LECTURE VI. LUMINESCENCE. 43. All methods of producing radiation, and more particularly light, other than the temperature radiation or incandescence, are generally comprised by the name luminescence. Some special cases of luminescence have already been discussed in the phe- nomena of fluorescence and phosphorescence, represented by the conversion of the radiation absorbed by a body into radiation of a different wave length. Usually luminescence at ordinary temperature, or at moderate temperatures, that is, temperatures below incandescence, is called fluorescence or phosphorescence. Fluorescence and Phosphorescence. Fluorescence is the production of radiation from the energy supplied to and absorbed by the fluorescent body, while phos- phorescence is the production of radiation from the energy stored in the phosphorescent body. This energy may be derived from internal changes in the body, as slow combustion, or may have been received by the body at some previous time — as by exposure to light a calcium sulphide screen absorbs the energy of incident radiation, stores it in some form, and afterwards radiates it. Fluorescence and phosphorescence usually occur simulta- neously : the energy supplied to such a luminescent body brings about certain changes in the body — as vibrations of the atoms, or whatever it may be — which cause the body to send out radia- tion. As long as this energy is supplied, the radiation of the body continues, that is, it fluoresces. The changes in the body which make it luminesce, represent energy storage — the kinetic energy of the luminescent vibration, etc. — and when the energy supply to the body ceases, the radiation issuing from the body does not instantly cease, but continues, with gradually decreasing intensity, until the stored energy is dissipated : the body phos- 94 LUMINESCENCE. 95 phoresces. Inversely, fluorescent radiation probably does not appear instantly at full intensity, as energy has first to be stored. The persistence of the luminescence after the power supply has stopped, as phosphorescence, is very short, except with a few substances, where it lasts for days. Where the energy of phos- phorescent radiation is supplied by the energy of chemical change in the body — as with yellow phosphorus — obviously the phosphorescence persists as long as these chemical changes can occur. The different forms of luminescence may be distinguished by the character of the energy which is converted into radiation. The conversion of radiation energy into radiation of different wave length, either immediately, or after storage in the body, thus may be called radio-fluorescence and radio-phosphorescence. It was discussed in Lecture II. The same bodies, exposed to an electric discharge in a vacuum (Geissler tube or Crooke tube) show electro-luminescence, fluores- cence as well as phosphorescence, and usually with the same color as in radio-luminescence. Thermo-luminescence is exhibited by some materials, as the violet colored crystals of fluorite (CaFl2), which, when slightly warmed, luminesce — it is this which gave the name "fluores- cence" to the phenomenon. Some solutions, when crystallizing, show light during the formation of crystals, and thus may be said to exhibit a physical phosphorescence. Chemical phosphorescence is exhibited by yellow phosphorus and its solutions, which in the air glow by slow combustion, at ordinary atmospheric temperature. As the ignition point of phosphorus, that is, the temperature where it spontaneously ignites, is little above atmospheric temperature, the chemical phos- phorescence of phosphorus occurs at temperatures a few degrees below ignition; it ceases, however, at very low temperature. The chemical luminescence, as shown by phosphorus, is not an exceptional phenomenon, but many substances exhibit chemi- cal phosphorescence at temperatures a few degrees below their ignition temperature, as the result of slow combustion. With those substances which have an ignition point above incandes- cence, this cannot be observed, but it is observed, for instance, in carbon bisulphide, CS2, which ignites spontaneously at about 96 RADIATION, LIGHT, AND ILLUMINATION. 180 deg. cent., and a few degrees below this temperature phos- phoresces in air, by slow combustion. A biological phosphorescence is shown by many forms of life: some bacilli of putrefaction phosphoresce, and are the cause of the faint glow occasionally observed in decaying food, especi- ally fishes. Amongst insects, numerous sea animals of dif- ferent classes, especially deep-sea animals, phosphorescence is frequently met, but its origin, that is, the mechanism of light production by the firefly, etc., is still unknown. When splitting a sheet of mica, or shaking a well-exhausted tube containing mercury, flashes of light are seen in the darkness. This, however, is not real phosphorescence but due to electrostatic flashes of frictional electricity. The light given by fluorescence and phosphorescence of solids or liquids, gives a continuous spectrum, that is, is a mixture of all frequencies, just as is the case with temperature radiation; it differs, however, from temperature radiation by the distribu- tion of the energy in the spectrum, which is more or less charac- teristic of the luminescent body, and to some extent, also, of the method of exciting the luminescence. Thus crystalline calcium tungstate, W04Ca, fluoresces white in the X-ray, light blue with ultra-violet light; the aniline dye, rhodamine, 6 G, in alcoholic solution fluoresces green in daylight, crimson in the light of the mercury lamp; willemite (calcium silicate) shows a maximum fluorescent radiation in the green, some chalcites in the red, etc. So far, fluorescence and phosphorescence nave not yet found any extended industrial application. 44. Some of the characteristic forms of luminescence at higher temperatures are pyro-luminescence, chemical-luminescence, and electro-luminescence. As pyro-luminescence or heat-luminescence, must be considered all radiation, produced by heat, which exceeds at some wave length the intensity of the black body radiation at the same temperature. Whether real pyro-luminescence exists, is uncertain, but by an extension of the definition any colored temperature radiation may be considered as heat luminescence of a grey body of an albedo which as normal temperature radiation would give the same total radiation at the same temperature as the colored radiator. Heat luminescence has been discussed already under colored radiation. LUMINESCENCE. 1 97 Chemical Luminescence. Whenever intense chemical changes take place at higher tem- perature, luminescence frequently occurs. I have here an ordi- nary, non-luminous bunsen flame. I dip a platinum wire into a solution of lithium chloride, LiCl, and then hold it into the lower edge of the flame: the flame colors a bright red, and through the spectroscope you see a bright deep red line and a less bright orange line, the spectrum of Li. After a little while, the color- ing disappears by the LiCl evaporating from the wire, and the flame again becomes non-luminous. I repeat the same experi- ment, but dip the platinum wire into sodium chloride, NaCl, solution, and you see the flame colored brightly yellow, and the spectroscope shows one yellow line, the sodium line D. Dipping the platinum wire into thallium chloride, T1C1, I color the flame a bright deep green, the characteristic Tl spectrum, which has one bright green line. As you see, the green coloring disappears more rapidly than the yellow did, and the flame turns yellow; the Tl salt is more volatile than the sodium salt, evaporates more rapidly, and as it contains some Na as impurity, the latter be- comes visible as yellow flame coloring after the Tl has evap- orated. In the bunsen flame these salts are evaporated, split up into their elements by the flame gases, and recombine, and by these chemical changes the atoms of Li, Na or Tl are set in vibration, and as vapors, being free to vibrate without mutual interference, they vibrate with their characteristic frequency, that is, give a definite frequency and thus color of the light, independent of the temperature; if we introduce the same salts into the carbon arc we get the same color and the same spectrum lines, only much brighter, as at the much higher temperature of the arc flame the vibration is far more intense; but it is of the same frequency, and in this respect essentially differs from temperature radiation which varies in frequency with the temperature. In the same manner by introducing Sr, Ba or Ca salts in the bunsen flame, the flame is colored with other character- istic colors; bright red, green, orange The spectroscope shows in every case a spectrum having a number of definite lines which are brightest and most numerous in the red for Sr, in the green for Ba, and in the orange yellow for Ca. In general, 98 RADIATION, LIGHT, AND ILLUMINATION. metal spectra show a number, frequently very many lines in the visible range. As Sr, Ba, Ca, are much less volatile than Li, Na, Tl, to get good effects in the bunsen flame, instead of the chlorides, the nitrates, or preferably the chlorates or perchlorates are used, which are more unstable, and thus easier split up and carried into the flame. At the much higher temperature of the carbon arc, the chlorides, or even the still more refractory oxides are used. Chemical luminescence is used industrially in fireworks and colored signal lights; salts of these metals with acids which con- tain a large amount of easily split off oxygen, as nitrates, or more commonly chlorates and perchlorates, are mixed with some com- bustible material, as charcoal, sugar, sulphur, antimony sulphide, etc. When ignited, the combustible burns with the oxygen given off by the nitrates or chlorates, and in the focus of this intense chemical action, intense luminescence of the metal is produced. Thus Sr gives a bright red, Ba a green, Ca an orange yellow, copper ammon a blue coloring. Electro-luminescence of Gases and Vapors. 45. Industrially this is the most important form of lumines- cence. Solids and liquids can be made to luminesce only indi- rectly by exposure to electric discharges, as electrical fluorescence. Gases, however — and under gases here and in the following we include vapors as, for instance, the carbon vapor, which is the conductor in the carbon arc — become electro-luminescent by being used as conductors of the electric current. It is a charac- teristic of electric conduction of the gases that this conduction is accompanied by the production of radiation, and in the electric conduction of gases we thus find the means of a more direct conversion of electric energy into radiation, and thus into light. It is, therefore, in this direction that a radical advance in the efficiency of light production would be possible, and the subject of electric conduction of gases (including vapors) thus is of the highest importance. Two forms of electric conduction in gases exist: disruptive conduction, as represented by the Geissler discharge or the elec- trostatic spark, and continuous conduction, as represented by the electric arc. LUMINESCENCE. 99 Disruptive Conduction. In disruptive conduction the conductor is the gas which fills the space between the terminals, and in carrying the current is made luminous. The color of the light and its spectrum is that of the gas which fills the space, and the electrode material has no effect on the phenomenon, is immaterial (in the Geissler tube, or the spark gap, any material may be used as terminal, if it •otherwise is suitable, that is, is not destroyed by whatever heat is produced at the terminals, or by the chemical action of the gas in the space, etc.) usually, however, the electrodes gradually dis- integrate in disruptive conduction. Disruptive conduction is discontinuous; that is, no current exists below a certain definite voltage, while above this voltage there is current. The voltage at which conduction begins is called the disruptive voltage. It is the minimum supply voltage at • which current exists : if the supply voltage rises above this value there is current; if it drops below the disruptive voltage the current ceases, but begins again spontaneously as soon as, the voltage rises above the disruptive value. Disruptive conduction thus occurs equally well with unidirectional, with alternating, or with oscillating currents. It is best studied with alternating or oscillating voltage supply, as with a steady unidirectional voltage, the disruptive conduction, that is, conduction by the gas filling the space between the electrodes, tends to change to continuous conduction, by vapors forming at the negative elec- trode and gradually bridging the space between the electrodes, and thereby replacing the gas which fills the space, by the elec- trode vapor as conductor. This is usually expressed by saying: the electrostatic spark between two terminals starts, or tends to start, an arc. Disruptive conduction, thus, does not follow Ohm's law; it is zero below the disruptive voltage, while with a supply voltage exceeding the disruptive voltage of the gas between the terminals, current exists, but the terminal voltage is apparently indepen- dent of the current, that is, if the other conditions as temperature, gas pressure, etc., remain the same, the terminal voltage of the Geissler tube or the spark gap remains the same and independent of the current, and the current is determined by the impedance between the. Geissler tube or spark gap and the source of 100 RADIATION, LIGHT, AND ILLUMINATION. e.m.f., or by the available power of the supply source. A Geissler tube, thus, cannot be operated directly on a constant potential supply of unlimited power, but requires a current limiting im- pedance in series with it, or a source of limited power, that is, a source in which the voltage drops with increase of cur- rent, as a constant current transformer or an electrostatic machine, etc. The disruptive voltage essentially depends on the gas pressure in the space between the electrodes, and also on the chemical nature, and on the temperature of the gas. It is over a wide range, directly proportional to the gas pressure. Thus, at n atmospheres pressure the voltage required to jump a spark between two terminals is n times as great as at one atmosphere. This law seems to hold from the highest pressures which have been investigated down to pressures of a few mm. mercury, that is, down to about T^ atmosphere. When coming to still lower pressures, however, the disruptive voltage decreases less, ulti- mately reaches a minimum — usually somewhere between 1 mm. and 0.1 mm. mercury pressure — and then increases again and at extremely high vacua becomes much higher than at atmos- pheric pressure, so that it seems that it is infinite in a perfect vacuum, that is, no voltage can start conduction through a perfect vacuum. As the gas filling the space is the conductor in disruptive conduction, it is easily understood that in a per- fectly inert space, or an absolute vacuum, no disruptive con- duction would exist. The visible phenomena of disruptive conduction very greatly change with the change of gas pressure; from the electrostatic spark at atmospheric pressures to the Geissler tube glow in the vacuum; but the change is gradual, thus showing the identity of the two phenomena. At atmospheric pressure, disruptive conduction occurs by a sharply denned, relatively thin and noisy spark of very high brilliancy, which traverses the space between the electrodes in an erratic zigzag path, not unlike in appearance to the mechanical fracture of a solid material; and, indeed, the spark is an electrostatic rupture of the gas. If the electrostatic field is fairly uniform, as between parallel plates, or between spheres of a diameter 1.5 or more times their distance, with gradual rising voltage, the spark occurs when the disruptive voltage is reached, without being preceded, at lower voltage, by LUMINESCENCE. 101 any other phenomenon. If, however, the electrostatic field is not uniform, as, for instance, between needle points or small spheres or wires, with increasing voltage the disruptive strength of the gas is exceeded at those places where the field intensity is highest, as at the needle points, before the disruptive voltage of the spark gap is reached, and then a partial break down occurs at the points of maximum field intensity, as at the needle points, or at the surface of high potential conductors, etc. A blue glow, then, appears at the needle points followed by violet streamers (in air, the color being the nitrogen spectrum; in other gases other colors appear), and gradually increases in extent with increasing voltage, the so-called " brush discharge," or " corona." Between needle points the brush discharges increase in extent, and approach each other until they bridge nearly 60 per cent of the gap, and then the static spark occurs. At higher gas pressures the spark increases in brilliancy, in noisiness, but gets thinner. If, however, we gradually decrease the gas pressure, the spark gets thicker, less brilliant, and less noisy, its edges are less sharply defined, that is, get more diffused, and ultimately it passes between the terminals as a moderately bright, thick and noiseless stream, gradually fading at its outside, and at still higher vacua it fills the entire space of the vacuum tube. At the same time the required voltage is decreased with decreasing gas pressure, as discussed above. 46. I show you here (Fig. 31) the gradual change from the static spark to the Geissler tube glow: in a closed glass tube G, I have two needle-shaped terminals, 5 cm. distant from each other, and supply them with energy from a small 33, 000- volt trans- former. You see the oscillating static spark at atmospheric pressure. By now exhausting the tube, while the voltage is maintained at the terminals, you can watch the gradual change from the static spark to the Geissler tube glow. In this experi- ment, a small condenser, a Leyden jar, is shunted across the high- potential terminals of the transformer, to guard against the disruptive conduction changing to continuous conduction, that is, to an arc, and a reactance inserted into the low-tension pri- mary of the step-up transformer, to limit the discharge current, as shown diagrammatically in Fig. 31. If the Geissler tube has a considerable diameter, 3 to 5 cm., the Geissler discharge with alternating current is striated; that 102 RADIATION, LIGHT, AND ILLUMINATION. is, disk-shaped bright spots with diffused outlines alternate with less luminous spaces, about as shown in Fig. 32. The distance between the luminous disks increases with decrease of the gas pressure. Two sets of such disks exist, one issuing from the one, the other from the other terminal. They are stationary FIG. 31. only if the gas pressure is perfectly constant, but separate and contract with the slightest change of pressure, hence are almost never at rest, but constantly moving through each other. The two sets of disks, by passing through each other during their motion, give rise to a number of different appearances. Some of the successive shapes are shown in Fig. 32. The voltage distribution in the space between the terminals, in disruptive conduction, also: changes with the pressure- at LUMINESCENCE. 103 atmospheric pressure, practically all the voltage is consumed in the space between the terminals, and between needle points for distances of 10 cm. and over very closely 4000 volts effective alternating per cm. (10,000 volts per inch) are required (a 2-cm. gap between needle points, however, requires 10,000 volts). With II II II FIG. 32. decreasing gas pressure the voltage consumed in the space be- tween the terminals decreases,but the voltage consumed at the terminals increases, and in a good Geissler tube vacuum with nitrogen gas filling the space between the terminals, from 1000 to 3000 volts may be consumed at the terminals, while the voltage consumed in the space between the terminals may drop as low as 2 volts per cm. The voltage consumed at the terminals seems to decrease v ith increase of their size. The voltage consumed in the space be- 104 RADIATION, LIGHT, AND ILLUMINATION. tween the terminals, that is, in the luminous stream of the Geiss- ler tube, seems to be practically independent, not only of the current, but also of the size of the tube, as should be expected with a disruptive discharge. It varies, however, with the tem- perature, and is different with different gases, that is, different gases have different disruptive strength. The light given by the Geissler tube shows the spectrum of the gas, and thus is very bright and fairly efficient with a gas as nitro- gen, which gives a large number of spectrum lines in the visible range, and less efficient with a gas as carbon dioxide or hydrogen, in which the lines in the visible range represent only a small part of the radiated energy. The industrial use of the electro-luminescence of discontinuous conduction, that is, Geissler tube lighting, is still very limited (Moore tube). So far only nitrogen gives a fairly good efficiency, reaches apparently values between the tungsten lamp and the tantalum lamp ; or, a specific consumption of two watts per mean spherical candle power. The color of the nitrogen spectrum is a golden yellow. As the range of gas pressure in which the voltage is near the minimum is very narrow, and the gas pres- sure changes during operation, by absorption at the electrodes, etc., means have to be provided to maintain constant gas pressure by automatically feeding gas into the tube whenever the pres- sure drops below the minimum voltage or maximum efficiency point. The greatest disadvantage of Geissler tube lighting, however, is the high voltage required at the terminals. To get fair efficiency the tube must be so long that the voltage con- sumed in the stream — which represents the power converted into light — is much larger than the voltage consumed at the terminals — which represents wasted power. With a terminal drop of 2000 volts, and two volts per cm. in the conducting gas stream, to use half of the supply voltage for light production, thus 2000 requires a tube length of — — - = 1000 cm. = 10 m. or 33 feet, 2 and to use 80 per cent of the supply voltage for light production, that is, waste only 20 per cent of the supplied power in heating onrvrj the terminals, requires a tube length of — — = 40 m. or 133 feet. of the mercury spectrum. The terminals are quiet, as they do not participate in the conduction. I now connect terminals 1 and 2 through a resistance, to 'a direct current supply, and tilt the tube momentarily to let some mercury run over from 2 to 1, and by thus momentarily connecting these terminals, establish the current and so start the arc, and you see the mercury arc pass between terminals 1 and 2, and see at one terminal — the negative one — a rapidly moving bright spot, which marks the point from which the vapor stream issues which carries the cur- rent. We have here in one and the same vacuum tube, and with the same material — thus, the same color and spectrum of light, both types of conduction — the continuous high current and low voltage conduction of the mercury arc, and the striated high voltage low current disruptive conduction of the Geissler dis- charge through mercury vapor. The conducting vapor stream which carries the current in the arc, at least in all arcs which so far have been investigated, issues 108 RADIATION, LIGHT, AND ILLUMINATION. from the negative terminal or cathode, and is in rapid motion from the negative towards the positive. The character of the arc, therefore, is determined by the material of the negative terminal, the temperature of the arc stream in general probably is the temperature of the boiling point of the negative terminal, 3J=10 OHMS FIG. 34. and the spectrum of the arc is the spectrum of the negative ter- minal. An exception herefrom, occurs only in those cases in which the positive terminal contains material which boils below the temperature of the arc stream (flame carbons) and the posi- tive terminal is made so small that its tip is raised to the temperature of the arc stream, and at this temperature heat evaporation of the material of the positive occurs. These vapors enter the arc stream, and there become luminous, possibly by chemical luminescence, and add their spectrum to that of the arc conductor, that is, the negative material. In this case the arc spectrum shows the negative as well as the positive material, or at least the more volatile components of the positive material. LUMINESCENCE. 109 With the exception of this case of heat evaporation from the positive terminal, the material of the positive terminal does not participate in the phenomena occurring in the arc. Thus the positive can be made of any conducting and refractory material, and if made sufficiently large not to get too hot, does not con- sume; only the negative terminal of the arc consumes in feeding the arc flame, that is, supplying the vapor conductor, but the positive is inherently non-consuming, and may be made a perma- nent part of the arc-lamp mechanism. On the contrary, if the positive is made so large that its temperature remains very much below the arc temperature, condensation of the arc vapor occurs at it, and it builds up, that is, increases in size. Consumption of the positive terminal is thus due merely to the heat produced at it by combustion or heat evaporation. While the arc conductor issues from the negative terminal, in general more heat is produced at the positive terminal. Thus with both terminals of the same size and material, as usual in the carbon arc, the positive gets hotter, and therefore in open air burns off faster, which has led to the erroneous assumption that the positive feeds the arc. While carbon is the material most commonly used as termi- nals, the carbon arc is not a typical arc, but is an exceptional arc. (1) Because carbon is one of the very few substances which change directly from the solid to the vapor state, that is, do not melt at atmospheric pressure, but boil below the melting point. (2) Carbon is the most refractory substance and the tempera- ture of the carbon arc higher than the boiling point of any other substance. Any material existing in the terminals of a carbon arc thus evaporates, and by entering the arc stream shows its spectrum, so that luminescent material can be fed into the carbon arc from either terminal. (3) At the temperature of the carbon arc all gases and vapors have become good conductors, and a carbon arc thus can operate equally well on alternating current as on direct current; that is, the voltage required to maintain the carbon arc is. sufficient, after the reversal of current, to restart it through the hot carbon vapor. A typical arc is shown in Fig. 35 as the magnetite arc, with a lower negative terminal M consisting of magnetite, the non-consuming upper terminal C of copper, and of such 110 RADIATION, LIGHT, AND ILLUMINATION. size that it does not get so hot as to oxidize or evaporate, but sufficiently hot to avoid condensation of magnetite vapor on it. The arc flame consists of an inner cylindrical core A, of bluish white color and high brilliancy, slightly tapering at both ends, which is surrounded by a less luminous shell B, of more yellowish color, narrowest at the negative end, and increasing in diameter towards the positive, surrounding the latter. The inner core A is the arc conductor, or con- ducting vapor stream, while the outer shell B is non-conducting luminous vapor, possibly containing particles of solid material floating in it as incan- descent bodies. The arc conductor A issues from a depression S in a melted pool P formed on the surface of the terminal M. This depression S is in a rapid and erratic motion, and thereby causes a constant and rapid flickering of the arc. It is this flickering, inherent to all arcs in which the negative terminal is fusible (which therefore does not exist in the carbon arc), which has retarded the industrial development of the more efficient metal arcs until late years. Its cause is the reaction exerted by the velocity of the vapor blast from the negative, which presses the surface of the liquid pool down at the point from which the current issues. The starting point of the current con- tinuously climbs up the side of this depression, in shortening the arc, but, in doing so, depresses its new starting point, that is, the depression S, and thereby the negative end of the arc stream moves over the surface the faster the more fluid the surface is. In the mercury arc, this phenomenon of the running spot at the negative terminal is also very marked, but not so objectionable, as the arc stream is so long that the flicker at the negative terminal has no effect on the total light. This flickering disappears in the magnetite arc if we destroy the fluidity of the melted magnetite by mixing with it some much more refractory material, as chromite. The chromite remains solid and holds the melted magnetite like a sponge. The reaction of the vapor blast, then, cannot depress its starting point, and no tendency exists of shifting the starting point, and the arc becomes FIG. 35. LUMINESCENCE. Ill steady. In this manner such arcs have now been made steady and thereby suitable for industrial use. 49. Since the arc conduction issues as a rapidly moving vapor stream from the negative terminal or cathode, it must be con- tinuous at the cathode; if interrupted even for a very short time at the cathode, a break exists in the continuity of the conductor and conduction ceases, that is, the arc extinguishes. At any other point of the arc stream, however, a break in the continuity of the stream may exist, provided that current continues from the negative, since such a break in the continuity of the con- ducting vapor stream is bridged again, and conduction re-estab- lished by the vapor stream coming from the negative. Thus the FIG. 36. arc can be started by merely starting a conducting vapor stream from the negative, as by an auxiliary arc. As soon as this con- ducting vapor reaches the positive terminal, it closes the circuit and establishes conduction. An arc can be shifted or jumped from one positive terminal to another one, but cannot be shifted from negative to negative; the negative terminal, as the source of the conducting vapor stream, must be continuous. To illustrate this, I have here (Fig. 36) in a hand lamp two copper rods A and B of about 5 mm. diameter, as arc terminals, separated by 2.5 cm., and connected into a 220- volt direct-cur- rent circuit, with sufficient resistances in series to limit the current to about 4 amperes. A third copper rod of the same size, C, is connected by a flexible lea.J to the upper terminal B. I close the reversing switch S so as to make A negative, and B and O , . '.. i \. , : ; ,- ,. . - 112 RADIATION, LIGHT, AND ILLUMINATION. positive, and start an arc between A and C by touching C to A. I draw this arc to about 4 cm. length, and without touching C with B, as soon as the conducting vapor stream of the arc AC (the inner core A of Fig. 35) touches B, as shown in Fig. 36, the arc leaves C and goes to B, that is, by the arc AC I have started arc AB. If I had separate resistances in series with the terminals B and (7, the arc AC would also continue to exist after it started arc AB- otherwise, as two arcs cannot run in parallel, the longer arc, AC, goes out as soon as the shorter arc AB starts. I now reverse the circuit by throwing switch S, and make A positive, and B and C negative, again start AC by contact, and draw it out until the arc flame wraps itself all around terminal FIG. 37. B} but the arc does not transfer. I even insert 10 ohms resist- ance rl in series with C (Fig. 37), so that the voltage AB is about 40 volts higher than AC, that is, B by 40 volts more negative than C, and still the arc does not transfer. I now touch C with B and separate it again; if during contact the negative spot during its motion happens to run over to terminal B, the arc con- tinues between B and A; if, however, the negative spot has remained on C, when separating again, the arc remains at C as negative, although B is more negative by 40 volts. An arc therefore can be started at its normal starting voltage by an auxiliary arc having the same negative, but not by an auxiliary arc with the same positive, and an arc can be shifted from one positive to another, but not from one negative to LUMINESCENCE. 113 another. The cause is, as explained above, the necessity of the continuity at the negative terminal as the source of the conduct- ing vapor stream. Still more startling is the following demonstration : I shift the resistance rl from C to B, and start the arc from A to B, with B as negative, by bringing these terminals into contact with each other, and then separating them. The auxiliary terminal C (Fig. 38) now is by 40 volts more negative than the negative terminal B of the arc. I now cut slowly through the arc stream by moving C across it between A and B, as shown in Fig. 38: the arc AB remains, but no current goes to C, although more r =20 OHMS ws FIG. 38. negative, that is, at a higher potential difference and a shorter distance against A than B is. I even hold C for some time in the conducting core of the arc AB, and still the current does not shift from the negative B to the still more negative terminal C. This experiment is interesting in demonstrating that a conductor immersed into the arc flame does not assume the potential of the arc flame, but may differ therefrom by considerable voltage, and that it therefore is not feasible to determine the potential dis- tribution in an arc by means of exploring electrodes, as has frequently been attempted. Obviously, if I now reverse the circuit, and make B and C positive, A negative, the current leaves B and goes to C as soon as C touches the conducting core of the arc AB. 50. The electric arc, therefore, is a unidirectional conductor, that is, the vapor stream is conducting between its negative 114 RADIATION, LIGHT, AND ILLUMINATION. terminal A in Fig. 36, that is, the starting point of the arc stream, and any point reached by it which is positive to A, but is non-conducting for any point which is negative with respect to A. If, now, in Fig. 38, with the terminal C immersed in the arc stream, I connect A and C to a source of alternating voltage, as shown in Fig. 39, while a direct-current arc flows from A to B, with A as negative, then during that half-wave of the alternating voltage, for which C is positive to A, there is current between A and (7, while for the reverse half-wave, in which C is negative to A, there is no current. The arc thus rectifies the alternating voltage, and the rectification is complete, that is, there is T= 30 OHMS MAA FIG. 39. current during one half-wave only, but no current at all dur- ing .the other. I show you this experimentally, using 50 volts alternating between A and (7. With this arrangement, to maintain the rectification continuously, obviously the ter- minal C would have to be cooled. Alternating voltage thus can be rectified by means .of the unidirectional character of the arc : if a continuous vapor stream is maintained from one terminal, either by direct-current ex- citation or by overlapping several waves of alternating cur- rent, current is in that direction only in which this exciter terminal is negative, but not in the opposite direction. . Such arc rectifiers — of which the mercury arc rectifier is the most commonly used — have been developed and extensively introduced in the industry, of la.te years, for operating, low-volt- LUMINESCENCE. 115 age constant direct potential and high-voltage constant direct- current circuits from a source of alternating voltage. Regarding the electrical phenomena occurring in arc rectification, see " Theory and Calculation of Transient Electric Phenomena and Oscillations/' Section II, Chapter IV. The inability of an alternating voltage to maintain an arc, I show you here on the same apparatus by connecting the two terminals (Fig. 40) A and B to the 1000-volt terminals of a transformer — with sufficient resistance in series to limit the current. While 220 volts direct current easily maintained a steady 2-cm. arc between these terminals, with 1000 volts alternating between the terminals, if I try to produce an alternating arc by gradually separating the terminals, the circuit opens before the terminals have separated 1 mm.; that is, 1000 volts alter- nating cannot maintain an arc of 1 mm. between these copper Fh i = iu unM5 v— W 110 VJOLTS 60 CYCLES FIG. 40. terminals. The cause is obvious: to maintain an arc between two terminals, a voltage is required sufficiently high to restart the arc at every half-wave by jumping an electrostatic spark between the terminals through the hot residual vapor of the preceding half-wave. The voltage required by an electro- static spark, that is, by disruptive conduction, decreases with increase of temperature: for a 13-mm. (0.5-in.) gap, it is about 10,000 volts at atmospheric temperature, 7000 volts at the boiling point of mercury (360 deg. cent.), 2500 volts at the boiling point of zinc (1000 deg. cent.), 500 volts at the boiling point of magnetite (2000 deg. cent.), 100 volts at the boiling point of titanium carbide (3000 deg. cent.), 40 volts at the boiling point of carbon (3500 deg. cent.). The voltage re- quired to maintain a 13-mm. alternating arc must therefore be 116 RADIATION, LIGHT, AND ILLUMINATION. Sit least as high as given by a curve somewhat like curve I in Fig. 41 * (to bring the values of voltage within the scale of the figure, the logarithm of voltage, as ordinate, is plotted against the temperature as abscissa). The voltage required to maintain an arc, that is, the direct- current arc voltage, increases with increasing arc temperature, and thereby increasing radiation, etc. For a 13-mm. (0.5-in.) FIG. 41. arc it is approximately shown as Curve II in Fig. 41 : 20 volts for the mercury arc, 40 volts for the zinc arc, 60 volts for the * As the disruptive voltage also depends on the chemical nature of the vapor, that is, some gases and vapors have a higher disruptive strength than others, as discussed above, the arrangement of the different materials regard- ing their alternating arc voltages is not entirely determined by their boiling points, but modified by individual characteristics. It further depends on the current: at higher currents and thus larger amounts of residual vapor, the voltage is lower. It further depends on the frequency: the lower the fre- quency and the greater, therefore, the cooling effect during the reversal of current the higher is the required voltage. LUMINESCENCE. 117 magnetite arc, 75 volts for the titanium carbide arc, 80 volts for the carbon arc.* As seen from Fig. 41, the curves I and II intersect at some very high temperature, near the boiling point of carbon, and materials which have a boiling point above the temperature of intersection of these curves require a lower voltage for restart- ing the arc than for maintaining it, and a voltage sufficient to maintain the arc restarts it at every half-wave of alternating current, that is, such materials can maintain a steady alternat- ing arc at the same voltage as a direct-current arc. Even materials like titanium carbide, in which the starting voltage is not much above the running voltage, maintain a steady alter- nating arc, as in starting, the voltage consumed during running in the steadying resistance or reactance is available. Alternating arcs thus can be maintained at moderate volt- ages only by a few materials of extremely high boiling points, as carbon and carbides, but by far the largest number of materials cannot be used as terminals of an alternating-current arc. In Fig. 41 the range between the curves I and II is the " rectify ing range," as in this range unidirectional current is produced from an alternating source of voltage through the arc, if the arc conductor is maintained by excitation of its negative terminal. The voltage range of rectification thus is highest in the mercury arc, which has the lowest temperature, and vanishes in very high-temperature arcs. The carbon arc thus cannot give complete rectification, while the mercury arc, or zinc arc, etc., can do so. The mercury arc, having the greatest recti- fication range, thus is practically always used for this purpose. Below curve II of Fig. 41 no conduction occurs, between curves I and II, unidirectional conduction takes place, and above curve I, disruptive conduction and alternating current can exist. 51. The light, and in general the radiation given by the arc proper, that is, by the vapor conductor which carries the cur- rent between the terminals, is due to luminescence, that is, to a more or less direct transformation of electric energy into * This voltage also is not merely a function of the arc temperature, but modified somewhat by the chemical individuality of the material. It is a function of the current and decreases with increase of current, so that above values are approximate only, corresponding to about 4 amperes. 118 RADIATION, LIGHT, AND ILLUMINATION. radiation, without heat as intermediary form of energy. The quality or color of the light, or its spectrum, that is, the fre- quency or frequencies of radiation given by the arc stream, thus are not a function of the temperature, as in the radiation produced by heat energy, but the frequencies are those at which the luminescent body is capable of vibrating, that is, are determined by the chemical nature of the luminescent body or vapor conductor. The efficiency of light production thus does not directly depend upon the temperature, does not in- crease with increase of temperature, as in temperature radia- tion, but to some extent rather the reverse. We have the same relation as in other energy transformations: when converting heat into other forms of energy, the more intense the heat, that is, the higher the temperature, the higher efficiency we may expect. When transforming, however, some form of energy differing from heat, into another form of energy, as mechanical into electrical energy, the heat produced repre- sents a waste of energy, and the lower the temperature, the higher in general, other things being equal, would be the effi- ciency. The efficiency of light production by the arc thus is not a function of the temperature, but the lowest temperature arc, the mercury arc, is one of the most efficient. The light given by the arc contains only a finite number of definite wave lengths, that is, gives a line spectrum: very few lines in the ordinary mercury arc, many thousands in the tita- nium arc. The color of the light is essentially characteristic of the nature of the luminescent body. For instance, it is white in the titanium arc, as the lines of the titanium spectrum are fairly uniformly distributed over the entire visible range. The light of the calcium arc is orange yellow, as the spectrum lines of calcium are more frequent and more intense in the orange- yellow range of radiation, etc. Frequently a change of the color of the luminescent light of the arc occurs with the temperature, but it does not follow a definite law, as in temperature radiation, but is a char- acteristic particularity of the luminescent body: some of the spectrum lines increase more rapidly in intensity, with increas- ing temperature, than others, and the resultant color of the light changes thereby. For instance, the ordinary iron arc, as produced by 4 amperes direct current across a gap of 2 cm. LUMINESCENCE. 119 between iron or magnetite terminals, and requiring about 75 volts, is white and very brilliant, that is, has a spectrum with many lines about uniformly distributed over the visible range. We can greatly increase the temperature of the arc by using a high-frequency condenser discharge: in this case very large currents of very short duration exist as oscillations between the terminals, with periods of rest between the oscillations, very long compared with the duration of the current. In this case the duration of the current is too short to feed a large volume of electrode vapor into the arc stream, and as the current is very large during the short moment of the discharge, the vapor between the terminals is very greatly overheated. Oscil- lating condenser discharges thus offer a means of increasing the temperature of the arc stream very greatly beyond the boiling point of the material. When using a condenser discharge be- tween iron terminals, we thus get an iron arc of very much higher temperature, and this arc gives very little visible light, but a very large amount of ultra-violet radiation. It is this arrangement which we have used in the preceding to produce ultra-violet light by the so-called " ultra-violet iron arc." In the iron arc the average wave length of the radiation thus shifts with increasing temperature to shorter wave lengths, or higher frequencies, similar as in temperature radiation. The reverse is the case with the mercury arc: the ordinary mercury arc in an evacuated glass tube, with ample condensing chamber, gives practically no red light; only a very powerful spectroscope can discover some very faint red lines. If now the condensation of the mercury vapor is made insufficient, by obstructing ventilation, or greatly raising the current, or omitting the condensing chamber in the construction of the lamp, and the mercury vapor pressure and thereby the tem- perature increased, at least three red lines located about as shown in Fig. 42 become visible in the mercury spectrum even in a low-power spectroscope, Ill Oi i\J W IJ\J W Cl Ok-'CLiLlWO^V^lJC, and increase in intensity with _J [ I m v. ^ ,/ YELLOW <^^-^_ ^ BLUE VlO increasing vapor pressure. To show you this I use a U-shaped FlG- 42- mercury lamp constructed as shown half size in Fig. 43. I con- nect the lamp into a 220-volt direct-current circuit, with an inductive resistance in series thereto, to limit the current, and 120 RADIATION, LIGHT, AND ILLUMINATION. start the arc by pouring some mercury over from one side to the other. Immediately after starting the lamp you see no red lines in the low-power spectroscope which I have here. As with the large current which I use — 3 amperes — the mercury vapor cannot freely condense, the mercury vapor pressure rises and FIG. 43. presses the mercury level down in the center tubes, up in the outside tubes, as indicated at b in Fig. 43, and thereby enables us to measure the mercury pressure. Gradually you see the three red lines appear, and increase in intensity, and when the vapor pressure has risen to about 5 cm., the three red lines are fairly bright, and numerous other red and orange mercury lines have appeared. At this pressure we are so close to the softening point of the glass that we cannot go further, but by operating the mercury arc in a quartz tube, vapor pres- sures of several atmospheres can be produced, and then the red lines are very much more intense, many more lines have be- come visible in the mercury spectrum, and the light is far less greenish than the low-temperature mercury arc, more nearly white. Still much higher temperatures can be reached in the mercury arc in an ordinary glass tube by using the condenser discharge. I have here, in Fig. 44, a mercury-arc tube with four LUMINESCENCE. 121 terminals — the same which I used in Fig. 34 for showing simultaneously the mercury arc and the Geissler discharge. I connect terminals 3 and 4 to the high potential terminals of a step-up transformer, but shunt a small condenser C across 3 and 4; you see, in the moment where I connect the condenser, the previously existing green and striated Geissler discharge changes r=so OHMS 3 r % \ \ -f- JL2JUL&L J>UP00Q 0000 roooo T III 1-^-40 1-7-40 1 r=.io OHMS 110 VOLTS 60 CYCLES 2 = 10 OHM 110 VOLTS 60 CYCLES *m FIG. 44. to a bright pinkish-red arc, and the spectroscope shows that the spectrum lines in the red and orange have greatly increased in number, and have increased in intensity beyond that of the lines in the green and blue, and the color of the light therefore has changed from green, to pinkish red. We have here in the same mercury tube shown diagram- matically in Fig. 44 all three forms of luminescence of mercury vapor: the high-current low-voltage, low-temperature arc of 122 RADIATION, LIGHT, AND ILLUMINATION. uniform green color, from 1 to 2; the green high-voltage low- current striated Geissler discharge, from 2 to 3, and the red high- voltage mercury arc, from 3 to 4. In the mercury arc, as result of the more rapid increase of intensity of the red lines, the color of the light thus changes with increase of temperature from bluish green at low tempera- ture to white to red at very high temperature, that is, the aver- age frequency decreases with increase of temperature, just the reverse from what is the case with temperature radiation. The change in the distribution of the power of radiation between the different spectrum lines, with change of tempera- ture, may increase the efficiency of light production — if the lines in the visible range increase faster than in the ultra-red and ultra-violet — or may decrease — if the visible lines in- crease slower — or may increase in some temperature range, decrease in some other temperature range, but all these changes are characteristic of the luminescent material, and do not obey a general law. Thus in the mercury arc the efficiency of light production, with increase of temperature, rises to a maximum at about 150 deg. cent., then decreases to a minimum, and at still higher temperature increases to a second maximum, higher than the first one, possibly between 600 and 800 deg. cent., and then decreases again. 52. Essentially, however, the efficiency of light production by the arc is a characteristic of the material of the arc stream, and thus substances which give a large part of their radiation as spectrum lines in the visible range — as calcium — give a very efficient arc, while those substances which radiate most of their energy as lines in the invisible, ultra-violet or ultra-red — - as carbon — give a very inefficient arc. The problem of efficient light production by the arc therefore consists in selecting such materials which give most of their radiation in the visible range. Carbon, which is most generally used for arc terminals, is one of the most inefficient materials : the carbon arc gives very little light, and that of a disagreeable violet color; it is practi- cally non-luminous, and the light given by the carbon arc lamp is essentially incandescent light, temperature radiation of the incandescent tip of the positive carbon. The fairly high effi- ciency of the carbon arc lamp is due to the very high tempera- ture of the black body radiator, which gives the light. LUMINESCENCE. 123 The materials which give the highest efficiencies of light production by their spectrum in the arc stream arcs mercury, calcium and titanium. As mercury vapor is very poisonous, the mercury arc has to be enclosed air-tight, and has been developed as a vacuum arc, enclosed by a glass or quartz tube. Its color is bluish green. Calcium gives an orange-yellow light of very high efficiency, and is used in most of the so-called "flame-carbon arcs," or "flame arcs." Titanium gives a white light of extremely high efficiency. It is used in the so-called "luminous arc," as the magnetite arc in direct-current circuits, the titanium-carbide arc in alternating- current circuits. 53. Two methods exist of feeding the light-giving material into the arc stream : (1) By electro-conduction, that is, using the material as the vapor conductor which carries the current. In this case, it must be used as negative, as the vapor conductor is supplied from the negative; such arcs are called "luminous arcs." (2) By heat evaporation; in this case, a very hot arc must be used, and thus usually a carbon arc is employed. As the posi- tive terminal is the hottest, the material is mixed with the car- bon of the positive terminal, and as negative terminal either a plain carbon, or also an impregnated carbon used; such arcs are called "flame arcs." The method of heat evaporation is always used witri calcium, since no stable conducting calcium compound is known which may be used as negative arc terminal. With titanium, usually electro-conduction is employed, that is, a titanium oxide-mag- netite mixture, or titanium carbide, used as negative terminal, and any other terminal, as copper or carbon, as positive ter- minal. Titanium can also be introduced by heat evaporation by using a titanium-carbon mixture as positive terminal or as both terminals of the flame-carbon arc. Both methods of feeding — electro-conduction and heat evapo- ration — have advantages and disadvantages. Electro-conduction has the great advantage that the tem- perature of the terminals is immaterial, as heat plays no part in feeding the luminescent material into the arc flame. The posi- tive terminal of the arc can be made sufficiently large and of 124 RADIATION, LIGHT, AND ILLUMINATION. such material as not to consume at all, and the trimming of the lamp thus reduced to the replacing of one electrode only — the negative. The negative electrode also can be made so large as to remain fairly cold, and therefore consumes only at the very slow rate required to supply the arc vapor, but does not con- sume by combustion or heat evaporation. Thus its rate of con- sumption can be reduced to 1 mm. or less per hour (while the open carbon arc of old consumes about 5 cm. of electrodes per hour), and thereby even with a moderate size of electrode a life of electrodes of 100 to 200 hr. or even much more secured. This method of feeding thus lends itself very well to long-burning arcs, as they are almost exclusively used for American street lighting. By electro-conduction higher efficiencies can be reached than by heat evaporation, as the arc vapor stream when produced by electro-conduction can be made to consist entirely of the vapor of the luminescent material, as when using metallic titanium as negative terminal. A disadvantage of the method of feeding the arc by electro- conduction is the much greater limitation in the choice of materials: the material must be an electric conductor, which is stable in the air, and reasonably incombustible. In the method of feeding by heat evaporation any material can be used, as it is mixed with carbon, and the conductivity is given by the carbon. Thus, in the titanium arc, either metallic titanium or titanium carbide or sub oxide must be used, but the most com- mon titanium compound, Ti02, or rutile, is not directly suitable, since it is a non-conductor. In the direct-current titanium arc, the so-called magnetite arc, a solution of Ti02, or rutile, in mag- netite, Fe304, which is conducting, is used, that is, a mixture of rutile with a considerable weight of magnetite. While mag- netite also gives a luminous arc, — the white iron spectrum, — the efficiency of the iron arc is lower than that of the titanium arc, and the efficiency of the magnetite arc thus lower than that of the pure titanium arc, though much higher than that of the carbon arc. Calcium cannot be used at all by electro-conduction: the only more common conducting calcium compound is calcium carbide. As negative terminal calcium carbide gives an arc of an efficiency far superior to that of the flame-carbon arc, but, as calcium carbide disintegrates in the air, it cannot be used. LUMINESCENCE. 125 Still greater is the limitation for alternating current; in this case the material, in addition to its other qualifications, must have such a high boiling point as to maintain a steady alternat- ing arc, as discussed above. Of the titanium compounds only titanium carbide seems to fulfill this requirement; of the iron compounds, apparently none. 54. The most serious disadvantage of the use of electro- conduction for feeding the arc, however, has been the inherently greater unsteadiness of metal arcs compared with the carbon arc. It is this feature which has retarded the development of true luminous arcs until recent years, that is, until means were found to produce steadiness by eliminating the nickering of the negative spot by the admixture of a more refractory material, — chromite in the magnetite arc, — and eliminating the unsteadiness due to the occasional momentary fading out of the luminous inner core of the arc by the admixture of a very small amount of some more volatile material. The great advantage of the method of feeding the luminescent material into the arc flame by heat evaporation, mainly from the positive, is the possibility of using carbon as arc conductor, which gives the inherent steadiness of the carbon arc, and thus has led to the development of this type of high efficiency arc, the flame arc, before the development of true luminous arcs. A further advantage is the possibility of using alternating current equally well and with the same electrodes as used with direct current, as the arc is a carbon arc and thus operative on alternating current Another advantage is the great choice of materials available, since practically any stable compound, whether conducting or not, can be used in the flame carbon. Thus in the yellow-flame arc, calcium fluoride, oxide and borates are used; in the tita- nium arc, the oxide (rutile) or the carbide may be used. The most serious disadvantage of the method of feeding by heat evaporation, which has so far excluded the flame arc from general use for American street illumination, is the rapid con- sumption of the electrodes and their consequent short life. Since the luminescent material is fed into the arc by heat evap- oration, the electrodes must be so small that their ends are raised to arc temperature, and thus rapidly consume by the combus- tion of the carbon. The combustion cannot be reduced by 126 RADIATION, LIGHT, AND ILLUMINATION. excluding the air by enclosing the arc with an almost air-tight globe, as in the enclosed carbon arc, since the luminescent material leaves the arc as smoke, and by depositing on the globe rapidly obstructs the light. The rate of consumption of the electrodes thus is the same as in the open carbon arc, 3 to 5 cm. (1 to 2 in.) per hour, and the flame-carbon arc even with very great length of carbon thus lasts only one night, that is, requires daily trimming. To some extent this difficulty may be reduced by using the same air again, after passing it through a smoke-depositing chamber in a so-called " circulating" or "regenerative" flame lamp. The mercury arc, being enclosed in a glass tube, necessarily must always be fed by electro-conduction from the negative. The calcium arc is always fed by heat evaporation from the carbon positive, with a carbon negative, or from positive and negative, by using flame carbons for both electrodes. The titanium arc is usually fed by electro-conduction from the neg- ative, but also by heat evaporation from the positive by using a titanium-flame carbon. 55. As, by electro-luminescence, electric energy is converted more directly into radiation, without heat as intermediary form of energy, no theoretical limit can be seen to the possible effi- ciency of light production by the arc, and in the mercury, cal- cium and titanium arcs, efficiencies have been reached far beyond those possible with temperature radiation. Thus, specific con- sumptions of 0.25 watt per mean spherical candle power are quite common with powerful titanium or calcium arcs, and even much better values have been observed. It is therefore in this direction that a radical advance in the efficiency of light pro- duction appears most probable. At present, the main disad- vantage of light production by the arc is the necessity of an operating mechanism, an arc lamp, which requires some atten- tion, and thereby makes the arc a less convenient illuminant than, for instance, the incandescent lamp, and especially the limitation in the unit of light: the efficiency of the arc decreases with decrease of power consumption, and, while the arc is very efficient in units of hundreds or thousands of candle power, its efficiency is much lower in smaller units, and very small units cannot be produced at all. Thus, for instance, while a 500- watt flame arc may give 10 times as much light as a 500-watt LUMINESCENCE. 127 carbon arc, to produce by a flame arc the same amount of light as given by a 500-watt carbon arc requires very much more than one-tenth the power. So far no way can be seen of maintaining the efficiency of the arc down to such small units of light as represented by the 16- or 20-eandle power incandescent lamp.