Dimensions of integrated circuits. What is an integrated circuit (IC)

Konyaev Ivan Sergeevich, 3rd year student of the Armavir Mechanical and Technological Institute (branch) of the Federal State Budgetary Educational Institution of Higher Professional Education Kuban State Technical University, Armavir [email protected]

Monogarov Sergey Ivanovich, candidate of technical sciences, associate professor of the department of in-plant electrical equipment and automation of the Armavir Mechanical and Technological Institute (branch) of the Federal State Budgetary Educational Institution of Higher Professional Education Kuban State Technical University, Armavir [email protected]

Principles of construction of large integrated circuits

Annotation. This article is devoted to the principles of construction of large-scale integrated circuits (LSI). Key words: LSI, large integrated circuit, basic matrix crystals, programmable logic devices.

Currently, microelectronic equipment uses both specialized and universal microcircuits of varying degrees of integration. At the same time, there is a certain trend in the widespread use of integrated circuits with a high degree of integration - large integrated circuits (LSI), which will be discussed in this article. Universal microcircuits are produced in large quantities and are used in a wide range electronic devices, while specialized microcircuits are produced in limited editions and have a strictly defined scope. Specialized LSIs made on basic matrix crystals (BMCs) and programmable logic devices (PLUs) have a particularly wide application. Such widespread use is due to the fact that automated design of such LSIs takes a relatively short period of time: about several weeks for LSIs based on BMK, several days for LSIs based on PLUs. Let us consider the principles of construction and parameters of basic matrix crystals. The BMK includes a pre-formed matrix of base cells (located in the central part), as well as a group of buffer cells, which are located along the periphery of the crystal (Fig. 1). In turn, the cells include groups of unconnected elements (transistors, capacitors, resistors ) and sections of semiconductor buses intended for the implementation of intersecting electrical connections.Various functional elements (triggers, counters, registers, etc.), buffer elements, as well as connections between them are formed from cell elements using electrical connections in the form of metal (conductor) and semiconductor buses.

A) b) c) Figure 1 – Typical structures of BMK: a) with a continuous array of homogeneous cells; b) with an array of homogeneous cells or macrocells separated by vertical and horizontal channels for conductors; c) with an array of heterogeneous cells separated by horizontal channels; 1 – matrix of basic cells; 2 – matrix of buffer cells; 3,5,8 – matrix cells, 4,7,10 – buffer cells, 6,9 – macrocells; 11,12 – horizontal channels; 13 – vertical channels

IN this type LSIs, as a rule, the main functional elements consume a small amount of energy, sufficient to ensure the required speed. In turn, the buffer elements that carry out external communications of the matrix LSI consume higher power, which is due to the need to match logic voltage levels of a certain value, load capacity and noise immunity. The cells include a variety of active and passive elements. At the same time, the parameters of passive elements are subject to requirements of fairly high accuracy and stability. The composition of BMKs intended for the manufacture of analog-digital LSIs usually includes two matrices of cells to form analog and digital devices, respectively. Basic matrix crystals for digital and analog LSIs are formed on the basis of bipolar transistors and field-effect transistors with an insulated gate. In analog LSIs, bipolar transistors with a high slope current-voltage characteristic are widely used. In turn, matrices can consist of homogeneous or inhomogeneous cells. In BMK intended for the implementation of digital LSIs with non- high degree integration (about 1000 logic elements) homogeneous cells are used, while for digital LSIs with a high degree of integration (about 10,000 logical elements) and digital-to-analog LSIs – matrices with heterogeneous cells. Two methods of organizing the cells of the BMK matrix are used: 1. Based on the elements of the cell, one basic logical element can be formed that performs an elementary function (NOT, INE, OR with branches for inputs and outputs). To implement more complex functions multiple cells are used. The number, types and parameters of elements are determined by the electrical circuit of the basic logical element. 2. Based on the elements of the cell, any functional element of the library can be formed. The types of elements and their number are determined by the electrical circuit of the most complex functional element. With the first method of constructing cells, it is possible to obtain a fairly high coefficient of their use in the matrix, a coefficient of utilization of the BMK area and, accordingly, an increased degree of integration of the LSI. With the second method of constructing BMK cells, the system is simplified computer-aided design LSI, since the seats of cells of the same shape and size are predetermined. However, if the designed LSI uses quite a lot of simple functional elements of the library with a low utilization rate of cell elements, the utilization rate of the chip area decreases, and hence the degree of integration of the LSI. In matrix LSIs electrical connections are performed using metal (conductor) and semiconductor (mono- and polycrystalline) buses. The power and grounding busbars are usually made of aluminum, which is characterized by low resistivity. Doped semiconductor buses with high resistivity are mainly used to implement short, low-current signal circuits. To create electrical connections between elements, single and multi-level metallization is used. Upon completion of the design, the set of parameters and characteristics of the BMK should be sufficiently complete for the consumer. Typical parameters and characteristics of BMK include: 1. manufacturing technology; 2. number of cells in the crystal; 3. structure (set of elements) of the cell; 4. name, standard electrical parameters, circuits and fragments of standard functional elements formed on the basis of cell elements; 5. parameters of input/output elements; 6. number of peripheral contact pads; 7. requirements for the power source; 8. instructions for the location and use of contact pads for power and grounding circuits, etc.; BMK can serve as the basis for digital, analog, digital-to-analog and analog-to-digital large integrated circuits. At the same time, the set of BMK elements intended for use in analog LSIs allows the formation of amplifiers, comparators, analog digital switches and other devices. Not so long ago, the main application of BMK was computer technology and control systems technological processes. Some BMKs, for example T34VG1 (KA1515ХМ1216), were used in Soviet clones of the ZX Spectrum computer as a controller external devices. An analogue of the BMK is the ULA microcircuit in Sinclair computers. Currently, BMKs in most applications have been replaced by FPGAs (programmable logic integrated circuit – author’s note), which do not require factory production process for programming and allowing reprogramming. Next, we will consider programmable logic matrices. Programmable logic devices have a matrix structure and a bus organization of elements (each element is connected by vertical and horizontal buses). The PLU uses programmable matrices AND, OR and their combinations: non-programmable AND - programmable OR; programmable AND - non-programmable OR; programmable AND - programmable OR. There are two types of programmable logic devices:

programmable in the conditions of production of specialized LSIs based on semi-finished crystals using one custom photomask according to technology, similar technology production of matrix LSIs;

programmed by the consumer-manufacturer of the equipment by “loading” (entering information) internal registers or physical impact to individual elements of the matrices (burnout of jumpers, breakdown of diodes, changing operating modes of semiconductor devices). Consumer-programmable logic devices are universal microelectronic devices that are “customized” to a given function using automatic programmers. In practice, such types of PLUs as programmable logic matrices (PLMs) and programmable read-only memories (PROMs). The use of PLMs makes it possible to reduce the number of logical elements and connections in logical devices, which is especially important for regular structures implemented on LSI chips. Single-time programmable PLMs and repeatedly programmable-reprogrammable ones have been developed and are used PLM (RPLM). Methods are being developed for the design and production of matrix LSIs with reconstructed connections (MaBISRS) and with programmable architecture (MaBISPA) - subsystems on wafers. Programming using masks (photomasks) of metallization or contact windows in the oxide is widely used in PLMs based on bipolar transistors and diodes. Figure 2 shows a diagram of the connections of elements in a diode PLM. Input signals of positive polarity are supplied to inputs a – e, the products M0 – M2 are removed from load resistors R. The advantages of diode matrices are simplicity and small area occupied on the crystal, but the disadvantage is the significant currents consumed at the inputs of the matrix. The use of multi-emitter transistors instead of diodes allows significantly reduce input currents (BN times, BN is the normal current transfer coefficient of the transistor) and increase the speed of the PLM. Figure 3 shows a diagram of a PLM fragment on bipolar multi-emitter transistors. Matrices based on MOS transistors provide the most high density layout of elements, have minimal power consumption, but are inferior in performance to matrices on bipolar transistors The advantages of PLMs with mask programming are their small area and high reliability, which has led to their widespread use as part of specialized and microprocessor LSIs. Such PLMs are programmed once by the manufacturer during the production of the microcircuit, which narrows the scope of their application. Electrically programmable PLMs have greater flexibility, especially when used in peripheral devices, the “tuning” of which to implement specified functions is carried out by the user.

Figure 2 – Fragment of a diode PLM

Figure 3 – Fragment of PLM on BT

Figure 4 shows the most common electrically programmable matrix elements. Programming is carried out by melting jumpers (usually nichrome or polysilicon) or breaking down diodes (pn junctions or Schottky barriers).

Figure 4 – PLM elements with electrical programming

The jumpers have a resistance of about 10 Ohms and melt (open) when a current pulse is passed through them, the amplitude of which is significantly greater than the amplitude of the read current. To destroy nichrome or polysilicon jumpers, a current of 20...50 mA is sufficient; the melting time is 10...200 ms. The diodes break down (short-circuit) when a pulse is applied reverse voltage from a source with a small internal resistance, providing sufficient current (200...300 mA). This causes avalanche and thermal breakdown of pn junctions (Schottky barrier) and migration of metal particles into the semiconductor with the formation of a reliable low-resistance contact (dashed lines in Fig. 4). The circuit formation time is 0.02…0.05 ms. For electrical programming and control of the PLM, special installations controlled by a computer are used. The initial information for programming and control is: truth table; sign of burnout (breakdown) log. ones or zeros (depending on the initial information of the unprogrammed PLM); parameters of programming pulses. The control program enumerates the addresses at the inputs from 00...0 to 11...1. Supply voltages are supplied to the PLM, and if there are signs of programming in the initial information, a burnout (breakdown) pulse is applied. After programming, control is performed and the test result, indicating a match (non-match) with the truth table, is printed. PLMs are used in modern peripheral and main computer devices of expansion boards in the Plug and Play system, which have a special chip - FPGA. It allows the board to report its identifier and a list of required and supported resources. To create VLSI (super large-scale integrated circuits) and subsystems on wafers, regular structures (Fig. 5) with a matrix of cells of a sufficiently large degree of integration are used. Programming of connection elements is performed by creating or breaking them.

Figure 5 – Fragment of LSI with reconstructed connections

Matrix LSIs with reconstructed connections are usually created on the basis of CMOS transistors, characterized by minimal power consumption. All types of jumpers are applicable for such transistors. The use of matrix LSIs with reconstructable connections for the construction of multiprocessor subsystems is promising. Contacts between connecting conductors of different levels are programmed with a laser beam (the dielectric melts), some connections are cut. Laser reconstruction when controlled by a computer lasts about 1 hour. Such microsystems can contain up to 100 million transistors. The layout density for VLSI with a minimum element size of 0.5 ... 2 microns reaches 20 thousand transistors per square millimeter. Currently, there are memory elements that retain information when the supply voltage is turned off, which makes it possible to create PLMs with erasing and rewriting of implemented functions - reprogrammable logic matrices (RPLMs). MOS transistors with a floating gate and avalanche gate are widely used in RPLMs. injection (Fig. 6). The structure of such a transistor is similar to a conventional MOS transistor with a polysilicon gate, which is not galvanically coupled to the rest of the circuit. In the initial state, the transistor does not conduct current (see Fig. 6, a). To transition to a conducting state (write), a sufficiently large voltage (about 50 V) is applied between the source and drain of the transistor for about 5 ms. This causes an avalanche breakdown of the source (drain) pn junction and injection of electrons into the polysilicon gate. A charge approximately equal to 107 C/cm2 captured by the gate (see Fig. 6b) induces a channel connecting the source and drain, and can persist for a long time (10...100 years) after the voltage is removed, since the gate is surrounded by an oxide layer , having very low conductivity. Information is erased by irradiation with ultraviolet rays with energy sufficient to knock electrons out of the gate and transfer them to the substrate (Fig. 6). Erasing can also be carried out using ionizing radiation, such as X-rays. Reading information from the matrix is ​​performed by applying a supply voltage of 5...15 V and monitoring the current flowing through the transistor. To organize the selection of certain cells into the matrix (see Fig. 6, c) in series with floating gate transistors include conventional MOSFETs.

Fig.6. PLM on MOS transistors with a floating gate: a) switched off (erased) storage transistor; b) switched on storage transistor; c) fragment of the matrix (sampling transistor Tv, storage transistor Tz); 1 – source; 2 – floating gate made of polycrystalline silicon; 3 – drain; 4 – injected charge; 5 – depletion region

Along with LSIs with reconstructable connections, a direction is developing related to the creation of LSIs and VLSIs with programmable architecture and implemented in the form of subsystems on wafers. Reorganization of the subsystem architecture is carried out using built-in memory switching elements. Moreover, memory elements can be implemented both on standard MOS or CMOS transistors, and on transistors with avalanche injection. Figure 7 shows a block diagram of a matrix LSI with a programmable architecture. The control bus (CB) is used to write codes for setting up (programming) the subsystem architecture for a specific task into distributed memory blocks (C). The decision blocks of the matrix (M) are connected to each other by distributed switches (K) through a switching bus (SC).

Figure 7 – Block diagram of a matrix LSI with programmable architecture

The use of VLSI with programmable architecture makes it possible to obtain a very high layout density and automate the assembly process.

Links to sources1. Educational site www.studfiles.ruURL: http://www.studfiles.ru/dir/cat39/subj1381/file15398/view155035/page2.html2. Free encyclopedia Wikipedia URL: http://ru.wikipedia.org /wiki/%D0%91%D0%9C%D0%9A3. Free encyclopedia WikipediaURL: http://ru.wikipedia.org/wiki/%D0%9F%D0%9B%D0%98%D0%A1

Konyaev Ivan Sergeyevich,3rd year student of Armavir Institute of Mechanics and Technology (branch) Kuban State University of Technology, ArmavirMonogarov Sergey Ivanovich,Candidate of Technical Sciences, Associate Professor of inplant electrical equipment and automation, Armavir Institute of Mechanics and Technology (branch) Kuban State University of Technology, ArmavirPrinciples of building largescale integrated circuitsAbstract:This article focuses on research of the principles of construction of largescale integrated circuits (LSIs).Keywords:BIS, a large integrated circuit, the base matrix crystals, programmable logic devices.

In early electrical computers, the circuit components that performed the operations were vacuum tubes. These tubes, which resembled light bulbs, consumed a lot of electricity and generated a lot of heat. Everything changed in 1947 with the invention of the transistor. In that small device used a semiconductor material named for its ability to both conduct and block electricity, depending on whether there is an electric current in the semiconductor itself. This new technology made it possible to build all kinds of electrical switches on silicon chips. Transistor circuits took up less space and consumed less power. For more powerful computers integrated circuits, or ICs, were created.

Nowadays, transistors have become microscopically small, and the entire IC circuit fits on a 1-inch square piece of semiconductor. Small blocks mounted in rows on a computer circuit board are integrated circuits enclosed in plastic cases. Each microcircuit contains a set of simple circuit elements, or devices. Most of them are occupied by transistors. An IC may also include diodes, which allow electrical current to flow in only one direction, and resistors, which block the current.
Fixed parts. In the interior of a computer, rows of integrated circuits in protective housings, as shown below, are mounted on the computer's circuit board (green). Each pale green line represents a path along which electric current flows; together they form “highways” through which electric current is carried from circuit to circuit.

Tiny messengers. Along the edge of the chip, highly magnetized wires resembling human hairs send electrical signals from electrical circuit(name above). These gold or aluminum wires are virtually resistant to corrosion and are good conductors of electricity.

Anatomy of a transistor
Transistors, the basic microscopic elements of an electronic circuit, are switches that turn electrical current on and off. Small metal tracks ( grey colour) conduct current (red and green colors) from these devices. Organized into a combination called logic gates, transistors respond to electrical impulses in a variety of preset ways, allowing the computer to perform a wide range of tasks.

Logic diagram. If the incoming electrical current (red arrows) activates the base of each transistor, the supply current (green arrows) will rush to the output wiring.

Classification of integrated circuits

According to the design and technological design, they are distinguished semiconductor, film and hybrid ICs.

Semiconductor devices include SMCs (semiconductor integrated circuits), all elements and inter-element connections of which are made in the volume or on the surface of the semiconductor. Depending on the insulation methods individual elements a distinction is made between PMSs with insulation by p-n junctions and microcircuits with dielectric (oxide) insulation. SLM can also be produced on a substrate made of dielectric material based on both bipolar and field-effect transistors. Typically, in these circuits, transistors are made in the form of three-layer structures with two p-n junctions (n-p-n-type), and diodes are made in the form of two-layer structures with one p-n junction. Sometimes, instead of diodes, transistors are used in diode connection. PMS resistors, represented by sections of a doped semiconductor with two terminals, have a resistance of several kilo-ohms. The reverse resistance of a p-n junction or the input resistance of emitter repeaters are sometimes used as high-resistance resistors. The role of capacitors in the PMS is performed by reverse biased p-rt junctions. The capacity of such capacitors is 50 - 200 pF. It is difficult to create chokes in PMS, so most devices are designed without inductive elements. All PMS elements are produced in a single technological cycle in a semiconductor crystal. The connections of elements of such circuits are made using aluminum or gold films produced by vacuum deposition. The circuit is connected to external terminals using aluminum or gold conductors with a diameter of about 10 microns, which are attached to films using thermal compression and then welded to the external terminals of the microcircuit. Semiconductor chips can dissipate power of 50 - 100 mW, operate at frequencies up to 20 - 100 MHz, and provide a delay time of up to 5 ns. The installation density of electronic devices on the PMS is up to 500 elements per 1 cm3. A modern group technological cycle allows the simultaneous processing of dozens of semiconductor wafers, each of which contains hundreds of PMSs with hundreds of elements in the crystal, connected in given electronic circuits. This technology ensures high identity electrical characteristics microcircuits

Film integral(or simply film circuits PS) are called ICs, all elements and inter-element connections of which are made only in the form of films. Integrated circuits are divided into thin- and thick-film. These schemes may have quantitative and qualitative differences. ICs with film thicknesses up to 1 micron are conventionally classified as thin-film, and ICs with film thicknesses above 1 µm are classified as thick-film. The qualitative difference is determined by the film manufacturing technology. Thin-film IC elements are deposited onto the substrate using thermal vacuum deposition and cathode sputtering. Elements of thick-film ICs are manufactured primarily by silk-screen printing followed by burning in.

Hybrid integrated circuits(GIS) are a combination of mounted active radio elements (microtransistors, diodes) and film passive elements and their connections. Typically, GIS contain: insulating bases made of glass or. ceramic, frames, on the surface of which film conductors, resistors, capacitors are not formed large capacity; mounted open-frame active elements (diodes, transistors); mounted passive elements in miniature design (chokes, transformers, high-capacity capacitors), which cannot be made in the form of films. Such a manufactured GIS is sealed in a plastic or metal case. Resistors with resistance from thousandths of an ohm to tens of kilo-ohms in GIS are made in the form of a thin film of nichrome or tantalum. The films are applied to an insulating base (substrate) and subjected to thermal annealing. To obtain resistors with a resistance of tens of megaohms, metal-dielectric mixtures (chromium, silicon monoxide, etc.) are used. The average dimensions of film resistors are (1 - 2) X 10 ~ 3 cm2. Capacitors in GIS are made of thin films of copper, silver, aluminum or gold. These metals are sprayed with a sublayer of chromium, titanium, and molybdenum, ensuring good adhesion to the insulating material of the substrate. A film of silicon oxide, beryllium, titanium dioxide, etc. is used as a dielectric in capacitors. Film capacitors are made with a capacity from tenths of a picofarad to tens of thousands of picofarads with sizes ranging from 10~3 to 1 cm2. GIS conductors, with the help of which inter-element connections are made and connections to output terminals, are made in the form of a thin film of gold, copper or aluminum with an underlayer of nickel, chromium, titanium, which ensures high adhesion to the insulating base. Hybrid integrated circuits, in which the thickness of the films formed during the manufacture of passive elements is up to 1 micron with a width of 100 - 200 microns, are classified as thin film. Such films are produced by thermal spraying on the surface of substrates in a vacuum using stencils and paints. Hybrid integrated circuits with a thickness of 1 micron or more are classified as thick film and are manufactured by sputtering conductive or dielectric pastes onto substrates through mesh stencils, followed by burning them into the substrates at high temperatures. These circuits are large in size and have a mass of passive elements. Mounted active elements consist of flexible or rigid “ball” leads, which are connected to a film chip by soldering or welding.

The density of passive and active elements with their multilayer arrangement in a GIS made using thin-film technology reaches 300 - 500 elements per 1 cm3, and the installation density of electronic devices on a GIS is 60 - 100 elements per 1 cm3. With such an installation density, the volume of the device containing 107 elements is 0.1 - 0.5 m3, and the trouble-free operation time is 103 - 104 hours. -

The main advantage of GIS is the possibility of partial integration of elements made using various technologies (bipolar, thin- and thick-film, etc.) with a wide range electrical parameters(low-power, powerful, active, passive, high-speed, etc.).

Hybridization is currently promising various types integrated circuits. With small geometric dimensions of film elements and a large area of ​​passive substrates, tens or hundreds of ICs and other components can be placed on their surface. In this way, multi-chip hybrid ICs are created with a large number (several thousand) of diodes and transistors in an indivisible element. In combined microcircuits it is possible to place functional units with different electrical characteristics.

Comparison of PMS and GIS. Semiconductor microcircuits with a degree of integration of up to thousands or more elements in one chip have received priority. spreading. The volume of production of PMS is an order of magnitude higher than the volume of production of GIS. In some devices it is advisable to use GIS for a number of reasons.

GIS technology is relatively simple and requires lower initial equipment costs than semiconductor technology, which simplifies the creation of non-standard, non-standard products and equipment.

The passive part of the GIS is manufactured on a separate substrate, which makes it possible to obtain passive elements High Quality and create high frequency ICs.

GIS technology makes it possible to replace existing methods of multilayer printed circuit assembly when placing unpackaged ICs and LSIs and other semiconductor components on substrates. GIS technology is preferred for high-power power ICs. It is also preferable to use a hybrid design of integrated circuits of linear devices that provide a proportional relationship between input and output signals. In these devices, signals vary over a wide range of frequencies and powers, so their ICs must have a wide range of ratings that are not compatible in a single manufacturing process of passive and active elements. Large LSI integrated circuits allow the combination of various functional units, and therefore they are widely used in linear devices.

Advantages and disadvantages of integrated circuits.

• The advantages of ICs are high reliability, small size and weight. The density of active elements in LSI reaches 103 - 104 per 1 cm3. When installing chips in printed circuit boards and connecting them into blocks, the density of elements is 100 - 500 per 1 cm3, which is 10 - 50 times higher than when using individual transistors, diodes, and resistors in micromodular devices.
• Integrated circuits are inertia-free in operation. Due to the small size of the microcircuits, the interelectrode capacitances and inductances of the connecting wires are reduced, which allows them to be used at ultra-high frequencies (up to 3 GHz) and in logic circuits with low delay times (up to 0.1 ns).
• The microcircuits are economical (from 10 to 200 mW) and reduce electricity consumption and the weight of power supplies.

The main disadvantage The IC is low power output (50 - 100 mW).

Depending on the functional purpose, ICs are divided into two main categories - analog (or linear-pulse) and digital (or logical).

Analog integrated AIS circuits are used in radio technical devices and serve to generate and linearly amplify signals that vary according to the law of a continuous function over a wide range of powers and frequencies. As a result, analog ICs must contain passive and active elements with different ratings and parameters, which complicates their development. Hybrid microcircuits reduce the difficulties of manufacturing analog devices in microminiature design. Integrated microcircuits are becoming the main element base for radio-electronic equipment.

Digital integrated CIS circuits are used in computers, discrete information processing and automation devices. With the help of digital information systems, digital codes are converted and processed. A variant of these circuits are logical chips that perform operations on binary codes in most modern computers and digital devices.

Analog and digital ICs are produced in series. The series includes ICs that can perform various functions, but have a single design and technological design and are intended for joint use. Each series contains several different types, which can be divided into nominal values ​​having a specific functional purpose and symbol. The combination of standard nominal values ​​forms an IP type.

INTEGRATED CXEMA (IC, integrated circuit, microcircuit), a functionally complete microelectronic product, which is a set of electrically interconnected elements (transistors, etc.) formed in a semiconductor monocrystalline wafer. ICs are the elemental base of all modern radio-electronic devices, computer devices, information and telecommunication systems.

Historical reference. The IC was invented in 1958 by J. Kilby (Nobel Prize, 2000), who, without dividing the germanium monocrystalline wafer into the individual transistors formed in it, connected them together with the thinnest wires, so that the resulting device became a complete radio-electronic circuit. Six months later, the American physicist R. Noyce implemented the so-called planar silicon IC, in which metallized areas (the so-called contact pads) were created on the surface of the silicon wafer for each area of ​​bipolar transistors (emitter, base and collector), and connections between them were made with thin-film conductors. In 1959, industrial production of silicon ICs began in the United States; mass production of IP in the USSR was organized in the mid-1960s in Zelenograd under the leadership of K. A. Valiev.

IS technology. The structure of a semiconductor IC is shown in the figure. Transistors and other elements are formed in a very thin (up to several microns) surface layer of a silicon wafer; a multi-level system of inter-element connections is created from above. As the number of IS elements increases, the number of levels increases and can reach 10 or more. Inter-element connections must have low electrical resistance. This requirement is satisfied, for example, by copper. Insulating (dielectric) layers (SiO 2, etc.) are placed between the layers of conductors. Up to several hundred ICs are simultaneously formed on one PP wafer, after which the wafer is divided into individual crystals (chips).

The technological cycle of IC manufacturing includes several hundred operations, the most important of which is photolithography (PL). The transistor contains dozens of parts, the contours of which are formed as a result of PL, which also determines the configuration of the interconnections in each layer and the position of the conducting areas (contacts) between the layers. In the technological cycle, PL is repeated several dozen times. Each PL operation is followed by operations for the manufacture of transistor parts, for example, deposition of dielectric, PP and metal thin films, etching, doping by implantation of ions into silicon, etc. Photolithography determines minimum size(MR) of individual parts. The main PL tool is optical projection stepper-scanners, which are used to perform step-by-step (from chip to chip) image exposure (illumination of the chip, on the surface of which a photosensitive layer is applied - photoresist, through a mask called a photomask) with a reduction (4: 1) in size images in relation to the mask dimensions and with scanning of the light spot within one chip. MR is directly proportional to the wavelength of the radiation source. Initially, PL installations used the g- and i-lines (436 and 365 nm, respectively) of the emission spectrum of a mercury lamp. The mercury lamp was replaced by excimer lasers using KrF (248 nm) and ArF (193 nm) molecules. Improvement of the optical system, the use of photoresists with high contrast and sensitivity, as well as special high-resolution technology when designing photomasks and stepper-scanners with a light source with a wavelength of 193 nm make it possible to achieve MR equal to 30 nm or less on large chips (with an area of ​​1-4 cm 2) with a capacity of up to 100 plates (diameter 300 mm) per hour. Advancement into the region of smaller (30-10 nm) MRs is possible using soft X-rays or extreme ultraviolet (EUV) with a wavelength of 13.5 nm. Due to the intense absorption of radiation by materials at this wavelength, refractive optics cannot be used. Therefore, EUV steppers use reflective optics on X-ray mirrors. Patterns should also be reflective. EUV lithography is an analogue of optical projection lithography, does not require the creation of new infrastructure and provides high productivity. Thus, by 2000, IC technology crossed the 100 nm (MR) barrier and became nanotechnology.

Integrated circuit structure: 1-passivating (protective) layer; 2 - top layer of conductor; 3 - dielectric layer; 4 - inter-level connections; 5 - contact pad; 6 - MOS transistors; 7 - silicon wafer (substrate).

Directions of development. ICs are divided into digital and analog. The main share of digital (logical) microcircuits consists of processor ICs and memory ICs, which can be combined on one crystal (chip), forming a “system-on-chip”. The complexity of an IC is characterized by the degree of integration determined by the number of transistors on the chip. Before 1970, the degree of digital IC integration doubled every 12 months. This pattern (it was first noticed by the American scientist G. Moore in 1965) was called Moore’s law. Moore later refined his law: the complexity of memory circuits doubles every 18 months, and that of processor circuits doubles every 24 months. As the degree of IC integration increased, new terms were introduced: large IC (LSI, with the number of transistors up to 10 thousand), ultra-large IC (VLSI - up to 1 million), ultra-large IC (ULSI - up to 1 billion) and giant LSI (GBIS - more 1 billion).

There are digital ICs based on bipolar (Bi) and MOS (metal-oxide-semiconductor) transistors, including in the CMOS configuration (complementary MOS, i.e., complementary r-MOS and w-MOS transistors connected in series in the “source” circuit supply - a point with zero potential"), as well as BiCMOS (on bipolar transistors and CMOS transistors in one chip).

An increase in the degree of integration is achieved by reducing the size of transistors and increasing the size of the chip; this reduces the switching time of the logical element. As the size decreased, the power consumption and energy (the product of power times the switching time) spent on each switching operation decreased. By 2005, IC performance had improved by 4 orders of magnitude and reached fractions of a nanosecond; the number of transistors on one chip was up to 100 million.

The main share (up to 90%) in global production since 1980 has been made up of digital CMOS ICs. The advantage of such circuits is that in any of the two static states (“0” or “1”) one of the transistors is closed, and the current in the circuit is determined by the current of the transistor in the off state I OFF. This means that if I OFF is negligible, the current from the power supply is consumed only in switching mode, and the power consumption is proportional to the switching frequency and can be estimated by the relation Ρ Σ ≈C Σ ·Ν·f·U 2, where C Σ is the total load capacitance at the output of the logical element, N is the number of logical elements on the chip, f is the switching frequency, U is the supply voltage. Almost all power consumption is released in the form of Joule heat, which must be removed from the crystal. In this case, the power consumed in the switching mode is added to the power consumed in the static mode (determined by the I OFF currents and leakage currents). With a decrease in the size of transistors, static power can become comparable to dynamic power and reach an order of magnitude of 1 kW per 1 cm 2 of crystal. The problem of high power dissipation forces the maximum switching frequency of high-performance CMOS ICs to be limited to the range of 1-10 GHz. Therefore, to increase the performance of systems-on-chip, additional architectural (so-called multi-core processors) and algorithmic methods are used.

At channel lengths of MOS transistors of the order of 10 nm, the characteristics of the transistor begin to be affected by quantum effects, such as longitudinal quantization (an electron propagates in the channel as a de Broglie wave) and transverse quantization (due to the narrowness of the channel), direct tunneling of electrons through the channel. The latter effect limits the possibilities of using CMOS elements in ICs, since it makes a large contribution to the total leakage current. This becomes significant at a channel length of 5 nm. CMOS ICs will be replaced by quantum devices, molecular electronic devices and etc.

Analog ICs comprise a wide class of circuits that perform the functions of amplifiers, oscillators, attenuators, digital-to-analog and analog-to-digital converters, comparators, phase shifters, etc., including low frequency (LF), high frequency (HF), and microwave ICs. Microwave ICs are circuits with a relatively small degree of integration, which can include not only transistors, but also film inductors, capacitors, and resistors. To create microwave ICs, not only the traditional silicon technology is used, but also the technology of heterojunction ICs based on Si - Ge solid solutions, A III B V compounds (for example, gallium arsenide and nitride, indium phosphide), etc. This makes it possible to achieve operating frequencies of 10-20 GHz for Si-Ge and 10-50 GHz and higher for microwave ICs on A III B V connections. Analog ICs are often used together with sensor and micromechanical devices, biochips, etc., which ensure the interaction of microelectronic devices with humans and the environment, and can be enclosed with them in the same housing. Such designs are called multi-chip or system-in-a-package.

In the future, the development of IP will lead to the merging of two directions and the creation of microelectronic devices of great complexity, containing powerful computing devices, environmental control systems and means of communication with humans.

Lit. look at Art. Microelectronics.

A. A. Orlikovsky.

Large integrated circuit

Modern integrated circuits designed for surface mounting.

Soviet and foreign digital microcircuits.

Integral(engl. Integrated circuit, IC, microcircuit, microchip, silicon chip, or chip), ( micro)scheme (IS, IMS, m/skh), chip, microchip(English) chip- sliver, chip, chip) - microelectronic device - an electronic circuit of arbitrary complexity, made on a semiconductor crystal (or film) and placed in a non-separable case. Often under integrated circuit(IC) refers to the actual crystal or film with an electronic circuit, and by microcircuit(MS) - IC enclosed in a housing. At the same time, the expression "chip components" means "surface mount components" as opposed to traditional through-hole soldered components. Therefore, it is more correct to say “chip microcircuit”, meaning a surface-mount microcircuit. IN currently(year) most microcircuits are manufactured in surface-mount packages.

Story

The invention of microcircuits began with the study of the properties of thin oxide films, which manifest themselves in the effect of poor electrical conductivity at low electrical voltages. The problem was that where the two metals came into contact there was no electrical contact or it had polar properties. Deep studies of this phenomenon led to the discovery of diodes and later transistors and integrated circuits.

Design Levels

• Physical - methods of implementing one transistor (or a small group) in the form of doped zones on a crystal.
• Electrical - circuit diagram (transistors, capacitors, resistors, etc.).
• Logical - logical circuit (logical inverters, OR-NOT, AND-NOT elements, etc.).
• Circuit and system level - circuit and system design (flip-flops, comparators, encoders, decoders, ALUs, etc.).
• Topological - topological photomasks for production.
• Program level (for microcontrollers and microprocessors) - assembler instructions for the programmer.

Currently, most integrated circuits are developed using CAD, which allows you to automate and significantly speed up the process of obtaining topological photomasks.

Classification

Degree of integration

Purpose

An integrated circuit can have complete, no matter how complex, functionality - up to an entire microcomputer (single-chip microcomputer).

Analog circuits

• Signal generators
• Analog multipliers
• Analog attenuators and variable amplifiers
• Power supply stabilizers
• Switching power supply control chips
• Signal converters
• Timing circuits
• Various sensors (temperature, etc.)

Digital circuits

• Logic elements
• Buffer converters
• Memory modules
• (Micro)processors (including the CPU in a computer)
• Single-chip microcomputers
• FPGA - programmable logic integrated circuits

Digital integrated circuits have a number of advantages over analog ones:

• Reduced power consumption associated with the use of pulsed electrical signals in digital electronics. When receiving and converting such signals, the active elements of electronic devices (transistors) operate in the “key” mode, that is, the transistor is either “open” - which corresponds to a high-level signal (1), or “closed” - (0), in the first case at There is no voltage drop in the transistor; in the second, no current flows through it. In both cases, power consumption is close to 0, in contrast to analog devices, in which most of the time the transistors are in an intermediate (resistive) state.
• High noise immunity digital devices is associated with a large difference between high (for example 2.5 - 5 V) and low (0 - 0.5 V) level signals. An error is possible with such interference when a high level is perceived as low and vice versa, which is unlikely. In addition, in digital devices it is possible to use special codes that allow errors to be corrected.
• The big difference between the high and low level and a fairly wide range of their permissible changes makes digital technology insensitive to the inevitable dispersion of element parameters in integrated technology, eliminating the need to select and configure digital devices.