The future of electronic materials

Episode Summary

Executive Summary: Russ Altman and Eric Pop discuss how chip materials are reaching physical limits, prompting research into alternatives to copper and silicon. Pop explains that future electronics may rely on ultra-thin, surface-dominated conductors like niobium phosphide and 2D semiconductors like molybdenum disulfide, with AI increasingly helping discover and manufacture next-generation materials.

Main Topics: Why electronics need new materials (Priority: 5/5): The episode frames a shift in electronics as miniaturization pushes copper, silicon, and current architectures toward limits, especially below about 10 nanometers. Copper’s limits in ultra-thin interconnects (Priority: 5/5): Pop explains that copper wiring becomes less effective when narrowed to sub-10-nanometer dimensions because surface scattering and diffusion barriers reduce performance. Silicon’s dominance and emerging alternatives (Priority: 5/5): Silicon remains the main semiconductor for low-power electronics, but contenders such as carbon nanotubes and 2D semiconductors are being explored for future devices. Universal memory and logic-memory bottlenecks (Priority: 4/5): The conversation highlights the separation between logic and memory as a major performance bottleneck and defines universal memory as fast, non-volatile memory co-located with computation. Surface-dominated materials and topological behavior (Priority: 4/5): Pop describes niobium phosphide as a candidate material whose surfaces may conduct unusually well, enabling better performance at extreme nanoscale dimensions. AI in materials discovery and manufacturing (Priority: 4/5): The discussion closes with how AI may accelerate both prediction of new materials and the harder problem of figuring out how to manufacture them reliably.

Key Arguments: Electronics have relied on mostly the same materials—silicon and copper—for decades, but shrinking dimensions are exposing fundamental limits. The main bottleneck in modern computing is not just transistor speed, but the cost of moving data between separate logic and memory chips. Copper performs well in bulk, but at widths below about 10 nanometers electron scattering from surfaces and diffusion issues reduce its advantage. New materials can outperform copper only in very specific nanoscale regimes, especially when their surfaces are intrinsically conductive. Silicon remains practical and cheap for many uses, but future low-power transistors may require atomically thin materials to keep improving. Carbon nanotubes and 2D semiconductors are promising because they are only a few atoms thick and can switch at low voltage. AI is beginning to move materials science from trial-and-error discovery toward guided prediction, but manufacturing remains a separate major challenge.

Data Points: Timeline of silicon-based electronics: about 60 years - Silicon has been the dominant semiconductor material for mainstream electronics for decades. Timeline of copper interconnects in chips: about 20-25 years - Copper replaced aluminum in chip wiring roughly two decades ago. Hafnium oxide adoption: about 10 years - Hafnium-based insulators were introduced relatively recently compared with earlier silicon oxide. Transistor voltage: about 1 volt - Pop notes that today’s low-power transistors switch at roughly one volt and zero volts. Early transistor size: about 10 microns - Late-1960s transistors were around the diameter of a human hair or a red blood cell. Virus-scale transistor era: about 100 nanometers - About 15-20 years ago, transistor dimensions reached roughly virus size. Current transistor scale: about five times less than a virus - Modern individual transistors are now substantially smaller than virus-scale dimensions. Weirdness threshold for materials: about 10-20 nanometers - Below this scale, conventional silicon and copper become increasingly difficult to use effectively. Copper electron travel distance: tens of nanometers - Electrons in copper travel only a few tens of nanometers between collisions. Niobium phosphide composition: about 50% niobium and phosphorus - Pop describes niobium phosphide as the candidate conductor under study. 2D semiconductor thickness: 3 atoms thick - Molybdenum disulfide is described as a three-atom-thick sheet. Carbon nanotube diameter: about 1 nanometer - Carbon nanotubes are presented as nanoscale semiconductor contenders. Vertical silicon fin size: about 6 nanometers across and 40-50 nanometers tall - Current silicon transistors are being fabricated as fins, though the industry is moving toward nanosheets.

Pivotal Quotes: "the materials that we build our electronics with, silicon and copper, may be on their way out as we introduce better and newer materials" — Russ Altman: Opening framing of the episode’s central thesis "there's hope now that at least for sub-tens of nanometers of conventional metals, of which copper is the best, there may be a class of other materials which could carry electricity a little bit better" — Russ Altman: Introduction to the motivation for exploring copper alternatives "I think we're just starting to scratch the surface of these new materials. We are at the beginning of a materials revolution" — Eric Pop: Pop’s closing assessment of the field’s trajectory

Implications: Chip performance gains will increasingly depend on new materials, not just shrinking old ones. Expect advances in ultra-thin conductors, 2D semiconductors, and AI-guided materials discovery to reshape electronics and manufacturing.