All opportunities

Offers : 92

Transformation to simplify classification functions

Mail Sélection

Start date : 22 February 2020

offer n° SL-DRT-20-0871

Many of the problems addressed by Artificial Intelligence are problems of classifying complex input data into different classes. The functions transforming the complex input space into a simpler, linearly separable space are done either by learning (deep convolutional networks) or by projecting input data into a high-dimensional space in order to obtain a “rich” non-linear representation of the inputs, then having a linear matching between the high-dimensional space and the output units (the “reservoir computing” approach). These concepts are also linked to the Support Vector Machines (work of Vapnik 1966-1995). The objective of the thesis is to study this type of transformations that can be applicable for real applications, and to define an optimized, generic architecture for a given application domain, allowing data to be pre-processed in order to prepare them for a classification requiring a minimum of operations and which can, for example, be done on the fly (continuous learning).

The targeted research results are multiple:

– From a theoretical point of view, an approach unifying the transformations done by deep learning networks, “reservoir computing” and approaches that transform a complex input space into an essentially linearly separable space.

– Define which transformations should be done in practice for a given class of problems (e. g. object recognition) by simplifying them as much as possible (depending on the error rate, false positives, etc.).

– Propose optimized architectures, making the best use of advanced technologies (semiconductor, 3D stacking, photonics, etc.).

The final result would be the proposal of an optimized module, which could be used as preprocessing unit, to help efficiently perform transfer learning, one shot learning and continuous learning functions for example.

  • Keywords : Computer science and software, DACLE, Leti
  • Laboratory : DACLE / Leti
  • CEA code : SL-DRT-20-0871
  • Contact : marc.duranton@cea.fr

Innovative Mixed RF and low power devices integration in view of advanced fdsoi SOC

Mail Sélection

Start date : 1 October 2020

offer n° SL-DRT-20-1027

Connected mobile devices are becoming a strategic imperative in order to remain attractive, improve efficiency and competitive for advanced electronic applications. The wireless revolution where Laptops, Smartphone’s, tablets, TVs, vehicles and enterprises are connected in a cloud style environment makes possible communication anywhere at any time. Recent developments in wireless communications with the emergence of advanced radio-frequency standard such as LTE, LTE-A and 5 G have brought numerous challenges. The most critical challenge is to provide higher levels of integration with more power efficiency and cost-effective solutions on the same-chip. In parallel to the development of nanometer CMOS as well as beyond-CMOS device technologies for switching, memory and analog functions, the increasing need to integrate various (heterogeneous) technologies (e.g. RF communication, power control, passive components, sensors, actuators) helps to migrate from the system board-level into the system-in- package (SiP) or to the system-on- chip (SoC). In fact, mobile System-on-Chip (SoC) with heterogeneous integration of multiple technologies has truly revolutionized the semiconductor industry. Thanks to the trap-rich Silicon-on-Insulator (SOI) substrate invented at UCL and developed in collaboration with SOITEC, RF SOI presents outstanding RF performance. In addition, the presence of the buried oxide layer not only reduces the junction capacitance but also offers the opportunity of using high resistivity substrate to reduce substrate related RF losses and coupling. However in case of SoC integration the trap-rich is not suitable all across the wafer and localized solutions should be envisaged.

Fabrication on a 28FDSOI 300mm platform of specific RF stuctures

Characterization of the substrate impact (HR, trap rich, etc …) on the RF figure of merit

Imagine and integrate new technological process schemes to implement localized ‘trap-rich like’ area before or after FDSOI device realization.

Integrate some technological modules on new designed structures and electrical caracterization

  • Keywords : Electronics and microelectronics - Optoelectronics, DCOS, Leti
  • Laboratory : DCOS / Leti
  • CEA code : SL-DRT-20-1027
  • Contact : claire.fenouillet-beranger@cea.fr

Innovative package using ultra-thin chip transfer on substrate (UTCoS)

Mail Sélection

Start date : 1 July 2020

offer n° SL-DRT-20-0703

The subject is in the field of advanced microsystems, which is a strategic axis for CEA-LETI associated with current packaging trends: extreme compactness, conformability and functionalization. The approach developed is unique and allows the carrying of ultra-fine chips (“substrate-less “) on any type of host substrates with a process adaptable to different bonding solutions including direct bonding or with an intermediate layer. It uses CEA-leti’s expertise in ultra-thinning of layers, temporary and permanent bonding techniques, thin layer transfer on temporary carrier and advanced cutting techniques (plasma, laser). In addition, it uses advanced packaging technologies with thin FLEX substrates, molding encapsulation and additive technology interconnections (RDL, 3D printing and screen-printing). The proposed solution is generic and addresses many applications as CMOS image sensors, MEMS (sensors and piezo-electrics actuators), RF ICs (filters, switches, antenna arrays).

The focus of the study will be on CMOS image sensor with passive and active focal plan curvature with the objective of realizing a first functional demonstrator.

  • Keywords : DCOS, Leti
  • Laboratory : DCOS / Leti
  • CEA code : SL-DRT-20-0703
  • Contact : gabriel.pares@cea.fr

Lead free piezoelectric transducers and reliability

Mail Sélection

Start date : 1 October 2020

offer n° SL-DRT-20-0875

The LCMA, laboratory of micro-actuator components, is working on the integration of piezoelectric materials into microsystems that make it possible to obtain the electromechanical transduction function. Lead Zirconate Titanate (PZT) is the best performing piezoelectric material for micro-actuator applications. However, the introduction in the near future of a new standard for the lead level allowed in chips (European RoHS Directive) leads us to evaluate alternative lead-free materials to the PZT for piezoelectric actuator applications. The development of lead-free materials has in fact become a major focus of piezoelectric research. This research led to revisiting and modifying some classical piezoelectric materials such as KNbO3 and BaTiO3. In particular, the KNaxNb1-xO3 (KNN) family has been identified as promising.

The aim of the thesis is therefore to evaluate lead-free piezoelectric materials and to compare their properties with that of the reference material, PZT. Simple test vehicles will be made in the LETI clean room and then characterized using different techniques available in our laboratories. The electrical and piezoelectric properties of the materials as well as the reliability of the realized devices will be investigated throughout the integration process to optimize the lead-free piezoelectric (PZE) technology developed. Methodologies for characterization and reliability studies of PZE will be also developed. In order to carry out this work, the doctoral student will be able to rely on a solid experience of LETI experts developed since almost 20 years on thin film piezoelectric materials and on the characterization and reliability of the components.

  • Keywords : Materials and applications, DCOS, Leti
  • Laboratory : DCOS / Leti
  • CEA code : SL-DRT-20-0875
  • Contact : gwenael.le-rhun@cea.fr

Mechanical and electrical contact study of micro-inserts hybridization

Mail Sélection

Start date : 20 February 2020

offer n° SL-DRT-20-0974

In microtechnology, hybridizations are carried out, in particular optical sensor pixel matrix hybridizations on their read circuitry (CMOS). Connections, at each pixel, must be relevant mechanically as well electrically. During the thesis, an innovative process of hybridization based on micro-insertion will be developed and implemented. You should have to design and manufacture micro-inserts (typically nail-shape micro-insert) which have “to be driven” in metal pads (eg aluminum) with good electrical contact and sufficient mechanical strength not to risk any deshybridization.

During the thesis, you will try to model and simulate this micro-insertion. In parallel, in order to evaluate the insertion and de-insertion forces and to evaluate the electrical resistance of connection, two characterization benches should be developed. You will also be involved in the process of hybridization and in the joint optimization of the design – process point.

  • Keywords : DOPT, Leti
  • Laboratory : DOPT / Leti
  • CEA code : SL-DRT-20-0974
  • Contact : natacha.raphoz@cea.fr
More information
X