Industrial automation increasingly depends on electronics that can sense, process, store and communicate information within demanding size and energy constraints. Developments in silicon technologies offer new ways to combine these functions, from advanced CMOS devices to microsensors, emerging memories and power components. Research reported by CEA Tech and Leti in 2017 illustrates how these technologies were being explored across applications including the Internet of Things, automotive systems, energy and environmental monitoring.
For automation engineers and system integrators, the significance is not a single device or process. It is the expanding set of options for building compact systems that bring computation closer to sensors and connect complementary functions within one package or stack.
Silicon technologies for more integrated systems
Silicon research spans both process development and component design. CMOS scaling continues to pursue faster, more efficient computing, while “More than Moore” approaches add capabilities such as sensing, power handling, imaging and new forms of memory. These directions are relevant to industrial equipment, where control, measurement and communication functions may need to coexist in limited space.
Leti’s research infrastructure reflects the breadth of this work. Its platforms included 200 mm and 300 mm CMOS processing, a dedicated MEMS platform, and a 3D integration line. The facilities operated around the clock and supported process development, prototyping and characterization. A separate nano-characterization platform provided tools for examining materials and devices, including electron microscopy, X-ray diffraction and scanning-probe methods.
Such capabilities matter because a component’s practical value depends on more than its electrical behavior. Process compatibility, reliability, packaging and the ability to integrate it with other devices all influence whether a research concept can progress toward industrial use.
3D semiconductor integration connects complementary functions
Conventional circuit design places many functions side by side on a chip. 3D integration adds the option of stacking devices or components, potentially bringing sensing, storage, processing, actuation, communications and energy-related functions into a compact system. The research program described several approaches, including sequential integration, in which device layers are fabricated one above another.
Sequential integration presents a key process challenge: forming the upper devices without damaging the layers beneath them. That makes thermal budget a central design constraint. The reported work investigated low-temperature process steps, dopant activation, gate-stack reliability and intermediate interconnect layers. One reported process window was identified between 525°C and 550°C for several hours, balancing the stability of lower layers with the performance and reliability requirements of upper devices.
The potential applications extend beyond conventional high-performance computing. The research discussed 3D architectures for neuromorphic computing and processing-in-memory concepts, as well as co-integration of MOSFETs and sensors. For automation designers, these are possible routes to tighter coupling between data capture and local processing; they are not, by themselves, evidence of a ready-to-deploy industrial product.
Emerging memory and edge processing
Resistive RAM (RRAM) and phase-change memory (PCM) were among the emerging memory technologies covered in the report. These devices are being investigated as alternatives or complements to established memory structures. Their relevance includes non-volatile storage and potential roles in new computing architectures, including neuromorphic systems.
Memory behavior and array-level operation both matter. In RRAM arrays, programming energy and latency can be affected by pulse duration, voltage, current and device variability. The reported work evaluated programming patterns and described a circuit approach that detects switching during operation, with the aim of reducing programming consumption. This highlights an important engineering consideration: evaluating a memory technology at the individual-cell level is not enough. Array design, control circuitry and operating conditions also shape system-level outcomes.
For industrial edge devices, memory research is worth watching where equipment must retain information without continuous power or where processing and storage architectures are being reconsidered. Selection for a real application still requires assessment of reliability, integration needs and the operating conditions of the intended system.
Sensors, RF and power electronics
Microelectromechanical systems (MEMS) and nanoelectromechanical systems (NEMS) can provide sensing and actuation functions. The research program also addressed RF components, including switches and filters, as well as embedded RF functionality for future sensing systems. Together, these areas point toward sensor platforms that combine transducers with communications and supporting electronics.
Power electronics form another important strand. The report highlighted work on gallium nitride (GaN) power components and integration approaches for power devices. It also described a silicon H-bridge motor driver developed through collaboration. These efforts reflect continuing work to develop compact power and energy components, although the specific benefits of any implementation depend on its design and application.
Packaging and heterogeneous integration are essential to these developments. Combining different materials or device types can support functions that are difficult to achieve in a single process, but it also introduces integration and reliability requirements. The report’s examples—from image sensors and RF filters to power converters—show the range of components being considered for hybrid integration.
What automation teams should take from the research
For technical decision-makers, these developments suggest several practical points:
- Evaluate the system, not just the component. Consider interfaces, packaging, memory control and power requirements alongside device specifications.
- Track integration constraints early. In stacked devices, thermal budget and interconnect design can determine which combinations are feasible.
- Match technology to the application. A research direction suited to dense computing may not be the right choice for a sensor node, motor-control circuit or RF subsystem.
- Distinguish research results from deployment readiness. Process demonstrations and component studies are steps toward industrial transfer, not substitutes for application-specific qualification.
Conclusion
Silicon technologies are evolving along multiple paths: continued CMOS development, 3D integration, emerging memory, microsystems, RF components and power electronics. For industrial automation, the practical opportunity lies in understanding how these building blocks could be combined to support connected sensing and control systems. The engineering task is to assess the full integration chain—process, device, packaging, reliability and application requirements—before committing to a technology direction.