A thermometer smaller than an ant's antenna can track chip temperatures in 100 nanoseconds

Every one of the billions of transistors on a modern processor is a potential hotspot. When transistors overheat under computational load, chip performance drops sharply. But current temperature sensors sit outside the chip, providing only indirect and delayed measurements. What if you could embed thousands of thermometers directly among the transistors, each one small enough to fit between circuit elements and fast enough to catch thermal spikes before they cause damage?

A team led by Saptarshi Das at Penn State has built exactly that. Their sensors, made from a class of two-dimensional bimetallic thiophosphates never previously used for thermal sensing, measure just one square micrometer, roughly several thousand times narrower than a human hair. They respond to temperature changes in 100 nanoseconds and consume a fraction of the power that conventional silicon-based sensors require. The work was published March 6, 2026, in Nature Sensors.

The material nobody else was using

Penn State has established itself as a center for two-dimensional materials research, and Das's group recognized an opportunity in bimetallic thiophosphates that other teams had overlooked. These materials have a distinctive property: ions within them continue to move effectively even when exposed to electrical currents. In most semiconductor applications, this ionic mobility is considered a nuisance, something engineers try to eliminate. Das's team turned it into an advantage.

Ions are highly sensitive to temperature changes. By coupling the movement of ions (which respond to heat) with the flow of electrons (which carry the signal to be read), the sensors achieve strong temperature dependence at extremely small scales. The same electrical currents powering the chip can drive the sensors, meaning they add negligible overhead to chip operations.

One hundred times smaller, eighty times more efficient

The numbers paint a clear picture. Each sensor is more than 100 times smaller than other leading thermal sensor designs. Power efficiency improves by up to 80 times compared to traditional silicon-based temperature monitoring, because the sensors do not need the additional circuitry or signal converters that conventional approaches require.

The team manufactured the sensors and placed thousands on a single chip using advanced fabrication equipment at Penn State's Materials Research Institute Nanofabrication Laboratory. The manufacturing process is compatible with existing semiconductor fabrication methods, which is important for any technology aiming for integration into commercial chip production.

From parasitic ions to precise measurements

Dipanjan Sen, the paper's first author and a doctoral candidate in engineering science, explained the physics behind the design. The 2D material couples two types of particle transport: ionic and electronic. While industry typically views ionic transport in transistors as a problem that degrades device performance, those same ions become extremely useful thermometers when their temperature sensitivity is deliberately exploited.

The coupling allows each sensor to read temperature through ion behavior while transmitting that reading through electron flow, all within the same tiny device. No external conversion hardware is needed, which is what keeps the sensors so compact and power-efficient.

Proof of concept, not a product

Das is straightforward about where this technology stands. It is a proof of concept. The sensors have been demonstrated on test chips in a laboratory environment, not in commercial processors running real workloads. The transition from laboratory demonstration to integration in production chips would require collaboration with semiconductor manufacturers, extensive reliability testing, and validation across different chip architectures and operating conditions.

There are also open questions about long-term stability. Two-dimensional materials can degrade when exposed to air and moisture, and any on-chip sensor would need to survive the harsh conditions of chip packaging and years of continuous operation. The paper does not address accelerated lifetime testing.

Beyond temperature

Das sees the thermal sensors as a first application in a broader program. The same design framework, using 2D materials to exploit ion-electron coupling, could be adapted to sense chemical, optical, or mechanical signals in similarly compact formats. If the approach scales as hoped, it could enable a new generation of sensors embedded throughout electronic systems, providing real-time monitoring of conditions that currently go unmeasured.

The research team included collaborators from Northwestern University and the University of Chemistry and Technology Prague.