IRDS: International Roadmap
for Devices & Systems

download IRDS™
Join IRDS Community

Technical Articles

This section of the IEEE IRDS™ website presents articles that address some considerations of the IRDS teams.

The IRDS international focus teams present the following articles to highlight the role of application requirements on system, component and then device technical requirements for the IRDS assessments. Today’s microelectronics applications rely on a complex set of systems technologies, innovative architectures, components, integrated devices and that deliver goods and services in a never-ending variety of products and services. Computing innovations continuously promise the next “big” thing through discovery/invention/development.

The IRDS continues to focus research, development, and manufacturing considerations for the various industries it serves and does this through the strength of the roadmap teams and their global expertise. While the IRDS end products—the roadmap reports—present assessments of pre-competitive/shared technical challenges and opportunities for solutions (which promotes healthy industrial competitiveness), our teams first determine the various technical influences that result in such assessments.

To do such assessments effectively mandates understanding the dependencies of systems and architectures on applications, and the requirements of components and devices on system requirements throughout an application’s lifecycle. For example, the design of devices demands an understanding of not only device physics, but what is possible to manufacture at historical high volume. The manufacturing of complex architectures demands, for example, that metrology tools and methods can adequately measure and ensure reliability and quality. Sustainable industrial practices comprehend environmental health for people, the planet, and a profitable industry.

Predicting Computing’s Co-Designed Future: The Applications Benchmarking and Systems & Architectures Chapters of the IRDS
Kirk Bresniker, Tom Conte, and Vladimir Getov

Prominent both in the name and the organizational structure of the International Roadmap for Devices and Systems is what differentiates the IRDS from prior and peer road mapping efforts: a focus on Systems design. As was recognized across both the International Technology Roadmap for Semiconductors and the IEEE Rebooting Computing communities, we no longer have the luxury of independent advancement of the technology elements. If we want to continue to use computation to tackle the ever increasingly challenging problems of gaining insight in a time that matters and with energy consumption that enables us to sustainably provide the benefits it offers equitably to all, we must embrace a co-design methodology. To this co-design challenge, Applications Benchmarking (AB) brings the critical understanding of “what do and what will we need to be able to compute?” and Systems and Architectures (SA) brings the boundary conditions of “what are the size, weight, power, privacy, security and sustainability criteria at each design envelope from the edge device to the exascale data center?” What makes this both more challenging and vitally important is the near constant disruption and innovation in both areas. An algorithm that used to occupy the relatively benign environment of the data center now needs to run under the hood of an autonomous vehicle, exposed to harsh environmental conditions, physical tampering and advanced persistent threats. The monolithic scientific simulator is being replaced by an intersectional workflow of modelling, massive data analytics and pervasive AI/ML. Rapid evolution punctuated by discontinuous shifts is the new normal and the AB and SA topic areas bring this understanding to the roadmap. Read more…

The International Roadmap of Devices and Systems (IRDS) presents device characteristics for extending Moore’s Law with “More Moore.” Today’s systems have characteristics that compel new device designs and manufacturing technologies that encompass considerations for future devices Beyond CMOS. These also include developments and opportunities with Cryogenic Electronics and Information Processing. Read more…

The Internet of Everything (IoE) is continuing to expand in applications that demand higher volumes of higher performance communication. The IoE was initially defined as a wide range of Internet of Things (IoT) devices communicating with cloud computing that store data and which was analyzed with applications and actions communicated. As IoE was used for a broader range of applications, some applications had unacceptably slow performance due to the latency of communicating using the cloud. To overcome this latency limitation, some applications added local storage and processing close to the IoT devices and local network, which is referred to as “edge computing.”1 Many IoT products are manufactured with More than Moore technology to integrate multiple devices manufactured with different technologies.

“More than Moore” refers to the incorporation into devices of functionalities that do not necessarily scale according to Moore’s Law, but provide additional value in different ways.2 The More-than-Moore approach allows for the non-digital functionalities (e.g., RF communication, power control, passive components, sensors, actuators) to migrate from the system board-level into the package (SiP) or onto the chip (SoC). Read more…

For more than 30 years the typical improvements in computer performance (teraflops, petaflops) were mainly accomplished by shrinking the features of the IC to a new technology node at great expenses. In recent times this approach has become limited to performance improvements to less than 30%, however the adoption of advanced packaging (AP) technologies for dense off chip integration (DOCI) of computing modules/systems has resulted in the DOUBLING of performance and a 75% REDUCTION in energy required to transfer data between die (e.g., Google Tensor). This result has led to widespread adoption of DOCI via AP not just for conventional computing but it is also adopted by quantum computing. However, the existing AP technologies were actually developed and used in the U.S. over three decades ago and then aggressively adopted by offshore foundry organizations that realized very tangible profits. Up to now these AP technologies have not cycled back to be manufactured in the U.S. Representatives of offshore providers who merely replicated U.S.-developed AP technologies have coined the term “heterogeneous integration” that is now used widely for marketing purposes. To date, all the commercial implementations do NOT incorporate the essence and benefits of DOCI. If improvements in future high-performance computing (HPC) are expected to continue it is mandatory to
further develop and incorporate all the benefits of DOCI. Read more…

The International Roadmap for Devices and Systems (IRDS) predicts the integration of 2D materials into high-volume manufacturing as channel materials within the next decade, primarily in ultra-scaled and low-power devices. While their widespread adoption in advanced chip manufacturing is evolving, the need for diverse characterization methods is clear. This is necessary to assess structural, electrical, compositional, and mechanical properties to control and optimize 2D materials in mass-produced devices. Although the lab-to-fab transition remains nascent and a universal metrology solution is yet to emerge, rapid community progress underscores the potential for significant advancements. This paper reviews current measurement capabilities, identifies gaps in essential metrology for CMOS-compatible 2D materials, and explores fundamental measurement science limitations when applying these techniques in high-volume semiconductor manufacturing. Read more…