Maciej J. Ogorzalek

Poland

Towards New Architectures for 3D Integration

ABSTRACT:
For the past 50 years advances in integrated circuit design were based on the assumptions of planar process introduced by Jean Hoerni - one of the fathers of IC developments. All circuit elements were built onto a slice of flat material surface and successively covered by layers of other materials to introduce transistors, insulation layers, metal layers for interconnects and further semiconductor layers for enhanced functionality.

Looking at many generations of chips offering different functionalities the “flat-world” assumptions and technologies used are omnipresent. Even more, the shapes of building blocks are predominantly rectangular with interconnect lines crossing at right angles. Such geometries bring many limitations. Some attempts were made to make so-called X-interconnects going along the diagonals of the rectangles looking for shortening the lengths of wires.

Further developments in ICs introduced 3D architectures. Here the flat slices are being superimposed in a regular vertical construction - several specific techniques have been proposed, all based on principles of planar design - like constructing a building floor after floor. This kind of IC architectures are relatively easy to design and fabricate. Efficient technologies exist for their realisation. They are however far from being optimal! Building layered structures with limited variety of shapes introduces several very strong constraints. Power efficiency is limited, heat removal is difficult, wire length is not optimal, which results in timing problems and signal delays. Interconnect remains a major problem - so-called fan-out remains unsolved limitation. Looking at the solutions visible in living organisms one can immediately spot that there are no flat surfaces, no rectangles, connection and distribution systems in plants, organism tissues of animals and humans form complicated shapes and structures far from regular. Interconnects between eg. brain cells offer fan-out many levels superior to best ICs.

With the latest technological developments we can now think about constructing chips with unusual architectures and new properties in some cases resembling the nature but also proposing new concepts. This presentation will give some insights into these new directions, new ideas, techniques, and technologies comparing them with latest "flat-world" achievements.

Maciej J. Ogorzalek is Professor and Head of the Department of Information Technologies, Jagiellonian University Krakow, Poland. He held many visiting positions in Switzerland, US, Spain, Japan, Germany. Between 2006-2009 he held the Chair of Biosignals and Systems, Hong Kong Polytechnic University under the Distinguished Scholars Scheme. In 2019/2020 he was visiting professor at the Waseda University, Tokio, Japan and later at the Integrated Systems Laboratory at EPFL.

Author or co-author of over 380 papers published in journals and conference proceedings and a book Chaos and Complexity in Nonlinear Electronic Circuits. He gave over 60 plenary and keynote lectures at major conferences world-wide. He served as Editor-in-Chief of the IEEE Circuits and Systems Magazine (2004-2007), member of the editorial boards of the IEEE Transactions on Circuits and Systems Part I, Proceedings of the IEEE, International Journal of Bifurcation and Chaos, International Journal of Circuit Theory and Applications also the NOLTA Journal IEICE Japan. Dr. Ogorzalek is IEEE Fellow (1997). He was IEEE 2008 Circuits and Systems Society President. He served as IEEE Division 1 Director, Member of the IEEE Board of Directors (2016-2017). He is Member of the Polish Academy of Sciences (PAN) and Member of the European Academy of Sciences (Academia Europaea).