Demonstration of High Detectivity of Microbolometer Array with High-Purity Semi-Conducting CNT Networks

Uncooled infrared sensors of bolometer type have a wide range of applications such as security, military, food inspection, health care, and automotive night vision system. Currently, the highly sensitive device development is the key issue for further expansion of demand. The bolometer is an infrared detector of long wave infrared (LWIR) region for radiant heat by means of an infrared absorber having a temperature sensitive electrical resistance material. Infrared radiation strikes the absorber material heating it and thus changing resistor material resistance. Therefore, bolometer’s performance is strongly limited by temperature coefficient of resistance (TCR) of the resistor. The conventional resistor is generally based on vanadium oxide (VO_x) with TCR of about -2%/K^[1] and an outstanding resistor is essential to achieve highly sensitive infrared detectors. Recently, single-walled carbon nanotubes (SWCNTs) have been expected as promising materials with high TCR and high chemical stability. We have reported that semi-conducting SWCNT networks extracted by the “Electric-field inducing Layer Formation (ELF)” method^[2] show high TCR which is close to -6%/K^[3]. The pristine SWCNTs are fundamentally a mixture of both semi-conducting and metallic SWCNTs. The ELF method is the remarkable promising technique to extract semi-conducting SWCNTs with high purity, which show stable device performance and excellent electrical transportation property. Previously, we showed TCR of semi-conducting SWCNT network was responsible for the increased sensitivity of the microbolometer. We have established a fabrication process for Video Graphics Array (VGA) format microbolometer focal plane allays (FPAs) with semiconducting SWCNTs^[4]. And the responsivity of a single pixel in FPAs as test element group (TEG) was estimated to be over 10⁵ V/W at 0.2 V bias voltage. This is about three times higher than that in FPAs with VO_x ^[4].<br/><br/>In this study, the low-frequency noise of the TEG adjacenet to the FPA was measured, and the black body detectivity, D* was estimated. Device chip which was mounted on the ceramic carrier was installed and infrared-shielded in a vacuum dewar. For the low-frequency noise measurement, bias voltage was applied in series with a load resistor and the TEG. The lectric potential between the TEG and the load resistor was amplified by a current amplifier, and the frequency dependence of the noise power density was obtained by a signal analyzer. At frequencies up to 200 Hz, the noise power density was inversely proportional to frequency. This shows that the noise component is dominated by 1/f noise at the operating frequency of a typical microbolometer camera. The D* calculated from the noise and the responsivity was equal to or better than microbolometer using VO_x fabricated by a similar fabricating process.<br/><br/>In conclusion, we showed the high-purity semi-conducting SWCNT networks extracted by ELF method effectively work for higher detectivity of microbolometer FPAs. To further increase detectivity, we will improve the TCR and reduce the 1/f noise in the semi-conducting CNT networks.<br/><br/>Acknowledgments: Part of this study was supported by Innovative Science and Technology Initiative for Security Grant No. JPJ004596, ATLA, Japan.<br/><br/>Reference<br/>[1] C. Chen, et. al., Sen. Act. A. Phys. 90, 2001, 212.<br/>[2] K. Ihara, et. al., J. Phys. Chem. C, 115, 2011, 22827.<br/>[3] T. Tanaka. et. al., 2022 MRS Fall Meeting, NM02.09.08.<br/>[4] T. Tanaka. et. al., 2023 MRS Fall Meeting, EL07.06.10.