Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Long-Wave Infrared Spintronic Poisson Bolometers with High Sensitivity

Ziyi Yang, Sakshi Gupta, Jehan Shalabi, Daien He, Leif Bauer, Angshuman Deka, Zubin Jacob

TL;DR

This work addresses the challenge of achieving cryogenic-level sensitivity in uncooled LWIR detection, leveraging a spintronic Poisson bolometer in which signal and noise follow Poisson counting statistics. It integrates a broadband LWIR plasmonic absorber with a stochastic MTJ to boost absorption, achieving >60% absorptance across the LWIR band and aligning the peak with the blackbody spectrum near 300 K. Experimentally, the device delivers a best NEDT of 35 mK at 50 Hz, with multiple measurements near or below 100 mK, and an active pixel size of roughly 2 μm × 2 μm; the NEDT is related to the count-rate via $NEDT = \frac{C_n}{dC/dT}$. This approach moves uncooled LWIR detection toward cryogenic-like sensitivity, enabling high-speed imaging and remote sensing at room temperature with potential impact across imaging, sensing, and diagnostics.

Abstract

High-sensitivity long-wave infrared (LWIR) detection is crucial for observing weak thermal radiation. Recently, the spintronic Poisson bolometer was proposed as a promising platform for uncooled infrared detection. The Poisson bolometer operates in a probabilistic regime dominated by Poissonian noise, establishing a new detection paradigm. In contrast to traditional analog detectors, where signal and noise are continuous currents or voltages, the Poisson bolometer has both signal and noise governed by Poissonian counting statistics regardless of the light source, with the mean count rate modulated by incident radiation. In this work, we integrate a broadband plasmonic absorber optimized for LWIR absorption onto a spintronic Poisson bolometer to enhance thermal coupling and temperature rise in the sensing layer. The plasmonic absorber achieves over 60\% absorptance across the LWIR spectrum, matching the blackbody radiation peak at room temperature. The device exhibits a best noise-equivalent temperature difference (NEDT) of 35 mK at a 50 Hz frame rate and multiple results close to or below 100 mK, demonstrating room-temperature performance among the most sensitive uncooled LWIR detectors reported to date. This work advances uncooled infrared detection toward cryogenic-level sensitivity through the innovation of integrating spintronic materials and plasmonic materials, opening pathways to high-sensitivity LWIR sensing and imaging applications such as remote sensing, high-speed imaging, cryogenic system diagnostics, and industrial monitoring.

Long-Wave Infrared Spintronic Poisson Bolometers with High Sensitivity

TL;DR

This work addresses the challenge of achieving cryogenic-level sensitivity in uncooled LWIR detection, leveraging a spintronic Poisson bolometer in which signal and noise follow Poisson counting statistics. It integrates a broadband LWIR plasmonic absorber with a stochastic MTJ to boost absorption, achieving >60% absorptance across the LWIR band and aligning the peak with the blackbody spectrum near 300 K. Experimentally, the device delivers a best NEDT of 35 mK at 50 Hz, with multiple measurements near or below 100 mK, and an active pixel size of roughly 2 μm × 2 μm; the NEDT is related to the count-rate via . This approach moves uncooled LWIR detection toward cryogenic-like sensitivity, enabling high-speed imaging and remote sensing at room temperature with potential impact across imaging, sensing, and diagnostics.

Abstract

High-sensitivity long-wave infrared (LWIR) detection is crucial for observing weak thermal radiation. Recently, the spintronic Poisson bolometer was proposed as a promising platform for uncooled infrared detection. The Poisson bolometer operates in a probabilistic regime dominated by Poissonian noise, establishing a new detection paradigm. In contrast to traditional analog detectors, where signal and noise are continuous currents or voltages, the Poisson bolometer has both signal and noise governed by Poissonian counting statistics regardless of the light source, with the mean count rate modulated by incident radiation. In this work, we integrate a broadband plasmonic absorber optimized for LWIR absorption onto a spintronic Poisson bolometer to enhance thermal coupling and temperature rise in the sensing layer. The plasmonic absorber achieves over 60\% absorptance across the LWIR spectrum, matching the blackbody radiation peak at room temperature. The device exhibits a best noise-equivalent temperature difference (NEDT) of 35 mK at a 50 Hz frame rate and multiple results close to or below 100 mK, demonstrating room-temperature performance among the most sensitive uncooled LWIR detectors reported to date. This work advances uncooled infrared detection toward cryogenic-level sensitivity through the innovation of integrating spintronic materials and plasmonic materials, opening pathways to high-sensitivity LWIR sensing and imaging applications such as remote sensing, high-speed imaging, cryogenic system diagnostics, and industrial monitoring.

Paper Structure

This paper contains 3 sections, 4 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Results
  3. Conclusion

Figures (4)

  • Figure 1: LWIR detector sensitivity landscape, spintronic Poisson bolometer structure, and operation principles. (a)-(b) Visionary comparisons of state-of-the-art LWIR detectors based on NEDT. (a) Cooled detectors, with NEDT measured in vacuum. (b) Uncooled detectors, with NEDT measured in air unless labeled “vacuum”. In this work, the best NEDT achieved with our spintronic Poisson bolometer is 35 mK at 50 Hz (measured in air), which ranks among the most sensitive uncooled LWIR detectors. (c) Spintronic Poisson bolometer device structure schematic. The device is composed of a magnetic tunnel junction (MTJ) and a plasmonic antenna array absorber on top. The MTJ is composed of a sensing layer, an insulating layer, a readout layer, and a synthetic antiferromagnetic layer (SAF). The SAF pins the magnetization of the readout layer to the perpendicular direction, while the sensing layer has two stable states of magnetization ($M_1$ and $M_2$) that are separated by an energy barrier. The device exhibits tunnel magnetoresistance (TMR) depending on the relative orientation of the magnetization of the sensing layer and the readout layer. The plasmonic absorber is optimized for broadband LWIR absorption, enabling a higher temperature increase in the sensing layer. (d) Operation mechanism of spintronic Poisson bolometers. When no light is incident, transitions at a low rate between M1 and M2 exist in the sensing layer due to natural heat, which are read out through the device’s resistance change and are referred to as dark counts; when light is incident, the temperature of the sensing layer rises, increasing the rate of transitions, which are referred to as bright counts.
  • Figure 2: Design and characterization of the LWIR plasmonic antenna array absorber. (a) Plasmonic absorber unit cell structure and dimensions. The absorber is a metal-insulator-metal (MIM) type of metasurface with circular Ti nanoantennas on Ge-Ti layers. The dimensions are engineered to achieve enhanced broadband LWIR absorption. (b) SEM image of the fabricated spintronic Poisson bolometer device. The MTJ nanopillar bridges two top gold contacts and two bottom gold contacts. The top contact strip area above the MTJ is covered with the plasmonic absorber. The left inset shows the plasmonic nanoantenna array. The MTJ is located near the nanoantenna to obtain higher heat transfer. (c) Left vertical axis: Simulated and measured absorptance of the plasmonic absorber. The design enables broadband LWIR absorption higher than 60% with a peak around 11 µm wavelength. Right vertical axis: Simulated blackbody spectral radiance at 300 K. At room temperature, the blackbody radiation peaks at the LWIR spectrum (red shaded region). The absorption of the plasmonic antenna array peaks at the same region to maximize the temperature increase in the sensing layer of the MTJ.
  • Figure 3: NEDT measurement setup and baseline count rate measurement. (a) NEDT measurement setup schematic. The spintronic Poisson bolometer is on a chip wire-bonded to a PCB board with a readout circuit. A magnet is placed 8 mm behind the PCB board to apply an external bias field perpendicular to the MTJ layer stack. A ZnSe LWIR lens (f/0.5) is placed at 1.5 cm in front of the PCB board to focus blackbody radiation onto the device. The 18cm by 18cm large-area blackbody is placed at 5 cm from the lens and set to different temperatures for NEDT measurement. (b) Tunnel magnetoresistance (TMR) measurement of this spintronic Poisson bolometer device. A TMR as high as 38% enables high voltage contrast between the two different magnetization states and reliable signal readout. (c) Voltage waveforms of the baseline transitions when the blackbody temperature is set to 293.15 K (temperature of the environment). A continuous tuning of the count rate by the applied field is observed. (d) Field dependence of the baseline count rate with 20 ms integration time. The dashed line suggests a good fit to the Neel-Arrhenius law. The offset in peak count rate from H = 0 is due to stray fields induced in the sensing layer by the nearby readout and SAF layer.
  • Figure 4: NETDT measurement results. (a) Measured NEDT of a FLIR A325sc LWIR camera. This result is measured with an f/0.6 camera lens and 30 Hz frame rate. No external LWIR lens is used for the camera. The distance between the camera lens and the blackbody is 5 cm. The 79 mK result is close to the datasheet's 50 mK sensitivity specification, validating the measurement setup’s effectiveness, and is used as a benchmark of a commercial uncooled LWIR detector based on VOx microbolometers. The error bars in this plot and the following plots represent the standard deviation of 5 frames. (b) NEDT result measured with dT = 0.3 K. For this applied field and bias voltage, the device shows a linear response between 24.7 and 25.3 ° C. Afterward, the count rate stops increasing due to device saturation. (c) Repeated NEDT measurement with dT = 0.1 K with the same applied field and bias voltage as (b), but with a focus on the linear response region to achieve the best possible sensitivity. The result shows an improved NEDT of 35 mK. (d) Repeated NEDT measurement with decreased bias voltage and slightly increased applied field. The result shows a decreased count rate and higher NEDT due to both the current-induced spin-transfer torque effect and the field dependence of the count rate. With the bias voltage and the applied field properly selected, the spintronic Poisson bolometer is able to achieve close to, well below 100 mK NEDT in multiple measurements on different days. (e)-(f) NEDT measured with negative bias voltage. The result shows the tunability of device performance with the bias voltage and applied field. Each of these 6 measurements is done on a different day. The integration time for (b) - (d) is 20 ms (50 Hz frame rate). The integration time for (e) and (f) is 40 ms (25 Hz frame rate), all comparable to the frame rate of state-of-the-art commercial LWIR cameras.