电子工程师手册（英文版二）：ch031

Hecht, J., Watkins, L.S., Becker, R.A. “Optoelectronics” The Electrical Engineering Handbook Ed. Richard C. Dorf Boca Raton: CRC Press LLC, 2000 31 Optoelectronics 31.1 Lasers Differences from Other Light Sources?The Laser Industry 31.2 Sources and Detectors Properties of Light?Absorption?Coherence?Geometric Optics?Incoherent Light?Detectors, Semiconductor?Detectors, Photoemissive?Imaging Detectors?Noise and Detectivity 31.3 Circuits Integrated Optics?Device Fabrication?Packaging?Applications 31.1 Lasers 1 Jeff Hecht The word laser is an acronym for “light amplification by the stimulated emission of radiation,” a phrase that covers most, though not all, of the key physical processes inside a laser. Unfortunately, that concise definition may not be very enlightening to the nonspecialist who wants to use a laser and cares less about its internal physics than its external characteristics. From a practical standpoint, a laser can be considered as a source of a narrow beam of monochromatic, coherent light in the visible, infrared, or ultraviolet parts of the spectrum. The power in a continuous beam can range from a fraction of a milliwatt to around 25 kilowatts (kW) in commercial lasers, and up to more than a megawatt in special military lasers. Pulsed lasers can deliver much higher peak powers during a pulse, although the power averaged over intervals while the laser is off and on is comparable to that of continuous lasers. The range of laser devices is broad. The laser medium, or material emitting the laser beam, can be a gas, liquid, glass, crystalline solid, or semiconductor crystal and can range in size from a grain of salt to filling the inside of a moderate-sized building. Not every laser produces a narrow beam of monochromatic, coherent light. Semiconductor diode lasers, for example, produce beams that spread out over an angle of 20 to 40 degrees, hardly a pencil-thin beam. Liquid dye lasers emit at a broad or narrow range of wavelengths, depending on the optics used with them. Other types emit at a number of spectral lines, producing light that is neither truly monochromatic nor coherent. Table 31.1 summarizes important commercial lasers. Practically speaking, lasers contain three key elements. One is the laser medium itself, which generates the laser light. A second in the power supply, which delivers energy to the laser medium in the form needed to excite it to emit light. The third is the optical cavity or resonator, which concentrates the light to stimulate the emission of laser radiation. All three elements can take various forms, and although they are not always immediately evident in all types of lasers, their functions are essential. Figure 31.1 shows these elements in a ruby and a helium-neon laser. Laser-like devices called optical parametric oscillators have come into increasing use. They are more costly and complex than lasers, but can be tuned across a broad range, with wavelengths from 0.2 to 4 micrometers. 1 Modified from J. Hecht, The Laser Guidebook, 2nd ed., New York: McGraw-Hill, 1991. With permission. Jeff Hecht Laser Focus World Laurence S. Watkins Lucent Technologies R.A. Becker Integrated Optical Circuit Consultants ? 2000 by CRC Press LLC TABLE 31.1 Important Commercial Lasers Wavelength (mm) Type Output Type and Power 0.157 Molecular fluorine (F 2 ) Pulsed, avg. to a few watts 0.192 ArF excimer Pulsed, avg. to tens of watts 0.2–0.35 Doubled dye Pulsed 0.235–0.3 Tripled Ti-sapphire Pulsed 0.24–0.27 Tripled alexandrite Pulsed 0.248 KrF excimer Pulsed, avg. to over 100 W 0.266 Quadrupled Nd Pulsed, watts 0.275–0.306 Argon-ion CW, 1-W range 0.308 XeCl excimer Pulsed, to tens of watts 0.32–1.0 Pulsed dye Pulsed, to tens of watts 0.325 He-Cd CW, to tens of milliwatts 0.337 Nitrogen Pulsed, under 1 W avg. 0.35–0.47 Doubled Ti-sapphire Pulsed 0.351 XeF excimer Pulsed, to tens of watts 0.355 Tripled Nd Pulsed, to tens of watts 0.36–0.4 Doubled alexandrite Pulsed, watts 0.37–1.0 CW dye CW, to a few watts 0.442 He-Cd CW, to over 0.1 W 0.45–0.53 Ar-ion CW, to tens of watts 0.51 Copper vapor Pulsed, tens of watts 0.520–0.569 Kryption ion CW, >1W 0.523 Doubled Nd-YLF Pulsed, watts 0.532 Doubled Nd-YAG Pulsed to 50 W, or CW to watts 0.5435 He-Ne CW, 1-mW range 0.578 Copper vapor Pulsed, tens of watts 0.594 He-Ne CW, to several milliwatts 0.612 He-Ne CW, to several milliwatts 0.628 Gold vapor Pulsed 0.6328 He-Ne CW, to about 50 mW 0.635–0.66 InGaAlP diode CW, milliwatts 0.647–0.676 Krypton ion CW, to several watts 0.67 GaInP diode CW, to 10 mW 0.68–1.13 Ti-sapphire CW, watts 0.694 Ruby Pulsed, to a few watts 0.72–0.8 Alexandrite Pulsed, to tens of watts (CW in lab) 0.75–0.9 GaAlAs diode CW, to many watts in arrays 0.98 InGaAs diode CW, to 50 mW 1.047 or 1.053 Nd-YLF CW or pulsed, to tens of watts 1.061 Nd-glass Pulsed, to 100 W 1.064 Nd-YAG CW or pulsed, to kilowatts 1.15 He-Ne CW, milliwatts 1.2–1.4 InGaAsP diode CW, to 100 mW 1.313 Nd-YLF CW or pulsed, to 0.1 W 1.32 Nd-YAG Pulsed or CW, to a few watts 1.4–1.6 Color center CW, under 1 W 1.5–1.6 InGaAsP diode CW, to 100 mW 1.523 He-Ne CW, milliwatts 1.54 Erbium-glass (bulk) Pulsed, to 1 W 1.54 Erbium-fiber (amplifier) CW, milliwatts 1.75–2.5 Cobalt-MgF 2 Pulsed, 1-W range 2.3–3.3 Color center CW, under 1 W 2.6–3.0 HF chemical CW or pulsed, to hundreds of watts 3.3–29 Lead-salt diode CW, milliwatt range 3.39 He-Ne CW, to tens of milliwatts 3.6–4.0 DF chemical CW or pulsed, to hundreds of watts 5–6 Carbon monoxide CW, to tens of watts 9–11 Carbon dioxide CW or pulsed, to tens of kilowatts 40–100 Far-infrared gas CW, generally under 1 W ? 2000 by CRC Press LLC Several general characteristics are common to most lasers that new users may not expect. Like most other light sources, lasers are inefficient in converting input energy into light. Efficiencies range from less than 0.001 to more than 50%, but except for semiconductor lasers, few types are much above 1% efficient. These low efficiencies can lead to special cooling requirements and duty-cycle limitations, particularly for high-power lasers. In some cases, special equipment may be needed to produce the right conditions for laser operation, such as cryogenic temperatures for lead salt semiconductor lasers. Operating characteristics of individual lasers depend strongly on structural components such as cavity optics, and in many cases a wide range is possible. Packaging can also have a strong impact on laser characteristics and the use of lasers for certain applications. Thus, there are wide ranges of possible characteristics, although single devices will have much more limited ranges of operation. Differences from Other Light Sources The basic differences between lasers and other light sources are the characteristics often used to describe a laser: the output beam is narrow, the light is monochromatic, and the emission is coherent. Each of these features is important for certain applications and deserves more explanation. Most gas or solid-state lasers emit beams with divergence angle of about a milliradian, meaning that they spread to about 1 m in diameter after traveling a kilometer. (Semiconductor lasers have much larger beam divergence, but suitable optics can reshape the beam to make it much narrower.) The actual beam divergence depends on the type of laser and the optics used with it. The fact that laser light is contained in a beam serves to concentrate the output power onto a small area. Thus, a modest laser power can produce a high intensity inside the small area of the laser beam; the intensity of light in a 1-mW helium-neon laser beam is comparable to that of sunlight on a clear day, for example. The beams from high-power lasers, delivering tens of watts or more of continuous power or higher peak powers in pulses, can be concentrated to high enough intensities that they can weld, drill, or cut many materials. Figure 31.1 Simplified views of two common lasers, (a) ruby and (b) helium-neon, showing the basic components that make a laser. Electrode Mirror (defines laser cavity) laser beam Helium - neon gas mixture (laser medium) Power supply (drives discharge through laser gas) Power supply (drives flash lamp) (b) (a) Mirror (defines laser cavity) Electrode Ruby rod (laser medium) Flash lamp (excites laser rod) Mirror laser beam Mirror ? 2000 by CRC Press LLC The laser beam’s concentrated light delivers energy only where it is focused. For example, a tightly focused laser beam can write a spot on a light-sensitive material without exposing the adjacent area, allowing high- resolution printing. Similarly, the beam from a surgical laser can be focused onto a tiny spot for microsurgery, without heating or damaging surrounding tissue. Lenses can focus the parallel rays in a laser beam to a much smaller spot than they can the diverging rays from a point source, a factor that helps compensate for the limited light-production efficiency of lasers. Most lasers deliver a beam that contains only a narrow range of wavelengths, and thus the beam can be considered monochromatic for all practical purposes. Conventional light sources, in contrast, emit light over much of the visible and infrared spectrum. For most applications, the range of wavelengths emitted by lasers is narrow enough to make life easier for designers by avoiding the need for achromatic optics and simplifying the task of understanding the interactions between laser beam and target. For some applications in spectroscopy and communications, however, that range of wavelengths is not narrow enough, and special line-narrowing options may be required. One of the beam’s unique properties is its coherence, the property that the light waves it contains are in phase with one another. Strictly speaking, all light sources have a finite coherence length, or distance over which the light they produce is in phase. However, for conventional light sources that distance is essentially zero. For many common lasers, it is a fraction of a meter or more, allowing their use for applications requiring coherent light. The most important of these applications is probably holography, although coherence is useful in some types of spectroscopy, and there is growing interest in communications using coherent light. Some types of lasers have two other advantages over other light sources: higher power and longer lifetime. For some high-power semiconductor lasers, lifetime must be traded off against higher power, but for most others the life vs. power trade-off is minimal. The combination of high power and strong directionality makes certain lasers the logical choice to deliver high light intensities to small areas. For some applications, lasers offer longer lifetimes than do other light sources of comparable brightness and cost. In addition, despite their low efficiency, some lasers may be more efficient in converting energy to light than other light sources. The Laser Industry Commercial Lasers There is a big difference between the world of laser research and the world of the commercial laser industry. Unfortunately, many text and reference books fail to differentiate between types of lasers that can be built in the laboratory and those that are readily available commercially. That distinction is a crucial one for laser users. Laser emission has been obtained from hundreds of materials at many thousands of emission lines in laboratories around the world. Extensive tabulations of these laser lines are available [Weber, 1982], and even today researchers are adding more lines to the list. However, most of these laser lines are of purely academic interest. Many are weak lines close to much stronger lines that dominate the emission in practical lasers. Most of the lasers that have been demonstrated in the laboratory have proved to be cumbersome to operate, low in power, inefficient, and/or simply less practical to use than other types. Only a couple of dozen types of lasers have proved to be commercially viable on any significant scale; these are summarized in Table 31.1. Some of these types, notably the ruby and helium-neon lasers, have been around since the beginning of the laser era. Others, such as vibronic solid-state, are promising newcomers. The family of commercial lasers is expanding slowly, as new types such as titanium-sapphire come on the market, but with the economics of production a factor to be considered, the number of commercially viable lasers will always be limited. There are many possible reasons why certain lasers do not find their way onto the market. Some require exotic operating conditions or laser media, such as high temperatures or highly reactive metal vapors. Some emit only feeble powers. Others have only limited applications, particularly lasers emitting low powers in the far-infrared or in parts of the infrared where the atmosphere is opaque. Some simply cannot compete with materials already on the market. ? 2000 by CRC Press LLC Defining Terms Coherence: The condition of light waves that stay in the same phase relative to each other; they must have the same wavelength. Continuous wave (CW): A laser that emits a steady beam rather than pulses. Laser medium: The material in a laser that emits light; it may be a gas, solid, or liquid. Monochromatic: Of a single wavelength or frequency. Resonator: Mirrors that reflect light back and forth through a laser medium, usually on opposite ends of a rod, tube, or semiconductor wafer. One mirror lets some light escape to form the laser beam. Solid-state laser: A laser in which light is emitted by atoms in a glass or crystalline matrix. Laser specialists do not consider semiconductor lasers to be solid-state types. Related Topic 42.1 Lightwave Waveguides References J. Hecht, The Laser Guidebook, 2nd ed., New York: McGraw-Hill, 1991; this section is excerpted from the introduction. M. J. Weber (ed.), CRC Handbook of Laser Science and Technology (2 vols.), Boca Raton, Fla.: CRC Press, 1982. M. J. Weber (ed.), CRC Handbook of Laser Science and Technology, Supplement 1, Boca Raton, Fla.: CRC Press, 1989; other supplements are in preparation. Further Information Several excellent introductory college texts are available that concentrate on laser principles. These include: Anthony E. Siegman, Lasers, University Science Books, Mill Valley, Calif., 1986, and Orzio Svelto, Principles of Lasers, 3rd ed., Plenum, New York, 1989. Three trade magazines serve the laser field; each publishes an annual directory issue. For further information contact: Laser Focus World, PennWell Publishing, Ten Tara Blvd., Nashua, NH 03062; Lasers & Optronics, PO Box 650, Morris Plains, N.J. 07950-0650; or Photonics Spectra, Laurin Publishing Co., Berkshire Common, PO Box 1146, Pittsfield, Mass. 01202. Write the publishers for information. 31.2 Sources and Detectors Laurence S. Watkins Properties of Light The strict definition of light is electromagnetic radiation to which the eye is sensitive. Optical devices, however, can operate over a larger range of the electromagnetic spectrum, and so the term usually refers to devices which can operate in some part of the spectrum from the near ultraviolet (UV) through the visible range to the near infrared. Figure 31.2 shows the whole spectrum and delineates these ranges. Optical radiation is electromagnetic radiation and so obeys and can be completely described by Maxwell’s equations. We will not discuss this analysis here but just review the important properties of light. Phase Velocity In isotropic media light propagates as transverse electromagnetic (TEM) waves. The electric and magnetic field vectors are perpendicular to the propagation direction and orthogonal to each other. The velocity of light propagation in a medium (the velocity of planes of constant phase, i.e., wavefronts) is given by ? 2000 by CRC Press LLC (31.1) where c is the velocity of light in a vacuum (c = 299,796 km/s). The denominator in Eq. (31.1) is a term in optics called the refractive index of the medium (31.2) where e is the dielectric constant (permittivity) and m is the magnetic permeability. The wavelength of light, l, which is the distance between phase fronts is (31.3) where l 0 is the wavelength in vacuum and u is the light frequency. The refractive index varies with wavelength, and this is referred to as the dispersive property of a medium. Another parameter used to describe light frequency is wave number. This is given by (31.4) and is usually expressed in cm –1 , giving the number of waves in a 1-cm path. Group Velocity When traveling in a medium, the velocity of energy transmission (e.g., a light pulse) is less than c and is given by (31.5) In vacuum the phase and group velocities are the same. FIGURE 31.2 Electromagnetic spectrum showing visible and optical wavelengths. v c = em n =em l l u == 0 n v s l = 1 uv dv d =-l l ? 2000 by CRC Press LLC Polarization Light polarization is defined by the direction of the electric field vector. For isotropic media this direction is perpendicular to the propagation direction. It can exist in a number of states, described as follows. Unpolarized. The electric field vector has a random and constantly changing direction, and when there are multiple frequencies the vector directions are different for each frequency. Linear. The electric field vector is confined to one direction. Elliptical. The electric field vector rotates, either left hand or right hand, at the light frequency. The magnitude of the vector (intensity of the light) traces out an ellipse. Circular. Circular is the special case of the above where the electric field vector traces out a circle. Absorption Light in traveling through media can be absorbed. This can be represented in two ways. The light flux propagating through a medium can be written as (31.6) where x is the distance through the medium with incident light flux I 0 . a is the absorption coefficient, usually stated in cm –1 . An alternative way of describing absorption is to use the imaginary term in the media refractive index. The complex refractive index is (31.7) where k is the attenuation index. a and k are related as (31.8) Coherence Light can be partially or fully coherent or incoherent, depending on the source and subsequent filtering operations. Common sources of light are incoherent because they consist of many independent radiators. An example of this is the fluorescent lamp in which each excited atom radiates light independently. There is no fixed phase relationship between the waves from these atoms. In a laser the light is generated in a resonant IIe x = - 0 a nn ik=+()1 a p l = 4 0 nk ? 2000 by CRC Press LLC cavity using a light amplifier and the resulting coherent light has well-defined phase fronts and frequency characteristics. Spatial and Temporal Coherence. Spatial coherence describes the phase front properties of light. A beam from a single-mode laser which has one well-defined phase front is fully spatially coherent. A collection of light waves from a number of light emitters is incoherent because the resulting phase front has a randomly indefinable form. Temporal coherence describes the frequency properties of light. A single-frequency laser output is fully temporally coherent. White light, which contains many frequency components, is incoherent, and a narrow band of frequencies is partially coherent. Laser Beam Focusing The radial intensity profile of a collimated single-mode TEM 00 (Gaussian) beam from a laser is given by (31.9) where w 0 is the beam radius (1/e 2 intensity). This beam will diverge as it propagates out from the laser, and the half angle of the divergence is given by (31.10) When this beam is focused by a lens the resulting light spot radius is given by (31.11) where l is the distance from the lens to the position of the focused spot and w d is the beam radius entering the lens. It should be noted that l @ f, the lens focal length, for a collimated beam entering the lens. However, l will be a greater distance than f if the beam is diverging when entering the lens. Geometric Optics The wavelength of light can be approximated to zero for many situations. This permits light to be described in terms of light rays which travel in the direction of the wave normal. This branch of optics is referred to geometric optics. Properties of Light Rays Refraction. When light travels from one medium into another it changes propagation velocity, Eq. (31.1). This results in refraction (bending) of the light as shown in Fig. 31.3. The change in propagation direction of the light ray is given by Snell’s law: (31.12) where n 1 and n 2 are the refractive indices of media 1 and 2, respectively. Critical Angle.When a light ray traveling in a medium is incident on a surface of a less dense medium, there is an incidence angle q 2 , where sin q 1 = 1. This is the critical angle; for light incident at angles greater than q 2 the light is totally internally reflected as shown in Fig. 31.3(b). The critical angle is given by q c = sin –1 (n 1 /n 2 ). Ir I r w () exp= - ? è ? ? ? ÷ é ? ê ê ù ? ú ú 0 2 0 2 2 q l p 12 0 / = w w l w f d = l p nn 12 sin sin 12 qq= ? 2000 by CRC Press LLC Image Formation with a Lens Many applications require a lens to focus light or to form an image onto a detector. A well-corrected lens usually consists of a number of lens elements in a mount, and this can be treated as a black box system. The characteristics of this lens are known as the cardinal points. Figure 31.4 shows how a lens is used to form an image from an illuminated object. The equation which relates the object, image, and lens system is (31.13) The image magnification is given by M = s 2 /s 1 . When the object is very far away s 1 is infinite and the image is formed at the back focal plane. Incoherent Light When two or more incoherent light beams are combined, the resulting light flux is the sum of their energies. For coherent light this is not necessarily true and the resulting light intensity depends on the phase relationships between the electric fields of the two beams, as well as the degree of coherence. FIGURE 31.3(a) Diagram of a light ray in medium 1 incident at angle q 1 on the surface to medium 2. The ray is refracted at angle q 2 . (b) Diagram of the situation when the ray in medium 2 is incident at an angle greater than the critical angle q c and totally internally reflected. FIGURE 31.4Schematic of an optical system forming an image of an object. Light rays from the object are captured by the lens which focuses them to form the image. EFL, effective focal length, ?, of the lens; FFL and BFL, distances from the focal points to the outer lens surface. Principal planes are the positions to which the focal points, object distance, and image distance are measured; in a simple lens they are coincident. 111 12 fss =+ ? 2000 by CRC Press LLC Brightness and Illumination The flux density of a light beam emitted from a point source decreases with the square of distance from it. Light sources are typically extended sources (being larger than point sources). The illumination of a surface from light emitted from an extended source can be calculated using Fig. 31.5. The flux incident on a surface element dA from a source element dS is given by (31.14) The constant B is called the luminance or photometric brightness of the source. Its units are candles per square meter (1 stilb = p lamberts) and dE is the luminous flux in lumens. The total illumination E of the surface element is calculated by integrating over the source. The illuminance or flux density on the surface is thus (31.15) Two methods are commonly used for quantifying light energy, namely, the radiometric unit of watts and the photometric unit of candelas. The candela is an energy unit which is derived from light emission from a blackbody source. The two can be related using the relative visibility curve V(l), which describes the eye’s sensitivity to the visible light spectrum, it being maximum near a wavelength of 550 nm. The constant which relates lumens to watts at this wavelength is 685 lm/W. The luminous flux emitted by a source can therefore be written as (31.16) where V is the spectral response of the eye and P is the source radiant intensity in watts. The source radiance is normally stated as luminance in candle per square centimeter (1 lumen per steradian per square centimeter) or radiance in watts per square centimeter per steradian per nanometer. The lumen is defined as the luminous flux emitted into a solid angle of a steradian by a point source of intensity 1/60th that of a 1-cm 2 blackbody source held at 2042 K temperature (molten platinum). FIGURE 31.5Surface being illuminated by an extended source. Illumination of surface element dA is calculated by summing the effects of elements dS. dE BdA dS r = cos cosqy 2 I E dA = )(lumens/cm 2 FVPd= ò 685 ()() (llllumens) ? 2000 by CRC Press LLC Thermal Sources Objects emit and absorb radiation, and as their temperature is increased the amount of radiation emitted increases. In addition, the spectral distribution changes, with proportionally more radiation emitted at shorter wavelengths. A blackbody is defined as a surface which absorbs all radiation incident upon it, and Kirchhoff’s law of radiation is given by (31.17) stating that the ratio of emitted to absorbed radiation is a constant a at a given temperature. The energy or wavelength distribution for a blackbody is given by Planck’s law (31.18) T is in degrees Kelvin, l is in micrometers, and W is the power emitted into a hemisphere direction. Blackbody radiation is incoherent, with atoms or molecules emitting radiation independently. Figure 31.6 is a plot of the blackbody radiation spectrum for a series of temperatures. Very few materials are true blackbodies; carbon lampblack is one. For this reason a surface emissivity is used which describes the ratio of actual radiation emitted to that from a perfect blackbody. Table 31.2 is a listing of emissivities for some common materials. Tungsten Filament Lamp In the standard incandescent lamp a tungsten filament is heated to greater than 2000°C, and it is protected from oxidation and vaporization by an inert gas. In a quartz halogen lamp the envelope is quartz, which allows FIGURE 31.6Plot of blackbody radiation for a series of temperatures. Radiation is in watts into a hemisphere direction from a 1-cm 2 of surface in a 1-mm wavelength band. W a WB==constant W cc T c c = ? è ? ? ? ÷ - é ? ê ê ù ? ú ú =′ =′ - 1 5 2 1 4 4 1 37413 10 14380 10 l l mexp . . (watts/cm area per m wavelength) 2 1 2 ? 2000 by CRC Press LLC the filament to run at a higher temperature. This increases the light output and gives a whiter wavelength spectrum with proportionally more visible radiation to infrared. Standard Light Source—Equivalent Black Blackbody Because the emissivity of incandescent materials is less than 1, an equivalent source is needed for measurement and calibration purposes. This is formed by using an enclosed space which has a small opening in it. Provided the opening is much smaller than the enclosed area, the radiation from the opening will be nearly equal to that from a blackbody at the same temperature, as long as the interior surface emissivity is > 0.5. Blackbody radiation from such a source at the melting point of platinum is defined as 1/60 cd/cm 2 . Arc Lamp A gas can be heated to temperatures of 6000 K or more by generating an electric arc between two electrodes. The actual resulting temperature is dependent on the current flowing through the arc, the gas pressure and its composition, and other factors. This does provide a light source which is close to the temperature of the sun. Using an inert xenon gas results in essentially a white light spectrum. The use of a gas such as mercury gives more light in the UV as well as a number of strong peak light intensities at certain wavelengths. This is due to excitation and fluorescence of the mercury atoms. Fluorescent Lamp A fluorescent source is a container (transparent envelope) in which a gas is excited by either a dc discharge or an RF excitation. The excitation causes the electrons of the gas to move to higher energy orbits, raising the atoms to a higher excited state. When the atoms relax to lower states they give off energy, and some of this energy can be light. The wavelength of the light is characteristically related to the energy levels of the excited states of the gas involved. Typically a number of different wavelengths are associated with a particular gas. Low-pressure lamps have relatively low luminance but provide light with narrow linewidths and stable spectral wavelengths. If only one wavelength is required, then optical filters can be used to isolate it by blocking the unwanted wavelengths. Higher luminance is achieved by using higher gas pressures. The fluorescent lamp is very efficient since a high proportion of the input electrical energy is converted to light. White light is achieved by coating the inside of the container with various types of phosphor. The gas, for example a mercury–argon mixture, provides UV and violet radiation which excites the phosphor. Since the light is produced by fluorescence and phosphores- cence, the spectral content of the light does not follow Planck’s radiation law but is characteristic of the coating (e.g., soft white, cool white). Light-Emitting Diodes (LED) Light can be emitted from a semiconductor material when an electron and hole pair recombine. This is most efficient in a direct gap semiconductor like GaAs and the emitted photons have energy close to the bandgap energy E g . The wavelength is then given by (31.19) TABLE 31.2 Emissivities of Some Common Materials Material Temperature (°C) Emissivity Tungsten 2000 0.28 Nickel-chromium (80-20) 600 0.87 Lampblack 20–400 0.96 Polished silver 200 0.02 Glass 1000 0.72 Platinum 600 0.1 Graphite 3600 0.8 Aluminum (oxidized) 600 0.16 Carbon filament 1400 0.53 l@ hc E g ? 2000 by CRC Press LLC where h is Planck’s constant (6.626 ′ 10 –34 J-s) and c the velocity of light in vacuum. The spectral width of the emission is quite broad, a few hundred nanometers, and is a function of the density of states, transition probabilities, and temperature. For light emission to occur, the conduction band must be populated with many electrons. This is achieved by forward biasing a pn junction to inject electrons and holes into the junction region as shown in Fig. 31.7. Figure 31.8(a) shows the cross section of a surface emitting LED with an integral lens fabricated into the surface. The light from the LED is incoherent and emitted in all directions. The lens and the bottom reflecting surface increase the amount of light transmitted out of the front of the device. The output from the LED is approximately linear with current but does decrease with increasing junction temperature. Figure 31.8(b) shows an edge emitting LED. Here the light is generated in a waveguide region which confines the light, giving a more directional output beam. Various wavelengths are available and are obtained by using different bandgap semiconductors. This is done by choosing different binary, ternary, and quaternary compositions. Table 31.3 is a listing of the more common ones. The output power is usually specified in milliwatts per milliamp current obtained in a given measurement situation, e.g., into a fiber or with a 0.5 numerical aperture large area detector. Other parameters are peak wavelength, wavelength band (usually full width half max), and temperature characteristics. FIGURE 31.7Band structure of a double heterostructure LED. Forward bias injects holes and electrons into the junction region where they recombine and emit light. FIGURE 31.8Cross-sectional diagrams of (a) surface emitting LED and (b) edge emitting LED. The light output from the edge emitter is more directional because of confinement by the junction guide region. ? 2000 by CRC Press LLC LEDs for Fiber Optic Communications GaAs and InGaAsP LEDs are commonly used as sources for fiber optic communications. Since they are an incoherent source, it is only practical to use them with multimode fiber. Only multimode fiber has a large enough core diameter and numerical aperture (NA) to couple in enough light to be able to propagate any useful distance. Applications for LEDs in fiber optics are for short distance links using glass or plastic fiber at relatively low bandwidths, typically in the Mb/s rather than Gb/s. Primary applications of these are for low cost datalinks. The detector can be packaged two ways: first with a fiber pigtail directly attached to the detector package; or a more common package is to have a fiber connector molded in as part of the package so that a connectorized fiber can be plugged in to it. Many LEDs for fiber optics are now packaged with electronic drive circuits to form a transmitter module ready to receive standard format data signals. Detectors, Semiconductor When light interacts electronically with a medium, by changing the energy of electrons or creating carriers, for example, it interacts in a quantized manner. The light energy can be quantized according to Planck’s theory (31.20) where u is the light frequency and h is Planck’s constant. The energy of each photon is very small; however, it does increase with shorter wavelengths. Photoconductors Semiconductors can act as photoconductors, where incident light increases the carrier density, thus increasing the conductivity. There are two basic types, intrinsic and extrinsic. Figure 31.9 shows a simple energy diagram containing conduction and valence bands. Also indicated are the levels which occur with the introduction of donor and acceptor impurities. Intrinsic photoconduction effect is when a photon with energy hu, which is greater than the bandgap energy, excites an electron from the valence band into the conduction band, creating a hole–electron pair. This increases the conductivity of the material. The spectral response of this type of detector is governed by the bandgap of the semiconductor. In an extrinsic photoconductor (see Fig. 31.9), the photon excites an electron from the valence band into the acceptor level corresponding to the hole of the acceptor atom. The resulting energy hu is much smaller than the bandgap and is the reason why these detectors have applications for long wavelength infrared sensors. Table 31.4 is a list of commercial photoconductors and their peak wavelength sensitivities. The doping material in the semiconductor determines the acceptor energy level, and so both the host material and the dopant are named. Since the energy level is quite small it can be populated by a considerable amount by thermal excitation. Thus, for useful detection sensitivity the devices are normally operated at liquid nitrogen and sometimes liquid helium temperatures. The current response, i, of a photoconductor can be written as TABLE 31.3Common Light-Emitting Diode Compounds and Wavelengths Compound Wavelength (nm) Color GaP 565 Green GaAsP 590 Yellow GaAsP 632 Orange GaAsP 649 Red GaAlAs 850 Near IR GaAs 940 Near IR InGaAs 1060 Near IR InGaAsP 1300 Near IR InGaAsP 1550 Near IR Eh=u ? 2000 by CRC Press LLC (31.21) where P is the optical power at frequency u; h is Planck’s constant; v is drift velocity = mE, where m is mobility and E is electric field; h is quantum efficiency (at frequency u); t 0 is lifetime of carriers; and e is charge on electron. Charge Amplification. For semiconductor photoconductors like CdS there can be traps. These are holes, which under the influence of a bias field will be captured for a period of time. This allows electrons to move to the anode instead of recombining with a hole, resulting in a longer period for the conduction increase. This provides a photoconductive gain which is equal to the mean time the hole is trapped divided by the electron transit time in the photoconductor. Gains of 10 4 are typical. The charge amplification can be written as (31.22) where t d = d/n, the drift time for a carrier to go across the semiconductor. The response time of this type of sensor is consequently slow, ~10 ms, and the output in quite nonlinear. FIGURE 31.9 A simplified energy diagram for a photoconductive semiconductor, showing extrinsic effect of electrons into the acceptor level. TABLE 31.4 Photoconductor Materials and Their Peak Wavelength Sensitivity Photoconductor Peak Wavelength (mm) PbS 3 PbSe 5 HgCd 4 HgCaTe 10 HgCdTe 11 Si:Ga (4.2 K) 11 Si:As (4.2 K) 20 Si:Sb (4.2 K) 28 i Pev hd = ht u 0 t t 0 d ? 2000 by CRC Press LLC Junction Photodiodes In a simple junction photodiode a pn junction is fabricated in a semiconductor material. Figure 31.10 shows the energy diagram of such a device with a reverse voltage bias applied. Incident light with energy greater than the bandgap creates electrons in the p region and holes in the n region. Those which are within the diffusion length of the junction are swept across by the field. The light also creates electron–hole pairs in the junction region, and these are separated by the field. In both cases an electron charge is contributed to the external circuit. In the case of no bias the carrier movement creates a voltage with p region being positive. The maximum voltage is equal to the difference in the Fermi levels in the p and n regions and approaches the bandgap energy E g . PIN Photodiodes.The carriers which are generated in the junction region experience the highest field and so, being separated rapidly, give the fastest time response. The PIN diode has an extra intrinsic high field layer between the p and n regions, designed to absorb the light. This minimizes the generation of slow carriers and results in a fast response detector. The signal current generated by incident light power P is (31.23) The output current is linear with incident power plus a constant dark current due to thermal generation of carriers; h is the quantum efficiency. Avalanche Photodiodes When the reverse bias of a photodiode is increased to near the breakdown voltage, carriers in the depletion region can be accelerated to the point where they will excite electrons from the valence band into the conduction band, creating more carriers. This current multiplication is called avalanche gain, and typical gains of 50 are available. Avalanche diodes are specially designed to have uniform junction regions to handle the high applied fields. Detectors for Fiber Optic Communications A major application for junction photodioldes is detectors for fiber optic communications. Silicon detectors are typically used for short wavelength light such as with GaAs sources. InP detectors are used for the 1.3 and 1.5 mm wavelength bands. The specific type and design of a detector is tailored to the fiber optics application, depending on whether it is low cost lower frequency datalinks or higher cost high frequency bit-rates in the FIGURE 31.10Energy diagram of a pn junction photodiode showing the three ways electron–hole pairs are created by absorbing photons and the contribution to current flow in the circuit. i Pe h =+ h u dark current ? 2000 by CRC Press LLC Gb/s. The detector is packaged either with a fiber pigtail or with a fiber connector receptacle molded as part of the package body. Fiber optics detectors can also be packaged with pre-amplifier electronics or complete receiver and commu- nications electronics into a module. For very high frequency response it is important to minimize the capacitance of the detector and the attached preamplifier circuit. Solar Cells Solar cells are large-area pn junction photodiodes, usually in silicon, which are optimized to convert light to electrical power. They are normally operated in the photovoltaic mode without a reverse voltage bias being applied. Linear Position Sensors Large-area photodiodes can be made into single axis and two axis position sensors. The single axis device is a long strip detector, and the two axis is normally square. In the single axis device the common terminal is in the middle and there are two signal terminals, one at each end. When a light beam is directed onto the detector, the relative output current from each signal terminal depends on how close the beam is to the terminal. The sum of the output currents from both terminals is proportional to the light intensity. Phototransistors For bipolar devices the light generates carriers which inject current into the base of the transistor. This modulates the collector base current, providing a higher output signal. For a field effect device the light generates carriers which create a gate voltage. PhotoFETs can have very high sensitivities. SEEDs A self-electro-optic effect device (SEED) is a multiple quantum well semiconductor optical pin device and forms the combination of a photodiode and a modulator. It can operate as a photodetector where incident light will generate a photocurrent in a circuit. It can also act as a modulator where the light transmitted through the device is varied by an applied voltage. Devices are normally connected in pairs to form symmetric SEEDs as demonstrated in Fig. 31.11(a). These can then be operated as optical logic flip-flop devices. They can be set in one of two bistable states by application of incident light beams. The bistable state can be read out by similar light beams which measure the transmitted intensity. The hysteresis curve is shown in Fig. 31.11(b). These and similar devices are the emerging building blocks for optical logic and are sometimes referred to as smart pixels. FIGURE 31.11(a) S-SEED with voltage bias applied; (b) bistable outputs Q as a result of varying the input light power P 1 holding input power P 2 constant. ? 2000 by CRC Press LLC Detectors, Photoemissive In the photoemissive effect, light falls onto a surface (photocathode) and the light energy causes electrons to be emitted. These electrons are then collected at a positively biased anode. There is a threshold energy required for the electron to be emitted from the surface. This energy is called the work function, f, and is a property of the surface material. The photon energy hu must be greater than f, and this determines the longest wavelength sensitivity of the photocathode. Vacuum Photodiodes A vacuum photodiode comprises a negatively biased photocathode and a positive anode in a vacuum envelope. Light falling on the cathode causes electrons to be emitted, and these electrons are collected at the anode. Not all photons cause photoelectrons to be emitted, and quantum efficiencies, h, typically run 0.5–20%. These devices are not very sensitive; however, they have very good linearity of current to incident light power, P. They are also high-speed devices, with rise time being limited by the transit time fluctuations of electrons arriving at the anode. The photocurrent is given by (31.24) This kind of detector exhibits excellent short-term stability. The emissive surface can fatigue with exposure to light but will recover if the illumination is not excessive. Because of these properties, these devices have been used for accurate light measurement, although in many cases semiconductor devices are now supplanting them. Gas-Filled Tubes The light sensitivity of vacuum phototubes can be increased by adding 0.1 mm pressure of argon. The photo- electrons under the influence of the anode voltage accelerate and ionize the gas, creating more electrons. Gains of 5–10 can be realized. These devices are both low frequency, in the 10-kHz range, and nonlinear and are suitable only for simple light sensors. Semiconductor devices again are displacing these devices for most applications. Photomultiplier Tubes Photomultiplier tubes are the most sensitive light sensors, especially for visible radiation. Figure 31.12 is a schematic showing the electrical circuit used to bias it and form the output voltage signal. Light is incident on the photocathode, and the resulting photoelectrons are accelerated to a series of dynodes to generate secondary electrons and through this electron multiplication amplify the signal. Gains of 10 8 can be achieved with only minor degradation of the linearity and speed of vacuum photodiodes. The spectral response is governed by the emission properties of the photocathode. There are various types of photomultipliers with different physical arrangements to optimize for a specific application. The high voltage supply ranges from 700 to 3000 V, and the electron multiplication gain is normally adjusted by varying the supply voltage. The linearity of a photomultiplier is very good, typically 3% over 3 decades of light level. Saturation is normally encountered at high anode currents caused by space charge effects at the last dynode where most of the current is generated. The decoupling capacitors, C 1 , on the last few dynodes are used for high-frequency response and to prevent saturation from the dynode resistors. Photon Counting For the detection of very low light levels and for measuring the statistical properties of light, photon counting can be done using photomultipliers. A pulse of up to 10 8 electrons can be generated for each photoelectron emitted from the cathode, and so the arrival of individual photons can be detected. There is a considerable field of study into the statistical properties of light fields as measured by photon counting statistics. i Pe h =+ h u dark current ? 2000 by CRC Press LLC Imaging Detectors A natural extension to single photodetectors is to arrange them in arrays, both linear single dimension and two dimensions. Imaging detectors are made from both semiconductors and vacuum phototubes. Semiconductor Detector Arrays Detector arrays have been made using either photodiodes or photoconductors. The applications are for visible and infrared imaging devices. For small-sized arrays each detector is individually connected to an electrical lead on the package. This becomes impossible for large arrays, however, and these contain additional electronic switching circuits to provide sequential access to each diode. Figure 31.13 show an example of a charge-coupled device (CCD) linear photodiode array. The device consists of a linear array of pn junction photodiodes. Each diode has capacitance associated with it, and when light falls on the detector the resulting photocurrent charges this capacitance. The charge is thus the time integral of the light intensity falling on the diode. The CCD periodically and sequentially switches the charge to the video line, resulting in a series of pulses. These pulses can be converted to a voltage signal which represents the light pattern incident on the array. FIGURE 31.12The basic layout of a photomultiplier tube showing the dynodes and the electrical circuit to bias them. FIGURE 31.13Schematic diagram of a linear CCD diode array sensor. CCD shift register sequentially clocks out charge from each photodiode to the video line. ? 2000 by CRC Press LLC The location of the diodes is accurately defined by the lithographic fabrication process and, being solid state, is also a rugged detector. These devices are thus very suitable for linear or two-dimensional optical image measurement. The devices can be quite sensitive and can have variable sensitivity by adjusting the CCD scan speed since the diode integrates the current until accessed by the CCD switch. The spectral sensitivity is that of the semiconductor photodiode, and the majority of devices now available are silicon. Smaller arrays are becoming more available in many types of semiconductors, however. Image-Intensifier Tubes An image-intensifier tube is a vacuum device which consists of a photoemissive surface onto which a light image is projected, an electron accelerator, and a phosphor material to view the image. This device, shown in Fig. 31.14, can have a number of applications, for example, brightening a very weak image for night vision or converting an infrared image into a visible one. Light falling on the cathode causes electrons to be emitted in proportion to the light intensity. These electrons are accelerated and focused by the applied electric field onto the fluorescent screen to form a visible image. Luminance gains of 50–100 times can be achieved, and a sequence of devices can be used to magnify the gain even more. Image Orthicon Tube (TV Camera) There are two basic types of television (TV) camera tubes, the orthicon and the vidicon. The orthicon uses the photoemissive effect. A light image is focused onto the photocathode, and the electrons emitted are attracted toward a positively based target (see Fig. 31.15). The target is a wire mesh, and the electrons pass through it to be collected on a glass electron target screen. This also causes secondary electrons to be emitted, and they also collect on the screen. This results in a positive charge image which replicates the light image on the photocathode. A low-velocity electron beam is raster scanned across the target to neutralize the charge. The surplus electrons return to the electron multiplier and generate a current for the signal output. The output current is thus inversely proportional to the light level at the scanning position of the beam. The orthicon tube is very sensitive because there is both charge accumulation between scans and gain from the electron multiplier. Vidicon Camera Tube A simple TV camera tube is the vidicon. This is the type used in camcorders and for many video applications where a rugged, simple, and inexpensive camera is required. Figure 31.16 is a schematic of a vidicon tube; the optical image is formed on the surface of a large-area photoconductor, causing corresponding variations in the conductivity. This causes the rear surface to charge toward the bias voltage V b in relation to the conductivity image. The scanning electron beam periodically recharges the rear side to 0 V, resulting in a recharging current flow in the output. The output signal is a current signal proportional to the light incident at the position of the scanning electron beam. FIGURE 31.14Diagram of a simple image-intensifier tube. More complex ones use improved electron optics. ? 2000 by CRC Press LLC The primary disadvantages of the vidicon are its longer response time and smaller dynamic range. The recent availability of longer wavelength photoconducting films has resulted in new infrared cameras becoming avail- able. A recent advance in these types of image sensor is to replace the photoconductor with a dense array of very small semiconductor photodiodes. Photocurrent in the diode charges a capacitor connected to it. The raster scanned electron beam discharges this capacitor in the same way. Image Dissector Tube The image dissector tube is a photosensitive device which uses an electron deflection lens to image the electron from the cathode onto a pinhole in front of an electron multiplier. The image can be deflected around in front of the pinhole in a random access manner. The primary application of this kind of device is for tracking purposes. Noise and Detectivity Noise There are two primary sources of noise in photodetectors: Johnson noise due to thermal effects in the resistive components of the device and its circuits, and shot noise or its equivalent, which is due to the quantized nature of electro-optic interactions. In semiconductor devices noise is usually given in terms of noise current, (31.25) where i includes signal and dark currents, e is electron charge, M is avalanche gain (x depends on avalanche photodetector characteristics), D| is frequency bandwidth, k is Boltzmann’s constant, T is in degrees Kelvin, and R is the total circuit resistance at temperature, T. FIGURE 31.15 Schematic diagram of an image orthicon TV camera tube. FIGURE 31.16 Schematic of a vidicon TV camera tube. di eiM f kT f R x22 2 4 =+ + D D ? 2000 by CRC Press LLC For photoconductor devices (including effects of charge amplification) the noise current is given by (31.26) The first term is analogous to shot noise but includes the effects of carrier creation and recombination. t 0 is the carrier lifetime, t d is the drift time for a carrier to go across the photoconductor, and u is the light frequency. The noise for photoemissive devices is usually written as a noise voltage and is given by (31.27) where G is the current gain for the photomultiplier. Detectivity The performance of a detector is often described using the term D*, detectivity. This term is useful for comparison purposes by normalizing with respect to detector size and/or noise bandwidth. This is written as (31.28) where NEP is the noise equivalent power (for signal-to-noise ratio equal to 1) and A is detector area. The term D*(l) is used for quoting the result using a single-wavelength light source and D*(T) is used for the unfiltered blackbody radiation source. Defining Terms Charge-coupled device (CCD): A series of electronic logic cells in a device in which a signal is represented and stored as an electronic charge on a capacitor. The signal is moved from one cell (memory position or register) to an adjacent cell by electronically switching the charge between the capacitors. Electron multiplication: The phenomenon where a high-energy electron strikes a surface and causes addi- tional electrons to be emitted from the surface. Energy from the incident electron transfers to the other electrons to cause this. The result is electron gain which is proportional to the incident electron energy. Extended source: A light source with finite size where the source size and shape can be determined from the emitted light characteristics. The light is spatially incoherent. Light detection: The conversion of light energy into an electrical signal, either current or voltage. Light emission: The creation or emission of light from a surface or device. Point source: A light source which is so small that its size and shape cannot be determined from the characteristics of the light emanating from it. The light emitted has a spherical wave front and is spatially coherent. Television (TV): The process of detecting an image and converting it to a serial electronic representation. A detector raster scans the image, producing a voltage proportional to the light intensity. The time axis represents the distance along the raster scan. Several hundred horizontal scans make up the image starting at the top. The raster scan is repeated to provide a continuing sequence of images. Related Topic 42.2 Optical Fibers and Cables d tt put i ei f kT f R d2 0 22 0 2 4 14 4 = + + ()/ DD d v ei G f R kT f R 2 22 24=+DD D Af * = D NEP ? 2000 by CRC Press LLC References B. Crosignani, P. DiPorto, and M. Bartolotti, Statistical Properties of Scattered Light, New York: Academic Press, 1975. A.L. Lentine et al., “A 2 kbit array of symmetric self-electrooptic effect devices,” IEEE Photonics Technol. Lett., vol. 2, no. 1, 1990. Reticon Corp., subsidiary of EG&G, Inc., Application notes #101. Further Information W.J. Smith, Modern Optical Engineering, New York: McGraw Hill, 1966. M.J. Howes and D.V. Morgan, Gallium Arsenide Materials, Devices and Circuits, New York: John Wiley, 1985. M.K. Baroski, Fundamentals of Optical Fiber Communications, New York: Academic Press, 1981. C.Y. Wyatt, Electro-Optic System Design for Information Processes, New York: McGraw-Hill, 1991. S. Ungar, Fibre Optics—Theory and Applications, New York: John Wiley, 1990. 31.3 Circuits R.A. Becker In 1969, Stewart Miller of AT&T Bell Laboratories published his landmark article on integrated optics. This article laid the foundation for what has now developed into optoelectronic circuits. In it he described the concepts of planar optical guided-wave devices formed as thin films on various substrates using fabrication techniques similar to those used in the semiconductor integrated circuit (IC) industry. The attributes of these new circuits included small size, weight, power consumption, and mechanical robustness because all compo- nents were integrated on a single substrate. The field of optoelectronic circuits began as a hybrid implementation where optical sources (laser diodes) and detectors have historically been fabricated on separate semiconductor substrates, and waveguide devices, such as modulators and switches, have been fabricated on electro-optic single-crystal oxides such as lithium niobate (LiNbO 3 ). Often, the two dissimilar substrates have been connected using single-mode polarization preserving optical fiber. Now, although the hybrid concept is finding commercial applications, most active research is performed on monolithic implementations, where all devices are fabricated on a common semiconductor substrate. After a brief summary discussion of semiconductor, glass, and polymer material systems, we will deal exclusively with the most mature hybrid implementation of optoelectronic circuits based on LiNbO 3 . Because sources and detectors have been covered in previous sections, in this section the devices that are utilized in between, i.e., modulators and switches, will be discussed. Integrated Optics Integrated optics can be defined as the monolithic integration of one or more optical guided-wave structures on a common substrate. These structures can be passive, such as a fixed optical power splitter, or active, such as an optical switch. Active devices are realized by placing metal electrodes in close proximity to the optical waveguides. Applying a voltage to the electrodes changes the velocity of the light within the waveguide. Depending on the waveguide geometry and the electrode placement, a wide variety of technologically useful devices and circuits can be realized. The technological significance of integrated optics stems from its natural compatibility with two other rapidly expanding technologies: fiber optics and semiconductor laser diodes. These technologies have moved in the past 10 years from laboratory curiosities to large-scale commercial ventures. Integrated optic devices typically use laser diode optical sources, diode-pumped yttrium, aluminum, garnet (YAG) lasers, and transmit the modified optical output on a single-mode optical fiber. Integrated optic devices are typically very high speed, compact, and require only moderate control voltages compared to their bulk-optical counterparts. ? 2000 by CRC Press LLC In integrated optic devices, the optical channel waveguides are formed on a thin, planar, optically polished substrate using photolithographic techniques similar to those used in the semiconductor IC industry. Waveguide routing is accomplished by the mask used in the photolithographic process, similar to the way electrically conductive paths are defined in semiconductor ICs. The photolithographic nature of device fabrication offers the potential of readily scaling the technology to large volumes, as is done in the semiconductor IC industry. For example, the typical device is 0.75 in. ′ 0.078 in. in size. Dividing the substrate size by the typical device size and assuming a 50% area usage indicates that one can achieve 50 devices per 3-in. wafer. Substrate materials for integrated optics include semiconductors, such as GaAs and InP, glass, polymer coated glass or Si, and LiNbO 3 . Recently, primarily passive glass-based devices have been commercially introduced as replacements for passive all-fiber devices such as splitters and combiners. In addition, there are slow-speed switches (millisecond) now available that utilize the thermooptic effect in glass. Glass-based devices are fabri- cated by either depositing glass waveguiding layers on Si, or through the indiffusion of dopants into glass which results in a waveguiding layer. Both fabrication approaches are used in commercially available devices. Very recently, low-speed polymer-on-Si switches have been commercially introduced. These also operate via the thermooptic effect. However, since polymers can be engineered with electrooptic properties, high-speed devices may also be available in the future. The primary impediment to market penetration of polymer-based devices has been their relatively poor stability, especially at temperatures above 100°C. However, if polymers can be produced with both strong electrooptic properties and enhanced stability with temperature, they could be the material system of choice for many applications because of their low-cost potential. The area of semiconductor-based integrated optics has attracted much attention worldwide because if offers the potential of integrating electronic circuitry, optical sources and detectors, and optical waveguides on a single substrate. While being quite promising, the technology is still 5 years away from commercialization. Technical problems in semiconductor-based integrated optics include low electrooptic coefficients, higher optical waveguide attenuation, and an incompatibility of the processing steps needed to fabricate the various types of devices on a single substrate. However, considerable attention is being paid to these problems, and improve- ments are continually occurring. The primary substrate material in integrated optics is the widely available synthetic crystal, lithium niobate (LiNbO 3 ), which has been commercially produced in volume for more than 20 years. This material is transparent to optical wavelengths between 400 and 4500 nm, has a hardness similar to glass, and is nontoxic. LiNbO 3 -based devices have been commercially available since 1985 and have been incorporated in a large number of experimental systems. The basic LiNbO 3 waveguide fabrication technique was developed in 1974 and has been continually refined and improved during subsequent years. The material itself finds wide appli- cation in a number of electrical and optical devices because of its excellent optical, electrical, acoustic, and electro- and acousto-optic properties. For example, almost all color television sets manufactured today incor- porate a surface-acoustic-wave (SAW) electrical filter based on LiNbO 3 . In LiNbO 3 -based integrated optics, optical waveguides are formed in one of two ways. The first uses photo- lithographically patterned lines of titanium (Ti), several hundred angstroms thick, on the substrate surface. The titanium is then diffused into the substrate surface at a temperature of about 1000°C for several hours. This process locally raises the refractive index in the regions where titanium has been diffused, forming high- refractive index stripes that will confine and guide light. Because the diffusion is done at exceedingly high temperatures, the waveguide stability is excellent. The waveguide mechanism used is similar to that used in fiber optics, where the higher-index, doped cores guide the light. The exact titanium stripe width, the titanium thickness, and diffusion process are critical parameters in implementing a low-loss single-mode waveguide. Different fabrication recipes are required to optimize the waveguides for operation at the three standard diode laser wavelengths: 800 nm, 1300 nm, and 1500 nm. The second approach uses a technique known as proton exchange. In this approach, a mask is used to define regions of the substrate where hydrogen will be exchanged for lithium, resulting in an increase in the refractive index. This reaction takes place at lower temperatures (200–250°C) but has been found to produce stable waveguides if an anneal at 350–400°C is performed. Waveguides formed using the proton exchange method support only one polarized mode of propagation, whereas those formed using Ti indiffusion support two. Proton exchange waveguides are also capable of handling much higher optical power densities, especially at the shorter wavelengths, than are those formed by Ti indiffusion. More fabrication detail will be provided later. ? 2000 by CRC Press LLC Light modulation is realized via the electro-optic effect, i.e., inducing a small change in the waveguide refractive index by applying an electric field within the waveguide. On an atomic scale the applied electric field causes slight changes in the basic crystal unit cell dimensions, which changes the crystal’s refractive index. The magnitude of this change depends on the orientation of the applied electric field and the optical polarization. As a result, only certain crystallographic orientations are useful for device fabrication and devices are typically polarization dependent. The electro-optic coefficients of LiNbO 3 are among the highest (30.8 pm/V) of any inorganic material, making the material very attractive for integrated optic applications. Combining the concepts of optical waveguides and electro-optic modulation with the geometric freedom of photolithographic techniques leads to an extremely diverse array of passive and active devices. Passive components do not require any electric fields and are used for power splitting and combining functions. Two types of passive power division structures have been fabricated: Y-junctions and directional couplers. A single waveguide can be split into two by fabricating a shallow-angle Y-junction as shown in Fig. 31.17. An optical signal entering from the single-waveguide side of the junction is split into two optical signals with the same relative phase but one-half the original intensity. Conversely, light incident on the two-waveguide side of the junction will be combined into the single waveguide with a phase and intensity dependent on the original inputs. Directional couplers consist of two or more waveguides fabricated in close proximity to each other so that the optical fields overlap as shown in Fig. 31.18. As a result, optical power is transferred between the waveguides. The extent of the power transfer is dependent on the waveguide characteristics, the waveguide spacing, and the interaction length. A different type of passive component is an optical polarizer, which can be made using several different techniques. One such method is the metal-clad, dielectric-buffered waveguide shown in Fig. 31.19. In this passive device, the TM polarization state is coupled into the absorbing metal and is thus attenuated, while the TE polarization is virtually unaffected. Measurements of a 2-mm-long polarizer of this type have demonstrated TM attenuations exceeding 50 dB (100,000:1). Polarizers can also be fabricated in others ways. One interesting technique involves the diffusion of hydrogen ions into the LiNbO 3 . This results in a waveguide which, as discussed earlier, will only support the TE-polarized mode and, thus, is a natural polarizer. Active components are realized by placing electrodes in close proximity to the waveguide structures. Depend- ing on the substrate crystallographic orientation, the waveguide geometry, and the electrode geometry, a wide variety of components can be demonstrated. The simplest active device is the phase modulator, which is a single waveguide with electrodes on either side as shown in Fig. 31.20. Applying a voltage across the electrodes induces an electric field across the waveguide, which changes its refractive index via the electro-optic effect. For 800-nm wavelength operation, a typical phase modulator would be 6 mm long and would induce a p- phase shift for an applied voltage of 4 V. The transfer function (light out versus voltage in) can be expressed as I 0 (V) = I i exp(jwt + pV/V p ) (31.29) where V p is the voltage required to cause a 180-degree phase shift. Note that there is no change in the intensity of the light. Coherent techniques are used to measure the amount of phase change. Optical intensity modulators can be fabricated by combining two passive Y-junctions with a phase modulator situated between them. The result, which is shown in Fig. 31.21, is a guided-wave implementation of the classic Mach–Zehnder interferometer. In this device the incoming light is split into two equal components by the first FIGURE 31.17Passive Y-splitter. ? 2000 by CRC Press LLC Y-junction. An electrically controlled differential phase shift is then introduced by the phase modulator, and the two optical signals are recombined in the second Y-junction. If the two signals are exactly in phase, then they recombine to excite the lowest-order mode of the output waveguide and the intensity modulator is turned fully on. If instead there exists a p-phase shift between the two signals, then they recombine to form the second mode, which is radiated into the substrate and the modulator is turned fully off. Contrast ratios greater than 25 dB (300:1) are routinely achieved in commercial devices. The transfer function for the Mach–Zehnder modulator can be expressed as I 0 (V) = I i cos 2 (pV/2V p + f ) (31.30) FIGURE 31.18 Directional coupler power splitter. FIGURE 31.19 Thin-film optical polarizer. FIGURE 31.20 Electro-optic integrated optic phase modulator. ? 2000 by CRC Press LLC where V p is the voltage required to turn the modulator from on to off, and f is any static phase imbalance between the interferometer arms. This transfer function is shown graphically in Fig. 31.21. Note that the modulator shown in Fig. 31.21 has push-pull electrodes. This means that when a voltage is applied, the refractive index is changed in opposite directions in the two arms, yielding a twice-as-efficient modulation. Optical switches can be realized using a number of different waveguide, electrode, and substrate orientations. Two different designs are used in commercially available optical switches: the balanced-bridge and the Db directional coupler. The balanced-bridge design is similar to that of the Mach–Zehnder interferometer, except that the Y-junctions have been replaced by 3-dB directional couplers as shown in Fig. 31.22. Similar to the Mach–Zehnder, the first 3-dB coupler splits the incident signal into two signals, ideally of equal intensity. Once again, if a differential phase shift is electro-optically induced between these signals, then when they recombine in the second 3-dB coupler, the ratio of power in the two outputs will be altered. Contrast ratios greater than 20 dB (100:1) are routinely achieved in commercial devices. The transfer function for this switch can be expressed as I 0a = I i cos 2 (pV/2V p + p/2) (31.31) I 0b = I i sin 2 (pV/2V p + p/2) (31.32) and is graphically depicted in Fig. 31.22. In the other type of switch, the Db directional coupler, the electrodes are placed directly over the directional coupler as shown in Fig. 31.23. The applied electric field alters the power transfer between the two adjacent waveguides. Research versions of this switch have demonstrated contrast ratios greater than 40 dB (10,000:1); however, commercial versions typically achieve 20 dB, which is competitive FIGURE 31.21Mach–Zehnder intensity modulator and transfer function. FIGURE 31.22Balanced-bridge modulator/switch and transfer function. ? 2000 by CRC Press LLC with that achieved with the balanced-bridge switch. The transfer function for the Db directional coupler switch can be expressed as I 0a = sin 2 kL*sqrt(1 + (Db/2k) 2 )/(1 + (Db/2k) 2 ) (31.33) I 0b = 1 – I 0a (31.34) where k is the coupling constant and Db is the voltage-induced change in the propagation constant. This transfer function is depicted in Fig. 31.23. Another type of active component that has recently become available commercially is the polarization controller. This component allows the incoming optical polarization to be continuously adjustable. The device functions as an electrically variable optical waveplate, where both the birefringence and axes orientation can be controlled. The controller is realized by using a three-electrode configuration as shown in Fig. 31.24 on a substrate orientation where the TE and TM optical polarizations have almost equal velocities. Typical perfor- mance values are TE/TM conversion of greater than 99% with less than 50 V. One of the great strengths of integrated optic technology is the possibility of integrating different types or multiple copies of the same type of device on a single substrate. While this concept is routinely used in the semiconductor IC industry, its application in the optical domain is novel. The scale of integration in integrated optics is quite modest by semiconductor standards. To date the most complex component demonstrated is an FIGURE 31.23Directional coupler switch and transfer function. FIGURE 31.24Guided-wave polarization controller. ? 2000 by CRC Press LLC 8 ′ 8 optical switch matrix that uses 64 identical 2 ′ 2 optical switches. The most device diversity on a given substrate is found in fiber gyro applications. Here, components incorporating six phase modulators, two electrically tunable directional couplers, and two passive directional couplers have been demonstrated. Device Fabrication The fabrication of an integrated optic device uses the same techniques as used in the semiconductor IC industry. Device designs are first entered into a computer-aided design (CAD) system for accurate feature placement and dimensional control. This design is then output as a digitized tape that will control a pattern generation system for fabrication of the chrome masks that are used in device fabrication. A variety of equipment such as step-and-repeat and E-beam systems has been developed for the semiconductor IC industry for the generation of chrome masks. These same systems are used today for generation of masks for integrated optic devices. The waveguides can be fabricated by using either the Ti indiffusion method or the proton exchange method. The first step in fabricating a waveguide device using Ti indiffusion is the patterning in titanium. The bare LiNbO 3 surface is first cleaned and then coated with photoresist. Next, the coated substrate is exposed using the waveguide-layer chrome mask. The photoresist is then developed. The areas that have been exposed are removed in the development cycle. The patterned substrates are then coated with titanium in a vacuum evaporator. The titanium covers the exposed regions of the substrate as well as the surface of the remaining photoresist. The substrate is next soaked in a photoresist solvent. This causes all the residual photoresist (with titanium on top) to be removed, leaving only the titanium that coated the bare regions of the substrate. This process is known as lift-off. Finally, the substrate, which is now patterned with titanium, is placed in a diffusion system. At temperatures above 1000°C the titanium diffuses into the substrate, slightly raising the refractive index of these regions. This process typically takes less than 10 hours. This sequence of steps is depicted in Fig. 31.25. The proton exchange method is depicted in Fig. 31.26. Here a chrome masking layer is first deposited on the LiNbO 3 substrate. It is patterned using photoresist and etching. Next, the substrate is submerged in hot benzoic acid. Finally, the chrome mask is removed and the substrate is annealed. The regions that have been exposed to the benzoic acid will have an increased refractive index and will guide light. If the devices being fabricated are to be active (i.e., voltage controlled), then an electrode fabrication step is also required. This sequence of steps parallels the waveguide fabrication sequence. The only differences are that FIGURE 31.25Ti-indiffused LiNbO 3 waveguide fabrication. ? 2000 by CRC Press LLC an electrode mask is used and the vacuum-deposited metal used is chrome/gold or chrome/aluminum. This sequence of steps is shown in Fig. 31.27. In order to get the light in and out of the waveguide, the endfaces have to be lapped and polished flat with chip-free knife edges. This is currently accomplished using standard lapping and polishing techniques. After FIGURE 31.26Proton exchange LiNbO 3 waveguide fabrication. FIGURE 31.27Electrode fabrication via lift-off. ? 2000 by CRC Press LLC this step, the substrate can be diced into as many devices as were included on the substrate. Finally, the diced parts need to be electrically and optically packaged. Packaging To get the light in and out of an integrated optic waveguide requires a tiny optical window to be polished onto the waveguide’s end. Currently, the entire endface of the substrate is polished to a sharp, nearly perfect corner, making the whole endface into an optical window. An optical fiber can then be aligned to the waveguide end and attached. Typically, centration of the fiber axis to the waveguide axis must be better than 0.2 mm. Some devices require multiple inputs and outputs. In this case the fibers are prealigned in silicon V-grooves. These V-grooves are fabricated by anisotropic etching of a photolithographically defined pattern on the silicon. The center-to-center spacing of the fiber V-groove array can be made to closely match that of the multiple waveguide inputs and outputs. Integrated optic devices built on LiNbO 3 are inherently single-mode devices. This means that the light is confined in a cross-sectional area of approximately 30 mm 2 . The optical mode has a near-field pattern that is 5 to 10 mm across and 3 to 6 mm deep, depending on the wavelength. These mode spot sizes set limits on how light can be coupled in and out. There are a number of methods that can be used to couple the light into LiNbO 3 waveguides. These include prism coupling, grating coupling, end-fire coupling with lenses, and end- fire coupling with single-mode optical fibers. In general, most of these techniques are only useful for laboratory purposes. The most practical real-world technique is end-fire coupling with an optical fiber. In this case the optical fiber is aligned to the waveguide end. This is an excellent practical method since integrated optic devices are most often used in fiber optic systems. Therefore, the coupling problem is one of aligning and fixing a single-mode fiber to the single-mode LiNbO 3 waveguide. The size of the single-mode radiation pattern and its angular divergence set the alignment tolerances. A low-loss connection between a fiber and a LiNbO 3 waveguide requires <1/20 of a mode spot diameter (0.25–0.5 mm) in transverse offset, angular tilt of <2 degrees, and a longitudinal offset of <1 mode spot diameter (5–10 mm). These are very stringent alignment requirements, especially if they have to be maintained over a wide temperature range. Another aspect of the problem is that many integrated optic devices require a single, well-defined linearly polarized input. Ordinary single-mode fiber is not suitable in this case. The solution is to use polarization preserving fiber. This fiber is made such that it will maintain a single linearly polarized input over long distances. The use of polarization preserving fiber, however, adds another requirement to the coupling problem. This requirement is that the fiber must be rotationally aligned about its cylindrical axis so that the linearly polarized light coincides with the desired rotational axis of the LiNbO 3 waveguide. The rotational precision needed is <0.5 degrees. Many LiNbO 3 devices, such as fiber gyro components, require multiple input and/or output optical connec- tions. Thus, the packaging must be able to accommodate multiple inputs/outputs and maintain strict alignment for all connections. The method of end-fire coupling optical fibers to the LiNbO 3 waveguide is commonly called pigtailing. This is the only practical packaging method now used for integrated optic devices that operate in a real system and outside the laboratory. The reasons for this are quite logical. The end user installs the device in his system by connecting to the fiber pigtails. The connection can be made with single-mode connectors or by splicing. Flexibility is one of the big advantages of using fiber pigtails. The typical LiNbO 3 device is packaged in a metallic case with optical fiber pigtails connected at both ends. Electrical connections are provided by RF connectors or pins, which are common in the electronics industry. If hermetically sealed packages are desired, then the optical fiber pigtails must be hermetically sealed to the metallic package. Applications Many useful systems have been demonstrated using LiNbO 3 -based integrated optic devices. These system applications can be grouped into four broad categories: telecommunications, instrumentation, signal processing, and sensors. In some cases only a single integrated optic device is used, while in other applications a multi- function component is required. ? 2000 by CRC Press LLC Optical switches have been shown to be quite useful in the telecommunications area. High-speed 2 ′ 2 switches for time-domain mux/demux as well as lower-speed 4 ′ 4 switch arrays have been successfully demonstrated. In both cases a major advantage of optical switching is that the switch data transmission rate is not limited by the switch itself as is the case for electronic switches. Thus, it is possible to route optical signals at data rates exceeding terabits/per second (1,000,000,000,000 bits/s). Aside from the switching application in telecommunications, there also is the high-speed laser modulation application. Using an external LiNbO 3 -based optic intensity modulator, both analog and digital data transmis- sion systems have been demonstrated. Analog transmission systems using integrated optic devices are partic- ularly attractive as remote antenna links because of their high speed and ability to be driven directly by the received signal without amplification. Recently, the use of high-power diode-pumped YAG lasers operating at 1300 nm and external intensity modulators based on LiNbO 3 have found wide application in the cable TV industry. In addition, the ability of the Mach–Zehnder intensity modulator to control intensity with a controlled wavelength change (i.e., chirp) has allowed its use in long-haul telecommunications systems. Another demonstrated application of integrated optics in the telecommunications area is in coherent com- munication systems. These systems require both phase modulators and polarization controllers. Current optical fiber transmission systems rely on intensity modulated data transmission schemes. Coherent communication systems are attractive because of the promise of higher bit rates, wavelength division multiplexing capability, and greater noise immunity. In coherent communication systems the information is coded by varying either the phase or frequency of the optical carrier with a phase modulator. At the receiver, a polarization controller is used to ensure a good signal-to-noise ratio in the heterodyne detection system. One promising application of integrated optic devices in instrumentation is a high-speed, polarization- independent optical switch for use in optical-time-domain reflectometers (OTDR). OTDRs are used to locate breaks or poor splices in fiber optic networks. The instruments work by sending out an optical pulse and measuring the backscattered radiation returning to the instrument as a function of time. The next generation of OTDRs will possibly employ an optical switch, which will be used to rapidly switch the optical fiber under test from the pulsed light source to the OTDR receiver. Such an instrument could detect faults closer to the OTDR than currently possible, which is important in the short-haul systems now being installed. This feature is necessary for local area network (LAN) installations. Several types of sensors using integrated optic devices have been demonstrated. Two of the most promising are electric/magnetic field or voltage sensors and rotation sensors (fiber optic gyro). Electric field sensors typically consist of either a Mach–Zehnder intensity modulator or an optical switch that is biased midway between the on and off states. For small modulation depths about this midpoint the induced optical modulation is linear with respect to applied voltage. Linear dynamic ranges in excess of 80 dB have been accomplished. This is larger than that obtained using any other known technology. Perhaps the most promising near-term application of integrated optic devices in the field of rotation sensing is as a key component in optical fiber gyroscopes. A typical fiber optic gyro component is shown in Fig. 31.28. The device consists of a polarizer, a Y-junction, and two phase modulators, all integrated on a single substrate. In fiber gyro systems, the integrated optic component replaces individual, fiber-based components that perform the same function. The integrated optic component offers a greatly improved performance, at a significant reduction in cost, compared to the fiber-based components. Most fiber optic gyro development teams have done away with the fiber components and have adopted LiNbO 3 -based components as the technology of choice. Defining Terms Integrated optics: The monolithic integration of one or more optical guided-wave devices on a common substrate. Intensity modulator:A modulator that alters only the intensity of the incident light. Lithium niobate (LiNbO 3 ):A single-crystal oxide that displays electro-optic, acousto-optic, piezoelectric, and pyroelectric properties that is often the substrate of choice for surface acoustic wave devices and integrate optical devices. Optical guided-wave device: An optical device that transmits or modifies light while it is confined in a thin- film optical waveguide. ? 2000 by CRC Press LLC Phase modulator: A modulator that alters only the phase of the incident light. Polarization controller: A device that alters only the polarization state of the incident light. Related Topic 42.2 Optical Fibers and Cables References R. Alferness, “Waveguide electrooptic modulators,” IEEE Trans. Microwave. Theory Tech., vol. MTT-30, p. 1121, 1982. R.A. Becker, “Commercially available integrated optics products and services,” SPIE, vol. 993, p. 246, 1988. R. Childs and V. O’Byrne, “Predistortion Linearization of Directly Modulated DFB Lasers and External Mod- ulators for AM Video Transmission,” OFC’90 Tech. Dig., Paper WHG, 1990, p. 79. C. Cox, G. Betts, and L. Johnson, “An analytic and experimental comparison of direct and external modulation in analog fiber-optic links,” IEEE Trans. Microwave Theory Tech., vol. 38, p. 501, 1990. T. Findakly and M. Bramson, “High-performance integrated-optical chip for a broad range of fiber-optic gyro applications,” Opt. Lett., vol. 15, p. 673, 1990. P. Granestrand, B. Stoltz, L. Thylen, K. Bergvall, W. Doldisen, H. Heinrich, and D. Hoffmann, “Strictly non- blocking 8 ′ 8 integrated optical switch matrix,” Electron. Lett., vol. 22, p. 816, 1986. M. Howerton, C. Bulmer, and W. Burns, “Effect of intrinsic phase mismatch on linear modulator performance of the 1 ′ 2 directional coupler and Mach–Zehnder interferometer,” J. Lightw. Tech., vol. 8, p. 1177, 1990. S.E. Miller, “Integrated optics: An introduction,” Bell Syst. Tech. J., vol. 48, p. 2059, 1969. Further Information Integrated Optical Circuits and Components, Design and Applications, edited by Lynn D. Hutcheson (Marcel Dekker, Inc., New York, 1987) and Optical Integrated Circuits, by H. Nishihara, M. Haruna, and T. Suhara (McGraw-Hill Book Company, New York, 1989) provide excellent overviews of the field of integrated and guided-wave optics. Integrated Optics: Devices and Applications, edited by J. T. Boyd (IEEE Press, New York, 1991) provides an excellent cross section of recent publications in the field. In addition, the monthly magazine IEEE Journal of Lightwave Technology provides many publications on current research and development on integrated and guide-wave optic devices and systems. FIGURE 31.28 Fiber-optic gyro chip. ? 2000 by CRC Press LLC