Matter exists in various forms, such as solid, liquid, gas, plasma, etc. We usually refer to materials with poor electrical and thermal conductivity, such as diamond, artificial crystal, amber and ceramics, as insulators. Metals with good electrical and thermal conductivity, such as gold, silver, copper, iron, tin and aluminum, are called conductors. The material between conductor and insulator can be simply called semiconductor. Compared with metals and insulators, the discovery of semiconductor materials is the latest. It was not until the 1930s, when the purification technology of materials was improved, that the existence of semiconductors was truly recognized by academic circles.
In fact, the discovery of semiconductors can be traced back to a long time ago. 1833, ElBaradei, England, discovered for the first time that the resistance of silver sulfide varies with temperature, which is different from that of ordinary metal. Generally speaking, the resistance of metal increases with the increase of temperature, but ElBaradei found that the resistance of silver sulfide material decreases with the increase of temperature. This is the first time that a semiconductor phenomenon has been discovered. Soon, in 1839, Becker discovered that the junction formed by the contact of semiconductor and electrolyte would generate voltage under illumination, which was later called photovoltaic effect, which was the second characteristic of semiconductor discovered. In 1874, Braun observed that the conductivity of some sulfides is related to the direction of the applied electric field, that is, their conductivity is directional, and it will conduct when a DC voltage is applied across them; If the voltage polarity is reversed, it will not conduct electricity, which is the rectification effect of semiconductors and the third characteristic of semiconductors. In the same year, Schuster discovered the rectification effect of copper and copper oxide. 1873, Smith of Britain discovered the photoconductive effect of selenium crystal material with increased conductance under illumination, which is another unique property of semiconductor. Although these four effects of semiconductors were discovered before 1880 years ago, the term semiconductor was first used by Kauniberg and Weiss about 19 1 1 years ago. Summing up these four characteristics of semiconductors, Bell Laboratories didn't finish it until June1947+February. Many people will ask, why did it take so many years for semiconductors to be recognized? The main reason is that the materials at that time were impure. Without good materials, many problems related to materials are difficult to explain clearly.
Early application of semiconductor materials
The earliest application of semiconductor is to use its rectification effect as a detector, that is, a point contact diode (also known as a cat beard detector, that is, a metal probe contacts the semiconductor to detect electromagnetic waves). In addition to detectors, in the early days, semiconductors were also used as rectifiers, photovoltaic cells, infrared detectors and so on. And all four effects of semiconductors are used.
From 1907 to 1927, American physicists have successfully developed crystal rectifier, selenium rectifier and cuprous oxide rectifier. 193 1 year, Langer and Bergman successfully developed selenium photovoltaic cells. 1932, Germany successively developed semiconductor infrared detectors such as lead sulfide, lead selenide and lead telluride, which were used to detect aircraft and ships in World War II. During World War II, the Allied Forces also made great achievements in semiconductor research, and Britain used infrared detectors to detect German aircraft many times.
The invention of transistor
The invention of the transistor was actually six months before 65438+February 23rd. 1947. At that time, researchers at Bell Laboratories had seen the commercial value of transistors and kept them secret for half a year to write patents. It was not until 1947 and 1947 on February 23rd that Badin and Burton officially announced their invention, which became the official invention day of transistors. They used a very simple device, that is, on a germanium crystal, two very thin metal needles were glued to the surface of germanium, and a positive voltage was applied to one needle and a negative voltage was applied to the other probe. Now we call them emitter and collector respectively, and N-type germanium became the base, thus forming a PNP transistor with amplification effect.
Badin and Burton were working in a research group led by shockley. Although Shockley was the team leader at that time, he was very unhappy that his name was not on the invention patent. So in a very short time, that is,1948+June 23rd, 65438, shortly after the invention of the transistor, he proposed the structure of surface contact transistor instead of point contact. It turns out that this structure is really valuable.
Badin and Burton kept a secret for nearly half a year before announcing their invention. After the invention was published, the response was not as enthusiastic as expected. The New York Times put the news at the end of the 46th edition of the radio talk, with only a few short sentences; Academic magazines at that time were not very keen on this. Because the reaction at that time was not as strong as they thought, in April of 1952, in order to promote their invention, a public hearing was held again, just to announce their research results to the business community. At that time, they invited many American companies that manufacture vacuum tubes, and each company only had to pay $25,000 to attend the hearing, and promised that if his technology was adopted in the future, the admission fee of $25,000 for listening to the report could be deducted from it. At that time, about dozens of companies attended the hearing, but most of them were vacuum tubes and were not very interested in the significance of semiconductor transistors. Imagine that if the invention of the transistor is successfully applied, then the vacuum tube will slowly disappear. So from this perspective, it is understandable that their enthusiasm is not high. However, the scientific community still gave a high evaluation of this invention. 1956 Badin, Bourdon and shockley were awarded the Nobel Prize in Physics.
But today, the invention of transistor not only caused the revolution of electronic industry, but also completely changed our human production and lifestyle. Almost all the electrical appliances we use today do not use transistors, such as communications, computers, television, aerospace, aviation and so on.
semiconductor material
Today, semiconductors have been widely used in household appliances, communications, industrial manufacturing, aviation, aerospace and other fields. 194, the world market share of electronic industry was 69 1 100 million USD, and198 increased to 935.8 billion USD. Among them, due to the economic recession in the United States, the semiconductor market declined, from15 billion dollars in 1995 to13 billion dollars in 1998. After several years of wandering, the semiconductor market has rebounded.
Silicon single crystal and its epitaxy
At present, more than 90% of electronic components are made of silicon, and the global output value of silicon-related electronic industry is close to one trillion dollars. At present, silicon single crystal is mainly produced by Czochralski method. From 1950s to 1960s, the diameter of silicon single crystal was only two inches. At present, 8-inch, 12-inch and 1 meter long silicon single crystals have been produced on a large scale. 18 inch, that is, a silicon single crystal with a diameter of 45cm, has been successfully developed. The picture below is a photo of 12 inch Czochralski silicon, the length of which exceeds 1 m! (Editor's Note: Sketch)
At present, the annual output of monocrystalline silicon in the world has exceeded 1 10,000 tons. 8-inch silicon is mainly used in silicon integrated circuits, but the amount of 12-inch silicon is increasing year by year. It is estimated that by 20 12 years 18 inch silicon may be used in integrated circuit manufacturing, and the development of 27 inch silicon crystal is also planned.
Why does the diameter of silicon not develop from 8 inches, 10 inches, 12 inches and 14 inches, but from 8 inches to 12 inches, from 12 to 18 inches and from/kloc-0? The development of silicon integrated circuits follows Moore's law. Moore's Law means that the integration of integrated circuits doubles every 18 months, and its price decreases by half. So at present, in big cities, almost every household, even everyone has a PC, because the machine has good performance and low price. It is precisely because of the benefits brought by the increase of the diameter of silicon single crystal that the cost of the chip produced by using 12 inch silicon wafer in the production line is much lower than that produced by using 8 inch silicon wafer.
With the increase of silicon diameter, the distribution of impurities such as impurity oxygen in silicon ingot and silicon wafer becomes uneven, which will seriously affect the yield of integrated circuits, especially high integration circuits. In order to avoid the problems caused by oxygen precipitation, epitaxy can be used to solve them. What is extension? That is, a silicon single crystal wafer is used as a substrate, and then a layer of silicon, such as 2 microns, 1 micron, or 0.5 microns thick, is grown on it by gas phase reaction. The oxygen content in this layer of epitaxial silicon can be controlled below 10 16/cm3, and devices and circuits are made on epitaxial silicon instead of the original silicon single crystal, thus solving the problems caused by oxygen. Although the cost will increase, the integration and operation speed of integrated circuits have been significantly improved, which is an important direction of silicon technology development at present.
The development trend of silicon materials is to increase the diameter of silicon single crystal and develop in the direction of 12 inch and 18 inch from the point of view of improving the yield of integrated circuits and reducing costs; On the other hand, from the point of view of improving the speed and integration of silicon integrated circuits, it is the key to develop silicon epitaxial technology suitable for deep submicron or even nano-circuits and prepare high-quality silicon epitaxial materials. As mentioned above, the precipitation of oxygen in silicon single crystal will produce micro-defects. At present, the line width of integrated circuits has reached below 0. 1 micron. If the defect diameter is 1 micron or 0.5 micron, then a defect on a circuit chip will lead to the failure of the whole chip, which will seriously affect the yield of integrated circuits.
At present, the output of silicon single crystal in the world is about 10000 tons, and that in China is about 1000 tons per year. The raw material for preparing silicon single crystal is polysilicon, but the annual output of polysilicon in China is less than 100 ton, accounting for only a few thousandths of the world. Judging from the current development momentum of silicon materials in China, it is predicted that by 20 10, China's microelectronics technology will have a great development, which may reach about 20% of the world level. Judging from the line width of integrated circuits, the current level of integrated circuit technology in China is 0.35-0.25 micron, while the current international production technology has reached 0. 13-0.09 micron, and the 70-nanometer process in the laboratory has also passed the examination. The integrated circuit technology of SMIC, which was completed and put into production in Beijing last year, has reached 0. 13 micron and will be upgraded to 0.09 micron soon, so the gap between China's microelectronic integrated circuit technology and foreign countries has also been shortened to 1-2 generation.
Silicon microelectronics technology
Can silicon microelectronics technology develop forever according to Moore's law? At present, the mass production technology of silicon integrated circuits has reached 0. 13-0.09 micron, and will further reach 0.07 micron, that is, 70 nanometers or even less. It is predicted that by 2022, the line width of silicon integrated circuit technology may reach 10 nanometer, which is considered as the "physical limit" of silicon integrated circuits. In other words, if the size is reduced again, there will be many insurmountable problems. Of course, 10 nm mentioned here is not conclusive. With the development of technology, especially nano-machining technology, this "limit" size may be further reduced; But one day, contemporary silicon microelectronics technology will come to an end.
With the further reduction of integrated circuit linewidth, silicon microelectronics technology will inevitably encounter many insurmountable problems, such as the fluctuation of statistical distribution of doped atoms in CMOS device channels. For example, there are only about 65,438+000 doped atoms between the source electrode and the drain electrode with a length of 65,438+000 nm. How to ensure the distribution of 100 atoms in thousands of devices is obviously impossible, at least it is very difficult. That is to say, the fluctuation of impurity atom distribution will lead to different device performance and inconsistent properties, which makes it difficult to ensure the normal operation of the circuit. Another example is that the insulating layer under the gate of MOS device is silicon dioxide, and its thickness decreases with the decrease of device size. When the channel length reaches 0. 1 micron, the thickness of silicon dioxide is about one nanometer. Although the applied gate voltage is very low, such as 0.5 volts or 1 volt to 1 nanometer, the electric field intensity applied to it will reach more than 5- 10 megavolts per centimeter, which exceeds the breakdown voltage of the material. When this thickness is very thin, even if breakdown does not occur, the probability of electron tunneling is high, which will lead to device failure.
With the improvement of integrated circuit integration, the power consumption of the chip has also increased sharply, which is unbearable; At present, the power consumption of computer CPU is already very high. If it becomes a "nanostructure" in the future, that is, if we just follow Moore's Law and further improve the integration, the power consumption added to it may melt the silicon! Another problem is photolithography, which can reach about 0. 1 micron at present. Although there are some developing lithography technologies, such as X-ray lithography and ultra-ultraviolet lithography, they are still far from meeting the needs of nano-machining technology. Then there is the interconnection between circuit devices. Each chip has tens of millions to hundreds of millions of tubes per square centimeter, and the length of wires between tubes accounts for 60-70% of the device area. At present, there are as many as 8 layers to 10 layers of wires. Although the distance between the two tubes can be made small, the path of electrons from this tube to the other tube is not straight. We know that the narrower the line width, the smaller the cross section and the greater the resistance. Coupled with the distributed capacitance, it takes a long time for electrons to pass through the lead, which reduces the speed of CPU. In addition, the manufacturing cost of nano-machining is also very high. For these reasons, silicon-based microelectronics technology will eventually be unable to meet the growing information needs.
If people want to break through the above "physical limit", they must explore new principles and develop new technologies, such as quantum computing and optical computers. Their working principle is completely different from today, and they are still in the initial exploration stage. At present, in this transitional period, people hope to develop new semiconductor materials and new technologies, such as GaAs, InP and GaN-based material systems. Using these materials can improve the speed of devices and circuits and solve the problem of increased power consumption caused by the improvement of integration.
GaAs and InP single crystal materials
Compound semiconductor materials, taking gallium arsenide (GaAs) as an example, have the following characteristics: first, high luminous efficiency; Second, high electron mobility; At the same time, they can work in harsh environments such as high temperature, and are especially suitable for making ultra-high-speed, ultra-high-frequency and low-noise circuits; Another advantage of compound semiconductor material is that it can realize optoelectronic integration, that is, combining microelectronics with optoelectronics can greatly improve the function and operation speed of the circuit.
Wide band gap semiconductor material
Gallium nitride, silicon carbide, zinc oxide, etc. They are all wide band gap semiconductor materials, because their band gaps are all above 3 eV, and it is impossible to excite valence band electrons to the conduction band at room temperature. The working temperature of the device can be very high, for example, silicon carbide can work to 600 degrees Celsius; If diamond is made into semiconductor, the temperature can be higher, and the device can be used in oil drilling head to collect relevant information. They also have important applications in harsh environments such as aviation and aerospace. At present, the only high-power emitting tube in radio and TV stations is electron tube, which has not been replaced by semiconductor devices. This kind of electron tube has a life of only two or three thousand hours, and it is bulky and consumes a lot of electricity. If the high-power emitting device of silicon carbide is used, the volume can be reduced by at least tens to hundreds of times, and the service life will be greatly increased, so the high-temperature wide-band gap semiconductor material is a very important new semiconductor material.
The problem now is that this material is difficult to grow. Silicon grows on silicon and GaAs grows on gallium arsenide. It can grow well. But most of this material has no bulk material, so we have to use other materials as substrates to grow it. For example, when gallium nitride is grown on sapphire substrate, the thermal expansion coefficient and lattice constant of sapphire and gallium nitride are very different, and the grown epitaxial layer has many defects, which is the biggest problem and difficulty. In addition, the processing and etching of this material is also very difficult. At present, scientists are working hard to solve this problem. If this problem is solved, it can provide a very broad space for us to discover new materials.
Low-dimensional semiconductor materials
In fact, the low-dimensional semiconductor materials mentioned here are nano-materials. The reason why I don't want to use this word is mainly because I don't want to be confused with so-called nano shirts, nano beer bottles, nano washing machines and so on! In essence, an important purpose of developing nanotechnology is that people can control and manufacture powerful and superior nano-electrons, photoelectric devices and circuits, nano-biosensors and so on. Benefit mankind on the atomic, molecular or nano scale. It can be predicted that the development and application of nanotechnology will not only completely change people's production and lifestyle, but also change the social and political pattern and the form of war confrontation. This is why people attach great importance to the development of nano-semiconductor technology.
Electrons in bulk materials can move freely in three dimensions. However, when the characteristic size of the material is smaller than the average free path of electrons in one dimension, the movement of electrons in this direction is limited, and the energy of electrons is no longer continuous, but quantized. We call this material superlattice and quantum well material. Quantum wire material is that electrons can only move freely along the direction of quantum wire, but are restricted in the other two directions; Quantum dot material means that the size of the material in three-dimensional space is smaller than the average free path of electrons, electrons can't move freely in three directions, and energy is quantized in all three directions.
Due to the above reasons, the density of states function of electrons has also changed. The bulk material is a parabola on which electrons can move freely. If it is a quantum dot material, its state density function is completely isolated like a single molecule or atom. Based on this characteristic, powerful quantum devices can be manufactured.
At present, the memory of large-scale integrated circuits is realized by charging and discharging a large number of electrons. The flow of a large number of electrons requires a lot of energy, which causes the chip to heat up and limits the integration. If a memory composed of a single electron or several electrons is used, not only the integration level can be improved, but also the power consumption problem can be solved. At present, the efficiency of laser is not high, because the wavelength of laser changes with temperature. Generally speaking, with the increase of temperature, the wavelength will shift to red, so all lasers used in optical fiber communication now need to control the temperature. If the existing quantum well laser can be replaced by quantum dot laser, these problems can be solved easily.
Superlattice and quantum well materials based on GaAs and InP have been well developed and widely used in optical communication, mobile communication and microwave communication. Quantum cascade laser is a monopole device, a new type of medium and far infrared light source developed in recent ten years, which has important application prospects in free space communication, infrared countermeasure, remote chemical sensing and so on. It has high requirements for the preparation process of MBE. The whole device structure is several hundred to several thousand layers, and the thickness of each layer should be controlled at the accuracy of several tenths of a nanometer. China has made international leading achievements in this field; Another example is the interband quantum tunneling transport and optically coupled quantum well laser with multiple active regions, which has the characteristics of high quantum efficiency, high power and good beam quality and has a good research foundation in China; In the research of quantum dot (wire) materials and quantum dot lasers, it has also made achievements that attract the attention of international counterparts.
summary
From the development of semiconductor materials and information technology, the current information carrier is mainly electrons, that is, the charge (current) of electrons. There is another property of electrons, the spin of electrons, which we haven't used yet. If the spin of electrons is reused, it will increase a degree of freedom, which is also one of the research directions of people at present. We have developed from electronic materials such as silicon and germanium to photoelectric materials such as GaAs, InP and GaN. , is a material that can be used with electrons and photons, and photoelectric materials are more powerful than electronic materials. The next generation of materials is likely to be photonic materials. We only use the amplitude of photons now, but the polarization of light and the corresponding utilization of light have not yet been developed, so this has left us a very broad world for researchers. From the development of materials, from bulk materials to thin layers, ultra-thin layers, low-dimensional (nano) structural materials and functional chip materials; Functional chips may be a combination of organic and inorganic, or a combination of life and organic and inorganic, which also provides us with a very broad world of innovation. I believe people can make great achievements in this field in the future.