microelectronics

from Wikipedia, the free encyclopedia

The Microelectronics is a subfield of electronics , specifically the semiconductor electronics , and the microtechnology . Microelectronics is engaged in the design, development and manufacture of miniaturized , electronic circuits , today primarily integrated circuits . These semiconductor- based circuits use many electronic components as used in normal electronic circuits, such as transistors , capacitors , diodes, and resistors .

In the history of integrated microelectronics, various circuit families ( TTL , CMOS, etc.) have emerged that differ in terms of the functional principles used ( e.g. bipolar and unipolar components / transistors) and the associated circuit properties (power requirement, switching speed, etc.). Thanks to new design and manufacturing processes, users today have the option of having and using special, application-specific integrated circuits (ASIC) in addition to standard circuits ( microcontrollers , memory modules, etc.)

Integrated circuit as an example application from the field of microelectronics. The chip housing has been opened and allows a view of the actual semiconductor. The recognizable structures in the center are the implemented electronic circuit. Outside, the golden connection lines can be seen, which form the electrical wiring between the IC and the housing contacts.

Features of microelectronics

Microelectronic circuits are distinguished from conventional electronic circuits by two main features: integration and miniaturization.

Integration is understood to mean the combination of all components (transistors, but also resistors, capacitors and other semiconductor components ) and wiring to form an electronic circuit on a common substrate . In the case of microelectronic circuits, these are also manufactured in a joint work process.

In this context, miniaturization means that the individual components (and thus the circuit as a whole) are continuously reduced in size. In the early days, the dimensions of the most important component, the transistor, were over ten micrometers . On the other hand, the dimensions of transistors (physical gate length) nowadays are less than 30 nm (2017 approx. 30–24 nm for Intel and Samsung's 14 nm FinFET technology). This miniaturization enables integrated circuits with more than a billion transistors to be implemented on a piece of silicon with an edge length of a few (typically <10) millimeters. In addition, the smaller distance between the components allows the circuits to be operated at higher clock frequencies and, despite the higher computing power, the required electrical power to be increased only slightly.

Before the invention of integrated circuits, there were also intensive efforts towards miniaturization. Circuits with bulky electron tubes have been reduced in size through the development of battery tubes to such an extent that portable radio devices, for example, have become possible. The introduction of transistors brought a further step in miniaturization, with thick-film circuits as the smallest version before the integrated circuits.

Miniaturization of integrated circuits

The main driving force behind the downsizing of structures is the reduction in manufacturing costs. Microelectronic circuits are manufactured in batches on semiconductor wafers of a certain size (4 inches to 12 inches). Several hundred to a thousand chips are produced on one wafer at the same time. The production costs are mainly dependent on the number of chips per wafer, the sum of the costs for the structuring and property-changing processes and the production yield .

The number of chips per wafer can be achieved by downsizing the structures (transistor sizes, interconnects, etc.). In order to achieve approximately the same electrical properties, all dimensions of the chip (length, width and layer thicknesses) must be scaled equally. The number of chips increases (approximately) with the square of the scaling factor (the area reduction is equal to the length reduction squared plus better edge utilization minus non-linear effects), i.e. half the chip length (scaling factor 2) results in a four-fold increase in the number of chips with the same wafer size.

The costs of the structuring processes usually increase with increasing miniaturization. The reason for this lies on the one hand in the increasing complexity of the circuit and the associated increasing number of necessary process steps, on the other hand in higher demands on the manufacturing process itself, such as smaller tolerances, more complex manufacturing processes, etc.

Another cost reduction was achieved by using larger wafers. In the early years, manufacturers still used wafers with a diameter of 2 or 3 inches (corresponding to about 51 or 76 mm). In 2009, on the other hand, standard wafers used in industry were 200 millimeters in size, and some manufacturers such as AMD and Intel even used 300 mm wafers. The number of chips per wafer also increased as the wafer area increased. However, the production costs per wafer increased only comparatively slightly, despite the increased requirements, so that the total costs could be reduced. However, a corresponding changeover requires enormous investments in the production facilities.

Gordon Moore formulated the fact of permanent production cost reduction as early as 1965 - in the early phase of microelectronics - in the law named after him , by saying that the number of transistors on a chip doubles every twelve months (later eighteen months). Indeed, since then, microelectronics has made steady progress in terms of both integration density and the downsizing of structures.

The high constant reduction in manufacturing costs in microelectronics has been a major innovation driver in a large number of industries over the past thirty years - not just in electronics and computer technology (see section Applications ).

It is not certain whether this trend can be maintained in view of the increase in process costs when approaching physical limits and their compensation by saving chip area. In addition, some special circuit technologies can no longer be scaled down, for example those for achieving higher voltages than the supply voltage (s) of the chip.

Consequences of miniaturization

The miniaturization of integrated circuits has other effects in addition to lowering the cost of equivalent products.

Smaller transistors allow higher switching frequencies. Together with shorter conductor tracks, this leads to shortened signal propagation times . As a result of this effect, as the structures become smaller, ever faster and therefore usually more powerful circuits are possible. However, there are also opposing effects. As described in the previous section, miniaturization also reduces the layer thicknesses in the metallization (interconnects and intermediate insulation). The shortened distances between the conductor tracks lead to higher coupling capacities , which act as a delay time brake (see. RC element ).

The reduction in size of the gate insulation layer has a positive effect. The transistors can be operated at a reduced voltage and thus have a reduced power loss (the power loss per area increases, however → poorer heat dissipation). Furthermore, greater system integration (more functions on one chip) means fewer components on a circuit board and thus increased reliability due to fewer soldered connections. Without downsizing and integration, battery-powered, mobile electronics would be inconceivable as they are ubiquitous today: cell phones , notebooks , PDAs, etc.

History, development and people of microelectronics

prehistory

Microelectronics is often associated with or even equated with computer or computer technology , and the widespread use of integrated circuits in the form of processors and microcontrollers in almost all areas of life today has contributed to this. However, electronic calculating machines existed a few decades before the first transistors or integrated circuits.

The first electromechanically working computers - for example Colossus or Mark I - emerged in the Second World War in the early 1940s (see History of Computers ). The mainframe computer ENIAC ( Electronic Numerical Integrator and Calculator ), which was commissioned in 1946, was the first all-purpose purely electronic computer. However, these first calculating machines are not comparable with today's personal computers in terms of computing power or size.

What the transistor is for today's microelectronic circuits was for the ENIAC, which weighed around 27 tons, was the electron tube, which consisted of 17,468 electron tubes and is one of the tube computers . The basics of the electron tube go back to the discovery of glow emission (1883) by Thomas A. Edison (see history of the electron tube ). The first electron tube , a tube diode , was developed in 1904 by John Ambrose Fleming and modified in 1906 by Lee De Forest . Forest added a third electrode to the tube diode and thus created the triode tube , the tube electronics counterpart to the transistor.

As an alternative to digital computers, there were analog and hybrid computers until the 1970s , which were primarily used to calculate differential equations . For example, the Rockefeller Differential Analyzer worked with thousands of electron tubes and electrical relays and was the most powerful adding machine until the end of World War II.

The use of a large number of electron tubes in complex devices was opposed to some disadvantages of these components. Electron tubes were relatively error-prone, required a warm-up time and had very high power losses. The most important improvement goals of the developers after the commissioning of the first electronic calculating machines therefore included increased reliability and a reduction in manufacturing costs.

Miniaturization was also already an important topic; the first calculating machines filled entire rooms. However, electron tubes are considered to be hardly miniaturizable. Nevertheless, there were already intensive efforts towards miniaturization both in the construction of the overall circuit and in the tubes themselves. Circuits with bulky electron tubes were reduced in size through the development of battery tubes to such an extent that portable radio devices , for example, became possible. The introduction of transistors brought a further step in miniaturization, with thick-film circuits as the smallest version before the integrated circuits.

An early form of the working memory of computers was the core memory , in which many hard magnetic rings were used, which were strung on wires and which could be remagnetized and read out with a current surge.

Invention of the transistor

The fact that the computing power of the mainframe computers of that time can no longer keep up with that of today's pocket calculators is mainly due to the development of the transistor based on so-called semiconductors and integrated circuits.

The discovery of the transistor or the transistor effect are generally attributed to the American scientists John Bardeen , Walter Brattain and William B. Shockley . In 1956 they received the Nobel Prize in Physics “for their research on semiconductors and their discovery of the transistor effect” . Bardeen, Brattain and Shockley belonged to a group of the Bell Telephone Laboratories in Murray Hill (New Jersey) , which dealt with the investigation of field effects in solids. In one of the experiments carried out on December 16, 1947, Brattain observed a current gain, the transistor effect. The structure of an n-doped germanium crystal contacted with three electrodes was the first functioning bipolar transistor . The main change compared to earlier designs was that two electrodes were very close (approx. 50 μm) to each other, which was what made the transistor effect possible. This transistor, later known as the tip transistor, could not be produced reproducibly and its mode of operation was not sufficiently well known for a long time. Nevertheless, those responsible recognized the potential of this discovery very quickly. The main advantages over the electron tube were that no vacuum and no warm-up time were necessary and that no heat generation was observed. The possibility of miniaturization of electronic circuits was the starting point for a revolution in electronics that made many developments in microelectronics and computer technology possible in the first place.

From today's perspective, Bardeen, Brattain and Shockley were not the first or only researchers who dealt with the development of alternatives to the electron tube based on semiconductors. As early as 1925 ( Lilienfeld ), 1934 ( Heil ) and 1945 ( Heinrich Welker ) ideas for another transistor, the field effect transistor , were published. However, since the manufacturing processes (e.g., for cleaning the semiconductor substrates of foreign matter) were insufficient at that time, these ideas could not be realized. For this and other reasons, they were ignored by the public and were only known to a few experts.

The aspect of the substrate material is often disregarded in connection with the discovery of the transistor. However, the quality and purity of the semiconductors used are essential for the functioning of transistors. Producing semiconductor crystals with a sufficiently high purity was a major problem at that time (before 1950). Many of the germanium crystals used by the group at Bell Telephone Laboratories (BTL) came from WG Pfann, JH Scaff and HC Theuerer. They were made using a zone melting method by GK Teal and JB Little.

Independently of the BTL working group, Herbert F. Mataré and Heinrich Welker - at that time employees at Westinghouse in Paris - developed a transistor that worked on a similar principle. This bipolar transistor, also known as the "Transitron", was presented to the Americans a few months later (August 13, 1948). Mataré founded the company Intermetall in Germany in 1952 and was able to present the first prototype of a transistor radio with headphones ; a year before the first commercial US model.

In 1956, William Shockley opened a laboratory ( Shockley Semiconductor Laboratory ) in Mountain View near Stanford University in Palo Alto . The laboratory is considered to be the nucleus of Silicon Valley , it should be noted that at that time both research and industry were very concentrated on the east coast of the USA. Already in September 1957 eight employees ( Eugene Kleiner , Jay Last , Victor Grinich , Jean Hoerni , Sheldon Roberts , Julius Blank , Gordon E. Moore and Robert N. Noyce ) parted ways with Shockley because of differences of opinion . They started Fairchild Semiconductor Corporation with venture capital . Fairchild Semiconductor was one of the companies driving the development of microelectronics at that time, so Fairchild produced the first silicon-based transistor in series as early as 1958 and was significantly involved in the development of the planar transistor and the integrated circuit. Gordon Moore and Robert Noyce then founded the company Intel in 1968 , which today (2010) is the company with the highest sales in the field of microelectronics.

Silicon displaces germanium

Silicon became the dominant semiconductor material from the mid-1960s, although germanium was the leader in the early years of semiconductor electronics. In 2009 around 95% of all microelectronic circuits were produced on the basis of silicon.

The initial advantages of germanium were its better properties, such as a lower melting point and higher charge carrier mobility (enables higher switching frequencies ) and it was easier and better to clean than silicon until then.

The most important reasons for the success of silicon are the good properties of the material combination silicon and silicon dioxide . Silicon dioxide is used as an insulation material and, in addition to its good electrical properties ( breakdown field strength , band gap , etc.) shows very good layer adhesion on silicon. With the thermal oxidation of silicon , a simple manufacturing process of silicon dioxide layers on crystalline silicon is also available, which enables silicon dioxide layers with very good interface properties, such as a low concentration of interface charges. Unlike germanium dioxide, silicon dioxide is chemically stable to water, i.e. it does not dissolve in water. This enables the surfaces to be easily cleaned with water and various wet-chemical coating and structuring processes to be used. The higher melting point compared to germanium makes the material generally more robust and allows higher temperatures during production, for example in some diffusion and coating processes.

The good properties of thermally produced silicon dioxide enabled the development of the planar process and thus the development of integrated circuits as they are used today (see below). These important inventions in microelectronics were preceded by further significant improvements in the manufacture and stability of transistors through the use of thermally produced silicon dioxide. In addition to the suitability as selective doping masking , this includes above all the passivating effect and the very good electrical properties of the interface between thermal silicon dioxide and silicon. The passivation of the surface and the associated reduction in interface charges and environmental influences improved the electrical characteristics of the transistors (their characteristic curves) both in terms of reproducibility and their stability in use. In addition, the improved insulator-semiconductor interface enabled the production of the first functioning MIS field effect transistors (MIS-FET). After it was recognized that charges in silicon dioxide caused by alkali metal impurities also massively deteriorate the electrical properties of semiconductor components and that this has been taken into account in production, the fundamental problems in the manufacture of stable components based on semiconductors were solved.

Integrated circuits

The last step towards microelectronics was the transition from circuits made from discrete semiconductor components on a printed circuit board to integrated circuits (ICs). Integrated circuits are generally understood to mean circuits made from semiconductor components (mainly transistors) including the wiring on a substrate, also known as monolithic circuits . This concept was invented and patented independently by Jack Kilby , employee of Texas Instruments , and Robert Noyce, founding member of Fairchild Semiconductor, in the late 1950s . Kilby's patent from 1959 showed for the first time a circuit made up of various components (transistors and resistors) on a single substrate (made of germanium). This work resulted in the famous Kilby patent (rejected by the Japanese Patent Office and the Tokyo Supreme Court for lack of inventive step). This patent was disputed in court for around ten years because Robert N. Noyce had devised a very similar step, but later applied for a patent.

Unlike Kilby, who only thought up wiring on the substrate, Noyce's patent was based on the ideas and knowledge of the planar process , which was developed at the same time by Jean Hoerni (also Fairchild Semiconductor). Photolithographic and diffusion processes, which Fairchild Semiconductor had recently developed for the manufacture of the first modern diffusion bipolar transistor, have already been used for production. Among other things, the technical feasibility of such wiring revolutionized the manufacture of electronic circuits. As a result, many companies intensified their research and development in this area and an enormous miniaturization began.

In 1961 the first commercially available integrated circuit was presented. It was a flip-flop from Fairchild Semiconductors, was manufactured using planar technology and consisted of four bipolar transistors and five resistors. The bipolar transistors were quickly replaced by field effect transistors (FETs), mostly in the form of easier to manufacture MOSFETs ( metal-oxide-semiconductor field effect transistors ). The functional principle of the MOSFETs had been known for a few decades, but the first functional copies were only manufactured in 1960 by Martin M. Atalla and Dawon Kahng (see surface passivation in the article Thermal oxidation of silicon ). Further important developments in the 1960s were CMOS technology ( Frank Wanlass , 1963) and the first DRAM memory cell by Robert H. Dennard ( Thomas J. Watson Research Center / IBM , 1967, see DRAM ).

The complexity of the circuits increased rapidly and in 1970/71 the first microprocessors from three companies were introduced almost simultaneously: the Intel 4004 , the Texas Instruments (TI) TMS 1000 and the Garrett AiResearch "Central Air Data Computer" (CADC). At that time, circuits with transistor densities with several thousand components were implemented on one chip. This stage of development is as large scale integration (English: Large Scale Integration , LSI) called. Due to the rapid development of microelectronics, transistor densities were achieved in 1979 that were many times greater (around two orders of magnitude) than with LSI circuits. This level of ICs is known as Very Large Scale Integration (VLSI). This trend has essentially been maintained until today (2009) (cf. Moore's Law ), so that today over a billion transistors with clock frequencies of several gigahertz are possible. The size of the individual component is far less than a square micrometer. Increasingly, entire systems (combination of several assemblies such as processors , interface circuits and memories ) are being implemented on a single chip ( system-on-a-chip , SoC).

Future developments

If the structure sizes fall below the 100 nanometer limit (chips in mass production 2002 at 90 nm, 2008 at 45 nm, 2020 at 5 nm), one speaks formally of nanoelectronics or nanotechnology (definition of the US government). In the narrower sense, however, it is more likely to mean that special material properties are used that only appear when the structural dimensions are close to the size of the molecule or atom . Such structures include, for example, conductor tracks or transistors made of carbon nanotubes or insulation made of self-assembling monolayers .

New components are built with resonance tunnel diodes.

Integrated optoelectronics : In view of increasing signal propagation times, especially in long connecting lines (global interconnects) of large " system-on-a-chips ", consideration is being given to replacing these electrical lines with optical connections .

Organic electronics : In order to implement inexpensive “throwaway electronics” (for example electronic price labels), circuits based on organic and metallic materials are applied to organic substrates using thin-film technology (see organic field effect transistor ).

Interesting perspectives also arise from the fact that, due to the progressive scaling, the smallest structures that can be realized in microelectronics are reaching the order of magnitude of biological molecules . A convergence of biotechnology and microelectronics and the development of a new specialist science can currently be observed, which is often referred to as bioelectronics and will primarily concern the development of new types of biosensors .

Classification and components

In microelectronic circuits, a large number of semiconductor components (especially diodes and transistors ), electrical resistors, capacitors and rarely inductors are used and integrated, that is, joined together, on a semiconductor crystal (substrate). The microelectronic circuits can be divided into standard circuits or application-specific standard products (ASSP) and application-specific circuits (ASIC) on the basis of the application area, as well as into analog and digital ICs on the basis of the signals to be processed.

Standard circuits can be used in a large number of applications, are produced in large numbers and their characteristics are partly determined by standardization consortia. On the other hand, application-specific ICs are circuits that are designed and built for a specific application (e.g. engine control in a car) and their range of functions does not allow any other application. The demarcation of these two groups is fluid in practice: Some circuits called ASICs can still be reprogrammed, but not for any function or application.

Analog ICs are integrated circuits that can directly process analog signals - signals that can change continuously within a certain range of values, such as voltages or currents. So-called standard ICs with low functional integration, for example operational amplifiers and voltage regulators, are a large field of application .

Digital ICs, on the other hand, only process digital signals - signals whose value ranges have a finite number of signal values. They currently (2009) represent the largest group of circuits is Typical of digital ICs. Programmable logic devices (English: programmable logic devices PLD), memory devices (such as ROM , PROM , DRAM or flash memory ) and complex than microcode programmable circuits such digital signal processors , microcontrollers or microprocessors .

In addition to this rough division into analog and digital ICs, there are other circuits, such as the digital-to-analog or analog-to-digital converter , frequency-to-voltage converters and voltage-controlled oscillators (voltage -controlled oscillators) at the interface between the analog and digital areas. Frequency converter). Sensors are also often integrated directly into the microelectronics, often together with their adaptation electronics and possibly a digital interface for connection to a microprocessor. Temperature sensors are relatively simple . Large quantities of optical sensors are produced today, as image sensors or as parts of optoelectronic assemblies .

The various forms of component integration require different assembly configurations in order to combine the individual components into electronic assemblies . This resulted in a large number of complex and differentiated assembly and connection technologies . Thus components differ according to the mounting configuration in packaged on the circuit board attachable SMDs or pluggable wired components ( THT ) and unpackaged bare chip that directly or with an intermediate wiring substrate to the wiring substrate are placed. Today most components are mounted as SMDs. However, there are components for which attachable SMD designs are not yet available, or for which the electrical load capacity of the component is limited too much by the SMD design. In these cases, the older THT designs are still used.

Applications

Microelectronic components were originally developed to meet the requirements of space travel for small and light components. Today they can be found in a large number of technical devices and facilities:

At this point only an exemplary selection can be mentioned - there are a large number of other applications in the areas mentioned as well as a number of application areas not mentioned here, such as medical technology , building technology and much more.

Development of microelectronic components

On the , a single unpackaged semiconductor chip can billions of transistors and other basic elements of the microelectronic circuits are applied. The design of a functional chip is supported with computer-aided tools . The individual steps, which are usually divided into labor and taken several times with decreasing abstraction, are: design, simulation and verification. Production preparation concludes. Since the production of a chip is preceded by very high non-recurring engineering costs ( NRE costs ) (e.g. mask costs , see photolithography ) and an integrated circuit can only be repaired to a very limited extent and not productively practicable, it is of great importance that the draft only leads to the desired product with a few revisions (so-called redesigns). As a result, simulation and verification steps determine the development process to a large extent - in 2004 they account for around half of the development effort for the circuit - with an upward trend. In order to distribute these costs over a large number of chips, an attempt is made to distribute sub-steps of the development work over several projects. For example, logic gates or entire microprocessor architectures can be purchased as tested libraries and integrated into your own development. Another possibility is to use FPGAs (digital technology) and FPAAs (analog technology). These components contain freely programmable digital and analog circuits that only have to be configured according to your own circuit design.

In many cases, the chip designers (comparable describing the desired circuit in digital circuit blocks only in a "standard language" higher programming in computer science, common characteristics: VHDL , Verilog ), the computer calculates the switching networks ( Synthesis (English) synthesis ) and places the transistors (with human involvement and control).

For analog circuits, the designed circuits are characterized in a large number of simulation runs (for example with SPICE or similar), with many different combinations for operating temperatures and voltages, process variations and the like being simulated. Frequently, statistical variations are also simulated using a Monte Carlo analysis . For both digital and analog circuits, the layout can be extracted in order to take parasitic elements into account in the simulation and thus achieve more realistic results.

Ongoing miniaturization is pushing both the structuring processes and the implemented functional components, such as transistors and conductor tracks, to their physical limits. To counter the former, software is increasingly being used in the design process that simulates the physical boundary effects, such as optical diffraction in photolithography, and modifies the circuit design in such a way that these effects are compensated ( optical proximity correction , OPC). In order to counteract the miniaturization effects in the components, new simulation and modeling methods are continuously being added to the chip design process: for example, simulations of the voltage drop in long supply networks ( IR drop ), simulation of the parasitic capacitive coupling of neighboring conductor tracks, tools for the static analysis of the time relationships in a digital circuit (English static timing analysis , STA) etc.

In order to produce prototypes of a new chip reasonably inexpensively, the layouts of several chips can be put together on one wafer (English: multi project wafer , MPW), since the mask and production costs for the comparatively small number of prototypes can be distributed over several projects.

Manufacture of microelectronic circuits

The production of microelectronic circuits is carried out using methods of semiconductor technology (production of the components on a substrate and, in the case of monolithic circuits, the wiring) and assembly and connection technology (housing and wiring / linking of microelectronic and non-electronic components to form a system). Processes from thin and thick film technology are also used here, with the latter the components are applied to a film or embedded and connected; they are only relevant for special applications ( high frequency technology ).

Production takes place in an extremely clean environment, so-called clean rooms , with a very low density of dust particles. This is necessary because the structures to be produced are in the micro- and nanometer range and even the smallest particles (<0.1 µm) can cause manufacturing defects that result in the failure of a complete circuit. The production process itself can be divided into three areas (besides functional tests): the substrate production and the production of the components (transistors, diodes, etc.), the so-called front-end , and their "packaging" in housings , the so-called back-end .

Substrate production

Integrated circuits are manufactured on so-called wafers (monocrystalline semiconductor wafers), so that several integrated circuits can be manufactured in parallel and costs can be reduced. Hundreds of identical integrated circuits are produced in parallel on a wafer and, in the case of simple structures (for example individual transistors) thousands of identical integrated circuits.

So that the high requirements for the production of integrated circuits can be met, it is necessary to produce the substrate in the form of high-purity single crystals . The vast majority (more than 99%) of integrated circuits use the semiconductor silicon as substrate material . Other materials such as gallium arsenide are also used for very high-frequency or optical applications . For special applications, silicon is also used on the insulating substrate sapphire (English: Silicon-on-Sapphire , SOS).

In the case of silicon, a monocrystalline cylinder ( ingot ) is first drawn from a high-purity silicon melt (cf. extraction of pure silicon ). The so-called Czochralski method (CZ method) is primarily used for this purpose. An alternative process is zone melting , which can also be used for further cleaning of the CZ ingots; For some special applications, higher degrees of purity are necessary than one foreign atom on 10 9 atoms of the CZ ingot. Depending on the diameter, the ingots are sawn into 0.5–1.5 mm thin disks, the so-called wafers . The silicon wafers used today in mass production have diameters of 150, 200 or 300 mm (often referred to as 6, 8 or 12 inches). Through various etching, grinding and polishing processes, you get an almost perfectly flat surface with unevenness in the order of magnitude of less than a nanometer, i.e. with surface roughness of only a few atomic layers.

Manufacture of the components

Schematic structure of a CMOS chip in the 2000s (excerpt)

Front end

The so-called front-end in the manufacture of integrated circuits deals with the production of electrically active components (transistors, capacitors, etc.), the so-called front-end-of-line (FEOL), and their wiring (metallization), the so-called Back-End-of-Line (BEOL). A wide variety of semiconductor technology processes are used to build up layers ( epitaxy , sputter deposition , vapor deposition , CVD , etc.), layer removal and structuring ( photolithography ). Furthermore, methods for changing material properties ( e.g. doping ) are used.

Nowadays (2009), after the metallization, there is often a random or complete testing of the circuits with needle testers in the wafer assembly, primarily to determine the yield and as feedback on technological parameters. This saves you having to cap the sometimes considerable rejects. For the determination of technological parameters, the test (for example layer thickness test) is usually carried out directly after the respective process, here it is sometimes important to also include the respective systems, since identical systems with the same parameters produce deviations that can be outside the tolerance range.

Back end

In the subsequent production section, the back end , the ICs are then separated. This is generally done by sawing (rarely also by scoring and breaking) the wafer into dies (the so-called chips). In the subsequent packaging (English packaging ), the individual ICs are then introduced into a housing and contacted, the so-called bonding . Different processes are used depending on the type, for example chip bonding or wire bonding .

Capping ( Einhausen ) is used for hermetic sealing against environmental influences - for purely electrical circuitry, the enclosure must gas- and light-proof be - as well as better applicability: either the chip together with the bonding wires in a cavity (sheet metal, ceramics, possibly with window) included or coated with synthetic resin (cast). The connections to the outside are designed, for example, as a dual in-line package (DIL) or plastic leaded chip carrier (PLCC). Highly complex circuits (mostly for mobile applications) have recently (2009) also been used without a base housing and soldered directly onto the respective circuit boards (cf. Ball Grid Array ).

Finally, there is another function test, in which the guaranteed properties are checked on all circuits. The type test is carried out on a random basis or only in the development phase. The routine test is used to sort into circuits of different quality classes (for example according to offset voltage in operational amplifiers ). Test results and the type of capping determine the area of ​​application. In this way, high qualities are produced for extended operating temperatures and environmental requirements (so-called MIL standard for military and space applications). Higher tolerances and plastic capping are possible for mass applications (consumer goods).

In addition to the monolithic circuits, there are also so-called thick-film hybrid circuits . Individual functions of the circuit are implemented in different semiconductor chips and applied to an additional carrier material and wired for printing using the screen printing process. In this way, in addition to connecting lines, passive components can also be implemented.

If particularly compact components are required, for example in cell phones , several individual circuits are also electrically connected over the shortest possible route and housed in a common housing, see Multi-Chip Module .

Function test

In order to react to process fluctuations at an early stage, to correct faulty processes if necessary or even to remove wafers or lots from production, the still unfinished ICs are tested after many process steps. In the front-end, these are mostly samples. After the front end, all ICs are usually tested for their function before further processing. In some cases, certain functions ( high-frequency circuits or connections of the chip that are later not brought out to pins) can only be tested on the die. Above all, for reasons of cost it must be prevented that non-functional ICs are further processed in the subsequent manufacturing process.

Finally, the packaged chip is also subjected to a final test before delivery in order to identify errors in the back-end production. Some properties are also tested that change due to the packaging or whose measurement is not possible without a housing, such as bonding or certain high-frequency properties. The packaged chip can then go to the PCB assembly.

Although these measurements run fully automatically on special test systems ( Automatic Test Equipment ), the costs associated with highly integrated processor chips have almost reached the production costs. This is mainly due to the fact that economies of scale only take effect to a limited extent during testing (for example, parallelization is only possible with pure digital circuits) and newer ICs contain more and more functions that have to be tested one after the other.

Economy

The branch of industry that deals with the manufacture of microelectronic components - the semiconductor industry - exhibits two characteristics that distinguish it from others.

Large economies of scale: Semiconductor factories for the mass production of components with the smallest possible structural sizes are only profitable from a certain size. Furthermore, these factories are orders of magnitude more expensive than comparable manufacturing facilities in other industries: today (2004) the cost of building and equipping a high-volume, state-of-the-art factory is around USD 2 billion. Both together lead to the so-called pig cycle : There is only a comparatively small number of current semiconductor factories worldwide. When the industry is doing well, i.e. usually when the supply of semiconductor components is lower than the demand, it expands its manufacturing capacities because most companies can only then raise the sums for the expansion. Every new factory that goes into production increases the world market volume of available modules by percentage points, since they have to be very large for reasons of profitability. The sudden increase in the available volume leads to a correspondingly sharp drop in the price of components, which settles again as soon as demand has caught up with supply. Due to the drop in prices, many companies are unable to expand their production capacities for a while - the next shortage of supply is approaching. Then the cycle repeats itself.

Attribution of strategic importance: Many countries attribute strategic importance to the semiconductor industry. This is mostly due to the "germ cell effect" for other high technologies. In the area of ​​semiconductor industries, not only are highly qualified suppliers from the chemical and plant engineering sectors developing, but also from the customer industries for components, for example the computer and electronics industry. In some cases, the strategic importance is also justified militarily. The USA considers the importance of microelectronics for armaments programs so important that both devices for the production of current ICs as well as the circuit designs and even the circuit development software are the subject of their export control lists. This highly valued importance means that a large number of countries are promoting the establishment of the semiconductor industry in many ways: from start-up financing, special tax structures, government loan guarantees to government-funded research at university and industrial research centers, etc. These promotions are also occasionally the subject of economic ones Conflicts between states and companies - as last happened in 2003. At that time, the DRAM manufacturer Hynix was accused of having received excessive support from the South Korean state in its financial crisis. Hynix's competitors then obtained punitive tariffs on the import of this company's products in the USA, the European Union and, most recently, Japan, against which South Korea protested.

Business models: As in many other industries, there is also a full manufacturer - called Integrated Device Manufacturer (IDM). An IDM creates the product design, develops the manufacturing technology, manufactures the component and sells it itself. However, there are also the “Fabless Design Houses” and “ Foundries ”. Fabless Design Houses create the product design according to the specifications or in cooperation with the foundry , which will manufacture it later, and sell the finished product. The foundry develops the production technology, provides its customers with technology-specific tools for chip design ( EDA ) and manufactures the ICs. Combinations of these business models and niche models can also be found in practice.

See also

literature

  • Simon M. Sze : Physics of Semiconductor Devices. 2nd Edition. John Wiley and Sons, 1981, ISBN 0-471-05661-8 .
  • Ulrich Hilleringmann: Silicon semiconductor technology. Teubner, 2004, ISBN 3-519-30149-0 .
  • Ulrich Tietze, Christoph Schenk, Eberhard Gamm: Semiconductor circuit technology. 12th edition. Springer, 2002, ISBN 3-540-42849-6 .
  • Michael Reisch: Semiconductor components. Springer, 2004, ISBN 3-540-21384-8 .
  • Paul R. Gray, Paul J. Hurst, Stephen H. Lewis, Robert G. Meyer: Analysis And Design Of Analog Integrated Circuits. 4th edition. John Wiley and Sons, 2001, ISBN 0-471-32168-0 .

Web links

Videos

Individual evidence

  1. Kevin Gibb: Samsung's 14 nm LPE FinFET transistors: Page 4 of 4. In: eeNews Europe. January 20, 2016, accessed December 26, 2017 .
  2. Blaine Stackhouse, Sal Bhimji, Chris Bostak, Dave Bradley, Brian Cherkauer, Jayen Desai, Erin Francom, Mike Gowan, Paul Gronowski, Dan Krueger, Charles Morganti, Steve Troyer: A 65 nm 2-Billion Transistor Quad-Core Itanium Processor . In: IEEE Journal of solid-state circuits . No. 44 , 2009, p. 18–31 ( PDF [accessed May 25, 2010]). PDF ( Memento of the original from January 2, 2011 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.  @1@ 2Template: Webachiv / IABot / www.ece.ncsu.edu
  3. ^ Bo Lojek: History of Semiconductor Engineering . Springer, Berlin 2007, ISBN 978-3-540-34257-1 , pp. 18th ff .
  4. ^ Bo Lojek: History of Semiconductor Engineering . Springer, Berlin 2007, ISBN 978-3-540-34257-1 , pp. 24 .
  5. Patent CA272437 : Electric Current Control Mechanism. Published on July 19, 1927 , inventor: Julius Edgar Lilienfeld (Registered with the Canadian Patent Office ).
  6. Patent GB439457 : Improvements in or relating to electrical amplifiers and other control arrangements and devices. Registered on March 2, 1934 , published on December 6, 1935 , inventor: Oskar Heil.
  7. Patent DE980084 : Semiconductor arrangement for the capacitive control of currents in a semiconductor crystal. Registered April 6, 1945 , published August 2, 1973 , inventor: Heinrich Welker.
  8. ^ Bo Lojek: History of Semiconductor Engineering . Springer, Berlin 2007, ISBN 978-3-540-34257-1 .
  9. Patent FR1010427 : Crystal device for controlling electric currents by Means of a solid semiconductor. Registered on August 13, 1948 , published June 11, 1952 , inventor: Heinrich Welker.
  10. a b Bo Lojek: The MOS Transistor . In: History of Semiconductor Engineering . Springer, Berlin 2007, ISBN 978-3-540-34257-1 , pp. 317 ff .
  11. https://www.bsi.bund.de/cae/servlet/contentblob/486910/publicationFile/30685/Nanotechnologie_pdf.pdf
  12. ^ Bo Lojek: History of Semiconductor Engineering . Springer, Berlin 2007, ISBN 978-3-540-34257-1 , pp. 70 .
  13. Patent US2804405 : Applied December 24, 1954 , published April 27, 1957 , inventors: L. Derick, CJ Frosch.
  14. CJ Frosch, L Derick: Surface Protection and Selective Masking during Diffusion in Silicon . In: Journal of The Electrochemical Society . tape 104 , no. 9 , 1957, pp. 547 , doi : 10.1149 / 1.2428650 .
  15. D. Kahng, MM Atalla: Silicon-silicon dioxide field induced surface devices. In: IRE-AIEE Solid-state Device Res. Conf., (Carnegie Inst. Of Technol., Pittsburgh, PA). 1960.
  16. Patent US3102230 : Electric field controlled semiconductor device. Registered May 31, 1960 , published August 27, 1963 , inventor: D. Kahng.
  17. Patent US3056888 : Semiconductor triode. Registered on August 17, 1960 , published October 2, 1962 , inventor: MM Atalla.
  18. ^ P. Seidenberg: From Germanium to Silicon: A History of Change in the Technology of Semiconductors . In: Facets: New Perspectives on the History of Semiconductors . 1997, p. 36-74 ( PDF ).
  19. a b Shohei Oguri: Patent Infringement Litigation Case Study (1) . Ed .: Japan Patent Office. 2001 ( PDF in the Internet Archive ( Memento from September 1, 2004 in the Internet Archive )). Patent Infringement Litigation Case Study (1) ( Memento of the original from September 1, 2004 in the Internet Archive ) Info: The archive link has been inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.  @1@ 2Template: Webachiv / IABot / www.ircc.iitb.ac.in
  20. a b Patent US3138743 : Miniaturized electronic circuits. Applied on February 6, 1959 , published June 23, 1964 , inventor: Jack Kilby.
  21. a b Patent US2981877 : Semiconductor device and lead structure. Registered on June 30, 1959 , published April 25, 1961 , inventor: Robert N. Noyce.
  22. Klaus Hinkelmann: Consideration of the legal validity of patents in patent infringement proceedings: translation and commentary on the decision of the Supreme Court of April 11, 2000 . In: Journal of Japanese Law . No. 10 , 2000, pp. 266-273 ( online ). online ( Memento of the original from September 23, 2015 in the Internet Archive ) Info: The archive link has been inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.  @1@ 2Template: Webachiv / IABot / www.djjv.org
  23. ^ JS Kilby: Invention of the Integrated Circuit . In: IEEE Transactions on Electron Devices . ED-23, No. 7 , 1976, p. 648–654 , doi : 10.1109 / T-ED.1976.18467 ( PDF [accessed July 3, 2010]).
  24. IM Ross: The invention of the transistor . In: Proceedings of the IEEE . tape 86 , no. 1 , 1998, p. 7-28 .
  25. ^ RG Arns: The other transistor: early history of the metal-oxide semiconductor field-effect transistor . In: Engineering Science and Education Journal . tape 7 , no. 5 , 1998, pp. 233-240 .
  26. ^ Andrew Harter: Three-dimensional integrated circuit layout . Cambridge University Press, 1991, ISBN 0-521-41630-2 , pp. 4 .
  27. ^ Leslie Berlin: The man behind the microchip: Robert Noyce and the invention of Silicon Valley . Oxford University Press US, 2005, ISBN 0-19-516343-5 , pp. 135 ( limited preview in Google Book search).
  28. F. Wanlass, C. Sah: Nanowatt logic using field-effect metal-oxide semiconductor triodes . In: 1963 IEEE International Solid-State Circuits Conference (February 20, 1963). Digest of Technical Papers . tape 6 , 1963.
  29. Patent US3356858 : Low stand-by power complementary field effect circuitry. Registered on June 18, 1963 , inventor: Frank M. Wanlass.
  30. Patent US3387286 : Field-effect transistor memory. Registered on July 14, 1967 , inventor: Robert H. Dennard.
  31. Mario Birkholz: Convergence in Sight - on the common perspective of microelectronics and biotechnology (PDF; 665 kB). In: LIFIS ONLINE. July 29, 2009. Retrieved December 9, 2009.
  32. Wolfgang Scheel (Ed.): Assembly technology of electronics . Assembly. 2nd Edition. Verlag Technik, Berlin 1999, ISBN 978-3-341-01234-5 , preface to the first edition.
  33. Hans-Joachim Hanke (Ed.): Assembly technology of electronics . Printed circuit boards. 1st edition. Verlag Technik, Berlin 1994, ISBN 3-341-01097-1 , p. 82 .
  34. EU Commission imposes punitive tariffs on chip manufacturer Hynix. (News report) In: heise online. April 24, 2003. Retrieved August 12, 2009 .