Speaker
Description
We present the design and characterization of a CMOS pixel direct charge sensor, Topmetal-IIa, fabricated in a standard 0.35$\mu$m CMOS process. The sensor features a $45\times216$ pixel array with a 40$\mu$m pixel pitch which collects and measures external charge directly through exposed metal electrodes in the topmost metal layer. Each pixel contains a low-noise charge-sensitive preamplifier to establish the analog signal, which is accessible through a time-shared multiplexer. Initial tests show that the sensor achieved a $<10{e^-}$ analog noise per pixel. These characteristics enable its use as the charge readout device in future Time Projection Chambers without gas-avalnche gain, which has unique advantages in low background and low rate-density experiments.
Summary
We have successfully implemented a CMOS Integrated Circuit, Topmetal-IIa, that is uniquely suitable for charge collection and measurement in a Time Projection Charmber without gas-avalanche gain. It is a direct charge sensor with 40$\mu$m pitch between pixels fabricated in a standard 350nm CMOS technology without any post-processing. A metal node (Topmetal) is placed on the top of each pixel in a $45\times216$ pixel array for direct charge collection. Each pixel contains a low-noise charge-sensitive preamplifier (CSA) with a 1.3fF feedback capacitor to establish the analogue signal. The Topmetal electrode is directly connected to the input of the CSA. A ring electrode (Gring), which is in the same topmost metal layer as the Topmetal, surrounds the Topmetal while being isolated from it. In order to explore the charge collection behavior of different architectures between Gring and Topmetal as well as to realize the intended applications accordingly, we divided the Topmetal-IIa matrix in 3 sectors, each one implements a different Topmetal-Gring flavour. 3 different Topmetal-Gring architectures mean that only Topmetal is exposed, only Gring is exposed, and both Topmetal and Gring are exposed, respectively. A 2-bit in-pixel DAC is applied to the gate node of the feedback transistor of the CSA, in order to calibrate the total mismathch of the devices, hence improving the CSA's peaking and decay time uniformity. What's more, the CSA's sensitive voltage biases that have significant contributions to the output noise are individually provided by the peripheral Low-Pass Filter (LPF) with tunable cut-off frequency, aiming to improve the noise performance. The analog signal from each pixel is read out through the traditional ``rolling shutter'' style time-shared multiplexer controlled by the array scan unit, and then is fed to two array-shared analog buffers, of which one is capable of 50 Ohm driving strength aiming to elimilate the external buffer hence reducing the entire readout system noise. For its intended application, event-rate density is expected to be low and charge (both free electron and ion) drifting speed is expected to be slow; therefore, we tuned the CSA to have long signal retention and eliminated the in-chip pulse shaper while focusing on improving the noise performance. Some simulation and preliminary test results confirm the low-noise design and correct readout implementation. The Equivalent Noise Charge of the sensor is $<10{e^-}$ rms at a 15$\mu$s peaking time and 150ms decay time with a detector capacitance of 11.5fF.
To improve beyond Topmetl-IIa, besides optimizing the LPF design, we can further increase the working margin of CSA's sensitive voltage biases. We will investigate these options in future Topmetal sensor development. We will present the overall design and some initial test results of the Topmetal-IIa chip in the conference.
