Characterization of Random Telegraph Noise in an H2RG X-ray Hybrid CMOS Detector

William A. Bevidas Jr, Joseph M. Colosimo, Abraham D. Falcone, Timothy Emeigh, Lukas R. Stone, Kadri M. Nizam, Brynn Bortree, Jacob C. Buffington, David N. Burrows, Zachary E. Catlin, Killian M. Gremling, Md. Arman Hossen, Collin Reichard, Ana C. Scigliani, Anthony J. Tavana, Mitchell Wages · 2025

Read Characterization of Random Telegraph Noise in an H2RG X-ray Hybrid CMOS Detector by William A. Bevidas Jr, Joseph M. Colosimo, Abraham D. Falcone, Timothy Emeigh, Lukas R. Stone, Kadri M. Nizam, Brynn Bortree, Jacob C. Buffington, David N. Burrows, Zachary E. Catlin, Killian M. Gremling, Md. Arman Hossen, Collin Reichard, Ana C. Scigliani, Anthony J. Tavana, Mitchell Wages (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-16221.txt.

Introduction

William Bevidas and co-authors measure a flickering noise in an X-ray detector used for astronomy, in a 2025 paper. The detector is a hybrid design prized for fast readout and low power. The noise, which jumps between levels at random, can dull its sharpness. It had already been found in a newer detector of the same family.

The team mined stored data taken at two temperatures. They counted how much of the detector the noise affects and how it behaves. They then compared the pattern with the newer detector's. Knowing the noise's share and habits lets engineers correct for it or design around it. Clean detectors matter for faint X-ray sources in deep space.

Catalog record

CREATORWilliam A. Bevidas Jr, Joseph M. Colosimo, Abraham D. Falcone, Timothy Emeigh, Lukas R. Stone, Kadri M. Nizam, Brynn Bortree, Jacob C. Buffington, David N. Burrows, Zachary E. Catlin, Killian M. Gremling, Md. Arman Hossen, Collin Reichard, Ana C. Scigliani, Anthony J. Tavana, Mitchell Wages
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-16221

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Characterization of Random Telegraph Noise in an H2RG X-ray Hybrid CMOS Detector William A. Bevidas Jr, Joseph M. Colosimo, Abraham D. Falcone, Timothy Emeigh, Lukas R. Stone, Kadri M. Nizam, Brynn Bortree, Jacob C. Buffington, David N. Burrows, Zachary E. Catlin, Killian M. Gremling, Md. Arman Hossen, Collin Reichard, Ana C. Scigliani, Anthony J. Tavana, Mitchell Wages Abstract Hybrid CMOS detectors (HCDs) have several excellent features as high-performance X-ray detectors, including rapid readout, deep-depletion silicon for high quantum efficiency, radiation hardness, and low power. Random telegraph noise (RTN) is a type of noise that can reduce the performance of HCDs and other CMOS sensors. After finding and quantifying RTN in the recently developed engineering grade Speedster-EXD550 HCDs, this form of noise has also been found in other X-ray HCDs. This paper aims to investigate its presence and characteristics in the relatively mature H2RG X-ray HCD and to compare it with that of the Speedster-EXD550. We use archival data taken with an H2RG X-ray HCD at two different temperatures to determine the percentage of pixels that are being impacted by RTN. We identify RTN in 0.42% of pixels when the detector is operated at 140 K, while we are only able to identify RTN in 0.060% of pixels when the detector is operated at 160 K. We characterize RTN in two Speedster-EXD550 detectors, identifying 5.1% of pixels with RTN in one detector and 7.1% in another, which is significantly more than the H2RG. These results verify the difference between two different HCDs and provide techniques that can be applied to future hybrid CMOS detectors. keywords silicon imaging sensors, X-ray detectors, HCD, random telegraph noise, random telegraph signal 1 Introduction Modern charge-coupled devices (CCDs) are designed to achieve minimal noise and provide an accurate detection of photons, including individual photon detection and energy estimation in the X-ray energy band. [ 1 ] X-ray hybrid complementary metal-oxide semiconductor (CMOS) detectors (HCDs) are a newer type of photon detector composed of two separate silicon substrates that have been bump-bonded (“hybridized”) together. The top, X-ray absorbing, silicon layer is designed to optimize quantum efficiency (QE) through the use of high-resistivity silicon, allowing for the bulk to be fully depleted. The bottom substrate, ( i.e. , the readout integrated circuit (ROIC)), is optimized for fast and efficient readout schemes. [ 2 ] The absorbing layer allows each X-ray photon to interact with the silicon, causing a charge cloud to pass through the substrate. The charge, collected by the ROIC, is then amplified and the signal from each pixel is digitized using an analog-to-digital converter (ADC) to provide a digital number (DN) value that can be stored and analyzed. While these active pixel sensors offer faster readout rates, lower power consumption, and enhanced radiation hardness, reducing noise is still the most active area of research. [ 3 , 4 ] 1.1 The H2RG X-ray HCD The HAWAII-xRG (HxRG) HCDs, developed by Teledyne Imaging Sensors (TIS), are used for detecting photons over a wide range of energies. [ 5 , 6 ] In the HxRG line of detectors, the x signifies the size of the ROIC array in multiples of 1024 \times 1024 pixels. Hybrid Visible Silicon Imager (HyViSI) variations of HxRGs provide high sensitivity over the soft X-ray bandpass. The X-ray H2RG used in this investigation has a substrate with a 1024 \times 1024 pixel array of 36- \mu m pixel pitch bonded to a ROIC with a 2048 \times 2048 pixel array of 18- \mu m pixel pitch. One out of every four ROIC pixels is connected to the corresponding substrate pixel (see Figure 1 ). This specific X-ray H2RG design is meant to mitigate an issue of the H2RG’s predecessors, the X-ray H1RGs, which suffered from interpixel capacitance crosstalk (IPC), where the charge being integrated directly onto the detector node caused a voltage change that allowed for signal to couple with neighboring pixels. [ 7 ] The increase in pixel size on the substrate reduces the IPC due to the increased spacing between pixels. [ 8 ] HxRGs use source followers (SFs) as amplifiers for each pixel because of their simple design and low noise characteristics. [ 9 ] The SF uses a basic metal–oxide–semiconductor field-effect transistor (MOSFET) with a constant current supply; the voltage differential across the transistor gives a measurement of the charge integrated onto the sense node. The H2RG is read out using a process called up-the-ramp integration. This process begins by resetting the detector to acquire a data frame with effectively zero integration time. Charge is then integrated onto the detector over a short period of time and read out. After this, more charge is integrated onto the detector without resetting the charge, allowing it to accumulate throughout the ramp, which typically consists of around 100 and 300 samples. The charge is reset at the end of the ramp and this process repeats. In some detectors, correlated double sampling (CDS) is used to correct for reset noise through in-pixel sampling and correction for the variable reset level. The H2RG does not have in-pixel CDS circuitry; however, reset noise is not a contributing factor because up-the-ramp integration does not involve resetting the detector. To find the charge integrated between exposures and correct for fixed-pattern noise, a pseudo-CDS process is used in which subsequent frames from the integration process are digitally subtracted during analysis. Figure 1 : A 20 \times 20 segment of a raw image taken with the H2RG-122 at 140 K. The top left corner shows three pixels disconnected from the substrate and one connected to the substrate. This pattern repeats across the entirety of the detector. 1.2 Speedster-EXD550 The Speedster-EXD550 is another type of HCD that has a substrate with a 550 \times 550 pixel array of 40- \mu m pixel pitch bonded to a ROIC with a 550 \times 550 pixel array of the same pitch. [ 10 ] Capacitive transimpedance amplifiers (CTIAs) are used in the Speedster-EXD550 detectors. In this pixel design, the voltage at the sense node is kept constant (unlike in a source-follower design), eliminating IPC in these detectors. This detector can operate in an event-driven readout mode that uses a comparator in each pixel to identify pixels with signals above the background noise for readout and further processing. The SpeedsterEXD-550 uses in-pixel CDS circuitry to correct for reset noise. 1.3 Noise Amplifier noise is the uncertainty in the charge measurement from the amplifier. The MOSFET in the amplifier contains three fundamental transistor noises: Johnson noise, flicker (1/f) noise, and random telegraph noise (RTN). Johnson noise arises from thermal noise within the MOSFET and is a type of frequency independent noise that can impact pixel output variability. [ 11 ] Flicker noise has a power spectral density which falls off with a frequency of 1/f. This type of noise is thought to result from defects in the silicon-oxide interface resulting in charge carriers in the MOSFET becoming mobile and/or dense. [ 12 ] The read noise floor for CCDs is known to be typically limited by 1/f noise, while the read noise floor is limited by RTN in CMOS sensors. [ 13 ] Both noises are physically connected through the capture and emission of carriers in

Read the complete text