Chapter 1 AN ENVIRONMENT OF CHALLENGES 1.1 OVERVIEW Modern electronic materials and devices arguably are built upon nearly the entire periodic table (excluding only the actinides and a few other unusual or un- stable elements). These diverse materials are required to meet the intense challenges which electronic device applications present. In their full extent, electronic appli- cations range from simple copper wires, to high-performance magnetic materials for computer disks, to semiconductors for state-of-the-art microelectronic devices and many more. Likewise, the critical properties of the materials range from el tronic conductivity, to optical transmission, to diffusion-resistance or mechanical properties. It is not reasonable, nor is it particularly desirable, to cover all aspects of electronic materials in a single text. Consequently, the materials discussed here relate primarily to the most challenging applications, particularly with reference to microelectronic and optical devices. This volume is further restricted to the semiconducting materials used in active devices and leaves the metals, dielectrics and other materials used in microelectronic processes to other texts. a wide range considered including some traditional materials, such as silicon and some in their nancy, such as organic semiconductors. Readers may also wish to consider books on epitaxial growth, and other processes relevant to microelectronics manufacturing as supplements to this text The competitive nature of manufacturing of microelectronics has meant that completely new generations of devices and completely new electronic materials and processes have been required on a time scale of months rather than years or decades
Chapter 1 AN ENVIRONMENT OF CHALLENGES 1.1 OVERVIEW Modern electronic materials and devices arguably are built upon nearly the entire periodic table (excluding only the actinides and a few other unusual or unstable elements). These diverse materials are required to meet the intense challenges which electronic device applications present. In their full extent, electronic applications range from simple copper wires, to high-performance magnetic materials for computer disks, to semiconductors for state-of-the-art microelectronic devices, and many more. Likewise, the critical properties of the materials range from electronic conductivity, to optical transmission, to diffusion-resistance or mechanical properties. It is not reasonable, nor is it particularly desirable, to cover all aspects of electronic materials in a single text. Consequently, the materials discussed here relate primarily to the most challenging applications, particularly with reference to microelectronic and optical devices. This volume is further restricted to the semiconducting materials used in active devices and leaves the metals, dielectrics, and other materials used in microelectronic processes to other texts. A wide range is considered including some traditional materials, such as silicon, and some in their infancy, such as organic semiconductors. Readers may also wish to consider books on epitaxial growth, and other processes relevant to microelectronics manufacturing as supplements to this text. The competitive nature of manufacturing of microelectronics has meant that completely new generations of devices and completely new electronic materials and processes have been required on a time scale of months rather than years or decades. One consequence of this situation is that a book covering materials specific to the
The materials science of semiconductors current generation of devices would be out of date by the time it was written, let alone published and distributed. A more practical approach is to cover fundamental properties of classes of materials and processing methods that are relevant to both old and new device generations. This book attempts to cover these fundamentals but uses illustrations from current technology when reasonable As we will see, electronic applications have only one feature in common-they place higher demands for performance on their materials than any other class of products Working with these materials is truly working in an environment of challenges This chapter considers some of the issues currently facing the microelectronics engineer. It refers to constituents of the circuits without explanation, as this would disrupt the flow of the discussion. For the reader who is unsure of the terminology Chapter 2 presents a review of some of the basic physics of semiconductors. Selected active devices that are currently joined to produce an integrated circuit are described iefly in Chapter 3. Finally, Chapter 4 presents a brief review of concepts from aterials Science that are important in the discussions in later chapters. The details of the terminology are relatively unimportant for purposes of this chapter 1.2 A HISTORY OF MODERN ELECTRONIC DEVICES Modern electronic devices build on a long history of invention, discovery, and basic scientific research. The most critical devices are the control circuit elements-the diode and the switching devices. The latter began as triode vacuum tubes and are now generally transistors. Diodes pass current easily in only one direction. The riginal diode was invented in 1905 by J. Ambrose Flemming based on observations made in the laboratories of edison electric. This diode vacuum tube contained a hot lament, which emits electrons, and a metallic plate collector. Electrons flow only from filament to collector. The following year Lee DeForest created the triode vacuum tube and the electronic revolution was launched The vacuum triode consists of a heated cathode, an intervening wire grid and a plate or anode, and functions as a diode modified by the control grid. A small change in current at the grid produces a large change in current from cathode to anode. erefore, the triode allows amplification of weak signals. This ability to amplify is he essential element of both analog and digital circuits. Between 1906 and the mid 950s, vacuum tubes were developed and adapted to more and more specialized applications with more and more sophisticated internal structures to modify the electron current. Unfortunately, tubes, like incandescent light bulbs, have a very limited lifetime and consume large amounts of electricity, producing a large amount of heat. Even the development of miniature tubes could not overcome these problems. Vacuum tubes are still used in rare applications such as television picture tubes, very high-power amplifier stages in radio transmitters, and in environments where damage to transistors would occur and degrade their performance much faster than a vacuum-tube-based circuit would degrade. As devices included more and
2 current generation of devices would be out of date by the time it was written, let alone published and distributed. A more practical approach is to cover fundamental properties of classes of materials and processing methods that are relevant to both old and new device generations. This book attempts to cover these fundamentals but uses illustrations from current technology when reasonable. As we will see, electronic applications have only one feature in common – they place higher demands for performance on their materials than any other class of products. Working with these materials is truly working in an environment of challenges. This chapter considers some of the issues currently facing the microelectronics engineer. It refers to constituents of the circuits without explanation, as this would disrupt the flow of the discussion. For the reader who is unsure of the terminology, Chapter 2 presents a review of some of the basic physics of semiconductors. Selected active devices that are currently joined to produce an integrated circuit are described briefly in Chapter 3. Finally, Chapter 4 presents a brief review of concepts from Materials Science that are important in the discussions in later chapters. The details of the terminology are relatively unimportant for purposes of this chapter. 1.2 A HISTORY OF MODERN ELECTRONIC DEVICES Modern electronic devices build on a long history of invention, discovery, and basic scientific research. The most critical devices are the control circuit elements – the diode and the switching devices. The latter began as triode vacuum tubes and are now generally transistors. Diodes pass current easily in only one direction. The original diode was invented in 1905 by J. Ambrose Flemming based on observations made in the laboratories of Edison Electric. This diode vacuum tube contained a hot filament, which emits electrons, and a metallic plate collector. Electrons flow only from filament to collector. The following year Lee DeForest created the triode vacuum tube and the electronic revolution was launched. The vacuum triode consists of a heated cathode, an intervening wire grid and a plate or anode, and functions as a diode modified by the control grid. A small change in current at the grid produces a large change in current from cathode to anode. Therefore, the triode allows amplification of weak signals. This ability to amplify is the essential element of both analog and digital circuits. Between 1906 and the mid 1950’s, vacuum tubes were developed and adapted to more and more specialized applications with more and more sophisticated internal structures to modify the electron current. Unfortunately, tubes, like incandescent light bulbs, have a very limited lifetime and consume large amounts of electricity, producing a large amount of heat. Even the development of miniature tubes could not overcome these problems. Vacuum tubes are still used in rare applications such as television picture tubes, very high-power amplifier stages in radio transmitters, and in environments where damage to transistors would occur and degrade their performance much faster than a vacuum-tube-based circuit would degrade. As devices included more and The Materials Science of Semiconductors
An Environment of Challenges more tubes, their lifetime between servicing and their overall reliability decreased dramatically. This situation led researchers to seek new ways of producing a switching effect. The solution to the tube problem was found in the bipolar junction transistor, created in 1947 at the Bell Telephone Laboratories by John Bardeen, Walter H. Brattain, and William Shockley. The original device was produced from a lump of germanium and worked by virtue of diffusion of metals from the contacts into the Ge crystal. The device was found to control current effectively and yielded amplification as with vacuum tubes but contained no heated filament and consumed relatively little power As designs progressed the performance improved markedly. While the Ge transistor was revolutionary, it was not a practical solution for the long term. Ge has a relatively low energy gap, making it relatively conductive at room temperature. This allows current to leak backward through a device that is supposed to be turned off. Such leakage causes the entire circuit to consume large amounts of power at all times Selected Significant Events in the Development of Semiconductor Microelectronics. 1900-2000 1905 Vacuum tube diode invented by J. Ambrose Flemming 906 Triode vacuum tube invented by Lee DeForest 1916 Czochralski crystal growth technique invented by Jan Czochralski 1935 First patent issued on a field-effect transistor(Oskar Heil) 1938 Early reports of Si rectifiers by Hans Hollmann and Jurgen Rottgardt 1947 Transistor invented by Bardeen, Brattain, and Shockley 1951 First practical field effect transistor 1952 Single crystal Si produced 1954 SiO2 mask process developed 1958 First integrated circuit invented by Jack Kilby 1959 Planar processing methods, precursors of modern integrated circuit fabrication methods, created by Noyce and Moor 1960 First practical metal-oxide-silicon transistor 1960 First patent on a light emitting diode(Jw. Allen and P.E. Gibbons 1962 Transistor-transistor logic 1962 First practical visible light emitting diode 1962 First laser diode 1963 Complementary metal-oxide-silicon transistors provide lower power switching devices 968 Metal-oxide-semiconductor memory circuits introduced 1971 First microprocessor 1978 First continuously operating laser diode at room temperature 987 Polymer-based light emitters 1992 Er-doped fiber amplifier 1997 Introduction of Cu-based interconnects
An Environment of Challenges 3 more tubes, their lifetime between servicing and their overall reliability decreased dramatically. This situation led researchers to seek new ways of producing a switching effect. The solution to the tube problem was found in the bipolar junction transistor, created in 1947 at the Bell Telephone Laboratories by John Bardeen, Walter H. Brattain, and William Shockley. The original device was produced from a lump of germanium and worked by virtue of diffusion of metals from the contacts into the Ge crystal. The device was found to control current effectively and yielded amplification as with vacuum tubes but contained no heated filament and consumed relatively little power. As designs progressed the performance improved markedly. While the Ge transistor was revolutionary, it was not a practical solution for the long term. Ge has a relatively low energy gap, making it relatively conductive at room temperature. This allows current to leak backward through a device that is supposed to be turned off. Such leakage causes the entire circuit to consume large amounts of power at all times Selected Significant Events in the Development of Semiconductor Microelectronics, 1900-2000 1905 Vacuum tube diode invented by J. Ambrose Flemming 1906 Triode vacuum tube invented by Lee DeForest 1916 Czochralski crystal growth technique invented by Jan Czochralski 1935 First patent issued on a field-effect transistor (Oskar Heil) 1938 Early reports of Si rectifiers by Hans Hollmann and Jürgen Rottgardt 1947 Transistor invented by Bardeen, Brattain, and Shockley 1951 First practical field effect transistor 1952 Single crystal Si produced 1954 SiO2 mask process developed 1958 First integrated circuit invented by Jack Kilby 1959 Planar processing methods, precursors of modern integrated circuit fabrication methods, created by Noyce and Moore 1960 First practical metal-oxide-silicon transistor 1960 First patent on a light emitting diode (J.W. Allen and P.E. Gibbons) 1962 Transistor-transistor logic 1962 First practical visible light emitting diode 1962 First laser diode 1963 Complementary metal-oxide-silicon transistors provide lower power switching devices 1968 Metal-oxide-semiconductor memory circuits introduced 1971 First microprocessor 1978 First continuously operating laser diode at room temperature 1987 Polymer-based light emitters 1992 Er-doped fiber amplifier 1997 Introduction of Cu-based interconnects
The materials science of semiconductors and lowers the gain of the amplification that can be obtained. The solution to this dilemma was to exchange Ge for Si. The bipolar transistor has, following a switch from Ge to Si, become a ubiquitous element in modern circuits Another such device, the field effect transistor, was created at about the same time While patents on field-effect switching devices were filed as early as 1930, the first practical device was produced in 1951. The current control was based on the depletion of charge resulting from the presence of a reverse biased diode junction The field-effect transistor has a much higher control electrode resistance(that of a reverse biased diode or capacitor) compared to the bipolar transistor. This was a particular advantage because vacuum tubes can be configured to have a very high input resistance. Thus, the field-effect transistor provided a better replacement for vacuum tubes in many applications- for example, in amplifiers for low-power signal A switch to Si-based devices occurred rapidly as the technology for its produc tion improved. Silicon has a larger energy gap and consequently pure Si is less conductive at room temperature than Ge. This dramatically lowers reverse leakage current and circuit power. However, the major reasons why Si has remained the most popular semiconductor are the performance, stability and reproducibility insulating layers and contacts that can be produced on it. The developments needed for the use of Si in microelectronic devices included two major process improvements- methods for purification of the material, in particular the removal of problem impurities; and methods for growing large single crystals The latter had been solved as early as 1916 with the creation of the, now ubiquitous Czochralski method for bulk crystal growth, although many improvements and adaptations were needed before large Si crystals could be grown. As time has progressed and more has become known about the basic science of the Czochralski method, wafer sizes have increased from -25 mm in the 1960s to 300 mm today Purification presented a much greater problem. In the early 1950,s The Siemens Company developed a method based on reaction of Si with HCI to produce dichlorosilane, SiH2CI2, an easily evaporated liquid. The dichlorosilane is fractionally distilled and subsequently reduced in a reverse reaction to produce pure Si. (See also Chapter 4. )In 1952, w.G. Pfann created the zone refining method for further reduction of impurities, which was subsequently improved to reduce handling of the material between process steps. This method was used through the 1970s. Around that time, improvements in the initial purification process made zone refining largely unnecessary Miniaturization of electronic circuits took another major step forward with Jack Kilby's invention of the integrated circuit in 1958 at Texas Instruments and through he contributions of Robert Noyce and Gordon Moore at Fairchild Semiconductor (later founders of Intel Corporation). Together they developed methods for producing
4 and lowers the gain of the amplification that can be obtained. The solution to this dilemma was to exchange Ge for Si. The bipolar transistor has, following a switch from Ge to Si, become a ubiquitous element in modern circuits. Another such device, the field effect transistor, was created at about the same time. While patents on field-effect switching devices were filed as early as 1930, the first practical device was produced in 1951. The current control was based on the depletion of charge resulting from the presence of a reverse biased diode junction. The field-effect transistor has a much higher control electrode resistance (that of a reverse biased diode or capacitor) compared to the bipolar transistor. This was a particular advantage because vacuum tubes can be configured to have a very high input resistance. Thus, the field-effect transistor provided a better replacement for vacuum tubes in many applications – for example, in amplifiers for low-power signals. A switch to Si-based devices occurred rapidly as the technology for its production improved. Silicon has a larger energy gap and consequently pure Si is less conductive at room temperature than Ge. This dramatically lowers reverse leakage current and circuit power. However, the major reasons why Si has remained the most popular semiconductor are the performance, stability and reproducibility of insulating layers and contacts that can be produced on it. The developments needed for the use of Si in microelectronic devices included two major process improvements – methods for purification of the material, in particular the removal of problem impurities; and methods for growing large single crystals. The latter had been solved as early as 1916 with the creation of the, now ubiquitous, Czochralski method for bulk crystal growth, although many improvements and adaptations were needed before large Si crystals could be grown. As time has progressed and more has become known about the basic science of the Czochralski method, wafer sizes have increased from ~25 mm in the 1960’s to 300 mm today. Purification presented a much greater problem. In the early 1950’s The Siemens Company developed a method based on reaction of Si with HCl to produce dichlorosilane, SiH2Cl2, an easily evaporated liquid. The dichlorosilane is fractionally distilled and subsequently reduced in a reverse reaction to produce pure Si. (See also Chapter 4.) In 1952, W.G. Pfann created the zone refining method for further reduction of impurities, which was subsequently improved to reduce handling of the material between process steps. This method was used through the 1970’s. Around that time, improvements in the initial purification process made zone refining largely unnecessary. Miniaturization of electronic circuits took another major step forward with Jack Kilby’s invention of the integrated circuit in 1958 at Texas Instruments and through the contributions of Robert Noyce and Gordon Moore at Fairchild Semiconductor (later founders of Intel Corporation). Together they developed methods for producing The Materials Science of Semiconductors
An Environment of Challenges and interconnecting all of the basic elements of a circuit on a single piece of Si. This resulted in single electronic packages containing far more functionality in a compact form than could be obtained from discrete devices. It has proven much less expensive, much more reliable, and much faster to produce complex circuits from standard but individually complex integrated circuits than to wire the devices from discrete com- nents. Today, complete heterodyne radio receivers, video and audio signal proces- sors, amplifiers, and computing circuits are available on single integrated circuit hips. The power and complexity of these circuits has grown amazingly rapidly since the 1950s as discussed in the next section Not all of microelectronics is focused on integrated circuits, although that sector ertainly drives much of the field and receives most of the press attention. Optical devices are getting increasing notice as the information age drives a greater and greater need to transfer data. More mundanely, everywhere one looks; one finds small colored dots of light glowing to indicate readiness of electronic devices to serve our needs. These are primarily light-emitting diodes or LED's. The first patent on such devices was in 1955 by R. Braunstein describing electroluminescence from various semiconductors. The first practical visible light emitting diode had to wait for 1962 when N. Holonyak created a GaAs-GaP alloy or"Ga(As, P)device Likewise, the first patent on a semiconductor laser was in 1962, but a continuously operating laser diode that would run at room temperature waited until development of advanced heterojunction structures in the early 1970s. The progress in efficiency of light emitting devices is shown in Figure 1.1 The use of laser and light-emitting diodes has grown explosively in recent years with the current market estimated to be over 30 billion dollars. This growth has been fueled recently by the development of blue and green light emitters. These devices low permit the complete spectrum of colors to be generated, and hence allow full- color emissive displays to be produced. Early blue and green light emitters were based on the ll-VI family of compounds including materials such as ZnS and Cds hese proved unsatisfactory for a number of reasons. Several attempts were made to develop SiC-based devices. However, the true breakthrough was the discovery of a method for producing both p and n-type group Ill nitrides( Gan and related materials)by Nakamura and the demonstration of high-intensity blue and green light emitting devices in 1992. Laser diodes(based on GaAs and related compounds and primarily emitting in the infrared region) have proven essential in high capacity fiber-optic communications and compact-disk data storage systems. These devices are expensive, and require complex driving circuitry. One of the most recent breakthroughs in optical materials that greatly improved the performance of long-distance fiber-optic communication systems is the Er-doped silica amplifier. This device, at its root, is constructed simply by doping a section of optical fiber with Er atoms, providing higher energy light waves as a pump to excite the system, and allowing stimulated emission as
An Environment of Challenges 5 and interconnecting all of the basic elements of a circuit on a single piece of Si. This resulted in single electronic packages containing far more functionality in a compact form than could be obtained from discrete devices. It has proven much less expensive, much more reliable, and much faster to produce complex circuits from standard but individually complex integrated circuits than to wire the devices from discrete components. Today, complete heterodyne radio receivers, video and audio signal processors, amplifiers, and computing circuits are available on single integrated circuit chips. The power and complexity of these circuits has grown amazingly rapidly since the 1950’s, as discussed in the next section. Not all of microelectronics is focused on integrated circuits, although that sector certainly drives much of the field and receives most of the press attention. Optical devices are getting increasing notice as the information age drives a greater and greater need to transfer data. More mundanely, everywhere one looks; one finds small colored dots of light glowing to indicate readiness of electronic devices to serve our needs. These are primarily light-emitting diodes or LED’s. The first patent on such devices was in 1955 by R. Braunstein describing electroluminescence from various semiconductors. The first practical visible light emitting diode had to wait for 1962 when N. Holonyak created a GaAs-GaP alloy or “Ga(As,P)” device. Likewise, the first patent on a semiconductor laser was in 1962, but a continuously operating laser diode that would run at room temperature waited until development of advanced heterojunction structures in the early 1970’s. The progress in efficiency of light emitting devices is shown in Figure 1.1. The use of laser and light-emitting diodes has grown explosively in recent years with the current market estimated to be over 30 billion dollars. This growth has been fueled recently by the development of blue and green light emitters. These devices now permit the complete spectrum of colors to be generated, and hence allow fullcolor emissive displays to be produced. Early blue and green light emitters were based on the II-VI family of compounds including materials such as ZnS and CdS. These proved unsatisfactory for a number of reasons. Several attempts were made to develop SiC-based devices. However, the true breakthrough was the discovery of a method for producing both p and n-type group III nitrides (GaN and related materials) by Nakamura and the demonstration of high-intensity blue and green light emitting devices in 1992. Laser diodes (based on GaAs and related compounds and primarily emitting in the infrared region) have proven essential in high capacity fiber-optic communications and compact-disk data storage systems. These devices are expensive, and require complex driving circuitry. One of the most recent breakthroughs in optical materials that greatly improved the performance of long-distance fiber-optic communication systems is the Er-doped silica amplifier. This device, at its root, is constructed simply by doping a section of optical fiber with Er atoms, providing higher energy light waves as a pump to excite the system, and allowing stimulated emission as