on a substrate or plastically relaxed by the formation of dislocations. Another example concerns the transformation of a thin pseudomorphic layer (2D) into separate islands (3D) under the effect of the relaxation of the elastic energy accumulated in the 2D layer.  Our objective is to develop experimental and theoretical methods to describe the elastic and plastic relaxation phenomena under stress in III-V materials. We have thus developed a relatively simple 3D analytical model that allows us to deduce the local stoichiometry of InGaAs islands in a GaAs matrix from deformation images obtained by TEM (GPA on HREM) and thus to understand, simulate and control the optical properties of lasers using these quantum boxes (Fig. 2.4) [1, 2].

Figure 2.4 Out-of-plane strain (HR-DFEH), elastic strain (FEM), localisation potentials and hole wave functions (calculation) in an InGaAs/GaAs quantum box.

As another example, we have developed an island growth method in the InGaN/GaN system to control and optimise the electro-optical properties of light-emitting diodes [3].

[1] 1.   N. Cherkashin, S. Reboh, M. J. Hÿtch, A. Claverie, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, V. V. Chaldyshev, “Determination of stress, strain, and elemental distribution within In(Ga)As quantum dots embedded in GaAs using advanced transmission electron microscopy”, Applied Physics Letters , 102:173115 (2013).

[2] A. N. Kosarev, V. V. Chaldyshev, and N. Cherkashin, Experimentally-Verified Modeling of InGaAs Quantum Dots, Nanomaterials, MDPI, 12 (12), 1967 (2022).

[3] A. F. Tsatsulnikov , W. V. Lundin , A. V. Sakharov , A. E. Nikolaev , E. E. Zavarin , S. O. Usov , M. A. Yagovkina , M. J. Hÿtch , M. Korytov, and N. Cherkashin, “Formation of Three-Dimensional Islands in the Active Region of InGaN Based Light Emitting Diodes Using a Growth Interruption Approach,” Science of Advanced Materials 7(8):1629-1635 (2015).

3. Functional oxides 

The “Functional oxides for electronics and spintronics” theme aims to analyse, understand and predict the physical properties of functional oxides or oxide interfaces. These properties could be used in original devices, taking advantage of the great wealth and diversity of behaviour offered by this family of materials. They could be manipulated and tuned by fine engineering of their atomic structure.

Oxides form a family of materials with a wide range of chemical compositions and crystallographic structures, conducive to the emergence of a vast choice of physical properties (magnetism, ferroelectricity, superconductivity…). In addition to their multifunctional aspect, one of the attractions of these compounds is that their properties are generally easily manipulated through the control of intrinsic parameters (choice of chemical composition, structural defect engineering, interface design) or the application of external stimuli (electric or magnetic field or temperature variation). The theoretical activities we develop are based on the use of first-principles calculations applied to the physical chemistry of oxides. We are interested in the link between structural, electronic and magnetic properties of complex oxides in thin films, with a focus on the study of interface properties. This theme encompasses both studies of

– two-dimensional electron gases at interfaces with a polar oxide [1,2]

– spin-orbit effects in oxides [3],

– electronic structure at metal/oxide interfaces, in order to understand magnetoelectric coupling effects [4] or the mechanisms defining the Schottky barrier height [5].

Figure 3.1: Spin degeneracy lifts and Rashba spin-orbit parameters of the d-bands of Ti atoms associated with the formation of the two-dimensional electron gas at the LaAlO3/SrTiO3(001) interface [Thesis by J. Gosteau (].

Currently, we are participating in a collaborative project funded by the ANR (project MULTINANO,, grant number ANR-19-CE09-0036, led by A. Barbier, SPEC, CEA-Saclay), in which we would like to understand the physical mechanisms at the origin of the formation of conductive domains in ferrites with spinel structure, in the vicinity of their interfaces.

[1] R. Arras, et al., Spin-polarized electronic states and atomic reconstructions at antiperovskite Sr3SnO(001) polar surfaces, Phys. Rev. B 104, 045411 (2021).

[2] K. Rubi, et al., Aperiodic quantum oscillations in the two-dimensional electron gas at the LaAlO3/SrTiO3 interface, npj Quantum Mater. 5, 1 (2020).

[3] J. Gosteau, et al., Spin-orbit effects in ferroelectric PbTiO3 under tensile strain, Phys. Rev. B 103, 024416 (2021).

[4] R. Arras and S. Cherifi-Hertel, Polarization Control of the Interface Ferromagnetic to Antiferromagnetic Phase Transition in Co/Pb(Zr,Ti)O3, ACS Appl. Mater. & Interfaces 11, 34399 (2019).

[5] R. Arras, et al., Schottky barrier formation at the Fe/SrTiO3(001) interface : Influence of oxygen vacancies and layer oxidation, Phys. Rev. B 102, 205307 (2020).

contact :

The study of the scientific and technological challenges concerning (i) the fabrication of complex oxide nanostructures on semiconductors, (ii) their (photo)ferroelectric properties and (iii) their dynamics is at the heart of the French-German ANR PRCI project FEAT (see section “The Projects”). The nanostructures are prepared by RIE (Reactive Ion Etching) and FIB (Focused Ion Beam) of Neon ions (partner HZB) on BaTiO3 oxide, solid or fabricated as a thin film by MBE (Molecular Beam Epitaxy) on Silicon (via an intermediate layer of SrTiO3 oxide). At CEMES, we are trying to map the atomic structure, chemistry and atomic displacements in relation to the electrical polarisation (amplitude and direction of the polarisation at the local scale) with advanced methods of transmission electron microscopy (TEM and STEM) and adapted quantitative processing of the experimental imaging data. The formation of domains and their reversal are studied in situ in the dedicated microscope I2TEM (I3EM CEMES group).

contact :,

Emerging communication technologies (5G, NFC) require ferroelectric film (FE) varactors with significantly thinner thicknesses to operate at higher frequencies or lower working voltages. The metal/perovskite Ba1-xSrxTiO3 (BST)/metal solution is the most widely used at present, but reducing the thickness leads to the appearance of an interface FE “dead layer” and increased leakage current. The ANR BEPOLAR project (see section “The Projects”) aims to study the ability to modify and control the properties of this interface by introducing a few atomic layers of a perovskite oxide (Interface Control Layer ICL = La1-xSrxMnO3, Ba1-xSrxRuO3) by a systematic approach (study of the effect of the nature, thickness and chemical composition of the ICL). In this project, CEMES uses advanced transmission electron microscopy (STEM) methods and data processing using in-house pioneer scripts to provide structural and chemical information on the different interfaces in relation to their properties. These aberration-corrected TEM microscopies resolve the nature and individual positions of the atomic columns to an accuracy of a few tens of picometers. This will allow the determination of lattice distortion, octahedral rotations and interface chemistry of BST/ICL epitaxial heterostructures at the interfaces. DFT calculations help to understand the properties measured at the interfaces.

contact :,

In the field of mobile telecommunication, the most widespread radio frequency (RF) filters today are surface acoustic wave (SAW) filters. These are bandpass filters made up of a single crystal piezoelectric substrate on which a set of interdigitated electrodes are deposited. Today, the new constraints imposed by telecommunication networks, particularly 5G, require substrates with better performance. Lithium Tantalate (LiTaO3 or LTO) integrated in the form of a thin piezoelectric film on insulator (POI for Piezoelectric-on-Insulator) opens up prospects in this direction. This LTO/SiO2/Si heterostructure makes it possible to produce filters with a better-quality factor, a wider bandwidth, increased temperature stability and reduced insertion losses. The POI substrates are manufactured using the Smart Cut™ technology developed by the Soitec company, which allows the transfer of single crystal thin films. The main process steps are 1) Implantation of hydrogen into a single crystal LTO substrate. The result is an implanted and damaged layer with a peak hydrogen concentration at a certain depth below the surface; 2) Molecular bonding of the implanted LTO substrate to a SiO2/Si substrate; 3) Annealing step which will allow the precipitation of hydrogen and point defects in the implanted LTO layer, resulting in the growth of microcracks until fracture of the entire crystalline layer is achieved. As the fracture takes place at a certain depth below the LTO surface, a thin layer of LTO is transferred onto the SiO2/Si substrate; 4) Final treatment of the transferred layer by annealing and polishing to smooth the surface and recover the crystalline integrity of the LTO. This thesis work has allowed us to highlight the structural [1] and chemical property changes that take place within the LTO during these main steps at different scales. In addition, this work required the adaptation or development of new methodologies, notably to study the deformation in the implanted layers by X-ray diffraction (XRD) and more locally by high-resolution scanning mode transmission electron microscopy (STEM). The redistribution of chemical elements in the implanted layers was first studied locally by electron energy loss spectroscopy (EELS) and then completed on a more global scale by glow discharge optical emission spectroscopy (GDOES).

The links between deformation, hydrogen concentration (measured by SIMS) and redistribution of chemical elements, particularly Li, in the implanted LTO were first highlighted. The study of the evolution under annealing then provided some answers on the growth conditions of the nano-cracks at the origin of the fracture and their atypical morphology. Finally, a dark-field TEM and high-resolution STEM study showed that the transferred LTO layers, although similar to bulk LTO, present in some cases a local reversal of the polarisation at the interface with the SiO2. The cause of this phenomenon is still under discussion and may be the subject of a future study.

[1] A. Louiset, S. Schamm-Chardon, O. Kononchuk, and N. Cherkashin, Reconstruction of depth resolved strain tensor in off-axis single crystals: Application to H+ ions implanted LiTaO3, Appl. Phys. Lett. 118, 082903 (2021).

4. Van der Waals heterostructures for spintronics and spin-orbitronics


Magnetic skyrmions, pseudo-particles characterised by topologically protected chiral spin structures, have attracted considerable attention due to their potential applications in information storage and processing. So far, the scientific community has mainly studied these objects in massive three-dimensional magnets, ultrathin magnetic layers and multilayers. We numerically study skyrmions in newly discovered two-dimensional (2D) magnets and their van der Waals heterostructures. Our study, entirely guided by theoretical/numerical modelling, proposes to unveil the mechanisms of skyrmion stability, detection and manipulation in 2D magnets. These foundations are necessary to design 2D skyrmionic devices that can be used for memory and logic applications.  The key points of this project are:

  • the understanding of magnetism in low-dimensional systems
  • the calculation of topological spin structures
  • the calculation of quantum transport through non-collinear magnetic nanostructures

This study is funded under the “NanoX disruptive (2022-2025)” project.


Van der Waals heterostructures are complex nanomaterials, constituted by the association of several two-dimensional crystals of sub-nanometric thickness with potentially very different electronic and magnetic properties, linked to each other only by Van der Waals forces.  Within the framework of the ANR SYZMO2D project, we are interested in heterostructures associating a two-dimensional semiconductor of the transition metal dichalcogenide family (such as MoS2) and a magnetic layer with perpendicular anisotropy. This combination should in principle allow the injection of a carrier gas whose spins are perpendicular to the plane of the 2D semiconductor, without the need to apply an external magnetic field to impose this orientation.

We have used ab-initio methods based on DFT to calculate the physical properties of Fe/MgO/MoS2 multilayers; these stacks well combine a ferromagnetic electrode with perpendicular anisotropy (ultra-thin iron layer) and a MoS2 monosheet, separated from each other by a thin MgO layer whose thickness is adjustable. We have shown that this system is the seat of an electron transfer from the Fe/MgO interface to the MoS2 sheet. As a result of this transfer, the MgO layer is the seat of an electric field and the MoS2 conduction band is partially occupied by an electron gas. Although the spin polarisation of this electron gas remains very low, we observed very interesting spin-dependent effects in the band structure of the MoS2 sheet: The dispersion of the valence bands of MoS2 is influenced by the greater or lesser magnetic proximity of the iron layer and by the electric field due to the charge transfer and the proximity of the interfacing MgO layer. This results in a competition between the magnetic effects and those linked to the spin-orbit coupling which depends on the thickness of the MgO layer. Both types of effects modify the band energy and the spin texture, see figure below.

Figure 3.2: Band structure of the Fe(7MLs)/MgO(3MLs)/MoS2 multilayer. The colours give the spin projection on the axis perpendicular to the MoS2 sheet. For this small thickness of MgO, the gap is direct, the lifting of the valence band degeneracy is clearly visible and the spins are perpendicular to the layer in the centre of the Brillouin zone. The gap becomes indirect, the spin degeneracy weaker and the spin texture different, for higher MgO layer thicknesses

[1] Z. Zhou, P. Marcon, X. Devaux, P. Pigeat, A. Bouché, S. Migot, A. Jaafar, R. Arras, M. Vergnat, L. Ren, H. Tornatzky, C. Robert, X. Marie, J.- M. George, H. Jaffrès, M. Stoffel, H. Rinnert, Z. Wei, P. Renucci, L. Calmels and Y. Lu, Large perpendicular magnetic anisotropy in Ta/CoFeB/MgO on full-coverage monolayer MoS2 and first-principles study of its electronic structure, ACS Appl. Mater. Interfaces 13, 32579 (2021).

5. Molecular Spintronics


Controlling and sensing magnetism in compact and energy-efficient devices is paramount for the development of future spintronic devices. The use of single molecules as quantum units opens a new path to reach the physical limits of miniaturisation. Molecular spintronics is an emerging field combining the flexibility of molecular electronics and molecular magnetism with the advantages of spintronics. Its main objective is the manipulation of electron spin by a judicious combination of ad hoc molecules and inorganic substrate.

Figure 5.1: a) Schematic representation of a FeTPP molecule deposited on a graphene sheet (doped), located between an STM tip and a gate electrode. b) The two magnetic states (S=3/2 or S=1) of the molecule deposited on boron-doped graphene that can be obtained reversibly when the gate voltage is changed. c) Spin polarisation of the electronic transport in the graphene sheet: this polarisation is 10% without applying a gate voltage whereas it disappears when a negative voltage is applied. This principle makes it possible to write binary information (0 or 1 states) by applying a voltage and to read this information by measuring the current.

Our scientific objective is to propose general concepts that can be used to optimise the magneto-transport properties of materials/devices. For example, we have recently proposed a new way to achieve fully electrical control of molecular spintronic devices, without the need to apply external magnetic fields [1] (see Fig. 5.1). The important points of this project are:


  • the manipulation of magnetic properties in spin-crossing molecules [2]
  • quantum transport through metallic, magnetic and molecular nanocontacts [3].

[1] F. Gao, D. Li, C. Barreteau, and M. Brandbyge, Phys. Rev. Lett. 129, 027201 (2022) (Editor’s suggestion).

[2] D. Li et al, accepted in J. Phys. Chem. Lett (2022), arXiv:2206.13767.

[3] D. Li et al, Phys. Rev. B 99, 115403 (2019); A. Pal & D. Li et al, Nat. Commun. 10, 5565 (2019); D. Li et al, Phys. Rev. Research 3, 033017 (2021).

This work was carried out in the framework of the European project COSMICS.

6. Development of experimental methods and software for quantitative strain measurement:

The continuous miniaturisation of semiconductor heterostructure-based devices makes it necessary to study both deformation and composition in these structures from the nanoscale to the micrometer scale. While there are a few dedicated TEM methods for measuring deformation with nanoscale resolution in the vicinity of nanoscale objects, we felt it was necessary to propose new methods and develop new techniques to achieve sufficient accuracy, spatial resolution and field of view to meet the needs of some complex material problems.

contact :

For the study of nanocrystals embedded in amorphous matrices and epitaxial layers with very different crystal structures, we have been looking for a solution to measure the deformation from a high resolution image (HR-TEM or scanning probe TEM=HR-STEM) in the absence of a reference on the same image and/or the presence of random distortions. This is a problem we were facing, and is an always annoying constraint inherent to all other methods. To solve this problem, we proposed and developed a new method called “Absolute strain” (AbStrain) for HR-(S)TEM image processing allowing the measurement of absolute values of interplanar distances and angles in single crystal structures without the need to use a reference grating present on the same image. To this end, we developed a theory of “absolute correction” (as opposed to relative correction) of instrumental distortions affecting experimentally measured images and wrote a script to apply this theory to analyse nanoscale structures for which a reference grating could not be imaged. This approach can be applied not only in reciprocal space [1, 2]] (Fig. 6.1) (where the GPA technique operates) but also in real space (where the so-called peak findings technique operates).

Figure 6.1 AbStrain: (a) HR-TEM image of a SbxAs1-x nanoinclusion (hexagonal structure) embedded in GaAs (zinc-blend structure); (b)-(d) zero components of the strain tensor obtained by AbStrain with respect to an ideal hcp lattice of As0.1Sb0.9.

[1] N. A. Bert, V. V. Chaldyshev, N. A. Cherkashin, V. N. Nevedomskiy, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, V. I. Ushanov, M. A. Yagovkina, “Metallic AsSb nanoinclusions strongly enriched by Sb in AlGaAsSb metamaterial”, J. of Appl. Phys. 125, 145106 (2019).

[2] M. A. Luong, N. Cherkashin, B. Pécassou, Ch. Sabbione, F. Mazen, A. Claverie, Effect of Nitrogen Doping on the Crystallization Kinetics of Ge2Sb2Te5, Nanomaterials, MDPI, 11 (7), 1729 (2021).

contact :

We have sought to develop a method to quantify micrometer-scale deformation with nanoscale spatial resolution and ~10-4 X-ray accuracy in epitaxial heterostructures composed of multiple multilayers, i.e., deeply implanted ionic regions. For this purpose, we have invented a technique called “Moiré by Sample Design” (MoSD). This method allows the measurement of deformations in single-crystal structures, in cross-section and in plan view, via the formation of “controlled” moiré fringes while using conventional microscopes.  The idea is to make a sample in the form of a stack containing the structure to be measured and a known reference crystal. Images then reveal moiré fringes from which the deformation fields in the structure under study can be extracted. By adjusting the angle of rotation between the two blades, the period of these fringes can be adjusted and thus maps of the strain fields can be obtained with nanometer resolution [1, 2] (Fig. 6.2). Having developed a moiré image processing algorithm, MoSD demonstrates the possibility of mapping two-dimensional strain fields with nanoscale spatial resolution (≥ 2 nm), an ultimate accuracy of 1×10-4 and a field of view twenty times larger than offered by GPA; this technique can be implemented using a conventional transmission electron microscope of modest cost.

Figure 6.2 MoSD: In-plane exx and out-of-plane ezz strain measured by MoSD and calculated by FEM in: (a) Si1-xGex multilayers [110]: conventional image, moiré, and the profiles of exx and ezz; (b) 3D mechanically coupled In0.4Ga0.6P islands formed on (112)GaAs substrate; (c) 3D mechanically coupled SiGe islands formed on (001)Si substrate.

[1] N. Cherkashin, T. Denneulin, M. J. Hÿtch, “Electron microscopy by specimen design: application to strain measurements”, Scientific Reports 7, 12394 (2017).

[2] N. N. Ledentsov, V. A. Shchukin, Yu. M. Shernyakov, M. M. Kulagina, A. S. Payusov, N. Yu. Gordeev, M. V. Maximov, A. E. Zhukov, T. Denneulin, and N. Cherkashin, “Room-temperature yellow-orange (In,Ga,Al)P-GaP laser diodes grown on (n11) GaAs substrates,” Opt. Express 26 (11), 13985-13994 (2018).

7. Development of a theory/simulation platform for realistic ab initio transport in spin-orbitron devices

contact :

While early progress in spintronics relied mainly on phenomenological modelling (drift-diffusion and semi-classical Boltzmann transport equations), much more advanced quantum theoretical methodologies have made it possible to calculate magnetic and transport properties from realistic band structures obtained by ab initio methods based on density functional theory (DFT). However, due to the high computational cost, it is very difficult to model at the ab initio level magnetic skyrmions characterised by non-collinear spin textures of several nanometres width, induced by the spin-orbit coupling and resulting from the competition between different magnetic interactions. To this end, we will develop an ab initio large-scale spin-orbit transport platform that will allow us to model realistic quantum transport through skyrmions.

  • Theory and implementation of new features in DFT +NEGF (non-equilibrium Green’s functions) based codes: spin-orbit torque, spin-charge orbital pumping, effects beyond linear response.
  • Development of a large-scale (>1000 atoms) modelling platform (TBnc-trans) based on the NEGF formalism to calculate spin-orbit transport properties in systems involving non-collinear spin textures (e.g. magnetic skyrmions or vortices) and/or structural defects.



The EHIS project consists in using, for the first time, electron holography as a new experimental tool for imaging spin dynamics in an individual nanostructure. The principle is based on the modification of magnetic phase images due to the quasi-static components of the precessional magnetization. The major interests of this method are to benefit from the spatial resolution of electron microscopy (in the nanometre range) and to allow a direct link between the remanent magnetic configuration and the local amplitude of dynamic precession. Thus, this project will offer an alternative to current spin dynamics imaging techniques whose spatial resolution may be limited for the study of sub-100 nm systems. This new characterisation tool will allow the study of model and advanced magnonic and spintronic systems in relation to the development of spin-wave based logic and communication devices.

The EHIS project brings together three French academic partners: CEMES (Toulouse), LPCNO (Toulouse) and LSPM (Villetaneuse). It involves 8 permanent researchers and 3 non-permanent researchers for a duration of 48 months.

In the field of mobile telecommunication, the most widespread radio frequency (RF) filters today are Surface Acoustic Wave (SAW) filters. These are bandpass filters made up of a single crystal piezoelectric substrate on which a set of interdigitated electrodes are deposited. Today, the new constraints imposed by telecommunication networks, particularly 5G, require substrates with better performance. Lithium Tantalate (LiTaO3 or LTO) integrated in the form of a thin piezoelectric film on insulator (POI for Piezoelectric-on-Insulator) opens up prospects in this direction. This LTO/SiO2/Si heterostructure makes it possible to produce filters with a better quality factor, a wider bandwidth, increased temperature stability and reduced insertion losses. The POI substrates are fabricated using the Smart Cut™ technology developed by the company Soitec, which allows the transfer of single crystal thin films. This thesis aims to address the following two issues: 1) Elucidate the impact of the damage of the piezoelectric matrix by the implantation as a function of the implanted hydrogen dose and the transformation of the system during annealing to initiate material fracture; 2) Identify the impact of the Smart CutTM process steps on the recovery of the full intrinsic piezoelectric properties of the bulk LTO.

This CIFRE thesis is part of a collaboration between CEMES-CNRS (Toulouse) and Soitec (Bernin).

The subject of the thesis is the bonding of two SiC substrates to each other. This results in the presence, in the final product, of a bonding interface whose physicochemical and electrical properties will play an important role in the performance of the devices produced on the product. The objective is therefore to study and provide a quantitative description of the behaviour of this SiC/SiC interface at high temperature. More precisely, the following points will be addressed: to study the morphology and the evolution of the interfaces between a crystalline SiC thin film and any SiC substrate, to understand the evolution of the crystalline quality of the thin films, to characterize the electrical resistivity of these interfaces.

This CIFRE thesis is part of a collaboration between CEMES-CNRS (Toulouse) and Soitec (Bernin).

FEAT ANR-19-CE24-0027 – Mars 2020/Septembre 2023

  • CEMES Centre d’Elaboration de Matériaux et d’Etudes Structurales
  • IFOX-HZB Helmholtz Zentrum Berlin für Materialen und Energie / Institut IFOX
  • UP University of Potsdam / Institut für Physik und Astronomie

The essential functionality of ferroelectrics comes from the possibility of reversing the polarisation under the application of an electric field. As the size of the ferroelectric decreases, the polarisation can become unstable, non-reversible or extremely weak due to the absence or asymmetry in the surface screening charges and their redistribution during reversal. It has recently been shown that the ferroelectric polarisation can be reversed in BaTiO3 films as thin as 1.6 nm deposited on Si. In this project, we want to study the ferroelectricity of very thin nanostructures on semiconductors but also with submicrometer lateral dimensions. We will thus address new aspects of ferroelectricity at the nanoscale with a view to integrating ferroelectrics on semiconductors for applications in nanoelectronics and integrated photonics.

We wish to address several technological challenges related to the fabrication of complex oxide nanostructures on a planar Si substrate. ALD, reactive ion etching and Ne ion beam machining will be implemented to prepare cylindrical nanostructures of variable height and diameter (2-100nm x 50-800 nm). We will discuss different topics at the frontiers of current knowledge on semiconductor ferroelectricity. Since in the nanostructures the surface contributes significantly or even more than the volume, the chemical screening will compete with the formation of the domains in lowering the depolarization energy; we will study these fundamental aspects for polarization control. We will use piezoelectric force microscopy/spectroscopy to image the domains and explore the complex contribution of electrochemical and ferroelectric states that might even induce a ferro-ionic mixed state. Raman spectroscopy and geometric phase analysis from TEM images of atomic structures will be compared to X-ray diffraction to study strain in nanostructures. We will also develop advanced transmission electron microscopy methodologies to map polarization direction and amplitude at the nanometer scale to understand the effects of boundary conditions and nanostructure shape on polarization distribution (closure flow, vortex). Finally, we will focus on the duration of the polarization reversal of ferroelectric nanostructures by femtosecond pump-probe spectroscopy and ultrafast X-ray diffraction with a laboratory plasma source or with Bessy II synchrotron radiation. In addition, the formation and reversal of domains will be studied by in situ TEM under an applied electric field. This approach carried out at CEMES with a dedicated microscope is at the frontier of current technological development and has never been demonstrated on ferroelectrics on semiconductors. The coupling of in situ TEM and ultrafast diffraction will contribute in a unique way to the knowledge of the dynamics of BaTiO3 films and nanostructures.

To conclude, we wish to address topics at the frontier of current knowledge on nanoscale ferroelectricity on Si with the hope of making breakthrough discoveries. Furthermore, the nanostructures studied in FEAT constitute the building blocks of many potential applications in nanoelectronics and integrated photonics.

BEOLAR Projet-ANR-20-CE24-0008       Mars 2021/Février 2025

  • GREMAN Matériaux, Microélectronique, Acoustique, Nanotechnologies
  • CEMES Centre d’Elaboration de Matériaux et d’Etudes Structurales
  • SPEC Service de physique de l’état condensé

Emerging communication technologies (5G, NFC) require thinned ferroelectric (FE) film varactors to operate at higher frequencies or lower voltages. The interface FE “dead layer” and an increased leakage current limit this evolution. Recent ab initio calculations show, at perovskite electrode/FE interfaces, the influence of chemical, polar and structural disparities on polarization stabilization and Schottky barrier height (SBH). In the BEPOLAR project (University of Tours, Project leader), we propose a systematic interface engineering using combinatorial laser ablation deposition. Local investigation by advanced spectroscopy and microscopy of chemically modulated interfaces, combined with DFT calculations (CEMES & CEA/SPEC partners), will lead to the identification of optimized materials (minimal FE dead layer and increased SBH), as well as understanding of the underlying physical and chemical mechanisms. The best interfaces will be tested in industrial varactors (ST Tours).


Ensemble des publications du groupe MEM via Hal.

Annuaire en cours d’élaboration