image: Figure | Structure, working principle and photoresponse of the tunable-sensitivity phototransistor. a, Schematic of the phototransistor, consisting of a MoS2 FET integrated with a photodiode. h-BN serves as both the dielectric and protective layer, while graphite is used as the contact and gate electrode. b, Band structure of the device in the dark. c, Band structure of the device under illumination. d, Dependence of IDS on Pin under 516-nm light at different VGS. e, Variation of the current ratio (tunable-sensitivity phototransistor: I0/Ilight; conventional phototransistor: Ilight/I0) as a function of the light intensity ratio (Pin/P0) at different VGS, where P0 represents the minimum detectable light intensity, and I0 is the corresponding current. f, Comparison of responsivity (R = |ILight-IDark|/Pin) of the tunable-sensitivity phototransistor and conventional phototransistor under different Pin.
Credit: Shun Feng et al.
With the rapid development of the intelligent machine vision, the demand for high-performance sensing technologies has grown significantly, especially for accurate detection in real-world environments involving weak light, low contrast, and complex backgrounds. However, traditional photodetectors rely on fixed sensitivity and struggle to balance strong and weak signals simultaneously. They are easily overwhelmed by noise, limiting their potential in next-generation vision systems. Addressing this challenge, the research team has introduced a bioinspired phototransistor with “adaptive eyesight,” offering a new solution for high-performance machine vision.
In their latest study, a research team led by Prof. Dong-Ming Sun from the Institute of Metal Research, Chinese Academy of Sciences, developed a bioinspired phototransistor with tunable sensitivity, capable of actively adjusting its response range to different light intensities and achieving human-eye-like dynamic perception. The work was recently published in Light: Science & Applications. The device can selectively amplify or suppress specific light-intensity segments, enabling it to extract target signals from noise and perform exceptionally well in weak-light detection and low-contrast recognition. As the researchers explained, “Traditional detectors capture average brightness, while our device behaves more like an eye that pays extra attention to critical highlights.”
At the heart of this phototransistor lies its tunable-sensitivity mechanism. By modulating the gate voltage, the device exhibits pronounced current changes only within certain illumination ranges, enabling segmented light-intensity response. For instance, under one gate condition, it becomes highly sensitive to extremely weak signals, while under another, it filters out weak signals and focuses on strong ones—creating a multi-stage “adaptive sensitivity” similar to the pupil adjustment of the human eye. The researchers further summarized: “This means a single device can adapt to diverse lighting conditions, ensuring high-quality detection from dim to bright environments.”
Looking ahead, the team believes this technology will have a broad impact on intelligent sensing, robotic vision, and low-power imaging systems. “Tunable sensitivity gives machines a capability similar to the human pupil for the first time,” they noted. “In the future, we aim to further expand the device’s dynamic range, enable tunable spectral response, and integrate it into large-scale arrays to support applications such as autonomous driving, night-vision systems, and biomedical imaging.” The researchers believe that this bioinspired phototransistor will serve as a key foundation for advancing intelligent vision technologies toward higher precision and lower power consumption.
Article Title
Intelligent metasurface imager and recognizer