Speaker
Description
We present a Skipper-in-CMOS image sensor integrating Skipper-CCD non-destructive readout with CMOS pinned photodiode gain and in-pixel processing. Fabricated in a 180 nm process, a 200×200 array with 15×15 μm² pixels achieves 0.075e⁻ noise via multisampling. The SPROCKET2 ROIC, in 65 nm CMOS, supports 66.7 ksps high-speed readout with low DNL/INL and 10 μV resolution. A 20,000-pixel SPROCKET2 array with optical 10.24 Gbps links enables 4 kfps operation across 320,000 sensor pixels. We also demonstrate silicon photonic integration using microring modulators for 10.24 Gb/s optical output, enabling future detectors with tightly co-designed sensing, readout, and communication elements.
Summary (500 words)
We present recent progress in image sensor technology that integrates Skipper Charge Coupled Devices (Skipper-CCDs) with CMOS imaging architectures to deliver exceptionally low-noise and high-speed readout. This innovation, termed the Skipper-in-CMOS image sensor, leverages the non-destructive readout capability of Skipper-CCDs in combination with the high conversion gain of pinned photodiodes and in-pixel signal processing. Fabricated using Tower Semiconductor’s 180 nm CMOS Image Sensor process, the sensor features a 15 × 15 μm² pixel within a 200 × 200 pixel array.
Initial measurements from pixel test structures with off-chip readout demonstrate sub-electron readout noise levels of 0.15e⁻. With on-chip readout integration, noise performance improves further, reaching deep sub-electron noise levels of 0.075e⁻. These results confirm single photon counting capability, validating the potential of this architecture for ultra-low noise applications in both scientific and quantum imaging domains.
To support high-speed readout, we developed the second-generation Skipper CCD-in-CMOS Parallel Read-Out Circuit (SPROCKET2) in a 65 nm CMOS process. Each SPROCKET2 pixel, measuring 60 × 60 μm², interfaces with a cluster of 16 active image sensor pixels. The readout circuit employs correlated double sampling and analog-domain accumulation of ten consecutive samples, enabling low-noise digitization at a rate of 66.7 ksps. Electrical characterization shows excellent linearity, with Differential Non-Linearity (DNL) and Integral Non-Linearity (INL) of approximately 0.44 LSB and 0.58 LSB, respectively.
A large-scale array of 20,000 SPROCKET2 ADC pixels is currently being tested. Designed to operate with a 1:16 multiplexing ratio, this readout array is capable of interfacing with up to 320,000 image sensor pixels. The architecture employs a high-speed 10.24 Gbps optical link to stream digitized data, supporting a sustained frame rate of 4,000 frames per second (4 kfps) across large imaging areas. In simulations, SPROCKET2 achieves an input-referred voltage resolution of 10 μV in its highest gain mode and operates with a simulated power consumption of just 50 μW per pixel. The constant current design of the analog front-end minimizes power rail crosstalk and supports scalable integration.
In parallel, we have developed a tightly integrated solution for future detector systems by combining Skipper-in-CMOS sensors with silicon photonic data transmission. Pixel detector ASICs are co-designed with integrated silicon photonic micro-ring modulators (MRMs), enabling direct high-speed optical modulation from each pixel. This approach facilitates 10.24 Gbps optical data rates per channel, significantly enhancing bandwidth and scalability for high-throughput imaging applications.
Importantly, the integrated photonic-electronic architecture is being validated both at room temperature and at cryogenic temperatures (~100 K), making it suitable for deployment in low-background environments such as dark matter detection and/or quantum sensing. The resulting system architecture exemplifies a co-designed platform that tightly couples sensing, computing, and communication elements within a compact and scalable cryo-compatible imaging system.
These results mark a key milestone toward enabling large-scale, low-noise, and high-speed imaging arrays for advanced scientific instrumentation, quantum detectors, and beyond-CMOS computational imaging platforms.
