存在读出辉光时红外探测器的基本噪声极限
Fundamental Noise Limits of Infrared Detectors in the Presence of Readout Glow
AI总结：
该研究针对红外探测器读出噪声问题，揭示读出辉光导致平均降噪效果不及理论值，推导基本噪声极限并通过HxRG和LmAPD传感器实验验证，为相关天文观测设备设计提供依据。
AI中文摘要：
红外传感器阵列的读出噪声仍是地基和空间天文学的主要障碍。长期以来人们认识到，即将到来的极大望远镜若要充分发挥潜力，必须大幅降低读出噪声；这也是宜居世界望远镜（HWO）的限制性约束，该空间望远镜旨在探测和表征附近类地系外行星。降低读出噪声的主要策略是通过多次非破坏性读取进行平均，但通常降噪效果不如理论预测的1/√N缩放。本研究表明，这种较差的平均行为很大程度上可由读出辉光解释，读出辉光是传感器电子器件的光子发射，在读出期间在像素中产生光电子。由于辉光随读取次数累积而非平均，这会施加一个基本噪声底限σ_min≈1.5σ_RN^(1/2)G^(1/4)。这将HxRG型传感器的平均降噪限制在约2-3个电子噪声，线性模式雪崩光电二极管（LmAPD）限制在约0.5个电子。我们提供了两种传感器架构的实验室数据，在两个数量级的平均范围内，LmAPD的噪声值与预测值的偏差在0.1个电子以内。
英文摘要：
Read noise in infrared sensor arrays remains a major obstacle for ground- and space-based astronomy. It has long been recognized that the upcoming extremely large telescopes cannot meet their full potential unless read noise is significantly improved, and it is also a prohibitive constraint on the Habitable Worlds Observatory, a space telescope with the goal of detection and characterization of nearby Earth-like exoplanets. The main strategy for lowering read noise is averaging through multiple non-destructive reads. However, this typically results in less noise reduction than the 1/$\sqrt{N}$ scaling predicted by theory. In this work, we show the poor averaging behavior can largely be explained by readout glow, photon emission from the sensor electronics that generates photoelectrons in the pixels during readout. Because glow accumulates with reads rather than averaging, this imposes a fundamental noise floor of σ_{\rm min} ~ 1.5 sigma_RN^(1/2)G^(1/4). This limits averaging in HxRG-like sensors to about 2-3 e- of noise, and linear-mode avalanche photodiodes (LmAPDs) to about 0.5 e-. We present laboratory data using both sensor architectures, with the LmAPD following the predicted noise value to within 0.1 e- over two decades of averaging.
