Research Bits: Aug. 25

Bacterial transistors

Researchers from the Massachusetts Institute of Technology (MIT) engineered bacteria that can function as transistors and form circuits for environmental monitoring.

The team used the bacterium Pantoea agglomerans to create two types of transistors, one which can be switched on by a molecule called OC-6, and another which the molecule switches off. Each transistor also detects the presence of a target molecule called OC-12. Depending on whether that molecule is present, and whether the switch is active, the transistors produce a different output molecule, called OHC-14. Three strains of the same bacterium were used to create relays that translate the OHC-14 signal into an output that can be fed into another transistor.

Circuits are created by printing colonies of bacteria onto plates containing agar, a growth medium. Each colony is printed about 5 millimeters from the nearest one, allowing signals to travel only to the nearest colony, which then relays them to the next one, so information flows only in one direction.

“We’ve built some initial computer architecture components that are commonly used, but any operation can be built with these five strains,” said Hamid Doosthosseini, an MIT postdoc, in a press release.

In the study, the researchers demonstrated a transistor that can perform several types of logic operations depending on its location in the circuit layout, including “multi-input,” “or,” and “imply” gates. They also combined the transistors to create more complex circuits that can add up two signals, process more signals simultaneously, or function as a demultiplexer. The largest circuit contained 24 bacterial colonies wired together to add two inputs.

The researchers hope to design circuits that could be applied to plant leaves or roots to detect and respond to different environmental stressors. The circuits take about eight hours to perform each calculation, which is fast enough for biological applications, said Christopher Voigt, head of MIT’s Department of Biological Engineering, in a press release. “We’re not trying to replace computers, but rather put computational control into biology. If you have bacteria on the root of a plant, or the plant itself is doing the computing, running a simple calculation overnight is fast enough relative to a growth season.” [1]

Forgetful antiferroelectric device

Researchers from Seoul National University and Sungkyunkwan University designed an antiferroelectric-based AI semiconductor device that can retain recent inputs while autonomously forgetting older information, a characteristic that they leveraged as a core computational function for processing time-series data.

The team fabricated a two-terminal device combining zirconium dioxide with amorphous indium gallium zinc oxide (a-IGZO) and achieved distinct current differences from minimal state changes by optimizing the IGZO composition. The optimized device demonstrated an on/off current ratio of approximately 890 and successfully distinguished 16 combinations of 4-bit inputs into distinct current states. In handwritten digit recognition, it achieved a 90.4% recognition accuracy while reducing input data by 75%

“The core of this research is that we utilized the property of antiferroelectric materials returning to their original state after voltage removal not as a limitation, but as a computational function called ‘forgetting’,” said Min Hyuk Park, a professor in the Department of Materials Science and Engineering at Seoul National University, in a press release. “By simultaneously implementing non-linear transformation, short-term memory, and natural initialization in a single two-terminal device, we have expanded the device options for low-power time-series AI hardware. We plan to reduce the device area and conduct array-level circuit verification to expand it into an edge AI system that processes voice, biological, and environmental signals on-site.” [2]

High-entropy mixing

Researchers from Carnegie Mellon University and Penn State University transformed an insulating metal oxide into a high-performance semiconductor using high-entropy mixing.

The material, A⁶WO₄, has both semiconducting properties and ultra-low thermal conductivity. By developing a complex mixture of manganese, iron, cobalt, nickel, copper, and zinc into a tungsten oxide framework within a single crystal structure, called wolframite, the researchers intentionally created a state of high configurational entropy.

“What stood out experimentally is that this material combines semiconducting transport with exceptionally poor heat conduction—two properties that are difficult to achieve together in oxides,” said Zhiqiang Mao, professor of physics at Penn State, in a statement. “The measurements show that high-entropy mixing is not just a way to create chemical disorder; it can be used as a powerful knob to engineer electronic and thermal properties simultaneously.”

“This approach activates unique microscopic mechanisms that we believe could serve as a new set of guiding principles for future materials engineering and design of next-generation devices,” added Ismaila Dabo, professor of materials science and engineering at CMU, in a statement. The researchers anticipate the new material could have promise in thermoelectric devices that convert waste heat into electricity. [3]

References

[1] H. Doosthosseini, H. Chen, & C.A. Voigt. Living circuit boards built by printing bacterial transistors. Nat Chem Biol (2026). https://doi.org/10.1038/s41589-026-02300-3

[2] T. Kwon, M. Jeong, S. I. Hwang, et al. Volatile ZrO₂ Antiferroelectric Tunnel Junctions for Rapid, Energy-Efficient Physical Reservoir Computing. Advanced Science (2026): e76692. https://doi.org/10.1002/advs.76692

[3] R.A. Robinson, T. Karimzadeh Sabet, F. Marques dos Santos Vieira, et al. High-entropy design of transition metal oxide semiconductors with ultra-low thermal conductivity. Commun Mater 7, 198 (2026). https://doi.org/10.1038/s43246-026-01103-2

The post Research Bits: Aug. 25 appeared first on Semiconductor Engineering.