Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 History of Light Sources John F. Waymouth* GTE Lighting Products, Marblehead, MA, USA Glossary and Abbreviations Glossary Exhaust A small diameter tube fused into the stem, through which air is exhausted from the tube interior of the lamp and any gas filling is introduced Flare A short piece of tubing of which the bottom end is flared out in a skirt that is fused to the envelope the lamp in the sealing operation (usually fabricated from leaded glass) Hard glass A glass having a softening temperature greater than 700 C and a service temperature ca 200 C or less Leaded A “soft” glass containing lead oxide (commonly referred to as “lead glass” in the glass industry; this term is used to avoid confusion with the lead-in wire) Lead-in The connection from the circuit into the interior of the lamp (commonly referred to as wire the “lead wire” in the industry; this term is used to avoid confusion with the element Pb) Mount The completed assembly incorporating the stem, filament, and filament supports, ready to be sealed to the envelope of the lamp Press The part of the flare that is fused around and pressed on to the lead-in wires to make a hermetic seal Soft glass A glass having a softening temperature ca 700 C and a normal service temperature ca 100 C or less Stem A glass assembly comprising a flare, lead-in wires, and exhaust tube fused together in the press Abbreviations AEG Allgemeine Elektrische Gesellschaft (Germany) AEI Associated Electrical Industries (England) ANSI American National Standards Institute (USA) CFL Compact fluorescent lamp EMI Electric and Musical Industries (England) GE General Electric (Co) (USA) GEC General Electric Company (England) HPS High pressure sodium (lamp) LED Light-emitting diode LPS Low pressure sodium (lamp) John F. Waymouth is retired. *Email: Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 LTE Local thermodynamic equilibrium MH Metal halide (lamp) NEMA National Electric Manufacturers Association (USA) OEM Original equipment manufacturer TM PCA Polycrystalline alumina, branded “Lucalox ” by GE RGB Red-green-blue TH Tungsten halogen (lamp) Introduction There are many histories of discovery, invention, and development in the electric lamp industry. One of the best I have found, from the earliest days to 1947, is a book by Arthur Bright, “The Electric Lamp Industry” (Bright 1949). Several more recent ones are listed in the references (Zissis and TM Kitsinellis 2009; Gendre 2003). A Google search under the heading “History of Light Sources” turns up a number of websites, of which a sampling is listed (http://www.mts.net/~william5/history/ hol.htm; http://www.invsee.asu.edu/Modules/lightbulb/meathist.htm; http://www.en.wikipedia.org/ wiki/Light; http://www.edisontechcenter.org/incandescent.html; http://www.wired.com/gadgets/ miscellaneous/multimedia/2008/11/gallery_lights; http://www.inventors.about.com/library/inven tors/bllight2.htm; http://www.ies.org/lighting/history; http://www.nelt.co.jp/english/products/use ful/01.html). The world does not need another compendium of “who discovered/invented/developed which lamp, where, and when?” which would necessarily be just a copying from earlier sources. Therefore, I propose to discuss that history from several much broader perspectives: Section “The Role of Science in Light-Source Development and Optimization” Section “The Importance of Materials in Light-Source Development and Optimization” Section “The Contribution of Automated Light-Source-Manufacturing Machinery” Section “The Symbiosis of Light-Sources and the Electric Power System” Section “The Electric Lamp Market Through the Years” Because space is limited, this discussion will focus primarily on the history involving the major light sources in use today: incandescent, fluorescent, HPS, mercury, MH, TH lamps, and LEDs. Readers interested in other sources are referred to the bibliography, particularly Bright’s book. In addition, there is nothing in this summary about the history in Russia and China, because they made little or no contribution to what is discussed in sections “The Role of Science in Light-Source Development and Optimization,” “The Importance of Materials in Light-Source Development and Optimization,” “The Contribution of Automated Light-Source-Manufacturing Machinery,” “The Symbiosis of Light Sources and the Electric Power System,” and “I” know nothing about their internal markets. Since I had the good fortune to participate in some of (and observe more of) that history in the second half of the twentieth century, I trust it will not be taken amiss if I include a few personal reminiscences. Aword about references: the references listed herein do not necessarily indicate priority. They are frequently simply those which best support the point I wish to make. Page 2 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 The Role of Science in Light-Source Development and Optimization For the most part, the role of science in discovery, invention, and development of light sources has been indirect, background, or post-development explanation and optimization. For example, it has been known since the human race emerged from the Stone Age into the Bronze Age that heated objects emit light; blacksmiths have since that time used for temperature control the fact that that light shifted from dim and red at moderate temperatures toward bright and white at high tempera- tures. It was known since the work of Ohm that conducting wires got hot from the passage of current, to a degree that increased with their resistivity. Thus, early experimenters seeking to produce light by incandescence from electrically heated metals would have sought out conductors of high melting temperatures and high resistivities. Platinum and graphite were the obvious choices at the time: platinum because it could be used in air and graphite because it had a higher resistivity and high melting temperature. By far the most important contribution of science to the electric lamp industry was the discovery of the principle of electromagnetic induction byMichael Faraday in 1831, which was the foundation of the development of the “dynamo,” the DC electric generator, through the inventions of Hippolyte Pixii (1832), Antonio Pacinotti (1860), Werner Siemens (1867), Charles Wheatstone (1867), Zenobe Gramme (1871), and Charles Brush (1876) (http://www.en.wikipedia.org/wiki/Dynamo). Absent the dynamo, the only source of electric power was the chemical battery, and electric lighting would have been a mere laboratory curiosity. The numerous individuals active in inventing the incandescent light bulb were for the most part either gifted tinkerers or engineers, aware of the technical knowledge of the time, but far from scientists. They did experiments in their homes, in schools, and, in Edison’s case, in a working industrial laboratory: they repeated experiments to correct flaws and problems in previous experi- ments, over and over again. It is, after all, the light-source industry that has given the world of technology the term of “Edisonian Research.” Despite the relative absence of scientific research in the initial development of incandescent lamps, it contributed significantly to optimization and refinement. Research in Germany and Austria led to means of production of refractory metals which eventually resulted in a significant optimiza- tion of the commercially successful carbon-filament incandescent lamp by the replacement of graphite with tungsten (Bright 1949, pp. 183ff). Further optimization of the incandescent lamp resulted from the scientific work of Irving Langmuir at GE Research Laboratories in the convective heat loss from heated bodies in a gaseous atmosphere and the concomitant discovery of the “Langmuir sheath” (1912). This resulted in the development of the gas-filled incandescent lamp employing a coiled filament (to shorten the length of filament losing heat by convection). The reduction in evaporation rate contributed by the gas filling permitted higher-temperature operation of the filament for the same life and a corresponding significant improvement in luminous efficacy. The development of the fluorescent lamp traces back to the work of Geissler in 1856 and Faraday, Crookes, and others who discovered that AC current passed through low-pressure gases would generate light of a color specific to the gas being used (Bright 1949, pp. 218ff). The Moore tube followed, employing nitrogen for golden-yellow light or carbon dioxide for white light, but significantly more efficient than the carbon-filament incandescent lamps of the time. Cooper-Hewitt used mercury at a pressure of some torr to emit a ghastly blue-green light. Again, all of these were predominantly experimentation by technologists rather than the outcomes of scientific research. When the development of techniques for liquefaction of air and separation into components made neon available, neon filling was aggressively pursued by Georges Claude in Europe, who invented a Page 3 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 hollow cold cathode that had extremely long life and made a commercial reality of the use of “neon lamps” for advertising, first in Europe and then in the USA. Mercury vapor and internal phosphor coating (first innovated by Jacques Risler in France) for a greater variety of colors then set the stage for the invention at GE in the USA of the hot-cathode fluorescent lamp (Bright 1949, Chapter XIV), announced in 1938. GE engineers had explored experimentally ranges of values of parameters such as tube diameter, fill gas, fill gas pressure, mercury pressure, and discharge current to optimize the performance of the lamps. However, there was little or no guidance from science for this optimization or, for that matter, the reason for the astonishing efficiency of generation of 254-nmmercury resonance radiation by the rare-gas-mercury discharge (60+ %). That did not become available until the pioneering work of Carl Kenty, Mary Easley, and Bentley Barnes at GE (1950–1951). Using experimental probe measurements of electron temperature in the plasma by Easley, Kenty calculated excitation and 3 3 quenching rates of the mercury energy levels to show the importance of the P0 and P2 metastable states of mercury; his calculations showed that essentially all of the excitation into these states was 3 transferred to the P1 radiating state, effectively tripling its excitation cross section (Kenty 1950). Experiments and calculations by Kenty, Easley, and Barnes determined the elastic collision energy loss by electrons and showed the importance of the Ramsauer minimum in the elastic collision cross section of argon: it coincides with the maximum in the electron energy distribution, thus minimizing the elastic collision energy loss by the electron gas (Kenty et al. 1951). These two factors contribute notably to the efficiency of resonance radiation by the discharge. Kenty’s work was extended in 1956 (18 years after the commercial announcement) byWaymouth and Bitter, who developed a complete ab initio model of the positive column of the rare-gas-mercury discharge (employing two adjustable constants), with which a specification of the tube diameter and length, rare gas and its pressure, mercury vapor pressure, and discharge current produced results within 10 % of experiment for electron temperature, electron density, UVoutput, power consump- tion, and maintaining electric field (Waymouth and Bitter 1956). Whereas the W-B model treated radially averaged quantities, a later model by Cayless calculated the full radial dependence of all quantities directly (Cayless 1960, 1963). The most important aspect of these model calculations was that they permitted insight into the factors controlling efficiency. Waymouth used the W-B model to demonstrate that the sublinear dependence of UVoutput on power input of the standard fluorescent lamp (and consequent decrease 3 in efficiency with increasing power) resulted from approach of the mercury P manifold to thermodynamic equilibriumwith the electron gas. Therefore, by increasing the electron temperature, lamps could be made with 2.5 times the output while still maintaining 80+ % of the efficiency. This was accomplished by reducing fill pressure and substituting neon and/or helium for argon, to increase the ambipolar diffusion loss rate and the required ionization rate. This insight resulted in the development of a line of high-powered fluorescent lamps that found wide application in high-bay industrial and commercial applications until they were supplanted by metal-halide lamps (Waymouth et al. 1957; Gungle et al. 1967). The same principle is still used in compact fluorescent lamps (CFL’s) which are necessarily operated at high power density. The development of phosphors for fluorescent lamps has been almost entirely empirical, guided in general by scientific knowledge, but mainly the result of patient experiments, testing activators in a variety of host crystals, and testing host crystals with a variety of activators. An even greater passage of time (50 years!) was required before the #218 formula and process of Aladar Pacz at GE (1917) for producing non-sag tungsten wire was explained by the investigations of Ronald Koo at Westinghouse Research Laboratories (1967). See section “The Importance of Materials in Light-Source Development and Optimization” for details. Page 4 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 The high-pressure mercury vapor lamp was also developed empirically in the 1930s, initially at Philips in Holland, Osram in Germany, and GEC in England, first as a one-atmosphere-pressure discharge in an aluminosilicate hard-glass tube, with a luminous efficacy of ca 40 lum/W. The applied research of Willem Elenbaas at Philips through the latter half of the 1930s and the 1940s and summarized in his 1951 book (Elenbaas 1951) provided a complete understanding of this commer- cially important light source and demonstrated that efficiency increases with increasing power per unit length. Practically, this was realized by replacing the hard-glass burner by a shorter and smaller diameter quartz tube, resulting in an efficacy of 50 lum/W. In the course of his applied research, Elenbaas developed the Elenbaas-Heller equation for calculating the radial temperature distribution of a high-pressure plasma in local thermodynamic equilibrium (LTE), essentially a continuity equation for the transport of heat from discharge to the wall. The divergence of the radial heat flux is equal to the net local production of heat, the difference between local electrical power input and local radiation loss: 2 DivfgradðkTÞg ¼ PðheatÞ ¼ sðTÞE eðTÞ 2 in which k is heat conductivity, T is local temperature, s(T)E is electrical power input per unit volume, and e(T) is net radiation per unit volume (emission less absorption). This equation became very important in later years in the analysis of the more complex plasmas of high-pressure-sodium (HPS) lamps and metal-halide (MH) lamps. Another instance of science contributing many years after the fact to the improvement of a light source comes from the chemical photographic flashbulb, originally developed in the 1930s with aluminum combustible in oxygen by Philips in Holland. Licensed to Wabash in the USA, it was the foundation for a significant business for Sylvania, which acquired all the assets of Wabash in 1946. In the 1960s at the Bayside, New York, laboratories of Sylvania (by then a subsidiary of GTE), unpublished thermochemical calculations by Bernard Kopelman showed that the adiabatic reaction temperature of zirconium and oxygen was several hundred degrees higher than that of aluminum and oxygen. A following development program at Sylvania confirmed the superiority of Zr combustible, resulting in flashbulbs having equal output to the then-common Press-25 standard, but having less than one-quarter the volume. The new bulbs were small enough that they could be packaged in TM disposable magazines, of which the battery-ignited “Flashcube ” was the first (Fink and Shaffer TM 1971) of a series, followed by the “Magicube ,” using percussively ignited bulbs (Brooks and TM Kopelman 1970), and the “Flip-Flash ” employing piezoelectric ignition, a GE development (Blount 1976; Weber 1976). Together with inexpensive “point-and-shoot” cameras equipped with sockets for the magazines, these developments revolutionized popular photography and permitted taking of excellent pictures by complete photographic novices. The flashbulb business was very profitable for a number of years until the chemical flashbulbs were replaced by even smaller xenon-flash lamps integral to the camera. The original low-pressure-sodium lamps, introduced to the market by Philips in 1932, employed a discharge tube made of a special sodium-resistant glass, a neon fill at some tens of torr pressure, and a charge of sodium metal and operated at more than 100W. Since a cold-spot temperature ca 220 C was required to vaporize the sodium, the burner was enclosed in an unsilvered Dewar flask. The relatively high power density and fill pressure of neon was deliberately employed to increase the elastic collision loss by electrons in order to get the bulb hot. As a result, the luminous efficacy was only about 50 lm/W in spite of the high specific efficacy of the sodium yellow emission. Page 5 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 Post-World-War-II research at Philips led to greatly improved thermal insulation: the burner was enclosed in a vacuum outer jacket, which was coated with an indium tin oxide IR-reflecting film. As a result, the burner could be completely redesigned with lower fill pressures and operated at a much lower power density and still reach the 220 C cold-spot temperature. These changes yielded much greater efficiency of generation of sodium resonance radiation and lamps delivering 200 lm/W (Elenbaas et al. 1969). The very large resonance broadening of the sodium resonance lines at high pressure was not anticipated by the early investigators of discharges in sodium vapor, but was capitalized on when a suitable discharge tube material (translucent polycrystalline alumina, “PCA”) became available as a result of studies at GE Research Laboratories (see Part The Importance of Materials in Light‐ Source Development and Optimization for details). The commercial high-pressure-sodium (HPS) lamp introduced by GE in 1964 was optimized empirically as to tube dimensions, sodium pressure, discharge current, and fill gas and pressure; the final design also incorporated mercury vapor at ca one atmosphere as a buffer gas to increase arc resistance and reduce heat conduction loss to the wall (Schmidt 1966). Despite considerable subsequent research into its plasma properties and energy balance, summarized by de Groot and van Vliet (1986), its design has changed little since. It is somewhat difficult to track the influence of science on the development of tungsten-halogen (TH) lamps. There is an extensive summary of the scientific literature on the thermochemistry and vapor pressures of metal halides included in reports prepared as part of the Manhattan project and published as a book by L. L. Quill in 1950 (Quill 1950). In addition, van Arkel and deBoer had used an iodine cycle to purify metals in the 1920s; in a heated bulb containing a reservoir of impure metal, iodine transported metal as an iodide to decompose on a central hot filament (van Arkel and de Boer 1925). In addition, in the early days of carbon-filament lamps, the so-called “Novak” lamp employed an atmosphere of bromine to keep the bulb wall free of deposits of evaporated carbon. Necessarily, there is no mention of any of these sources in the earliest patent on TH lamps (Fridrich and Wylie 1959), nor in the first published article (Zubler and Mosby 1959), so that it is hard for an outsider to tell if any of these sources were used to guide the development. Metal-halide lamps were invented almost simultaneously at Osram in Germany by Kuhl and Krense (1964) and by Reiling (1966) at GE in the USA. Kuhl’s application date was Aug 12, 1960, whereas Reiling’s was Jan 23, 1961. There was prior art in the area of additions of halogens and halides to discharge lamps that both patents had to get around, but it was hardly scientific. Steinmetz (1911) patented the addition of halides to mercury-pool-cathode discharges to add additional spectral lines in 1911, Neunhoeffer and (Schulz 1954) patented the addition of a halogen to high- pressure-xenon arc lamps to keep the bulb clean, and Beese and Henry (1956) patented the use of metal halides in a non-mercury-containing vapor arc to generate ultraviolet light for signaling in 1956. Since the US patent laws at the time regarded a development predictable by scientific knowledge to be obvious to one of ordinary skill in the art and therefore not patentable, it was fortunate that there was none. Reiling (http://www.Americanhistory.si.lighting (click on: “Invention Factory twentieth century lighting,” then on “Laboratory: caution, inventors at work” then on “Metal halide scientific training”)) is quoted as saying that he was concerned that sodium iodide would be unacceptably corrosive to quartz until he used scientific knowledge to make thermochemical calculations con- vincing him that reaction would be miniscule. However, Kuhl’s patent makes a point to say that it was a surprise that sodium halides do not corrode quartz. Moreover, science may have been a bit misleading here. Although the chemical equilibrium at the quartz-salt interface may only result in a + few ppm, of Na in the quartz, it doesn’t stay there. Waymouth (1971) showed experimentally that negative charging of the outer surface of the quartz arc tube by photoelectrons emitted from the arc Page 6 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 tube mounting frame caused the positive sodium ions to migrate from the inner surface to the outer surface, become neutralized by capturing photoelectrons, and evaporate away as neutral atoms. The + depletion of Na ion concentration at the inner surface upset the thermochemical equilibrium and required further reaction of sodium iodide. By this process, essentially all the elemental sodium in the iodide dose would be lost in a few thousand hours, with disastrous consequences. In subsequent applied research and development of the enormous variety of possible halide additives, two sources of scientific information were extremely valuable (to me at least). One was the aforementioned book by Quill (1950) and the so-called JANAF tables of thermochemical data (Stull et al. 1971), and the other was the NBS tables of spectral line wavelengths and intensities by Meggers et al. (1961). The former provided grist for modeling calculations in determining radial temperature profiles and species concentrations, while the latter was useful in selecting elements to try. Even so, it could be misleading as well. Frederic Koury at Sylvania initially tested scandium in the belief that it would be an excellent UV emitter based on line intensities in the Meggers tables. + However, most of the UV emission comes from Sc , while Koury tested it in combination with + sodium iodide; the sodium reduced the arc temperature to the point that the Sc concentration was low, and the emission came primarily from neutral scandium, which is a lovely multiline spectrum of white light (Koury and Waymouth 1968). The sodium-scandium combination proved to be the preferred combination in the US market, because of its ~4,000 K color temperature, matching cool white fluorescent, whereas rare-earth blends of ~6,000 K color temperature were preferred in Europe. The widespread application of metal-halide technology generated a great deal of research (too voluminous to enumerate here) into thermodynamics and vaporization data of halide compounds, in particular into vaporization of complexes of mixed salts. There was a great deal of applied physics research in the 1970s, 1980s, and 1990s into the behavior of metal-halide discharges: highly sophisticated spectroscopic diagnostics and complex modeling calculations marrying determina- tions of species concentrations, including diffusional and convective transport, with Elenbaas-Heller temperature profile calculations incorporating detailed accounting of radiation emission and absorp- tion processes. While these have provided improved understanding of the multiplicity of processes active in such discharges, I am not aware that any of them resulted in lamps of improved performance in the marketplace. Those improvements mostly resulted from empirical investigations of additives, improved processing and materials, geometric tweaks to arc tube design, as well as the substitution of PCA for quartz as the arc tube material (see section “The Importance of Materials in Light-Source Development and Optimization”). Much of this work has been presented at conferences and published in the scientific and technical literature. Prior to WorldWar II, the lamp industry presented papers at conferences of, and published papers in the journals of, the several Illuminating Engineering Societies around the world, mostly using these channels to advertise new or improved products while revealing as little as possible of the technical details. Postwar, especially in the discharge lamp area, presentations at scientific conferences such as the Gaseous Electronics Conference in the USA and the periodic International Conferences on Ionization Phenomena in Gases provided avenues to exchange information with other technologists in the industry. Publications in journals such as the Journal of Applied Physics in the USA and the Journal of Physics D: Applied Physics in Europe served to advertise scientific and technological expertise and were not discouraged by the lamp companies. Other important sources are the in-house technical journals, such as the “Philips Technical Review,” the “Philips Research Reports,” the “Technische Wissenschaftliche Abhandlung der Osram Gesellschaft,” the “General Electric Review,” and the “GEC Journal.” Articles in these journals may be especially useful, since they are not so tightly constricted by page budgets as those in the general literature and hence able to Page 7 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 discuss the material in more detail. Since 1977, the International Symposium on the Science and Technology of Light Sources has provided a venue where the science of all light sources may be discussed by a truly international gathering of scientists, technologists, and yes, inventors. The first meeting at Loughborough University in England, organized by Prof. John Raffle and senior technologists at Thorn Lighting, has led to a continuing series meeting at intervals or 2–3 years, in 2012 for the thirteenth time at Rensselaer Polytechnic Institute in Troy, New York, USA. In sharp contrast to all of the above, the development of light-emitting diodes (LEDs) has rested entirely on the foundation of 50 years of semiconductor research and the technologies developed through that research. The development has proceeded almost entirely outside the traditional light- source companies, by people trained in semiconductor physics, chemistry, and technology. Without that background of science, LEDs would not have been possible. In recent years, the lamp companies have awakened to the threat to their established businesses and have taken a more active role in development and incorporating LEDs into marketable products. The Importance of Materials in Light-Source Development and Optimization The performance of every light source is limited by the capabilities of at least one of the materials of which it is composed. Accordingly, the electric lamp industry has been involved in the search for improved materials since its birth; it has been practicing “materials science” long before that has been recognized as a specialty. Although carbon in the form of graphite was selected as an emitter by nearly all the early inventors of incandescent lamps, there was a wide variety of precursor materials used to obtain the graphite filaments; Edison carbonized “Bristol Board” paper strips and Swan carbonized cotton thread and later, in 1883, carbonized extruded nitrocellulose. Although Edison’s original commercial lamps were derived from carbonized paper, he very shortly switched to bamboo fibers, which were used commercially as the filament precursor for a dozen years, because he could make more efficient longer-lived lamps with it. As a personal aside, some years ago on a visit to Japan, I was taken by my host, Makoto Toho, to a small park outside of Kyoto, a grove of bamboo. In the center of the park, reached by a short path, was a pedestal atop which was a bust of Thomas Edison. Toho-san told me that this commemorated the fact that this was the grove from which Edison got his original bamboo (Hachiman bamboo) (http://www.nelt.co.jp/english/products/useful/01.html). Incandescent bulbs in the early days were hand-blown from a leaded soft glass; this glass had a very wide working temperature range which made it ideal for manual glass-working activity, since it allowed the glass-blower a long time to complete the work. This was not an ideal glass for automatic glass-blowing processes because it takes too long for the glass to set before it can be discharged from the mold, making a high-speed machine uncomfortably large. After World War I, GE and Corning jointly developed an alternate soft glass from silica sand, soda ash, and limestone, “soda-lime glass” or just “lime glass,” with a narrower working temperature range much more suitable for automatic bulb-blowing machinery (see section “The Contribution of Automated Light-Source-Manufacturing Machinery”). This glass is still in use today for incandescent and fluorescent lamps. The most important material improvement in incandescent lamp technology was the replacement of carbon by tungsten as a filament material. This was merely the last refractory-metal substitute in a chain including osmium, tantalum, molybdenum, and their alloys. The early refractory-metal filaments were made by extruding metal powders in a binder into filaments, burning out the binder, Page 8 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 and sintering the metal to a solid. Although they were extremely fragile, commercial lamps using them were manufactured in Europe by a number of companies (the “Osram” firm’s name is a composite of “Osmium” and “Wolfram,” the German word for tungsten). Sintered-tungsten- filament lamps were introduced in the USA by GE in 1907 under patent rights purchased from the German Welsbach company (Bright 1949, p. 190). They were rather quickly superseded by filaments made from ductile tungsten drawn into fine wires. William D. Coolidge at GE Research Laboratories developed a process for preparing tungsten wires by hot-swaging a sintered-tungsten ingot to produce a fibrous grain structure oriented along the axis of the billet. The resulting rod could then be drawn in a narrow range of temperature (temperature decreased as drawing proceeded) through diamond dies to diameters less than a few micrometers. The drawing process further enhanced the fibrous grain structure to make the wire as flexible as a rope but which becomes rigid upon recrystallization by heating to service temperature. Lamps using this wire went on the market in 1911, and a patent covering this high-level-blacksmith process was issued in 1913 (Coolidge 1913). The primary advantage of tungsten over carbon is its lower vapor pressure, which permits the filament to be operated at a higher temperature for equal life, yielding a higher efficacy. An equally important materials development (although from a cost rather than performance perspective) was one of the earliest composite materials, the so-called “dumet” wire for glass-to- metal sealing, introduced into lamps in 1913 by GE in the USA and used around the world to this day for sealing to soft glass. Developed by Colin G. Fink of the GE Research Laboratory, this consisted of a core of a nickel-iron alloy having a thermal expansion coefficient less than that of glass, surrounded by a copper cladding having an expansion coefficient greater than that of the glass, in such proportions that the composite has the same expansion as the glass (Coolidge 1913). Prior to this development, the only satisfactory lead-in wire material was platinum, which was so expensive that burned-out bulbs were recycled to salvage the precious platinum wires. The introduction of the gas-filled incandescent lamp as a result of the work of Langmuir required that the tungsten filament be coiled to shorten the length of filament to reduce the loss of heat by convection. The ductile-tungsten filament produced by the Coolidge process had a distressing tendency to “sag” in operation, causing turns to short out, develop hot spots, and fail. This was due to grain growth at high temperatures with grain boundaries extending perpendicular to the axis of the wire. Slip could occur along these boundaries, “offsetting” the crystals causing the wire to elongate and sag. This problem was solved through the work of Aladar Pacz, at the GE Lamp Division in Cleveland, who tested many additives to prevent sag, in a classic “Edisonian Research” procedure. On the 218th test, incorporating a dopant including potassium and silica, he achieved success (Bright 1949, pp. 207, 325). Upon heat treatment at the recrystallization temperature, the crystals that grew were elongated along the axis of the wire and had irregular side walls which “interlocked” with adjacent crystals. Thus, there were no perpendicular grain boundaries, and the locking of the crystals together prevented sag. There are several apocryphal tales about this development. In one, Pacz had written the formula on the cuff of his shirt sleeve at the time of the experiment; the shirt had been sent to the laundry by the time he received the results of the wire test and had to be hastily retrieved before the formula was washed out. Another tale has the formula scribbled on the inside of a matchbook cover that had to be retrieved from the trash heap. In any case, the number “218” was used by GE as the commercial designation of its “non-sag” tungsten wire. Although Pacz received a US patent for this work (Pacz 1922), the patent was later invalidated in the courts, which claimed that the invention was anticipated in the Coolidge patent (1913). In the specification of the Coolidge patent, the inventor described as an essential part of his process the heating of tungsten oxide in Page 9 of 31 Handbook of Advanced Lighting Technology DOI 10.1007/978-3-319-00295-8_1-1 # Springer International Publishing Switzerland 2014 “Battersea crucibles” for the express purpose of contaminating it with substances evolved from the crucible. If the degree of contamination was less than 0.8 %, a repeat firing in a new crucible was required. Since the Battersea crucibles were glazed with a potassium silicate glaze, it was held that Coolidge had anticipated the invention of Pacz. There was no fundamental understanding of how the Pacz process worked until the work of Ronald Koo at Westinghouse Research Laboratories in 1967 (Koo 1967) (ca 50 years later). By transmission electron microscopy of thin sections of the recrystallized tungsten, Koo was able to demonstrate the presence of “strings” of nano-voids (ever after known as “Koo Voids”) or “bubbles” containing traces of potassium in the metal. The inclusion of potassium silicate in the doped tungsten oxide resulted in small “globules” of elemental potassium in the tungsten ingot after sintering in hydrogen. Potassium is not soluble in tungsten and does not form alloys with it. In subsequent swaging and drawing steps, the spherical globules are elongated into “needles” extending along the axis of the rod or wire. When the tungsten is annealed, these needles break up into strings of potassium drops. Further drawing and annealing operations elongate the drops into fibers and the fibers into nano-drops, continuing in repeated cycles of drawing and annealing; the resulting strings of potassium bubbles in the finished wire prevent grains from growing in the transverse direction when the wire is recrystallized and forces the crystals to grow along the axis to form the necessary interlocking structure. The voids themselves also serve as sinks for dislocations in the tungsten crystals themselves, increasing the integrity of the crystal structure. As noted already, the discovery and optimization of luminescent materials (“phosphors”) for fluorescent lamps is primarily an empirical process, testing various compounds (“host crystals”) and various luminescent centers (“activators”). In such materials the only limitation on the ingredients is that the host crystal must be transparent to the 254-nm ultraviolet radiation of the mercury-rare-gas discharge, whereas the activator must absorb it. There are literally millions of possible combinations. One of the most important phosphor developments was the calcium halophosphate phosphor activated with antimony and manganese, discovered by McKeag and Ranby at GEC in England in 1942 (McKeag and Ranby 1949). It literally saved the fluorescent lamp industry. Prior to this development, the phosphors employed were calcium tungstate (blue) and zinc beryllium silicate (yellow) blends to make white of various color temperatures. At the time of the commercial introduction of the fluorescent lamp, the toxic properties of beryllium compounds were not recognized. Hundreds of workers preparing phosphors and manufacturing lamps were afflicted with pulmonary beryllicosis (similar to silicosis but much more severe) and many died as a result. In addition, disposal of lamps at the end of life created a breakage hazard, releasing the phosphor to the environment. Had not the halophosphate become available, it is extremely likely that production of fluorescent lamps would have had to shut down. The halophosphate compounds are no more toxic than the enamel on your teeth, to which they are chemically similar. By varying the ratio of antimony (which fluoresces blue) to manganese (which fluoresces yellow), the same range of “white” color temperatures as before could be achieved. Moreover, the quantum efficiency of halophosphate phosphors was greater than that of the prior phosphors, and they had better maintenance of light output. Within a short time, all manufacturers changed to the new phosphor, and the episode remains mostly forgotten except by the relatives of the victims. Another major phosphor development was the so-called “three-band” phosphor system at Philips by Vrenken and co-workers in the mid-1970s (Vrenken 1978). These were aluminate compounds 2+ 3+ 3+ activated with rare-earth ions: Eu for blue, Tb for green, and Eu for red. Characteristically these luminescent centers radiate only in narrow bands located at or near the wavelengths for maximum stimulus of the RGB cones of the human visual system. In addition, the quantum efficiency of all of these phosphors was greater than 95 % and they responded well to the 185-nm Page 10 of 31