AI-Optimized Vanadium Oxide Multilayers for More Than 20-fold Enhancement in Bolometric Performance

Area of Science:

• Materials Science
• Nanotechnology
• Sensor Technology

Background:

• Phase-transition materials like vanadium dioxide (VO2) show non-linear, hysteretic behavior, limiting their use in infrared bolometric sensors.
• Nonstoichiometric VOx is used in bolometers but has degraded transitions and lower temperature coefficient of resistance (TCR).
• Achieving high TCR and linear, non-hysteretic response is a key challenge for microbolometer technology.

Purpose of the Study:

• To develop a multilayer approach using machine-learning-optimized WxV1-xOy thin films to achieve linear, non-hysteretic responses with high TCR.
• To overcome the limitations of traditional vanadium dioxide-based infrared sensors.
• To enhance the performance of microbolometers for advanced infrared detection.

Main Methods:

• Utilized a multilayer thin-film deposition strategy with varying tungsten (W) doping ratios in VxOy.
• Employed genetic algorithm optimization to tailor film properties for desired TCR profiles and reduced hysteresis.
• Grew WxV1-xOy thin films under complementary metal-oxide semiconductor (CMOS)-compatible conditions.

Main Results:

• Achieved tailored linear/flat TCR profiles and significantly reduced hysteresis in multilayer WxV1-xOy systems.
• Demonstrated simultaneous high TCR and low electrical noise.
• Reported a 23.6-fold improvement in universal bolometric performance compared to commercial materials.

Conclusions:

• The developed multilayer WxV1-xOy films offer a viable solution for linear, high-performance infrared bolometers.
• The machine-learning-guided approach provides a general methodology for optimizing materials with large, linear responses.
• This work has broad implications for microbolometer technology and other stimulus-responsive devices.