حسگرهای یکپارچه‌ پیکسل فعال (MAPS) در یک فناوری VLSI CMOS Monolithic active pixel sensors (MAPS) in a VLSI CMOS technology

۱٫ Introduction Silicon devices have been used since the 1960s for the detection of radiation (see Ref. [1] for a detailed review). The interest of MOS devices was immediately recognised and arrays were designed. But the world of solid-state imaging was going to be revolutionised by the invention of the ChargeCoupled Devices (CCD) at the Bell Laboratory in 1970 [2]. CCD took over all the competing technologies. In 1981, C. Damerell et al. proposed the use of CCD for the detection of Minimum Ionising Particles for precise vertex reconstruction (see, for example, Ref. [3]). In 1983, Hitachi and Sony introduced the first consumer camera and in the same year Texas Instruments introduced the first mega-pixel device [4]. During about twenty years, from the invention of CCDs till the late 1980s, CMOS sensors were confined to very specialised applications, namely to IR focal-plane detectors, where CMOS sensors were used as readout circuits of bump-bonded low-band gap semiconductor detectors [5]. Different amplifier architectures have been integrated and tested [6–۸]. In 1987, pixel sensors were also proposed for the detection of minimum ionising particles [9]. For this application, the detecting element is integrated in high-resistivity silicon in order to exploit the full depletion of the detector with reasonable voltages. Both the monolithic and the hybrid approach were proposed but in the following years, it was only the latter one, which gave interesting results [10]. Today many high energy-physics experiments have a vertex layer of hybrid pixel detectors [11]. Monolithic active pixel sensors (MAPS) based on high-resistivity silicon as a detecting element were demonstrated by S. Parker [12] in 1989. Good results were obtained only on small structures (about 1 mm2 active area) [13], but no further results have been published on usable size devices. In the late 1980s–beginning of 1990s, new developments on sensors based on a standard CMOS technology took place. CMOS technology uses low-resistivity substrates. In the literature, those sensors are normally referred to as CMOS sensors. The late 1980s developments took place at the University of Edinburgh, UK and were based on the so-called Passive Pixel Sensors (see Fig. 1a). These devices work much as amorphous silicon arrays. Only one selection transistor is integrated in the pixel together with the diode. The charge generated by the radiation is integrated in the diode. The readout is done by closing the selection switch and dumping the charge to a charge preamplifier, common to all the pixels in one column. This solution has the minimum amount of in-pixel electronics and thus has a very high fill factor, defined as the ratio between the detecting area and the total area of the pixel. It has however serious disadvantages in terms of speed and noise. In the early 1990s, the first Active Pixel Sensors (Fig. 1b) were introduced [14,15]. The development was mainly pushed by the requirements of low power and low weight for space applications. In the minimum configuration of an APS, three transistors are integrated in the pixel (Fig. 1b). The transistor MRST is used to reset the pixel by dumping the integrated charge to the positive power supply line. The transistor MSEL is activated to select the readout of the pixel and MIN is the input transistor of a source follower. The current source is common to all the pixels in one column. With respect to other competing imaging technologies, CMOS sensors have several potential advantages in terms of low cost, low power, lower noise at higher speed (see, for example, Ref. [16]), random access of pixels which allows windowing of region of interest, ability to integrate several functions on the same chip. This brings altogether to the concept of ‘camera-on-a-chip’ [۱]. The use of CMOS sensors in particle physics was proposed in 1999 [17]. The main difference with respect to visible light applications is that the sensor has to be 100% efficient. This can be achieved by using a structure which is readily available in most CMOS technologies and which was originally proposed for visible light detection [18]. A schematic view of the cross-section of a CMOS technology is shown in Fig. 2. In most modern CMOS process, n- and p-wells are fabricated on top of a thin p-doped epitaxial layer, with resistivity of the order of 1–۱۰ O cm. The epitaxial layer thickness ranges between a few and upto about 20 mm and it is lightly doped with respect to the underlying p-substrate, whose main function is for mechanical support. A p–n junction exists between the n-well and the p-epilayer and can be used as the detecting element. Because of the difference in doping between the epitaxial layer and the p-well and the p-substrate, a potential difference of a few times kT/q is created. The epitaxial layer acts as a shallow potential well for the electrons, which are the minority carriers. Electrons created by the radiation diffuse in the epitaxial layer till they are close enough to the nwell/p-epi diode, where they experience an electric field. They are then collected by the diode. Following the proposition of the concept, experimental results have shown the excellent properties of CMOS sensors as particle detectors, in terms of signal-over-noise, spatial resolution, detection efficiency [19–۲۱].