France

Biological networks of neurons and synapses have the capacity to process multiple operations through short electrical pulses, also referred to as “spikes”, with very low energy consumption. Spiking Neural Networks (SNN) are a new form of Artificial Intelligence (AI) that aim to replicate these cognitive features in order to improve next generation computer’s performance. We present the design and simulation of an individual spiking laser to be used as a photonic neuron for a SNN scale-up. The neuron is a semiconductor laser optimized for pulsed operation through modification of the CEA Leti III-V on Si integrated continuous wave laser&#39;s heterostructure. A first demonstration of its neural features connected with synaptic functions is also reported.

ESSCIRC ESSDERC Educational Events

Pulsed lasers neurons integrated on silicon for dense spiking neural network

iii-v on si process integration

mode locking

saturable absorber

semiconductor laser

spiking neural network

Naval Information Warfare Center Pacific is partnering with AFCEA San Diego chapter to host the workshop.

The NAML workshop is an opportunity for researchers from Navy research and development centers and our partner organizations to exchange technical ideas and to come together as a team.

The goals of NAML are:

* To spread awareness of current machine learning research relevant to Naval applications

* To connect machine learning researchers with experts in military requirements

* To build and strengthen collaborations within the defense research community

**Watch this video to hear about the AFCEA purpose.**

<iframe width=1280 height=706 frameborder="0" scrolling="no" src="https://screencast-o-matic.com/player/creleVVcvRh?v=6&ff=1&title=0&controls=1" allowfullscreen="true"></iframe>

NAML 2021

Biological networks of neurons and synapses have the capacity to process multiple operations through short electrical pulses, also referred to as “spikes”, with very low energy consumption. Spiking Neural Networks (SNN) are a new form of Artificial Intelligence (AI) that aim to replicate these cognitive features in order to improve next generation computer’s performance. We present the design and simulation of an individual spiking laser to be used as a photonic neuron for a SNN scale-up. The neuron is a semiconductor laser optimized for pulsed operation through modification of the CEA Leti III-V on Si integrated continuous wave laser's heterostructure. A first demonstration of its neural features connected with synaptic functions is also reported.

technical paper

The aim of ESSCIRC and ESSDERC is to provide an annual European forum for the presentation and discussion of recent advances in solid-state devices and circuits.
The level of integration for system-on-chip design is rapidly increasing.

This is made available by advances in semiconductor technology. Therefore, more than ever before, a deeper interaction among technologists, device experts, IC designers, and system designers is necessary. ESSCIRC and ESSDERC are governed by a common Steering Committee and share Plenary Keynote Presentations and Joint Sessions bridging both device and circuit communities.

Highlights:
* 3 joint keynote presentations
* 3 keynote presentations ESSCIRC
* 3 keynote presentations ESSDERC
* Invited papers with overall coverage of all aspects of advanced devices and circuits
* Presentation of IEEE and ESSCIRC/ESSDERC Awards
* Educational Events

JOINT PLENARY talks live broadcast: September 13, 2021<br>
ESSCIRC and ESSDERC KEYNOTE talks live broadcast: September 15, 2021<br>
LIVE Q&A INTERACTIVE sessions for all Conference Technical sessions: September 14 – 16 – 17, 2021<br>
LIVE Q&A INTERACTIVE sessions for all Educational Events: September 20-21-22, 2021<br>
(on demand videos available from September 6 until November 30, 2021)

**Please see here for the [ESSCIRC Guide](https://drive.google.com/file/d/10dghzuv4aQBty4QfTIY9Q8hXyqsue8Le/view?usp=sharing), [ESSDERC Guide](https://drive.google.com/file/d/10dx9y2tm9cHPDa4upNUYjfXCf44aPry5/view?usp=sharing), and the [Program at a Glance](https://drive.google.com/file/d/10eU0OLTXEpiNzI5hkqv4mSr43T8cedPP/view?usp=sharing)**

Please register for educational events at the following link: https://www.registerme.eu/event/esscirc-essderc-2021/

Please register for the ESSCIRC/ESSDERC Educational Events to join the event.

The Educational Events of ESSCIRC ESSDERC 2021. 

The usual amounts of data continue to grow exponentially. ICT currently consumes around 4% of the world's energy. Therefore one of the goals of the EU project “IPCEI” is a significant improvement the carbon footprint of digitization with the help of innovative Power semiconductors. In one example this goal will be addressed by increasing efficiency using the future generations of gallium nitride (GaN) technologies and housings for the electronics industry in Europe. In addition to circuit breakers for highly efficient energy conversion, inexpensive GaN on silicon high-frequency technologies will be developed, which accelerate the 5G expansion. Due to the large energy requirement, an efficiency improvement of only one percentage point results in a saving of approx. 10 megatons of CO2.
In this way, ICT can not only make many applications more “gree”, it can also become more “green” itself.

Power electronics for Green ICT

Power modules play a main role in traction inverter applications. They control engine speed and torque, regulating the energy flow between battery and motor. Due to the demanding electric and reliability performances requested by end costumers, semiconductor devices based on wide band gap compound are introduced in novel power modules. To support such device developments, there is the need to set a robust and accurate experimental methodology that shall be supported by numerical models. The targets are the prediction of thermal performance and the reliability forecast. In this framework, it has been proposed a dedicated flow to thermally characterize a multi-chip SiC MOSFET power module for automotive applications. Body-diode drop voltage has been selected as the temperature sensitive electric parameter to monitor the average junction temperature for thermal impedance measurements. Finite element model has been employed to predict temperature distribution due to device self-heating, referring to the previous characterization for the calibration of the numerical model. Once the thermal simulation methodology has been set, module temperature during inverter operations can be calculated, collecting temperature history profile into some loading conditions using a Rainflow-like approach. The several loading cases have been characterized by different temperature swing and cycle time. These classified data can be considered to evaluate lifetime duration due to selected failure mechanisms , supporting the definition of the most proper durability assessment for the application-relevant failure modes.

Structural characterization and simulation of a direct-cooled power module

After more than 35 years from its first conception and in the year of the recognition with the IEEE milestone, the Smart Power BCD technology is still evolving continuously and rapidly to serve at best the application requirements of new and changing markets.
As stated in the IEEE milestone for “Multiple Silicon Technologies on a Chip, 1985”, STMicrolectronics pioneered the super-integrated silicon-gate process combining Bipolar, CMOS and DMOS (BCD) transistors in single chips for power-demanding applications.
The markets currently requiring faster and higher technology evolutions are the Personal Electronics, bringing needs of wireless charging and power management, the Industrial, asking for more efficient power conversion, and the Automotive, where new challenges come from the battery management and inverter functions in Electric Vehicles and from Autonomous Driving. To meet those market requirements, the Smart Power BCD roadmap is following three key directions: digital processing capability increase (announced 40nm node with embedded non-volatile memory solutions), evolution to higher power and voltage (up to 10kV on-die galvanic isolation) and customization by application. 
We will walk along these three evolution paths, presenting their technology features and related product examples.

The Smart Power BCD evolution in the year of the IEEE milestone

AP&S wet process development on SiC wafers started in 2015 and was initiated by an inquiry from one of the worldwide leading semiconductor manufacturers in Europe. This one was faced with the task of introducing a new production line for microchips based on SiC for a new progressive automotive application. The goal to achieve efficient, high-yield and cost-effective production for SiC products thus has come in focus. One important step in this overall production process has been the reliable and cost-efficient wet processing. The joint development between the semiconductor manufacturer and AP&S, in particular the Demo Center, thus began. This journey took us from metal lift-off process to metal etching processing with endpoint detection on singlewafer application to high-volume batch processing. In this presentation both the changes in the semiconductor manufacturing processes and the solutions derived from them will be presented. Clear breakdowns of the various process data, parameters and conditions will be shown including a detailed overview of the development on SiC from 2015 to 2021.

Wet SiC Metal-Etch - From singlewafer to batch application

Silicon on Insulator is used extensively for its robustness in automotive applications.
X-FAB's BCD on SOI 180nm platform aims at high performance for HV applications and focuses on flexibility and ease of use.
It enables the integration of high voltage BCD devices for driving actuators and motors with high precision analog functions for sensor interface and signal processing. SOI in this context is relevant for its robustness such as latchup immunity and reduced high temperature leakage and also for its isolation properties.
In order to serve our foundry customers, the X-FAB 180nm BCD on SOI platform is highly modular and is fully supported by models, PDKs and design IPs.
When faced with particular operating range constraints, designers can select technology modules to optimize circuit performance for high power/high voltage interfaces.
Besides the traditional ON resistance for a given breakdown voltage, our recent development improved on the usability of the medium and high voltage options with devices ranging from 12V to over 300V. For each geometry the most suitable performance trade-off can be designed with fully characterized and modeled transistors for each voltage classe.
Our technologies are automotive qualified to grade 0 and characterized up to 175oC. We are extending the model validation up to 200oC allowing the design of sensor interfaces and drivers that will operate in a high temperature environment while allowing for dissipation and package thermal resistance.
To support the increase in digital content, non volatile memory IPs are provided either as embedded in the CMOS flow or with a dedicated SONOS module allowing denser implementation. For the SOI platform the memory IP designs are adapted for operation at hot temperature with a specific automotive friendly architecture. 


New results of Bipolar Cmos Dmos on Silicon On Insulator

High performance, low cost solid-state LiDAR systems require novel system architectures regarding sensor and readout circuits in combination with tailored illumination solutions. This talk will focus on performance trade-offs, suitable transmitting units, efficient operating modes and the integration in scalable systems. Elmos Solid State LiDAR concept heads forward to make environment detection smarter to meet upcoming requirements for new ADAS and autonomous driving.
We are presenting a scalable multi-channel laser-driver generating short high intensity pulses for direct time-of-flight measurements. Using a die stacking approach to minimize module size and parasitic inductance, peak currents of up to 50A are applied to the laser-diodes within a 1ns pulse. In addition, the laser driver has integrated functional safety diagnostics including monitoring of the average output power for eye-safe operation.
Implemented in the Elmos automotive-grade 350nm CMOS process, the laser driver is featured in our compact 905nm solid-state LiDAR system for demonstration purposes, which also includes an Elmos-designed SPAD imager.

Sub-nanosecond high-power pulsed multi-channel laser driver for time-of-flight applications

In the last 15 years, Fondazione Bruno Kessler (FBK) has developed a custom technology for single-photon avalanche diodes (SPADs) and silicon photomultipliers (SiPMs) optimized for several applications, from high-energy physics experiments to medical imaging or light detection and ranging (LiDAR). In the IPCEI framework, FBK is upgrading its microfabrication facility for the realization of 3D integrated radiation sensors. In the specific case of SiPMs, 3D integration technologies will allow to improve its capabilities by using back side illuminated (BSI) devices and Through Silicon Vias (TSV) interconnections. The BSI approach for SiPMs can be exploited to improve the current photon detection efficiency (PDE), even in the wavelength ranges far away from the visible spectrum, where typically it drops down to about 10% for the near infrared (NIR) and about 15% for vacuum ultraviolet (VUV). A first batch of BSI SiPMs for NIR detection was recently fabricated at FBK as a proof of concept. After wafer bonding and thinning, the characterization results showed an increase of the PDE of about 50% at 900 nm for BSI sensors with metal reflectors compared with the standard front side illuminated (FSI) technology. Research on the realization of the fabrication process modules optimized for near ultraviolet (NUV) and VUV BSI SiPMs is currently ongoing and will also be presented and discussed.

Back side illuminated SiPM from NIR to NUV light detection

Requirements towards chip separation are getting more and more complex in recent years. The increasing trend for lower chip thickness and different backside metal stacks leads to a big variety in processes and material combinations interacting with each other. Unwanted interactions can even block the overall manufacturing flow of chips in module or package assembly.
With the newly presented method on chip level it is now possible to measure in detail the release process of chip from the dicing tape during pick-up and to determine the influence of unit processes like dicing or UV irradiation and their interaction effects on the die bond process during assembly. In this presentation we will focus on the effect of different stress release processes and show the effect of different UV chambers on the UV curing of dicing tapes.

Characterization of Preassembly influence on Die Attach process

LYNRED product offer covers a large span of applications, in cooled and uncooled infrared domains. The presentation will focus on two detectors addressing two different infrared technologies:
 First - « Key elements for hybridization technology development of 2000x2000 pixels Infrared Focal Plane Arrays”..
 Then a LWIR camera module will be presented: an uncooled bolometers detector stacked with a custom IR image processor (ISP). The ISP is a dedicated ASIC built to apply a pipeline of image processing algorithms to the raw sensor data which corrects for pixel defects and non-uniformities in the sensor, without requiring to an external mechanical shutter. With this approach, LYNRED demonstrates an easy-to-use, plug-and-play solution


Lynred Advanced Developments in the Field of Infrared Imaging Technologies

For more than 50 years, Moore's Law has been driven by resolution improvements of lithography scanners which generate an image of the lithography mask on the semiconductor wafer. This image contains the patterning information needed to build up an integrated circuit.
The latest scanner generation uses 13.5 nm light at 0.33NA. Since the last 2 years they are now essential to produce leading edge semiconductor devices. To further increase the resolution ZEISS is working on next generation EUV optics. This optics consists of a highly flexible illumination system and projection optics with NA 0.55 enabling single-exposure sub 8nm half-pitch resolution to allow scaling beyond the next decade.
In this presentation we will explain the system design of the high-NA optical column and we will report on the manufacturing status of mirrors and frames. Furthermore, the buildup of the infrastructure including mirror polishing, coating, surface figure metrology, mirror handling, and integration tooling will be shown.

Pulsed lasers neurons integrated on silicon for dense spiking neural network

Downloads

Next from ESSCIRC ESSDERC Educational Events

Power electronics for Green ICT

Similar lecture

SpikePropamine: Meta-Learning Synaptic Plasticity for Spiking Neural Networks

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Pulsed lasers neurons integrated on silicon for dense spiking neural network

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ESSCIRC ESSDERC Educational Events

Power electronics for Green ICT

Similar lecture

SpikePropamine: Meta-Learning Synaptic Plasticity for Spiking Neural Networks

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads