Embedded innovative sensors and software in power electronic components – CAPTIF

Those nanosensors are made by deposition of 14 nm gold nanoparticles, functionalized with tri-phosphine ligands by Convective Self-Assembly. Their electrodes are composed of a thin conductive layer of gold. 50 nm deposition of alumina suppresses sensor drift. A spin-coating of liquid polyimide allows their integration on a power module previously defined as a target prototype. Electromagnetic sensors which are near-field loop antennas, were designed and characterized while taking into account their coupling mode. Analyses of the output sensor data, when integrated into a power demonstrator, were performed. The software sensors are linear functional observer, considering unknown inputs and developed for large size systems. So, it is possible to overcome the lack of knowledge of certain entries by treating them as unmeasured disturbances. The used aging indicator is the strain energy density. Measurement and indicator correlations led to formalize the lifetime prediction method and its link with measured data. Its actualization is made by the direct or indirect measurement of the temperature of a given point of the assembly.

This project had demonstrated how integrating sensors into power electronics modules helped predict thermomechanical behavior as well as health monitoring. Developped sensor design and assembly processes, highlighted a possible integration of a set of multi-physical sensors to a pre-industrial assembly process for conventional or innovative technologies. Software sensors, that used multi-physic sensor data, allowed to restore unmeasured information. Finally, the necessary link between the updating of an aging indicator and the physical and software sensors was revealed.
More specifically, the key results of the project are:
1-Strain gauges and nanosensors of temperature were manufactured and characterized. However, it has been noted: a high sensitivity of the strain gauges; a maximum operating temperatures limited to + 50 ° C and a drift in their electrical resistance that impacts performance. As a result, aPSI3D has oriented the demonstrators towards more conventional technology modules.
2-The electromagnetic sensors to integrate, loop type magnetic probes, were developed, modeled and characterized. Two objectives were achieved, the calibration of the antenna and the identification of its performance in terms of frequency band and measurement resolution. Following the instrumentation of an elementary converter, the real-time acquisition of the magnetic field was perfomed on the frequency band up to 80 MHz.
3- It has been demonstrated for the software sensors, that the linear functional observer estimates asymptotically the evolution of the temperature of a desired point on an experimental case study.
4-The aging indicator used was the deformation energy density that is accumulated in the viscoplastic joint assembly. It was also established that the temperature provided by the sensors was more efficient on the DED than the deformation measurements.

As a result of the project, aPSI3D has extended this concept of integrated sensor to the case of any other functionality added to power modules. In particular, the integration strategy of temperature sensors and decoupling capacitors is being study. Thus, within the framework of the Nano2022 program, aPSI3D is realizing the integration of fine capacitors, adapted to processes and constraints.
The prospects for improvement of the sensors developed by Nanolike mainly concern the improvement of the properties of nanosensors: increase the higher temperature limit; Remove the drift of the electrical resistance; Decrease the dispersion of the electrical resistance.
The two main perspectives of the work carried out on the software sensors are related on the one hand, to methods of characterizations, and on the other hand, to the necessary confrontation of the theory of the observers of linear functional with unknown inputs, to practical use and implementation.
In relation to aging monitoring of the assembly, the methodology proposed to meet the objectives is based on the energy dissipated in the plastic deformation during fatigue aging. This methodology is intrinsically adaptive to the different conditions of use of power modules. Nevertheless, the thermophysical parameters describing the viscoplastic behavior of the control material are not entirely relevant because they do not take into account the damage over time. Prospects thus seem natural: the taking into account of the damage of the materials in the models and the onboard adaptive approach, allowing to realize in situ a management of the reliability of the system.
The perspectives related to the integration of electromagnetic sensors are to use them in order to obtain a real time information of the occurrence of a mechanism of failure. That deals with signal theory (modeling, and time vs frequency characterization), and data packaging.

During the project, 13 valorization acts were identified. 3 are multi-partners and were proposed for international (2) and national (1) conferences or workshops, mainly for an industrial audience. The 10 single-partner publications were produced mainly for academics, for international (2 journals and 5 conferences) and national (2 conferences and 1 part of a paper of popularization) audience.

The joint emergence of Wide Band Gap materials (SiC, GaN, C) and new generation hybrid integration techniques significantly enhance performances of power electronic modules. Such modules should operate in severe environment and constraints: high temperature high power density, fast switching, etc. The challenges for driving these new modules are about developing, integration and signal processing techniques of several types of sensors. Industrial and academic partners, within this project, are proposing complementary skills and experiences in the technological and scientific domains: multi-physic sensors, signal processing, integration and reliability for power electronics. Nanoparticle-based strain gauges, temperature sensors and electromagnetic array sensors will be validated and integrated in power electronic device. The main interest of such sensors is their low power consumption, their miniaturization and their possible integration as well as their accuracy. A specific multi-physic model will be developed in order to specify the best location of interest within the power device module. Electromagnetic sensors will be issued from previous studies and adapted to dedicated device. This will result in a better knowledge of real time current density location, as well as current frequencies. For both sensor types, the data packaging will be challenging spin-offs. Finally, advanced data processing techniques – estimation as well as signal filtering – will be adapted to numerous sensor outputs. A clean room process flowchart will be established to guarantee an advanced pre-industrial prototype. The industrial partners will certify that project outputs will be perfectly adapted to future products. This will result in advanced health monitoring of power device.
The project is composed by 5 partners – 2 research laboratories and 3 companies. Research laboratories will develop electromagnetic sensors, will define the data processing techniques and the multi-physics modeling as well as perform reliability tests for sensors and functionalized power devices. Industrial partners will provide nano-technology sensors, adapt the clean room process and guarantee the technology readiness level. 6 work packages are defined. The program will be developed during 42 months. During the first stage, overall specifications will be established as well as a set of sensor will be studied and developed. Finally, they will be integrated in a demonstrator device. In the meanwhile, multi-physics modelling and preliminary reliability tests will be considered. This will provide overall information and recommendations for the final integrated prototype. In the second stage, data processing techniques will be studied and developed as far as data filtering and estimation are concerned. This will allow fulfilling the number and variety of signals to be analyzed. In the final step, a set of sensor would be manufactured and dedicated to a functional power electronic device. To conclude better optimization and behavioral knowledge, real time and off line health monitoring, will be issued from the CAPTIF results. The main objective is to build, for the first time, a complete multi-scale integrated set of sensors. The analysis of the multi-physic data should give relevant and real-time information to derive some appropriate and optimized designs, control and reliability rules on new power devices.