A new CMOS pixel detector structure - the smart diode array
CMOS pixel sensors vs. CCPDs
The use of pixel sensors in standard CMOS technologies has gained in popularity over the last decade. The use of CCPDs, previously dominant pixel-sensor technology, is now limited only to the special applications. The advantages of the CCPDs were a good pixel to pixel uniformity, a low noise operation, a good fill-factor - all the consequences of the simple design: The signal charge generated by the ionization is transported out of the sensor matrix, using an electronic bucket-chain principle. The signal is then measured by an external amplifier.
The principle of CMOS sensors is different: The signal charge is attracted to a positively biased pool of electrons (called n-well); it is converted to the voltage and amplified within the pixel itself. This leads to pixel-to-pixel nonuniformities if the in-pixel amplifiers do not match perfectly. Moreover, the clearing of the signal charge in CMOS sensors is usually not perfect, since only a part of the charge is removed in the clear process. Despite of these drawbacks, the recent CMOS technology improvements, especially the possibility to implement the on-chip signal processing, allowed high-performance CMOS sensors.
Fill-factor
CMOS sensors in their original form do not have a perfect fill-factor. This is caused by the pixel electronics that are placed near the signal-collecting n-well electrode. The signal charge, generated in the electronic area, gets attracted by the transistor structures and is lost. Such charge losses are not critical when visible light is detected, in such applications (e.g. digital-camera sensors) a large number of low-energy photons should be detected. An imperfect fill-factor leads only to a lower sensitivity that can be compensated by a longer exposition time.
In the case of particle detection in the high-energy physics, a relatively small number of particles that originate from rare decays should be detected. A nearly 100% fill-factor is of importance, otherwise an important event could be missed.
New type of CMOS detectors - electronics in an n-well
We have proposed a way to implement a CMOS pixel sensor with nearly 100% fill-factor. The idea is to place the pixel electronics inside the signal-collecting electrode i.e. inside the n-well. P-type transistors are placed directly inside the n-well, n-type transistors are placed in an p-well (an "island") inside the n-well. The input signal - a small voltage drop in the nwell is AC-coupled to the amplifier.
High depletion voltage
Since the electronic is surrounded by a common n-well and decoupled from the substrate, the substrate can be connected to a very low negative voltage without to damage the transistors. In such a way, the n-well to substrate diode is reversely biased and a relatively large zone around the n-well is depleted.
Depleted region
The large depleted area has a two good consequences:
1) The sensor is not only sensitive to photons that transfer all their energy in a single point than also to high-energy charged particles that transfer the energy uniformely along their tracks. The charge generated in the depleted area experiences a strong electric force and is collected the n-well.
2) A large depleted area leads to a small n-well capacitance and thus an effective conversion of the charge signal to voltage.
Smart diode array
In order to build a pixel array, we place the n-wells near each other in the way that their depleted regions overlap. In this way we construct a sensor with nearly 100% fill factor. Since the n-well/substrate diodes are equipped with electronics, we refer to them as "smart diodes" and name the structure a "smart diode array".
Fast charge collection
Due to strong electric field, the signal charge collection is very fast, which is of importance when particle hit time should be measured. The fast charge collection is also of importance for a high radiation tolerance. There is less chance for the radiation induced bulk damage to influence the signal collection. Both properties - a fast charge collection and a high radiation tolerance are important in high-energy physics applications.
Thinning
Thinning of SDA detetors is possible, since a large part of the substrate is undepleted and not relevant for particle detection and can be removed by thinning. Thin detectors are applied in the particle tracking, where too much detector material could distort the measurement of the particle trajectories.
Technology choices
SDA detectors can be implemented in any CMOS technology that offers the multiple-well option, however the best results are achieved if a high-voltage CMOS technology with a substrate resistivity of more than 20 Ohm cm is used.
