Beneath the screen of a smartphone lies a semiconductor chip threaded with billions of copper wires, each thinner than one-thousandth the width of a human hair. These wires connect the transistors and give rise to computation itself. But in recent years, the semiconductor industry has run up against a physical wall: copper gets slower as it gets thinner.
Electrons Slamming Into Walls: The Limits of Copper Wiring
Copper excels as a conductor because free electrons zipping through the metal can travel long distances even while repeatedly colliding with obstacles. But these electrons have an intrinsic travel distance called the "mean free path." For copper, this is roughly 39–40 nm.
Why does this number matter? At today's leading-edge process nodes, wire widths have shrunk below this mean free path. Once a wire becomes narrower than 40 nm, electrons begin slamming into the wire's sidewalls before they even collide with the crystal lattice, triggering repeated scattering. When this happens, copper's electrical resistance surges in a manner inversely proportional to scale.
That's not the only problem. Because copper is highly reactive, a diffusion barrier 2 to 4 nm thick must be placed around it to prevent it from diffusing into the surrounding insulating material. This non-conductive "shell" eats up a substantial portion of the cross-sectional area in thin wires, further shrinking the effective cross-section through which copper can carry current. The challenge facing copper interconnects at leading-edge nodes isn't a matter of manufacturing precision—it's a matter of physics.
Copper became the standard interconnect material in the late 1990s. Before that, aluminum was the mainstay, but IBM and AMD (then in a joint venture with Motorola) achieved copper wiring integration, dramatically boosting both performance and density. At the time, the shift to copper was hailed as a "megatrend." Now, a full 30 years later, that same copper faces the same question all over again.
The Paradox: The Thinner It Gets, the Better It Conducts
Research published in the journal Science on July 16, 2026, by a group led by Cornell University materials scientist Judy Cha has the potential to overturn this conventional wisdom. Lead author and PhD student Yeryun Cheon and colleagues focused their attention on niobium arsenide (NbAs), a material known as a "topological semimetal."
In NbAs, electrons flow not just through the bulk interior but also across the material's surface in a special quantum state. These surface electrons are protected from scattering by a quantum mechanical mechanism known as "topological protection." In other words, as a wire is made thinner and its surface-area-to-volume ratio rises, the contribution from these fast-moving surface electrons increases correspondingly—the exact opposite of copper's behavior.
Rather than climbing over the rugged mountain terrain of the crystal lattice like bulk electrons do, the surface electrons in a topological semimetal glide along a flat, high-speed highway.
"Unlike NbAs's surface electrons, in copper's case, making the wire thinner increases how often electrons hit the walls," Cha explains. "In NbAs, the surface electrons move fast and resist scattering. That's why it becomes more advantageous at the nanoscale."
The research team measured a room-temperature resistivity of 9.7±1.6 μΩ·cm—three to four times lower than that of bulk NbAs single crystals. This stands in stark contrast to copper's behavior, where resistance rises as the material thins. The material's stability is also noteworthy. While quantum materials generally only exhibit their special properties under low-temperature, vibration-free conditions, the research team confirmed that NbAs displays quantum mechanical effects even at room temperature and even in samples that aren't necessarily of the highest quality.
This value was obtained using the four-terminal method. The research team attached a total of four electrodes—at both ends of the nanowire and at two points along its interior—and independently measured current and voltage to eliminate the effects of contact resistance. Comparing resistivity while systematically varying the diameter revealed the "thinner equals lower resistance" behavior.
A Fabrication Technique That Opened a New Door
NbAs's electrical properties had already caught the attention of theoretical researchers before this, but there was a barrier to actually fabricating it as nanowires. Conventional synthesis methods fall into two categories—"vapor-liquid-solid growth" and "chemical vapor deposition"—but neither allows precise control over a nanowire's diameter or shape.
The solution Cha's team adopted is called "thermomechanical nanomolding." The process is completed in four steps. First, a bulk polycrystalline block of NbAs is placed on a mold made of porous aluminum oxide (alumina) containing an array of aligned nanoscale pores. Next, this is exposed to a high temperature just below the melting point while pressure is applied for several hours, forcing the material into the mold's pores. After the material crystallizes in the shape of the pores, the alumina is selectively dissolved away using an alkaline solution, leaving behind single-crystal nanowires of uniform diameter. Cha describes this as being "like a pasta maker. Just as you can switch out the front plate to make fettuccine, you can use molds with different pore diameters to control the diameter down to about 10 nm."
Thermomechanical nanomolding delivered more than just precision—it also dramatically accelerated the pace of material exploration. Where researchers previously could only study one or two material systems per year, thermomechanical nanomolding has enabled a pace of one material system per month—a tenfold increase in throughput.
"This is the essential breakthrough," Cha says. Being able to rapidly evaluate large numbers of candidate materials can significantly shorten the bottleneck between theory and experimental verification. Beyond NbAs's own significance, this method carries ripple effects for the entire field of future materials exploration. The research involved numerous institutions, including IBM and Johns Hopkins University. Professor Zhiting Tian of Cornell's Sibley School of Mechanical and Aerospace Engineering measured the thermal conductivity properties, and the work made use of infrastructure at the Cornell NanoScale Science and Technology Facility.
Cha's lab previously published research on molybdenum monophosphide (MoP), another topological semimetal, in Advanced Materials in 2023, demonstrating that it was more stable than copper at the nanoscale and had superior resistance to electromigration. However, while MoP was more stable than copper, its conductivity didn't improve as it got thinner. NbAs is positioned as the first material to achieve the next step—combining both stability and improved conductivity.
A comparison of the major candidate replacement materials is shown below.
| Material | Behavior of conductive layer as it "gets thinner" | Room-temperature stability | Diffusion barrier unnecessary | Barriers to practical use |
|---|---|---|---|---|
| Copper (Cu) | Resistance increases | Good | Not needed (but required in practice) | Size effects |
| Molybdenum monophosphide (MoP) | Resistance unchanged | Good | Not needed | No conductivity improvement |
| Niobium arsenide (NbAs) | Resistance decreases | Good (demonstrated at room temperature) | Not needed (theoretically) | Toxicity of arsenic |
The Arsenic Barrier—and What This Research Demonstrates
However, there is a fundamental problem with bringing NbAs into semiconductor fabs: arsenic.
Arsenic is a neurotoxin, and integrating it into current semiconductor manufacturing processes would require managing toxicity-related costs and complying with safety regulations. NbAs isn't the first arsenic-based compound to appear in the semiconductor field. Gallium arsenide (GaAs) has been used in high-frequency devices and solar cells for decades, and established procedures already exist for managing arsenic in manufacturing settings. That said, GaAs is more brittle and harder to process than silicon, which has limited its adoption in leading-edge logic chips. Whether NbAs follows the same path depends less on toxicity management issues than on whether it can be successfully integrated with silicon CMOS processes. Cha herself has stated, "I don't know whether niobium arsenide will become a practical replacement for copper," positioning this research as a proof of concept. The significance of this paper lies in the fact that it has, for the first time, opened the door to the possibility that the category of topological semimetals could become real device materials capable of engineering quantum effects for practical use.
At the same time, there is room to apply the screening efficiency of thermomechanical nanomolding toward the search for arsenic-free topological semimetals. At a pace of one material system per month, many candidates could be evaluated within just a few years.
Many questions remain unresolved. Does NbAs nanowires' high conductivity hold up at sizes below some threshold in nanometers? How would it perform in actual multilayer wiring structures? These have yet to be verified. Demonstrating that the quantum behavior of topological semimetals actually works as interconnect wiring in integrated circuits will require evaluation at smaller scales and in real device structures. As Cha points out, the true value of this achievement lies less in the material itself than in having established a starting point from which topological semimetals can be evaluated as genuine candidates for interconnect materials. What comes after copper remains a question without an answer—for now.