| Charge-coupled devices. Design and basic operating principles.
Sergey Ivanovich Neizvestny CHARGE-COUPLED DEVICES. Source: «Special Equipment» magazine History of development Charge-coupled devices (CCD) belong to the class of solid-state semiconductor receivers. The first receivers of this type were photodiodes and already at the dawn of their appearance they allowed to make a giant leap in the field of registration of light flows and images. It is enough to mention as an example the successful registration with the help of a photodiode of the phenomenon of a solar eclipse observed by Berlin scientists in Egypt in 1911. A lot of time has passed since then, photodiodes have been improved, but their main drawback — single-channel, still did not allow them to find wide application. From the end of the 30s among light receivers television tubes began to appear, which by the end of the 70s had won a leading position in this field. A relatively large number of devices of various types were developed: orthicons, isocons, secons, vidicons, plumbicons (in television broadcasting, tubes with a reverse beam), cremnicons and supercremicons, dissectors (specialized tubes with increased quantum efficiency), etc. All of them had a number of serious drawbacks: large dimensions, low quantum efficiency (at the level of 5-10%), small dynamic range, etc. A revolutionary change in the situation occurred with the advent of solid-state semiconductor receivers of the new generation. The quantum efficiency of modern semiconductor radiation receivers reaches 95-98%, i.e. almost every photon falling on the device is registered by the system with 100% probability. In 1970, the first devices with charge coupling were created, in which the technology of solid-state receivers was especially successful. Initially, CCDs were used as more efficient multi-channel substitutes for photodiodes and photodiode matrices. CCD matrices were most successful in recording weak light fluxes in such fields as microbiophysics, chemical physics, nuclear physics, and astrophysics. Since 1975, CCDs have been actively introduced as television light receivers. And in 1989, CCD detectors were already used in almost 97% of all television receivers. For comparison, 10 years earlier, CCDs were represented by only two percent. For a long time, the widespread use of CCD receivers in television technology was hampered by shortcomings in the technology of manufacturing light-sensitive elements — crystalline bases of the required size. The light-receiving region was non-uniform in quantum yield, there was noticeable geometric instability (floating low resolution), and various types of noise were present both on small scales (from pixel to pixel) and on large spatial scales (on scales of 10-100 pixels).
Only with the development and improvement of CCD technology and with a significant leap in the development of related electronic means and, above all, with the increase in the power and speed of the ADC, did it become possible to use CCDs more widely. By putting the production of initially expensive chips on the conveyor belt, many companies have achieved a sharp reduction in their cost. Cheaper CCD-based television cameras, smaller dimensions and weight, low energy consumption, simplicity and reliability in operation have allowed them to be used not only in professional studios, in scientific research, in expensive military systems. Today, CCD-based television cameras can be found in a variety of production areas, in various service sectors, in security systems, and in everyday life. The advent of miniature television cameras using CCD matrices with pixel sizes of several microns has made it possible to use them in microsurgery, microbiology, microvideo optics, which has led to the creation of special microvideo equipment. Today, serial production of CCD matrices is carried out by several companies: Texas Instruments, Thompson, Loral Fairchild, Ford Aerospace, SONY, Panasonic, Samsung, Philips, Hitachi Kodak. I would like to put in the same row with these mastodons and the Russian company — Scientific and Production Enterprise «Silar» (former department for the development of solid-state image receivers of the Central Research Institute «Electron») from St. Petersburg, which is the only manufacturer of CCD matrices in Russia, used for scientific, security and other purposes.
Physical principles of operation of the CCD matrix In simple terms, a charge-coupled device can be considered as a matrix of closely spaced MIS capacitors. Metal-insulator-semiconductor structures (MIS structures) were first produced in the late 1950s. Technologies were found and developed that ensured a low density of defects and impurities in the surface layer of the semiconductor. Thus, within 10 years, the prerequisites for the invention of charge-coupled devices were laid. From a physical point of view, CCDs are interesting because the electrical signal in them is not represented by current or voltage, as in most other solid-state devices, but by charge. With the appropriate sequence of voltage pulses on the electrodes of MOS capacitors, charge packets can be transferred between adjacent elements of the device. Therefore, such devices are called charge-transfer devices or charge-coupled devices. Fig. 1 shows the structure of one element, a linear three-phase CCD in the accumulation mode. The structure consists of a layer of p-type silicon (the substrate), an insulating layer of silicon dioxide, and a set of plate electrodes. One of the electrodes is biased more positively than the other two, and it is under it that charge accumulation occurs. A p-type semiconductor is obtained by adding (doping) acceptor impurities, such as boron atoms, to a silicon crystal. The acceptor impurity creates free positively charged carriers, holes, in the semiconductor crystal. Holes in a p-type semiconductor are the main charge carriers: there are very few free electrons there. If we now apply a small positive potential to one of the electrodes of the three-phase CCD cell, and leave the other two electrodes at zero potential relative to the substrate, a region depleted of the main carriers, holes, will form under the positively biased electrode. They will be pushed deep into the crystal. In the language of energy diagrams, this means that a potential well is formed under the electrode.
The operation of a CCD is based on the phenomenon of the internal photoelectric effect. When a photon is absorbed in silicon, a pair of charge carriers is generated — an electron and a hole. The electrostatic field in the pixel area «pulls apart» this pair, displacing the hole into the silicon. Minority charge carriers, electrons, will accumulate in a potential well under the electrode to which a positive potential is applied. Here they can be stored for quite a long time, since there are no holes in the depletion region and electrons do not recombine. Carriers generated outside the depletion region move slowly — they diffuse and, usually, recombine with the lattice before they come under the action of the field gradient of the depletion region. Carriers generated near the depletion region can diffuse to the sides and can get under the neighboring electrode. In the red and infrared wavelength ranges, CCDs have worse resolution than in the visible range, since red photons penetrate deeper into the silicon crystal and the charge packet is blurred. The charge accumulated under one electrode can be transferred under the adjacent electrode at any time if its potential is increased, while the potential of the first electrode is decreased (see Fig. 2). The transfer in a three-phase CCD can be performed in one of two directions (left or right, according to the figures). All charge packets of the pixel line will be transferred in the same direction simultaneously. A two-dimensional array (matrix) of pixels is obtained using stop channels dividing the electrode structure of the CCD into columns. Stop channels are narrow areas formed by special technological methods in the near-surface region, which prevent the charge from spreading under the adjacent columns.
Types and structure of CCD matrices for security television systems Most types of CCD matrices manufactured on an industrial basis are oriented towards television applications, and this is reflected in their internal structure. As a rule, such matrices consist of two identical areas — an accumulation area and a storage area. The device is shown schematically in Fig. 3.
According to the ratio of the sizes of the storage and accumulation areas, matrices are divided into 2 types:
There are also matrices that do not have a storage section, and then the line transfer is carried out directly along the accumulation section. Obviously, an optical shutter is required for the operation of such matrices. The storage area is protected from light by an opaque coating. During the reverse stroke of the frame scan beam of the television monitor, the image formed in the accumulation area is quickly transferred to the storage area and then, while the next frame is exposed, is read line by line at the line scan frequency into the output shift register. Parallel line transfer to the readout register occurs during the reverse stroke of the line scan. From the shift register, charge packets are output one after another, sequentially through the output amplifier located on the same silicon crystal. In this unit, the charge is converted into voltage for further signal processing by external electronic equipment. Such devices are called frame transfer CCDs. They are widely used in consumer video equipment, especially amateur, due to their low prices. Frame transfer devices can be used for filming in well-lit conditions. The use of such CCDs allows the use of video cameras without expensive mechanical shutters.CCDs designed for use in low-light conditions are usually made without a storage area and often have two shift registers on opposite sides of the device, such as the Tektronix TK512 CCD. The image can be shifted into either register, which may differ in the design of the output unit. Typically, one is optimized for slow readout speeds, the other for fast ones. During the signal output, such a matrix must be shielded from light. Mechanical shutters are most often used for this. CCDs with interlaced (interline) scanning of good quality of modern design are produced, for example, by Philips. Such matrices are supplied to TV cameras of the LTC 03, LTC 04 series. Thus, the LTC 0350 TV camera is supplied with an automatic electronic shutter of 1/50 — 1/100000 sec, working with a matrix format of 1/3 inch and a size of 752×582 pixels. The simplest CCDs consist of an electrode structure deposited directly on an insulator layer formed on the surface of a uniformly doped p-silicon wafer. The charge is accumulated and transferred directly in the near-surface layer of the semiconductor. Such devices are called surface-channel CCDs. The surface layer is characterized by a large number of defects, which negatively affects the efficiency of charge transfer. Charges are captured on surface-layer defects and are slowly released. This leads to image smearing. Surface-layer defects can also spontaneously emit charges, leading to an increase in the dark signal (current). Surface states are a factor limiting the performance of CCDs. It is impossible to completely get rid of surface states, but it is possible to significantly improve the characteristics of the device by storing and transferring charge packets at some distance from the crystal surface, i.e. by forming a volume transfer channel. This result can be achieved by creating a thin n-layer under the oxide on the p-type substrate. Such devices are called CCDs with a surround channel. Similar considerations are valid for the design of the output amplifier, since surface defects can greatly increase the amplifier noise. The output amplifier with a surround channel has significantly better characteristics. The working part of charge-coupled devices is a few microns thick. They are usually made from very thin semiconductor films grown on a relatively thick base — a substrate. Several methods have been developed for growing films on substrates, which are collectively called epitaxial.The term “epitaxy” is made up of two Greek words: “epi” (on, over) and “taxis” (arrangement in order). A very apt term, reminding us that we are talking about growing a monocrystalline (ordered) layer of material over a substrate. The grown epitaxial films are much less contaminated with foreign impurities. In the process of epitaxy, strictly controlled doping of the growing layer is possible. CCD-matrix electrodes For some time after their invention, CCD electrodes were most often made in a single metal layer. A layer of aluminum about 1 µm thick was deposited on the device by evaporation. Then the electrodes were formed by photolithography. The most critical step in the manufacturing process of a single-level structure of this type is the etching of the interelectrode gaps. To ensure good transfer of charge packets, the potential wells of adjacent electrodes must overlap. The depth of the potential well depends on the degree of doping of the silicon and the magnitude of the potential applied to the electrode. Typical values are units of microns. It follows that the interelectrode gaps should not be larger than units of microns. The total length of these narrow gaps in large devices is very large. For a lightly doped substrate material (acceptor atom concentration of about 1015 1/cm3, oxide thickness of 0.1 μm, and moderate clock pulse swing of about 10 V), the depletion layer penetrates into the silicon to a depth of about 1 μm. Recall that each cubic centimeter of solid material contains about 1022 atoms. A concentration of 1015 impurity atoms per 1 cm3 corresponds to 1 impurity atom per 10 million Si atoms. It is clear that any accidental short circuit of adjacent electrodes that occurs during one of the operations of the technological cycle will completely disable the device. Subsequent development of CCD technology was aimed at creating structures free from the shortcomings of the first technologies and operating with simpler control voltages. P3C for use as image receivers are manufactured with polysilicon electrodes (silicon deposited from the gas phase). After doping with boron or phosphorus to achieve a sufficiently low resistance, it can be used as a conductive layer. Thermal oxidation of polysilicon allows obtaining a high-quality interfacial dielectric, and its transparency facilitates the use of CCDs as image receivers. The use of this technology made it possible to record light not from the electrode side (this type of recording has many disadvantages, since the useful light signal is partially vignetted by the electrodes), but from the opposite side. Such matrices are called back illuminated.
Thanks to the use of the latest high-precision technologies in the manufacture of CCDs, these radiation receivers have now become dominant in television systems and brought them to a fundamentally new level, significantly expanding the functionality of CCDs and making them available at cost for widespread use. We will talk about the most important properties, basic characteristics of CCDs and television systems based on them in the next issues of the magazine. |
