Modern Light Sources*

E. Lax, M. Pirani, and R. Rompe, Berlin

Over the last twenty years, lighting engineering has truly become a branch of technology both with respect to the methods of producing light (light sources) and in the sense of its distribution. A third group of questions, comprising the physiological and psychological analysis of vision, at any rate as applied to lighting engineering, was understood in all its significance only during the past decade and is only now beginning to receive attention. This transformation may be illustrated by two examples from the field of problems concerning the distribution of light.

At the beginning of the twentieth century, the design of lighting installations was determined solely by specifying a certain illumination, defined as the illumination in the horizontal direction at a height of 1 m above the floor of the room. There existed simple rules by which the number and arrangement of arc lamps or incandescent lamps were determined (other light sources were scarcely used at that time). At present, in calculating illumination, one starts from the kind of work being carried out in the room to be illuminated. The tasks confronting vision are studied with respect to the necessary resolving power of the eye, taking into account the materials with which one has to deal. And with the same requirements placed on vision, depending on the material being processed, one arrives at different solutions. Thus, for example, in the textile industry, in order to obtain the best visual conditions, weaving looms for cotton cloth must be illuminated differently than the same looms when weaving artificial silk fabric1.

Such an approach to the problems of lighting technology, as has already been said, requires a profound study of the physiological and psychological foundations of vision. Our knowledge in this field is still very far from complete, and a great deal of work will be required before it will be possible to think of a complete solution to the problems that arise in developing questions of illumination.

The range of psycho-physiological problems that one has to encounter may be briefly outlined as follows: it is necessary to investigate what radiation can be perceived by a certain—

separate receiver—the eye—and how the properties of this radiation—with respect to intensity and spectral composition—affect the functioning of the eye.

Like the sensitivity of a photoelement, the sensitivity of the eye depends on the wavelength of the light incident upon it. The reading of a measuring instrument connected into the circuit of a photoelement having its maximum sensitivity, for example, in the green part of the spectrum, will, when illuminated with green light, be greater than when illuminated by an equal amount of energy from the red or blue region of the spectrum. In exactly the same way, under the action of green light, the human eye produces a stronger physiological excitation than under the action of red or blue radiation of the same intensity. This ability to produce excitation under the action of light, giving a measure for the intensity of radiation with account taken of the spectral distribution of sensitivity, is called the luminance sensitivity of the eye.

Fig. 1. Visibility curve of the human eye

Up to this point the analogy between the photoelement and the eye is complete. But whereas, when a photoelement is illuminated by radiation of different wavelengths, we obtain identical deflections on the measuring instrument (provided that the product of the spectral sensitivity of the photoelement and the radiation intensity remains unchanged), in the case of the eye we have something quite different. The eye can distinguish radiation not only by brightness, but also by spectral composition, owing to which, within the interval of sensitivity of the eye, it becomes possible to distinguish light by wavelengths. We already had this circumstance in mind above when speaking of green and red light.

Both the luminance and the color sensitivity of the eye are complex dependencies, the investigation of which presents great difficulties because of the considerable number of parameters that determine them. Both quantities depend on extraneous circumstances, for example, on the background—white, black, or colored—on which the given color appears (“relative colors” of W. Ostwald²).

When spectrally complex radiation falls on a photoelement, the resulting effect is obtained as the sum of the effects of the components (of course, with account taken of the spectral distribution of sensitivity). In a similar way, the luminance impression of the eye for mixed radiation is determined by the sum of the individual luminances. In this case, however, there also arises a definite color sensation corresponding to the perception of the given complex color.

Quantitative perception by the eye is determined by the visibility curve shown in Fig. 1. To obtain color perception, according to the Young–Helmholtz theory, one must sum the excitations produced by the three primary stimuli. The excitation curves for them are given in Fig. 2; a visual picture is provided by the color triangle shown in Fig. 3. Each color impression corresponds to one point. Mixed colors are given by points lying on the lines connecting the original colors.

Fig. 2. Excitation curves for the three primary stimuli

Between the coordinates of the visibility curve and the coordinates of the color triangle, quantitative relations can be established that make it possible to calculate the values of brightness perception from color excitation. These relations were first established by König and Dieterici ^3. Among the large number of subsequent works in this direction, we mention here the recently published study by Guild and Wright ^4.

Fig. 3. Maxwell–Helmholtz color triangle.
The wavelengths are given in \(10^{-5}\) cm. The curve gives the color of blackbody radiation (\(a = 100^\circ\), \(b = 1200^\circ\), \(c = 1400^\circ\), \(d = 1600^\circ\), \(e = 1880^\circ\), \(f = 2000^\circ\), \(g = 2400^\circ\), \(h = 2600^\circ\), \(i = 3000^\circ\), \(j = 3500^\circ\), \(k = 4000^\circ\), \(l = 5000^\circ\), \(m = 6000^\circ\), \(n = 8000^\circ\), \(o = 10000^\circ\), \(p = 14000^\circ\), \(q = 20000^\circ\), \(r = 28000^\circ\)). The color of the radiation of a tungsten lamp is given by the points of this curve lying between \(g\) and \(i\). In the triangle are marked the colors of the radiation of gas-discharge lamps: Ne—neon lamp, Na—sodium, Cs—cesium, He—helium, HgH—mercury high pressure, HgN—mercury low pressure, Tl—thallium, Mg—magnesium, Cd—cadmium, Zn—zinc.

From the color triangle we see that one and the same color impression can be produced by entirely different methods of mixing colors. Two light fluxes that seem to be colored alike, but differ in the way they are obtained, may, however, reveal strong differences under certain conditions of operation of the eye. Thus, for example, because of the well-known deficiency of achromatism in the optical system of the eye, in those cases where the greatest possible visual acuity is required,^5 yellow monochromatic light should be chosen. In the case of yellow light obtained by mixing red and green, having almost the same hue, no improvement in sharpness of vision will occur.

In addition to the physiological phenomena connected with vision that have been considered so far, we also know psychological effects caused by physiological perception and, in turn, acting upon it. The commonly used characteristics of light, “warm” and “cold,” indicate that psycho-physiological interaction plays a role in the perception of light. Nevertheless, in the field under consideration there is still no reliable material from a scientific point of view. The reason for this is the difficulties involved in systematic investigations of such borderline areas. The psychological picture of a phenomenon does not lend itself to objective study and varies from subject to subject. Therefore, in individual cases, for example with strongly colored light, it is impossible to establish whether the illumination is “good” in all respects. As a result, it has likewise so far proved impossible to exclude this highly individual psychological factor and to choose a light source in such a way that all test subjects without exception would not experience unpleasant sensations.

The present state of the technology of manufacturing light sources in most cases permits, for each individual purpose, several almost equally successful solutions. Therefore the choice of a light source in each particular case is made on the basis of technical and economic considerations, especially since this point is one of the essential parts of the general problem of light distribution. Let us consider, as an example, the problem of creating illumination close in spectral composition of radiation to daylight, which is necessary, for instance, when sorting products by color. In this case it is necessary to strive for the colors of objects in the light of the source to be the same as in daylight, i.e., for the distribution of energy in the spectrum of the source to be as close as possible to the distribution in daylight. Such a source is, for example, a lamp with carbon dioxide gas, and therefore it will be recognized as suitable, despite the fact that its output is low. If, however, the question is the elimination of double light, i.e., the equalization of the difference in chromaticity of the radiation of sources that add light to daylight, such equalization can be carried out, from the physical point of view, quite inaccurately. On the other hand, here it is essential that this improvement not too strongly

was reflected in the cost of illumination, for example, in comparison with illumination by incandescent lamps.

Here it should be noted that an approach to solving the above-mentioned problems first became possible only after a series of gas-discharge lamps had been developed, especially lamps giving monochromatic radiation. With the aid of these light sources, conditions of vision were first realized that differ substantially from the conditions of normal vision in “white” light. To what extent all the requirements imposed on light sources on the basis of physiological and psychological data can be met can be established only by knowing precisely all the possibilities for obtaining light. One may even say that any practical consideration of lighting-engineering problems presupposes the development of a technique for manufacturing light sources that permits the production of light of the desired intensity and spectral composition. In this lies the most important requirement imposed by lighting engineering upon physics.

The emission properties of existing or possible light sources can be established only on the basis of consideration of the physical aspect of the radiation process. After this has been done, technology is left with the very difficult task of practically developing the possibilities that have been found.

The technique of manufacturing light sources first received a serious scientific basis in the well-known works of Lummer and Pringsheim⁶ on the radiation of the absolutely black body. The calculation of the distribution of energy in the emission spectrum of a solid body and the subsequent study of the radiation of solid bodies led to the establishment of the limits of economy of light sources with incandescent solid bodies. The establishment of the fact that the economy of a thermal radiator increases with increasing temperature led to a search for substances with high melting temperatures. A number of refractory compounds were obtained (Table 1), and their applicability as materials for incandescent bodies was investigated. At the same time the possibility was studied of reducing the evaporation of metals (in particular tungsten) and their

TABLE 1

Melting points of refractory compounds

TABLE 2

Various possibilities for obtaining light and their technical use

Continuation

Continuation

crystallization at high temperatures. All this led to the creation of modern incandescent lamps.

The information accumulated by that time on the radiation of atoms and molecules of substances in the gaseous state again made it possible to use physics in lighting engineering, which in turn gave impetus to further investigation of radiation processes.

We have attempted to systematize the various possibilities for obtaining light in the form of a table (Table 2); from this table one can also see what has already entered the field of technical use by the present time. The further text is intended to serve only as an explanation of Table 2 and therefore, in itself, is no more than a schematic survey of the questions under consideration.

Fig. 4. Sodium resonance series

The basis of every method of producing light, as of every method of creating electromagnetic radiation, is the presence of a variable electric-field intensity. Such a field arises:

1) in periodic motions of electric charges,
2) in aperiodic, especially “sudden,” changes in the velocity of moving charged particles (“braking radiation”).

Systems in which periodic motion takes place possess characteristic natural frequencies—“lines.” The extent to which these natural frequencies give radiation depends above all on the coupling of the system with the surrounding medium and then on the character of the excitation. In what follows we shall be interested in the natural vibrations of atoms and molecules.

The ideal conditions for the occurrence of natural vibrations are realized when there exists a single isolated atom in an extremely rarefied gas. The fundamental vibration, giving radiation of the “resonance line,” can be excited in a perfectly pure form by acting on the atom with radiation of this very same frequency, or by impacts of electrons possessing a definite velocity.^7

In the presence of electrons with somewhat higher velocities, the entire “resonance series” is excited, which, in the case of atoms having a simple structure, represents a clear sequence of lines. Almost isolated atoms exist in rarefied gases at pressures down to \(10^{-3}\) mm Hg. These almost ideal conditions can be created under experimental conditions, and, under the action of radiation with frequencies corresponding to the resonance series or by bombardment with electrons of the proper velocities, emission of the corresponding lines of the resonance series arises. In Fig. 4 one can see the resonance series of sodium.

The spectrum of a diatomic gas may be imagined as arising from the spectrum of a monatomic gas in such a way that each line is split into a system of bands as a consequence of the superposition of the molecule’s internal vibrations and rotations. When resonance radiation is excited by the action of monochromatic light or by electron impacts, in this case, instead of the emission of a line, the emission of a series of lines is obtained. A typical picture of such splitting is shown in Fig. 5 (the resonance series of the iodine molecule).

Fig. 5. Resonance spectrum \(J_2\), excited by various lines

In the case of polyatomic gases, owing to the existence of a large number of possible vibrations, the spectrum proves to be still more complex, in some places changing from banded into continuous.

If one uses monatomic gases that are not very strongly rarefied, the vibrations resulting from the interaction of individual atoms prove to be “unsharp.” This is expressed in spectra in the broadening of lines, which, with increasing pressure, may become very considerable. In Fig. 6 this change in the spectrum of cesium vapor under the influence of a change in pressure is shown.\(^8\)

Fig. 6. Spectrum of the positive column of a discharge in a mixture of cesium vapor with a noble gas at various cesium-vapor pressures:
a) \(0.002\) mm Hg, b) \(0.02\) mm Hg, c) \(0.1\) mm Hg, d) \(100\) mm Hg

In the case of molecular gases, the interaction of individual molecules manifests זיך

also in the fact that, instead of series of resonance lines, as the pressure is increased, the emission of more or less complete bands arises.

The further “filling of the spectrum” is caused by the fact that, under strong excitation, the concentration of excited and ionized atoms may increase so much that their interaction with light quanta and electrons begins to play a significant role in the radiation. In this case “higher” lines appear in the spectrum, i.e., those which are absorption lines not of normal atoms, but of excited atoms and ions.

At the same time, a strong broadening of the lines is observed^9 and the appearance of regions of continuous spectrum (cf. Fig. 6). These regions

Fig. 7. Fluorescence spectrum of benzene under excitation by ultraviolet mercury lines: a) vaporous benzene, b) liquid, c) solid at 0° C, d) solid at −180° C.

are often observed in those cases where, owing to the high concentration of ions, recombination of ions with electrons frequently takes place.

In the case of liquids, the connection of molecules with one another is, generally speaking, so strong that isolated frequencies (lines) are not excited. Only in those cases when the emitting particle is isolated from the surrounding medium can it fluoresce, absorbing and then emitting radiation in a definite region of frequencies, in most cases in the form of a band of continuous spectrum. The emission and absorption then obey Stokes’ law, according to which the frequency of the emitted radiation is less than the frequency of the absorbed radiation or, in the limiting case, equal to it.^10 Among fluorescing liquids are, for example, solutions of fluorescein, eosin, and rhodamine.

Among nonmetallic solid substances there is a large number of such bodies, the individual atoms of which are in the same position as the “luminous” atoms of fluorescing liquids; these solids behave similarly, especially at low temperatures. As an example we give in Fig. 7 the spectrum

of benzene obtained when luminescence is excited by radiant energy. Regarding another group of fluorescent substances, the so-called phosphors, somewhat more will be said below.

Excitation of natural vibrations by electron bombardment is not observed in the case of liquids; in the case of nonmetallic solids it is observed only for phosphors. When electrons fall on a metal, radiation arises that is a consequence of the braking of electrons near the surface. This braking radiation, at electron velocities of 7 V, lies for W, Al, Be, Th, Pt, Cu, Ag in the visible and near ultraviolet regions of the spectrum[^11]. In X-ray tubes the braking of very fast electrons (up to \(10^6\) V) is used to obtain X-rays. The radiation of the already-mentioned luminous centers of phosphors, in addition to light quanta, can be excited by fast electrons and by impacts of heavy particles (\(\alpha\)-rays).

Until now, as causes of excitation of atoms and molecules, we have considered radiation of definite frequencies and electrons with definite velocities. A further possibility of imparting discrete quantities of energy to a radiating system is provided by the transfer of energy in collisions of atoms and molecules with one another, which is what occurs in chemical reactions in the gaseous and liquid states (chemiluminescence). As with other methods of excitation, in this case gases also give separate lines and bands, while liquids give separate regions of a continuous spectrum. Excitation of this kind is also possible on the surface of certain solids. For example, this can explain the special radiation of some oxides placed in a hydrogen flame: this radiation must be attributed to reduction processes taking place in the hydrogen flame[^12].

In the cases of excitation of luminescence considered so far, almost everywhere the discussion has concerned radiating systems whose mechanism of radiation is so independent of internal and external motions that the expenditure of energy on these processes has no effect on the process of emission.

Whereas for isolated and freely oscillating atoms the statistical treatment of temperature, which presupposes a definite possibility of equalizing the internal energies of individual systems, is inapplicable, in the case of densely packed atoms (for example, a solid) it can be used with full justification. In the state of thermal equilibrium, the emitted energy is distributed over all possible frequencies of vibration of the system according to laws determined by the total energy content of the body and by its properties as a radiator. In the case of gases (under high pressure) spectra are obtained that consist of a weak background of continuous spectrum, upon which a line spectrum is superposed. In Fig. 6, as an example, the spectrum of cesium at a pressure of 100 mm Hg is given. Fig. 8 shows the spectrum of a mercury lamp constructed by K. Böhm[^13], in which the mercury-vapor pressure reaches 150 atm;

this spectrum, as we see, has the same character. Spectra of the same type are given by vapors in the voltaic arcs of arc lamps, which in their time were widely used for street lighting.

The spectra of solid and liquid bodies are continuous. Sometimes, for example in the case of coal, the energy is distributed over a broad region of the spectrum, resembling the distribution of energy in the spectrum of a black body; in other cases (for example, the radiation of erbium^14) the spectrum of a solid body has a clearly expressed banded structure.

The choice that lighting engineering makes from the many possibilities presented is determined, on the one hand, by the physical constants of the materials, and on the other by the state of knowledge and by technical possibilities. Among the constants that determine the possibility of applying one process or another for obtaining light, one may indicate, for example, the following: vapor pressure, melting temperature, chemical activity. An essential role

Fig. 8. Spectrum of a mercury lamp (after K. Bohl): a) visible spectrum, b) ultraviolet (1—short exposure, 2—longer exposure)

is played, of course, also by the position of the emitted frequencies in relation to the visibility curve of the eye.

In addition to the constants characterizing the substance, the possibility of the appropriate delivery of energy to the technical model of the light source is very important. In the case of metallic solid bodies the best method is heating the metal by an electric current passing through it (incandescent lamps) or by an arc discharge (voltaic arc). In the case of nonconducting materials, heating in a flame is successfully applied (alcohol-kerosene and gas-mantle lamps).

For gaseous substances, excitation of the luminescence of atoms by means of a gas discharge is used. From the point of view of modern technology, this method has the important advantage that in it there is a direct conversion of electrical energy into light. In order to have the possibility of using the radiation of gases and vapors as fully as possible in lighting engineering, it was necessary to carry out a systematic study of the corresponding processes, in particular of the luminous efficiency, as a function of gas pressure and current density. Studies of this kind have to date been carried out only to a very small extent. The reason for this is that, because of great experimental difficulties, these investigations require the expenditure of a large amount of time and labor. Moreover, in many cases, before undertaking a physical investigation, it is necessary to develop the technology for producing the required materials.

An example of this may be the sodium lamp. The fact that the resonance line of sodium lies close to the maximum of the eye-sensitivity curve has long been known. The question of the possibility of obtaining this resonance radiation economically had not been investigated, and its clarification became possible only after a type of glass had been manufactured that is not affected by sodium vapor. After this, the discharge in sodium vapor was investigated and the efficiency coefficient for the conversion of electron energy into radiation energy was measured. At considerable vapor pressures and low current densities this coefficient proved to be equal to 80%. Such a high coefficient, which in subsequent experiments was raised almost to 100%, is attainable, however, only under laboratory conditions. Another physical constant—the pressure of sodium vapor—determines the temperature of the wall of the lamp bulb, which, as it turns out, must be equal to \(280^\circ\). This gives rise to losses due to thermal radiation, convection, and thermal conductivity. A technical light source can therefore operate only under conditions in which the energy supplied is spent not only on exciting the radiating atoms, but also on covering all the indicated losses.

In the radiation of the positive column of a technical sodium lamp the resonance line predominates, as can be seen from Table 3. What the overall energy balance is in a technical sodium lamp we shall indicate somewhat below.

TABLE 3

Relative intensity and relative brightness of lines in the positive column of a sodium lamp *

\[ \sum J_\lambda = 130.8;\qquad \sum J_\lambda \cdot V_\lambda = 77.90;\qquad \frac{\sum J_\lambda \cdot V_\lambda}{\sum J_\lambda}=0.595. \]

Luminous efficacy *** is equal to

\[ \frac{0.595}{0.00145}=410\ \mathrm{Lm/W}. \]

* The pressure of sodium vapor is about \(5\cdot 10^{-3}\) mm Hg, the current density is \(1\ \mathrm{A/cm^2}\), the buffer gas is neon at a pressure of several mm Hg.

** The intensity of the lines 5890—96 Å is arbitrarily set equal to 100%.

*** The radiation passing through the glass walls of the tube is meant.

In all other gas-discharge lamps used up to now (Hg, He, Ne, Zn, Cd, etc.), only the higher members of the series play an essential role in the production of light. Since the emission of a large number of lines occurs, many of them lie outside the visible region. Those of them that are situated in the ultraviolet may be used for obtaining light by irradiating the corresponding phosphors with them.

Taking into account the significance that phosphors, by virtue of their properties as “light transformers,” may have in illuminating engineering, we consider a brief digression into this field justified. In the phosphorescence of solid and liquid substances, the process in question, despite a certain outward resemblance to fluorescence, differs essentially from it. Whereas in fluorescence the absorption center is at the same time also the center of re-emission, in the case of phosphorescence the centers of absorption and emission are different.

The connection between the processes of emission and absorption, despite some successes in this direction, still cannot in any way be regarded as clarified. With respect to the practically most important group of phosphors—the Lenard phosphors^17—it is known that they are capable of storing the excitation energy for a certain interval of time. This capability is associated with the presence of so-called activators (for example, atoms of Ag, Cu, etc.), present in negligible concentrations \((\sim 10^{-6})\) in the host substance. Through processes still unknown to us, in which thermal motion plays a role (as follows from the temperature dependence of the “glow-up” of phosphors), the excitation energy is transferred to the emitting systems.

The spectrum of Lenard phosphors differs from the spectra of fluorescent gases and liquids. In the latter, owing to the identity of the absorbing and emitting systems, the absorbed and emitted wavelengths are connected by definite relations. The absorption and emission bands adjoin and overlap one another. In the case of Lenard phosphors there is no such direct connection between the absorbed and emitted wavelengths. However, from energy considerations there follows the general rule that shorter waves are absorbed than those that are subsequently emitted. The positions of the lines in the spectrum depend strongly on the composition and structure of the phosphor.

Fig. 9 shows how one can make visible the absorption spectrum of a phosphor lying in the short ultraviolet. The method consists in applying the phosphor to the surface of a glass plate, which is then illuminated by short ultraviolet. As a result, only the visible or long-wave ultraviolet radiation, transformed by the phosphor from the short-wave ultraviolet, reaches the photographic layer (on the reverse side of the plate).^18

As regards the efficiency of phosphors, it may be considered beyond doubt that the quantum yield (i.e., the number of emitted—

emitted quanta per one absorbed quantum is close to unity, just as occurs in the fluorescence of gases. Table 4 gives some relevant data. The maximum possible energy output of a phosphor is determined by the ratio \(\nu_2/\nu_1\), where \(\nu_1\) is the absorbed frequency and \(\nu_2\) the emitted frequency, or by the corresponding ratio of wavelengths \(\lambda_1/\lambda_2\).

Fig. 9. Fluorescence of potassium tungstate \(c\) and zinc sulfide \(d\); \(a\)—direct photograph of the excited spectrum, \(b\)—photograph of the same spectrum after filtration through a transparent glass plate.

TABLE 4

Properties of some phosphors

If there is a phosphor emitting in the green part of the spectrum ($\sim 5000\ \text{Å}$), then, when it is excited by the mercury resonance line $\lambda = 2537\ \text{Å}$, an output of 50% may be expected. Since, on the other hand, methods are known for exciting resonance lines with high output (gas discharge; see what was said above about sodium), then, with the aid of mercury lamps and the transformation of light by means of phosphors, a very high efficiency can be obtained (Table 5, next-to-last column).

The spectral composition of the light obtained from a lamp with phosphorescence is shown in Fig. 10. In the spectrum there are both mercury lines and continuous regions of phosphor radiation. There are phosphors whose emission bands lie in the yellow-red region, so that a combination of their radiation with the green and blue mercury lines can give a white-light source.

Fig. 10. Spectrum of a mercury lamp using phosphorescence. Taken with the aid of a stepped filter.

Lamps with phosphorescence, like gas-discharge lamps in general, are still at the initial stage of their development. The investigation of all the possibilities for producing light that are outlined schematically here could not, of course, have been carried out with sufficient completeness in the six years that have elapsed since work began in this field. Correspondingly small, too, is the number of new technical light sources now available, at any rate in relation to the variety of existing possibilities. The development of new light sources was also delayed by the circumstance that the properties of the new light sources based on gas-discharge phenomena differ greatly from those of incandescent lamps, both with respect to the spectral composition of the radiation and in the sense of their electrical characteristics. This made it necessary to seek new ways for their practical application. The methods of measuring luminous flux and power consumption also had to be changed. Measurements of the latter kind present particular difficulties when operating on alternating current and when the lamp is connected with a choke; the usual measurement methods do not give reliable results19. In the end, the calorimetric method proved to be the best.

The technical development of new light sources, as has already been indicated, is closely connected with economic considerations, dictated by factors quite insignificant from a scientific point of view, but practically decisive (for example, the cost of electro-

energy). Therefore, consideration of the question of light sources of the future seems inappropriate here.

Our brief review should be supplemented by an indication of certain details in the construction of various types of new light sources.

The starting point in the development of light sources with a gas discharge was the development of new incandescent oxide cathodes, stable under gas-discharge conditions[^20]. The chief advantage of these cathodes consists in the reduction of the “cathode drop” (which, in the case of cold metallic electrodes, reached 100–300 V) to a few volts. Thanks to this it became possible:

1) to construct lamps operating at the ordinary mains voltage, and

2) to increase the density of the discharge current without fear of absorption of the gas in the discharge tube by the sputtered material of the electrodes.

In some cases the electrodes are heated for a long time. In others only brief heating is necessary for ignition of the discharge. Finally, sometimes the electrodes are not heated from outside at all, and their heating is effected by the discharge itself (heating by means of the anode drop).

Lamps used for advertising purposes, in which, because of the great length of the discharge gap, it is necessary to use voltages much higher than the mains voltage—so that the magnitude of the cathode drop is small in relation to the total voltage—are for the most part constructed without oxide electrodes.

For lighting purposes at the present time two types of gas-discharge lamps are widely used: sodium lamps and high-pressure mercury lamps. These sources, in a certain sense, supplement one another, since the discharge in sodium vapor is used predominantly for low-power lamps, while the discharge in mercury vapor is used for large lamps. This division has a natural basis. The brightness of the glow and the current density in sodium lamps are small; when the current density is increased, the luminous efficiency of the sodium discharge decreases owing to increased excitation of lines lying outside the visible region. In order to obtain large luminous fluxes with sodium lamps, it is necessary to increase the dimensions of the lamps, which is inconvenient in their use.

In the case of high-pressure mercury lamps the brightness of the glow proves to be approximately 10 times higher. With increasing current density in this case both the brightness of the glow and the luminous efficiency increase, so that large lamps prove more economical.

The spectral composition of the radiation of sodium and mercury lamps determines the known limitations of their application. A sodium lamp (its spectrum is given in Table 3) may be used in all those cases where sufficiently bright illumination and great sharpness of vision are required and, on the other hand, distinction of the color of objects is unimportant; a high-pressure mercury lamp (for the spectrum see Fig. 11)—when stronger illumination is required ...

...and at the same time the change in the color of objects (owing to the absence of red lines in the spectrum of mercury) plays no role. Let us list some of the principal areas of application of sodium and mercury lamps. Sodium lamps: lighting of automobile roads, rail tracks, factory premises in which delicate work is carried out, and illuminated advertisements. High-pressure mercury lamps: lighting of places with heavy traffic and large factory premises, large lighting installations, especially for buildings and green areas (intense green lines in the spectrum

Fig. 11. Spectrum of discharge in mercury vapor: a—low pressure,
b—high pressure

of mercury). For lighting highways, mercury and sodium lamps have already acquired the rights of citizenship. The number of light points with the new lamps is still small in comparison with their total number, but it shows a tendency toward rapid growth.

Technical sodium lamps, operating on alternating current, with a power of 70 W consume 1.15 A at 50–55 V; the total luminous flux is 3,000 Lm. Besides this type, there exists another which, at a power of 120 W, gives 6,000 Lm. The luminous efficacy of sodium lamps lies between 43 and 50 Lm/W. The diameter of a 70-watt sodium lamp (having the form of a cylindrical tube) is 20 mm; its temperature in the operating state is equal to 280°. To keep heat losses at a low level, this tube is placed inside a double glass envelope, consisting of one open and one closed glass cylinder (Fig. 12); the space between the main tube and the cylinders is evacuated.

In the main tube, in addition to sodium, there is an inert gas under a pressure of several mm Hg. Until the tube has heated up, only the gas in it glows. The sodium glow appears gradually as the tube warms up and as the resulting increase...

tion of the sodium-vapor pressure. In the end, the glow of the gas disappears completely. One of the oxide electrodes of the lamp, when it is switched on, is heated externally. When operating from the mains, a step-down transformer is used for this purpose; it is connected, as may be seen in Fig. 12, in parallel with the lamp and in series with the choke coil \(D\). When the lamp is switched on, the line voltage falls on the primary winding of the transformer, and the electrode is heated. After ignition and the establishment of the normal operating regime of the lamp, only the normal operating voltage, equal to \(50\text{–}55\ \mathrm{V}\), falls on the transformer.

Fig. 12. Connection of sodium (1) and mercury high-pressure (2) lamps.

\(I\)—sodium lamp, \(II\)—mercury lamp,
\(D\)—choke, \(H\)—heating transformer

The power consumed by it in this case proves to be so small that the introduction of a relay into the circuit for switching off the transformer is unnecessary, especially taking into account the possibility of ignition failure when the transformer is switched off.

The power supplied to the sodium lamp is expended as follows: 14% is the radiation of sodium vapor passing through the glass (Table 3), 10% is the radiation of the oxide electrodes emerging to the outside, 35% goes to radiation of the glass itself; heat losses by convection and thermal conductivity amount to 41%.

Fig. 13. Luminous efficacy of a discharge in mercury vapor as a function of vapor pressure.

Mercury-vapor discharge lamps belong among the longest-known. Low-pressure lamps have been used for advertising purposes for more than 10 years; high-pressure quartz lamps have been used in therapy since 1908. Investigation of a discharge in mercury vapor from the standpoint of luminous efficacy showed that its maximum is observed at low pressure and low current density. With increasing pressure (Fig. 13), after a decrease in luminous efficacy, a new rise occurs. In this pressure region, extending from \(100\ \mathrm{mm}\ Hg to many atmospheres, high-pressure mercury lamps operate. <!-- source-page: 023 --> **TABLE 5** **Properties of the principal light sources** | Method of exciting luminescence | Types of lamps | Luminous flux in Lm* | Luminous efficacy in Lm/W* | Luminance of luminescence | |---|---|---:|---:|---:| | Temperature radiation of a solid body | Gas-filled lamps | 214—865 | 1,26 | 6 | | Temperature radiation of a solid body | Incandescent lamps — carbon filament | 50—500 | 3,3 | 71 | | Temperature radiation of a solid body | Incandescent lamps — lamps with a metallic filament — old lamps — Os | 160—200 | 6,7 | — | | Temperature radiation of a solid body | Incandescent lamps — lamps with a metallic filament — with a zigzag filament — Ta | 160—1 000 | 6,3 | 8,2 | | Temperature radiation of a solid body | Incandescent lamps — lamps with a metallic filament — with a zigzag filament — W | 125—1 250 | 10,9 | 240 | | Temperature radiation of a solid body | Incandescent lamps — lamps with a metallic filament — lamps with a spiral tungsten filament — vacuum | 70—720 | 7—12 | 145—318 | | Temperature radiation of a solid body | Incandescent lamps — lamps with a metallic filament — lamps with a spiral tungsten filament — gas-filled | 400—1 000 000 | 6,7—32 | 565—3 600 | | Simultaneous luminescence of an incandescent solid body and gas discharge | Amount of temperature radiation large — carbon arc | 400—10 000 | up to 30 | 2 000—3 000 | | Simultaneous luminescence of an incandescent solid body and gas discharge | Amount of temperature radiation small — carbon arc | 5 000—18 850 | 7—30 | 18 000 | | Simultaneous luminescence of an incandescent solid body and gas discharge | Amount of temperature radiation small — sunlight lamp | — | Mainly in the ultraviolet | — | | Gas discharge | Radiation of the positive column — low pressure, high voltage — Ne, Hg | 250 (per 1 m) | 3—6 | 0,05—0,1 | | Gas discharge | Radiation of the positive column — low pressure, high voltage — CO₂, N₂ | 200—400 (per 1 m) | 1—3 | 0,05—0,1 | | Gas discharge | Radiation of the positive column — low pressure, line voltage — Ne | depending on size | 10—15 | 1—2 | | Gas discharge | Radiation of the positive column — low pressure, line voltage — Hg | 3 000—6 000 | 10—19 | 1—2 | | Gas discharge | Radiation of the positive column — low pressure, line voltage — Na | depending on size | 40—50 | 14 | | Gas discharge | Radiation of the positive column — low pressure Hg with phosphors | depending on size | 25—40 | 0,5—2 | | Gas discharge | Radiation of the positive column — high pressure Hg | 10 000—50 000 | 36—50 | 180 | | Gas discharge | Negative glow Ne | 1,5 | 0,5 | 0,02—0,03 | \* Geffner lumens are meant. <!-- source-page: 024 --> These lamps are manufactured for powers of 275 W (10,000 Lm), 550 W (20,000 Lm), and 1,050 W (53,000 Lm). Their luminous efficacy is  **Fig. 14.** Distribution of energy in the spectrum of mercury lamps of low and high pressure at the same total radiation. from 35 to 50 Lm/W. The oxide electrodes of these lamps are not intended for heating from outside. The lamp is connected directly to the mains through a choke, as shown in Fig. 12, II. This type of lamp also has a gas filling. The density of mercury vapor under operating conditions is about \(27 \cdot 10^{19}\) atoms per \(1\ \mathrm{cm}^3\), which corresponds to a pressure of about 4 atm. The glass envelope of the lamp has a temperature above \(300^\circ\). The discharge has the form of a brightly luminous column passing through the middle of the tube. The discharge tube proper in lamps of this type is surrounded by a second tube, serving to protect against heat loss. The spectra of mercury lamps of low and high pressure are given schematically on p. 14.

Fig. 15. Sodium spectral lamp for alternating current.

Among the lamps intended for scientific and technical research, the so-called spectral lamps should be noted. They are small lamps filled with one or another gas or vapor. Fig. 15 shows a sodium spectral lamp. Lamps with vapors of cadmium, mercury, zinc, thallium, rubidium, cesium, potassium, or with neon, argon, and helium are made in a similar way. With the aid of these lamps it is possible to obtain monochromatic radiation in any region of the spectrum. \(^{22}\)

Among the other gas-discharge lamps that have come into technical use should be mentioned

mentioned above are the so-called point lamps, i.e., lamps with a narrowed luminous region,^23 filled with Ne, Hg, or N₂; then lamps with carbon dioxide for obtaining white light, sodium lamps for television, and also indicator lamps.

We shall conclude our survey by clarifying the properties of the principal light sources, as presented in Table 5.

Which of the many light sources described will play the chief role in the lighting technology of the future—and, in particular, whether new light sources will penetrate into the fields now belonging to incandescent lamps—cannot yet be said at present. In any case, one may be certain of one thing: that the enrichment of lighting technology with new light sources in the form of the gas-discharge family of lamps will lead to its further development, the foundation for which has been laid during the last ten years thanks to the collaboration of physics, chemistry, physiology, psychology, and economics.

References

1. N. Goldstein u. F. Putnoky, Licht u. Lampe 19, 12, 31, 1277, 1930; 20, 5, 25, 1931; 21, 347, 364, 381, 1932.
2. W. Ostwald, Physik. Farbenlehre, S. 50, Leipzig 1919.
3. A. König u. Dieterici, Z. Psychol. 4, 241, 1892.
4. W. D. Wright, Trans. Opt. Soc. Lond. 30, 141, 1928; 31, 201, 1929; J. Guild, Phil. Trasn. A 230, 149, 1931; D. B. Judd, J. Opt. Soc. 23, 359, 1933.
5. W. Arndt, Licht 3, 231, 1933; W. Arndt u. A. Dressler, Licht 4, 122, 143, 1934; P. J. Bouma, Ingenieur, A 49, 31, 243, 290, 1934.
6. Geiger u. Scheel, Handb. d. Phys, 19, 1, 1928; 21, 199, 1929.
7. B. M. Knoll, F. Ollendorf, R. Rompe, Gasentladungstabellen, Berlin, 1935.
8. H. Krefft, Naturwiss. 19, 269, 1931; Z. Physik, 77, 752, 1932.
9. Pirani u. R. Rompe. Kinotechnik, 15, 131, 1933.
10. P. Pringsheim, Fluoreszenz und Phosphoreszenz, S. 166, 3-e Auf. Berlin 1928.
11. F. L. Mohler u. C. Boeckener, Bur. Stand. Journ. Res. 7, 751, 1931.
12. E. L. Nichols and collaborators, Phys. Rev. 17, 453, 1921; 19, 300, 1922; 22, 420, 1923; 25, 375, 1925; M. L. Phillips, Phys. Rev., 32, 832, 1928.
13. C. Bol, Licht u. Lampe, 24, 177, 1935.
14. Geiger u. Scheel, Handb. d. Phys., 21, 247, 1929.
15. M. Pirani, Z. techn. Phys., 11, 482, 1930; H. Krefft, M. Reger u. R. Rompe, Z. techn. Phys. 14, 242, 1933.
16. R. W. Pohl, Naturwiss. 16, 477, 1928; W. Koch u. R. W. Pohl, Gött. Nachr. 1, 6 1929.
17. Wien-Harms, Handb. d. Experimentalphys. 23, T. 1 u. 2. Leipzig 1928.
18. G. Heyne u. M. Pirani, Z. techn. Phys., 14, 31, 1933; Techn. Wiss. Abh. a. d. Osram-Konzern, 3, 56, 1934.
19. H. Ewest, Z. techn. Phys., 14, 478, 1933; Techn. Wiss. Abh. a d. Osram-Konzern, 3, 57, 1934.
20. M. Pirani, Elektrotechn., Z. 51, 892, 1930.
21. H. Krefft u. E. Summer, Licht 4, 1, 23, 86, 105, 1934.
22. E. Lax u. M. Pirani, Unterrichtsbl. f. Math. u. Naturwiss. 41, 91, 1935.
23. M. Pirani, Techn.-Wiss. Abh. d. Osram-Konzern 2, 39, 1931.

1. Naturwiss., 23, 393, 1935. Translation by N. Khlebnikova. ↩