Ultra-High-Pressure Lamps

N. A. Kaptsov and D. A. Gokhberg

Introduction

Among the new light sources based on the radiation of an electric discharge in gases or vapors, ultra-high-pressure lamps—UHP—occupy a special place. This type of lamp compares favorably with other gas-discharge lamps by its much higher luminous efficiency, more favorable spectral characteristics, small dimensions, and high brightness values.

Along with this, UHP lamps possess an unquestionable advantage with respect to ease of operation.

The combination of these properties of UHP lamps makes it possible to use them successfully as a light source for a wide variety of optical instruments.

In addition to lighting-engineering data, the high output of ultraviolet radiation and the possibility of good modulation of the luminous flux up to comparatively high frequencies also considerably broaden the fields of application of the lamps. These properties can be used both in already existing instruments and for solving new problems in various fields of scientific and technical work.

UHP lamps are divided into those requiring forced cooling (air or water)—the so-called capillary lamps—and lamps operating under conditions of natural cooling—spherical UHP lamps.

Capillary UHP lamps, owing to a number of shortcomings in comparison with spherical ones, have not found any very significant application and are not considered in detail in this review.

At present, spherical UHP lamps may be divided into mercury and gas lamps. In the former, the discharge takes place in mercury vapor at ultra-high pressure; in the latter, in monatomic inert gases (argon, krypton, xenon).

In contrast to mercury UHP lamps, whose history of development already spans 15 years, gas UHP lamps are an achievement of recent times. Despite their inherent ...

“growing pains,” “children’s diseases”; because of certain specific properties they have every chance of further development and firm establishment.

Acquainting broad circles of workers in science and technology with all the properties of UHP lamps is extremely desirable and may significantly increase the utilization of the great possibilities embodied in these lamps.

I. MERCURY LAMPS OF ULTRA-HIGH PRESSURE

1. PHYSICAL PROCESSES IN MERCURY LAMPS OF ULTRA-HIGH PRESSURE

The search for light sources of great brightness, possessing high luminous efficacy and convenient for various kinds of applications, has led modern technology to the use for this purpose of an electric discharge in mercury vapor. At present mercury lamps of all types indisputably occupy first place, in breadth of distribution, among the new light sources. In the increasingly widespread fluorescent lamps, too, a discharge in mercury vapor is used.

Since the appearance in 1902 of the low-pressure lamp*) (pressure 0.1—0.01 mm Hg), light sources using an electric discharge in mercury vapor have undergone a substantial evolution, the most important stages of which are lamps with mercury-vapor pressure close to atmospheric (high-pressure lamps), and then lamps with pressure much higher than atmospheric (ultra-high-pressure lamps—UHP).

Experimental and theoretical work has shown that the electric arc in atmospheric air, discovered as early as 1802 by the Russian physicist V. V. Petrov¹, as well as the arc in gases and metallic vapors at still higher pressures, possesses properties sharply distinguishing it from other kinds of discharge. As is known, the very name “electric arc” owes its origin to the fact that in this case the discharge is concentrated in a narrow, brightly luminous channel connecting the two electrodes and taking the form of an arc under the action of convective gas currents. Determination of the gas temperature in the arc channel by spectral methods showed that this temperature is very high: about 5500—6000° K in Petrov’s arc between carbon electrodes in atmospheric air², about 6300° K in a discharge in mercury vapor at a pressure of one atmosphere³˒⁴, and above 8000° K in a mercury discharge at ultra-high pressure. The highest pressures and the most—

*) The first patent for a low-pressure mercury lamp was obtained in 1879 by the Russian professor Repiev.

the greatest brightness of the discharge cord’s glow were achieved by Boyle[^5]. In his experiments the pressure of mercury vapor in the lamp reached 250 atmospheres, and the brightness of the visible-light radiation reached 180,000 stilbs, i.e., greater than the brightness of the disk of the sun in clear weather and with the sun high above the horizon.

In Boyle’s experiments the discharge took place in a capillary quartz tube with an internal diameter of 1 mm. As direct measurements showed, the amount of energy emitted by the tube in 1 second, in all regions of the spectrum, reached 75% of the power of the electric current passing through the tube.

In all cases of discharge occurring at comparatively low pressures, the principal cause of ionization of the gas is the collision of fast electrons with gas particles. In cases of an arc discharge at atmospheric or at ultrahigh pressure, however, when the temperatures indicated above occur, the source of ionization is the energy of the fast gas particles colliding with one another—the so-called thermal ionization. Along with ionization there also occurs intense excitation of the gas atoms, accompanied by radiation. In the discharge cord there is an intensified exchange of energy among all the particles participating in the phenomenon: normal unexcited atoms, free electrons, positive ions, excited atoms, and photons. The average energies of motion of the atoms, ions, and electrons become equal to one another, and we are dealing with a so-called “isothermal plasma,” an electrical regime which, like its other characteristics, is determined exclusively by the temperature, pressure, and nature of the gas.

An isothermal plasma in a vessel with adiabatic walls, or bounded on all sides by a medium at an unchanged temperature, can exist for an indefinitely long time without changing its state.

An externally imposed electric field accelerating the electrons is not required for the creation and maintenance of an isothermal plasma under such conditions. In high- and ultrahigh-pressure lamps the matter is somewhat different. In this case the cord of the arc discharge is surrounded by a medium having a much lower temperature. Therefore energy leaves the cord for the surrounding medium by radiation, thermal conduction, and convection. In addition, electrons and positive ions diffuse from the axis of the discharge to its boundaries and recombine there, giving up the potential energy stored by them during ionization. For these reasons, to maintain the temperature regime of the isothermal plasma cord in HPM lamps, an electric current is necessary, releasing per unit length of the cord in one second an amount of energy equal to \(IE\), where \(I\) is the current, \(E\) is the longitudinal gradient of the potential in the cord. At the same time, the shape and position of the luminous cord

of the arc and the geometrical outlines of the region where ionization occurs almost exclusively owing to the high temperature of the gas depend on the boundary conditions. The very possibility of the existence of a stable discharge also depends on these same conditions.

An especially simple and convenient case for calculation occurs when the discharge at superhigh pressure takes place in a cylindrical tube of small diameter and when the temperature of the walls of this tube may be regarded as constant and specified by the relation between the energy liberated in the discharge and the cooling conditions of the outer surface of the tube. Starting from Saha’s equation, based on the laws of thermodynamics,

\[ \alpha^{2}p = AT^{\frac{5}{2}} e^{-\frac{eU_i}{kT}}, \tag{1} \]

which establishes the relation between the degree of ionization of the gas \(\alpha\), the gas temperature (in \({}^{\circ}\mathrm{K}\)), and the ionization potential of the gas \(U_i\), and from the Boltzmann equation, which gives the concentration of excited atoms

\[ n_{0}=nge^{-\frac{eU_i}{kT}} \tag{2} \]

(where \(n\) is the concentration of neutral gas particles, \(g\) is the ratio of the statistical weights of the excited and neutral states), as well as from the laws of thermal conductivity, it is possible to compose a differential equation relating the temperature at each point of the discharge arc to the distance of this point from the axis of the tube. In the general case this equation has the form\(^6\):

\[ \frac{E_z^{2}R_1}{g_1^{1/2}} f_1(T) = \frac{1}{r}\frac{d}{dr}\left(r\lambda_T\frac{dT}{dr}\right) + \frac{g_1}{R_1^{2}} f_2(T). \tag{3} \]

Here \(E_z\) is the longitudinal field gradient in the column, \(r\) is the distance of some given point from the axis of the tube, \(T\) is the gas temperature at this point, \(R_1\) is the radius of the cylindrical tube, \(\lambda_T\) is the heat-transfer coefficient (which is a function of \(T\)), \(g_1\) is the mass of gas per unit length of the cylindrical discharge tube, and \(f_1(T)\) and \(f_2(T)\) are functions of the temperature \(T\), namely:

\[ f_1(T)=C_1T^{\frac{3}{4}}e^{-\frac{eU_i}{2kT}}, \]

\[ f_2(T)=\frac{C_2}{T}e^{-\frac{eU_a}{kT}}, \]

where \(C_1\) and \(C_2\) are constants, and \(U_i\) and \(U_a\) are respectively the ionization and excitation potentials of the gas.

Boundary conditions of the problem:

for \(r=R_1\)
\[ T=T_{\text{st}}, \tag{a} \]
i.e., the temperature of the walls of the discharge tube;

for \(r=0\)
\[ \frac{dT}{dr}=0, \tag{b} \]
since on the axis of the tube the gas temperature must have a maximum value.

Differential equation (3) is too complicated for practical calculations, if only because it contains \(\lambda_T\), which is a function of temperature.

In solving practical problems the problem is simplified. Toward the edges of the luminous cord of the discharge the temperature rapidly decreases as the distance \(r\) increases. Therefore, instead of the tube radius \(R_1\), the cord radius \(R\) is introduced, and the temperature inside a cylinder of radius \(R\) is regarded as constant and close to the temperature that actually occurs on the tube axis. This method leads to certain algebraic relations useful in calculating the electrical and luminous parameters of HPM lamps\(^7\). The principal significance of equation (3) is that from it one can derive the so-called “laws of similarity,” which make it possible, by means of an appropriate recalculation, to transfer the results of measurements made for one case to a number of other cases, on the basis that for “similar” discharges, at points for which the ratio \(\frac{r}{R_1}\) is the same, the temperature is also the same.

This method of solving practical problems is analogous to the method used in hydro- and aerodynamics (for example, the modeling of ships in the practical solution of problems concerning the resistance offered to their motion by water).

We give two “laws of similarity” for mercury discharge at high and superhigh pressure in cylindrical tubes:

1) “Two high-pressure discharges in cylindrical tubes of different diameter, but filled with gas in such a way that for each centimeter of length of either there is one and the same quantity of gas, are similar if the discharge power per unit length of the tube is the same in both cases.”

2) “Two high-pressure discharges in mercury vapor in cylindrical tubes of different diameter, containing in each longitudinal centimeter different quantities of mercury vapor \(g_1\) and \(g'_1\), are similar if the discharge powers in each longitudinal centimeter of these tubes are in the ratio
\[ \frac{(8.5+5.75\,g_1)}{(8.5+5.75\,g'_1)}. \]
”

The numerical coefficients are established here empirically. Thus, the theory of a high-pressure mercury discharge makes it possible to solve practical problems in calculating new cylindrical mercury lamps of high and superhigh pressure, starting from experimentally determined parameters of already existing lamp types.

When passing to superhigh-pressure lamps intended to serve as sources of bright concentrated radiation, it proved in practice more expedient to manufacture them not in the form of cylindrical quartz tubes of small diameter, but in the form of spherical quartz bulbs, the so-called “spherical SHP lamps.” In these lamps the electrodes are located at a small distance from one another, usually not exceeding several millimeters. The radius of the sphere is calculated so that the quartz envelope of the lamp, with only natural cooling by convective currents of the surrounding air, has a temperature not exceeding that required for the specified pressure of mercury vapor in the lamp.

Further development of the theory⁸ showed that in these cases, as also in wide tubes of any shape, stabilization of the discharge by means of the “wall effect,” as occurs in narrow cylindrical tubes, is excluded.

Experience shows that with a sufficient distance between the electrodes and a large width of the discharge tube, and also with the complete absence of walls (an arc in air), the shape and position of the discharge cord are determined by convective gas currents. These currents not only modify the shape of the cord, but also cool it. Therefore, in the presence of convective currents, the boundary condition (a) must be replaced by another:

at \(r = R\), \(T\) is equal to the temperature of the gas in the convective flow,

where \(R\) is the radius not of the tube, but of the discharge cord. Instead of “arc stabilization by the influence of the walls,” we arrive at “arc stabilization by convective currents.” This stabilization is not as steady as stabilization by walls, since convective currents are always accompanied by turbulent motion of the gas. As a result, the cord of an arc “stabilized” by convection constantly changes its shape and the places where it comes into contact with the lamp electrodes.

With a small distance between the electrodes, as occurs in spherical mercury SHP lamps (up to 8 mm), the discharge cord, even in a horizontal position, already almost does not bend upward at all and retains a stable position. The shape of the luminous arc resembles an ellipsoid of revolution. Such behavior of the discharge arc cannot be explained either from the standpoint of stabilization under the influence of the walls, or on the basis of ideas about stabilization of the arc by convective gas currents, and is a consequence of the specific properties of the cathode and anode parts of the arc dis-

series at high and ultrahigh pressure. For the electric arc, the presence at the cathode of the so-called “cathode spot” is characteristic; in it, over a small portion of the cathode surface, all the discharge phenomena at the cathode are concentrated. The dimensions of the cathode spot correspond to a definite current density in the spot, depending on the nature and pressure of the gas and on the properties of the cathode. The cause of the formation of the cathode spot long remained mysterious. A detailed consideration of the processes taking place in the transition layer between the plasma of the arc discharge and the cathode, based on the thermal theory of the arc, showed that contraction of the discharge into a cathode spot with a definite current density must be due to a more favorable energy balance in the transition layer in this case ^9, 10^. Something similar also takes place in the transition layer at the anode. The constriction of the discharge column at the cathode and anode corresponds to the minimum of the power expended on the discharge, other conditions being equal, and consequently to the stable form of the discharge. At small distances between the electrodes, the constriction of the discharge channel at the cathode and anode spots leads to a spindle-shaped form of the arc column, close to an ellipsoid of revolution.

The boundary condition in this case is not the condition at the lateral boundaries of the discharge space, but the conditions at the boundaries between the arc and the electrodes. The theory of the “arc stabilized by electrodes” is given in work ^8^. In this theory elliptic coordinates $\mu$ and $\nu$ are introduced (see Fig. 1). $\nu=\mathrm{const}$ corresponds to the surface of a hyperboloid of revolution, $\mu=\mathrm{const}$ to the surface of an ellipsoid of revolution. It is assumed that on the surfaces $\mu=\mathrm{const}$ lie the current lines in the arc and that therefore the whole discharge column has the form of an ellipsoid of revolution, as a consequence of the constriction of the column in the cathode and anode spots.

In addition to the initial premises of the original thermal theory of the arc, created for the case of an arc in a narrow cylindrical tube, the theory of the “arc stabilized by electrodes,” in the general formulation of the problem, also considers the loss of energy in each element of volume of the column, caused by the process of ambipolar diffusion of charged particles to the boundaries of the discharge region.

The components of the density of the electron current $i_e$ and of the ion current $i_i$ are taken to be:

$$ i_{e\nu}=eb_e nE_\nu, $$

$$ i_{e\mu}=eb_e nE_\mu+eD_a \operatorname{grad}_\mu n, $$

$$ i_{i\nu}=eb_i nE_\nu, $$

$$ i_{i\mu}=eb_i nE_\mu-eD_a \operatorname{grad}_\mu n. $$

Here \(b_e\) and \(b_i\) are, respectively, the mobilities of electrons and positive ions; \(n\) is the concentration of electrons and ions, \(D_a\)—

Fig. 1. Elliptic coordinate system (for the theory of the arc stabilized by electrodes).

—is the coefficient of ambipolar diffusion, equal, as is known\(^{11,12}\),

\[ \frac{b_i D_e + b_e D_i}{b_e + b_i}. \]

In the case of a cylindrical form of the discharge arc, for the portion of the arc lying midway between the electrodes, one may again return to cylindrical coordinates. This leads to the equation

\[ \frac{1}{r}\frac{d}{dr}\, r\left(\lambda r\frac{dT}{dr}+eU_iD_a\frac{dn}{dr}\right) = -e(b_e+b_i)nE_\lambda^{2} \tag{4} \]

(where \(U_i\) is the ionization potential of the gas).

The term

\[ \frac{1}{r}\frac{d}{dr}\left(reU_iD_a\frac{dn}{dr}\right) \]

appears here as a consequence of the presence of ambipolar diffusion, not taken into account in equation (3) of the original theory.

In the case of an arc stabilized by electrodes, the similarity laws for discharges given above do not apply.

It is necessary to point out that arcs stabilized by the influence of electrodes are only those with a sufficiently small distance between the electrodes. This is clearly seen in Fig. I (see the insert at the end of the issue), where, as the distance between the electrodes is decreased while the mercury-vapor pressure and the value of the arc power density \((\mathrm{W}/\mathrm{cm})\) are kept constant, the contraction of the discharge into a narrow cord increases.

Attempts to explain the phenomenon of contraction of the discharge cord in SHP lamps solely by convective gas flows[^13] must be considered erroneous. If one observes or photographs the mercury discharge arc of an SHP lamp with the action of gravity excluded—for example, in a freely falling chamber or during rapid rotation of the arc about its axis, when convection does not occur—then one can note a certain widening of the discharge channel in comparison with a lamp operating in the same regime in the presence of convective flows. However, this widening, recorded by a photographic plate, is explained by the glow of the arc flame, which has no direct relation to the width of the discharge channel of the arc. Photographing the discharge arc through a red filter clearly confirms what has been said (Fig. II)[^8] (see the insert at the end of the issue).

2. DESIGN AND TECHNOLOGY OF MERCURY SHP LAMPS

A characteristic design feature of spherical mercury SHP lamps, determined by electrode stabilization of the arc, is the spherical or nearly spherical shape of the bulb, whose dimensions considerably exceed the distance between the electrodes. Figure 2 shows specimens of mercury SHP lamps. The main elements of the lamp are the bulb, the electrodes, and the lead-ins. The requirements imposed on them with respect to dimensions, temperature conditions, mechanical strength, and other physicochemical properties often contradict one another; the interdependence of these three elements is complex and therefore can be calculated only conditionally.

Bulb. The superhigh pressure in mercury lamps (after their ignition) is created by the complete evaporation of a definite, strictly metered quantity of mercury introduced into the lamps during their manufacture. In order to attain and maintain this specified pressure, the temperature at the coldest points of the bulb must be not lower than \(600\text{–}800^\circ\mathrm{C}\), and in individual cases about \(1000^\circ\mathrm{C}\). Naturally, only quartz glass (that is, fused quartz), possessing a softening temperature...

Lamps of ultrahigh pressure

…of about \(1200^\circ\text{C}\), makes it possible to withstand such a high thermal load without deformation of the lamp.

The lamp bulb must possess great mechanical strength in order, over the course of its service life in operating conditions, to withstand the bursting pressure. Secondary requirements include the correct geometrical shape of the bulb (a sphere) and the absence in it of bubbles and other defects that reduce transparency, since in most cases UHP lamps are used as a light source in optical instruments.

Since for quartz glass there is a definite upper limit of load which cannot be exceeded without the risk of rupture of the lamp, it is clear that the necessity of observing this limit is in contradiction with other requirements imposed on the bulb. Thus, the requirement of a short run-up time, i.e., in essence, of a small heat capacity of the lamp, dictates the necessity of reducing the dimensions of the lamp, which is incompatible with the requirement of thermal strength. On the other hand, the desire to increase the useful service life of the lamps, limited mainly by darkening of the bulb owing to the inevitable sputtering of the electrodes, gives rise to the necessity of increasing the dimensions of the bulb.

Fig. 2. Samples of mercury UHP lamps.

Taking into account that there is the closest connection between the dimensions of the bulb and the load on the walls, on the one hand, and between the principal parameters of the lamp—brightness, luminous efficacy, service life—on the other hand, the choice of the bulb dimensions must be made depending on which lamp parameters are of primary importance. To this it must be added that the large number of variables, some of which cannot be determined numerically, does not permit lamps to be designed by a purely computational method.

Experimental data have shown that the dimensions of the bulbs of UHP lamps must be chosen in accordance with the magnitude of the load on…

wall within the range from 25 to 40 W/cm². Since the bursting strength of the bulb under high gas pressure depends on the ratio of the outer and inner radii of the bulb, the value of the selected pressure is related to the dimensions of the bulb—the smaller the latter, the higher the pressure may be. Thus, the quartz bulb of a 100 W SVD lamp with an outside diameter of 12 mm withstands an operating pressure of about 70 atmospheres.

The most suitable wall thickness, depending on the dimensions of the bulb, is a thickness from 1.5 to 3 mm. Increasing the wall thickness beyond 3 mm, while increasing the heat capacity of the bulb (and, consequently, the warm-up period of the lamp), does not lead to an increase in mechanical strength, since in this case the nonuniform distribution of stresses in the bulb wall begins to have an effect.

Leads in quartz. In manufacturing quartz SVD lamps, as indeed any other quartz electrovacuum devices designed for a large current (from several units to hundreds of amperes), one of the most important technological problems is the creation of a vacuum-tight lead of metal into quartz. The difficulty of the problem is due to the extraordinarily small coefficient of thermal expansion of quartz \((\alpha \sim 6 \cdot 10^{-7})\), as a result of which it is impossible to make a vacuum seal of any metal into quartz.

In electrovacuum devices with quartz bulbs, a method of sealing into quartz, as leads, thin metal strips of molybdenum, 3–6 mm wide, is used rather widely. This method is based on the phenomenon of wetting of the metal by glass (in the present case, by quartz). A necessary condition for a vacuum-tight and thermally stable seal in this case is the small thickness (not more than 20–22 microns) of the metal sealed into the glass, at which the change of stresses in the glass is compensated by the deformation of the thin metal. If these deformations can follow the change of stresses in the glass, then the lead satisfies the requirements imposed. Figure 3 shows a vacuum lead into quartz with molybdenum foil. Such a lead, owing to the relatively large surface of contact with the quartz and, consequently, good

Fig. 3. Single vacuum lead into quartz with the aid of molybdenum foil 20–22 μ thick for a current of 8–16 amperes.

cooling, permits the passage of a considerably larger current than a wire of equal cross-section. Nevertheless, the magnitude of the current allowable for a single lead-in with molybdenum foil, depending on the width and thickness of the latter, lies within the range of 8–16 amperes.

A partial solution to the problem is the sealing-in of several molybdenum strips connected in parallel (Fig. 4),

Fig. 4. Multi-foil vacuum lead-in into quartz for a current of 100–200 amperes.

Labels in the figure: molybdenum foil; molybdenum leads; welding points; brass washer with holes according to the number of leads; washer bushing; copper tip.

In this way, at the cost of a considerable complication of the manufacturing technique, it is possible to make lead-ins up to 100 and even up to 200 amperes, which is the limit.

Quite recently, a fundamental solution to the problem has been found in the form of the so-called disk or cap lead-in, which in principle permits any current load (Fig. 5). The principle of this lead-in consists in separating the two functions performed by the molybdenum foil simultaneously: vacuum sealing and current conductor.

In the disk or cap-type lead-in, the vacuum seal is made by a molybdenum disk or cap with very thin (15–20 μ) edges fused into quartz, while the current passes through two tungsten rods welded to the disk on both sides perpendicular to it.

Disk and cap-type lead-ins have not yet found industrial application. There is no doubt, however, that only they will help solve the problem of high-power lead-ins into quartz and, thereby, the problem of designing UHP lamps with powers on the order of tens of kilowatts.

Fig. 5. Disk lead-in into quartz for a high current (hundreds of amperes).

Electrodes. The processes at the electrodes of a high-pressure arc have not yet received a complete explanation, despite numerous and varied investigations^14. Therefore, a practically suitable calculation of the electrode design of UHP lamps is impossible—their design and dimensions have to be determined purely empirically.

The electrodes are a very important element of UHP lamps, since in these lamps the radiation characteristics of the arc are closely connected with the radiation of a cathode spot of a discharge, and the latter depends on the form and properties of the electrodes. In addition, the reliability of ignition and the service life of the lamps also depend on the electrodes.

In accordance with the processes taking place during ignition and burning of the discharge, two basic requirements are imposed on the electrodes of mercury UHP lamps:

1) good electron emission during the short period of ignition of the discharge at comparatively low lamp supply voltages (120–220 volts) and, subsequently, during the warm-up of the lamps, i.e. during evaporation of the existing quantity of mercury in the lamp, when the pressure in it changes smoothly from tens of mm Hg to tens of atmospheres;

2) high resistance of the electrode to sputtering under the conditions of the established operating regime of the lamp, i.e. at the high local temperature of the electrode caused by the presence of the cathode spot.

Both of these requirements for the electrodes are to a certain extent contradictory. To fulfill the first of them it is necessary to use a coating of the electrodes with substances having a low

by a low electron work function, for example barium oxide, which facilitate ignition of the discharge and greatly reduce the magnitude of the cathode fall during the lamp’s run-up period.

The second requirement, in essence, rules out the first, since pure refractory metals, such as tungsten for example, are the most heat-resistant in a discharge. Since the magnitude of the cathode fall in SHP lamps in the operating regime depends almost not at all on the state of the cathode surface, electrodes made of pure tungsten could have been used in these lamps. However, the large cathode fall at low densities of mercury vapor in the lamps during the run-up process is the cause of severe sputtering of the electrodes and, consequently, considerably shortens the service life of the lamp. This is best illustrated by the generally known fact—the influence of the number of switchings-on of SHP lamps on their service life.

The contradiction between the two basic requirements for electrodes is resolved purely structurally, by creating a cathode that operates separately during the period of ignition and run-up of the lamp and during the steady-state regime. The design embodiment of such cathodes may vary, but the idea underlying them is one and the same: the arc during ignition of the lamp must arise on an activated section, and in the steady-state regime must transfer to pure tungsten. In this case use is made of the phenomenon that at high pressures the discharge tends to proceed by the shortest path, between the ends of the electrodes. Therefore the active coating is placed behind the end of the electrode, at a sufficient distance from the cathode spot of the arc. Arc formation at low pressure and a small voltage gradient occurs from the active parts of the electrodes, and after the lamp has run up the arc automatically transfers to the ends of the electrodes (Fig. IV, a and b, see insert). Examples of design solutions for such cathodes are shown in Fig. 6.

Fig. 6. Design embodiment of cathodes of mercury SHP lamps.

Labels in the figure: oxide paste; Ta cylinder; W wire; W core.

3. FEATURES OF HPM MERCURY LAMPS

It was indicated above that the temperature and its distribution, under conditions of thermal equilibrium, are the most important characteristics of the discharge, determining all its optical and electrical properties. Raising the temperature of the discharge can be achieved in several ways, but all of them are based on increasing the power released in \(1\ \mathrm{cm}^3\) of the arc. In this case, the relative weight of losses due to thermal conduction and convection decreases, while the brightness of the discharge and its efficiency increase. Substantial changes also occur in the character of the radiation spectrum.

First of all, an increase in the discharge power per \(1\ \mathrm{cm}^3\) can be achieved by increasing the pressure. Further, it is possible to achieve the same goal at relatively low pressures by increasing the current. A third possibility is narrowing the arc or considerably reducing the distance between the electrodes. However, this last approach leads, along with an increase in brightness, to a sharp fall in luminous efficiency.

The task of the design and technological development of mercury lamps has always been to increase, as much as possible, the power per unit volume of the discharge arc. Table I gives data showing the change in the characteristics of mercury lamps and, in particular, the increase in power per \(1\ \mathrm{cm}^3\) in the course of their development.

It is seen from Table I that capillary lamps with water cooling possess approximately the same light characteristics as spherical HPM lamps. However, a number of reasons—the chief of which are the relation between wall loading and specific power, the dimensions of the luminous field (\(\sim 0.8 \times 25\ \mathrm{mm}\)), and the undesirable combination of high supply voltage (up to \(2000\ \mathrm{V}\)) with water cooling—have led to the fact that capillary HPM lamps have found almost no practical application.

In spherical HPM lamps, where stabilization of the discharge by the walls is absent, the specific discharge power in lamps with a given bulb diameter, at one and the same pressure, can be considerably increased by reducing the distance between the electrodes. The magnitude of the wall loading in this case changes almost not at all and can be kept within permissible limits. Hence follows the possibility of operating lamps without artificial cooling.

Small distances between the electrodes, in addition to their advantages from the lighting-engineering standpoint (an exceptionally high brightness, exceeding the brightness of carbon arcs), lead—despite the presence of high pressures in the lamp and the accompanying large voltage gradients in the arc—to low voltages on the lamp and thus make it possible to supply the lamps from the mains

Table 1

Type of mercury lamps	Specific arc power (W/cm)	Power concentration in the arc (kW/cm³)	Operating pressure (atmospheres)	Gradient (V/cm)	Brightness (stilb)	Luminous efficacy (lm/W)	Wall loading (W/cm²)	Distance between electrodes (mm)
High-pressure lamps (made of refractory glass)	20–50	0.03–0.2	0.3–3.0	7–30	100–300	35–50	3–4	40–200
High-pressure lamps (made of quartz)	30–60	0.4–2	3–15	25–80	400–900	40–50	10–15	10–60
Super-high-pressure lamps, capillary, with water cooling	300–500	50–70	70–100	300–400	20000–40000	60–70	500–800	8–25
Super-high-pressure lamps, spherical, with natural cooling	100–4000	5–2000	20–100	100–400	5000–120000	50–75	25–45	0.5–15

220 volts. In addition, the small distances between the electrodes provide very good stabilization of the arc, which is also an advantage of the lamps when they are used in optical instruments.

Speaking of the peculiarities of SVD mercury lamps, one must not fail to mention the run-up time—the interval between ignition of the lamp and its attainment of the specified electrical and light-output regime. As is seen from Fig. 16, the run-up time reaches several minutes, which is a substantial operational shortcoming of SVD mercury lamps. The run-up process is determined by the heat capacity of the lamp. To create in it the specified pressure of mercury vapor, it is necessary that the lamp bulb acquire the temperature corresponding to this pressure. It should be remembered here that, first, during the run-up process the bulb is heated nonuniformly and, second, that the value of the mercury-vapor pressure is determined by the temperature of the coldest part of the bulb.

The slow course of the run-up process is also promoted by the fact that, when the lamp is ignited, the voltage across it is very low (about 15 volts), and therefore the power released in the lamp at the beginning of the starting period is small (not more than 30% of the power of the given lamp in the steady-state regime).

4. LIGHT CHARACTERISTICS

The light characteristics of SVD mercury lamps are, to a large extent, determined by the distance between the electrodes, on which depend the values of the specific power \((\text{W}/\text{cm})\), and of the power falling on \(1 \text{ cm}^3\) \((\text{kW}/\text{cm}^3)\) of the discharge.

If (at constant pressure and constant lamp power) the distance between the electrodes is reduced, the luminous efficacy increases. This increase is connected with the growth of the concentration of power in the discharge \((\text{W}/\text{cm}^3)\) and is observed down to a certain distance between the electrodes (about 6 mm). The maximum luminous efficacy attained at such distances is 75 lm/W. Further reduction of the distance leads to a steady fall in luminous efficacy, since the specific weight of the power loss at the electrodes \((V_a + V_k)\cdot I\) increases as the current strength increases and the voltage across the lamp decreases1. The power converted into radiation decreases more rapidly than the radiation increases owing to the increase in the temperature of the discharge. At extremely small distances between the electrodes, of the order of 0.3–0.5 mm, corresponding to voltages across the lamp of about 18–20 V, more than half of the power consumed by the lamp goes into losses at the electrodes. Therefore the luminous efficacy of such lamps is relatively small, although—

to the absolute value (20 lm/W) equal to the luminous efficacy of high-power incandescent lamps.

As for the brightness, as the distance between the electrodes decreases it at first slowly increases because of the increase in the power concentration in the arc, and then rises sharply. The sharp rise in brightness is explained by the stabilizing action of the electrodes, as a result of which contraction of the arc increases, leading to exceptionally high values of the power concentration in the arc and, consequently, to an increase in the arc temperature.

This circumstance makes it possible to obtain lamps with an enormous brightness, which at the same time are almost ideal point sources.

It is also of interest that, as the length of the discharge gap decreases, there is observed, although weakly, nevertheless an increase in the content of red radiation of the discharge. This can be explained by the growth of continuous radiation of the discharge, which occurs when the arc temperature increases with increasing power concentration.

Fig. 7. Dependence of brightness, luminous efficacy, and red-radiation content on the arc length.

In Fig. 7 are shown the dependences of the brightness, luminous efficacy, and red-radiation content on the length of the discharge gap for three types of mercury lamps of different power at one and the same pressure.

For the brightness values there are given the values of the so-called “average brightness,” corresponding to measurements made over the half-width of the luminous cord of the discharge. The maximum values

brightness is considerably higher than the average brightness, exceeding the latter by approximately 50%.

In accordance with the distribution of temperature in the discharge cord, decreasing from the axis to the periphery, the distribution of brightness across the arc has the same character. The ratio of the maximum value of brightness (on the arc axis) to the width of the arc depends on the distance between the electrodes and on the pressure and can be specified. Fig. 8 shows a typical distribution of brightness across the discharge cord for HPMV lamps.

Fig. 8. Distribution of brightness across the discharge arc of HPMV lamps.

Near the electrodes, in the places where the arc forms the cathode spot of the discharge, the brightness reaches enormous values. However, owing to the insignificant area of the cathode spot, practical use of the brightness of the near-electrode parts of the arc is impossible.

Among the distinctive features of mercury HPMV lamps one may also include their ability to modulate the luminous flux up to relatively high frequencies (up to 10,000 hertz), at which the degree of modulation of the light is equal to the degree of modulation of the current[^15].

5. SPECTRAL CHARACTERISTICS OF MERCURY HPMV LAMPS

Fig. 9 shows the distribution of the radiation energy of mercury HPMV lamps in the ultraviolet and visible regions of the spectrum. As can be seen, the radiation consists of several relatively weak lines in the middle ultraviolet, strongly broadened lines of the near ultraviolet (mainly lines with wavelengths 3125/3132, 3341, and 3650/3663 Å) and of the visible region (4047, 4358, 5461, and 5770/5790 Å).

Both in the ultraviolet and in the visible regions, a continuous background is also superposed on the line spectrum; at pressures of the order of several tens of atmospheres, almost as much energy is radiated in this background as in the lines. The filling of the intervals between lines by a continuous spectrum, as well as the presence of weak radiation in the red part of the spectrum, leads to some improvement in the color quality of HPMV lamps in comparison with high-pressure mercury lamps.

Nevertheless, color rendering under illumination by an HPMV lamp is strongly distorted because of the excess of radiation in the yellow and green parts of the spectrum and its deficiency in the blue and, chiefly, in the red part. Improvement of color quality is possible by several methods, of which only one—adding cadmium to the mercury—is practically implemented in lamps intended to serve as ...

Fig. 9. Distribution of the radiant energy of mercury SHP lamps in the ultraviolet (a) and visible (b) parts of the spectrum.

a light source for motion-picture projection. This method of correcting the color of mercury SVD lamps gives different results for lamps of different power. The degree of correction of the radiation may be judged from Table II, which gives data for two types of lamps.

Table II

Lamp type	Relative content: red	Relative content: yellow	Relative content: green	Relative content: blue	Relative content: violet
SVD 200 watt, mercury . . . . . .	0.29	1.9	1.07	0.59	1.16
The same with cadmium added . . . .	0.46	1.08	0.86	1.4	1.22
SVD 2000 watt, mercury . . . . . .	0.34	2.14	0.98	0.54	1.01
The same with cadmium added . . . .	0.5	1.47	0.79	1.14	1.1

As can be seen from Table II, in SVD lamps corrected by the addition of cadmium, the content of the red and blue parts of the radiation increases, while the content of the yellow decreases. The decrease in the brightness of the green part and the increase in the brightness of the violet part are very slight. When cadmium is added to mercury, the luminous efficacy of the lamps decreases.

In addition to radiation in the ultraviolet and visible regions, there is also radiation in the infrared part of the spectrum. Fig. 10 gives a comparison of the radiation intensities in the infrared region (per unit wavelength) of three types of SVD lamps of the same power, equal to 200 watts.

The explanation of the large difference in radiation intensity among these three types lies in the different values of the voltage gradient in the arc, the pressure, and especially the power released per unit volume of the discharge. Thus, for two types the value of the power concentration is \(80\ \mathrm{kW}/\mathrm{cm}^3\), whereas for the third type its value is \(\sim 2600\ \mathrm{kW}/\mathrm{cm}^3\).

The share of the mercury resonance line with wavelength 2537 Å in the total radiation of mercury SVD lamps is negligible owing to its strong absorption by mercury vapor.

Fig. 10. Radiation intensity of mercury HPM lamps of equal power in the infrared region as a function of the power incident on 1 cm³ of the arc.

6. DEVICES FOR SWITCHING ON AND IGNITING HPM MERCURY LAMPS

Like all gas-discharge devices with a falling volt-ampere characteristic, HPM mercury lamps can be connected to the supply network only with a ballast resistance, since otherwise the discharge cannot be stabilized. This resistance must be ohmic in the case of direct current and, preferably, inductive for alternating-current lamps. In the latter case, as a rule, chokes or transformers with iron cores with large leakage are used. The advantage of chokes is not only an increase in the efficiency of the installation, but also an improvement in the operating conditions of the lamp on alternating current. When operating with ohmic resistance, the current and voltage are always in phase. When the current curve passes through zero, the voltage curve also passes through zero. Therefore ignition of the lamp in each successive half-period is greatly impeded, the dark pause between half-periods increases, and the current curve is distorted.

Fig. 11. Principle of construction of a special choke for HPM mercury lamps, reducing the run-up time.

When chokes are used, owing to the presence of a phase shift between current and voltage, at the moment when the current is zero there is still a sufficiently large voltage on the lamp for the lamp to ignite easily in the new half-period.

To avoid an unstable regime in lamps operating with chokes, when the mains voltage fluctuates the operating point must lie on the rectilinear portion of the volt-ampere characteristic of the choke.

The comparatively long run-up time of HPM mercury lamps is the reason for special requirements imposed on switching devices, which amount to reducing the duration of the nonstationary period after the lamp is switched on. As an example of a switching device that considerably shortens the run-up time, a choke with two windings is of interest (Fig. 11). During ignition of the lamp the additional winding 3 (the number of turns of which and the inductive coupling with the main winding 2 are chosen with allowance for the maximum current that may be passed through the lamp) is connected in parallel with the lamp. When the voltage on the lamp reaches a certain value (\(\sim 55\) volts), relay 4, connected in parallel with the discharge gap, breaks the circuit of the additional winding, and the lamp operates in series with the main winding.

ULTRA-HIGH-PRESSURE LAMPS

choke. The run-up time of lamps with powers up to 500 watts can be reduced, by means of such a switching device, to 1 minute.

A similar principle is also used for direct-current lamps. In this case, a voltage relay connected in parallel with the lamp, when it is ignited, short-circuits part of the ohmic resistance of the ballast, reducing its value, and after the lamp has run up switches in the entire resistance fully.

For igniting UHP mercury lamps, devices are used that provide either a high-voltage current pulse or also a high-frequency one. The latter are most widely used and are usually made in the form of a miniature high-frequency transformer.

Fig. 12. Circuit for switching on and igniting UHP lamps with three electrodes.

A short ignition pulse is applied to the auxiliary electrode after the main electrodes of the lamp are already under the voltage of the supply network (120 or 220 volts). The switching circuit with an ignition device for UHP lamps is shown in Fig. 12.

In the absence of a third auxiliary electrode in the lamp, in the case of direct current, devices are used in which the ignition pulse is obtained by briefly short-circuiting the self-induction in the lamp circuit by means of a special vacuum switch (Fig. 13).

In the case of alternating current, analogous circuits are used, or circuits in which the high-voltage pulse is obtained due to the resonant properties of the circuit.

For direct-current lamps without an auxiliary electrode, an ignition device called a “capacitor” device is also used. Its principle is very simple: \(n\) capacitors connected in parallel are charged to the supply voltage of the lamp, for example to 220 V. By a simple switching operation (pressing a button), all the capaci-

tors are connected in series, and their total voltage ignites the lamp. Twenty paper capacitors in such a device are sufficient for reliable first ignition of any SVD lamp.

Fig. 13. Circuit for switching on and igniting SVD lamps with two electrodes.

As for repeated ignition of a burning and just-extinguished mercury lamp, it is possible only if an igniting electrode is present and high-frequency igniting devices are used.

7. FIELDS OF APPLICATION OF SVD MERCURY LAMPS

The fields of application of SVD mercury lamps are entirely determined by their properties and characteristics and therefore may be very diverse. A detailed description of all specific cases of application of these lamps is impossible; therefore only data of general interest are given below.

A. Owing to the high values of brightness and luminous efficacy, the use of SVD mercury lamps in optical instruments and projectors is fully justified. Let us give several examples.

a) Motion-picture projection. The possibility of increasing the illumination of the screen is easily achieved when alternating-current lamps are used, by shifting the dark pause (during which the luminous flux of the lamp decreases to 10% of the maximum) to the time of obturation*). If we assume that the current through the lamp is sinusoidal, that the luminous flux is proportional to the current, and that the time during which the obturator darkens is equal to the dark pause, then combining the dark pause with the darkening by the obturator makes it possible to reduce the loss of light due to obturation to 29%. By using chokes operating in the saturation regime, the magnitude of the losses can be reduced still further. As is known, when inertial light sources are used (for example, incandescent lamps), the loss of light due to obturation is not less than 50%.

For direct-current lamps, a gain in screen illumination can be achieved by briefly overloading the lamp

*) This assumes a transition from 24 frames per second to 25, which causes some increase in film consumption (by 4%).

in the intervals between the obturations and, accordingly, a reduction of its power during obturation. In this case the average value of the power remains normal and the service life of the lamp is not reduced. The intermittent mode of operation of the lamp is achieved by short-circuiting, synchronously with the obturator, part of the lamp ballast resistance. By this method it is possible to increase the illumination of the screen by 60–70%.

Speaking of the use of mercury SHP lamps in motion-picture projection, we have in mind the projection of black-and-white films. Mercury lamps are unsuitable for the projection of color films. The suitability for color cinema of mercury lamps with additions of zinc and cadmium (see above) is very conditional and is determined rather by the operational advantages of SHP lamps in comparison with the carbon arc than by the spectral composition of the radiation.

b) Projectors. The scale of application of mercury SHP lamps in searchlights is limited by the power of the lamps produced. Thus, lamps with a power of 2000 watts can be used in searchlights with a mirror diameter of up to 1 meter. The advantages of using SHP lamps in searchlights, due to their high brightness and better economy, are evident from the following examples. A 200-watt SHP lamp in a searchlight 200 mm in diameter gives a luminous intensity 9 times greater than the best 100-watt incandescent searchlight lamp, and 3 times greater than a 500-watt incandescent lamp with a mirror 360 mm in diameter. Thus only half the power and a much lighter searchlight are required in order, with an SHP lamp, to obtain a luminous intensity 3 times greater than with an incandescent lamp.

For SHP lamps of high power (several kilowatts) almost the same ratios hold, the comparison being made already with carbon arcs. Thus, for a 2-kilowatt SHP lamp and a carbon arc of equal power in a searchlight with a 60 cm mirror, the following figures are characteristic:

Light source	Comparative luminous intensity (candle/watt)	Effective brightness (stilb)
Mercury lamp, 2000 watts . . . . . .	0.0443·10⁶	43 700
Carbon arc, 90 amperes . . . . . .	0.017·10⁶	39 900

In addition to their purely optical advantages, SVD lamps also possess a number of operational advantages in comparison with carbon arcs (no need to adjust or replace carbons, arc stability, the possibility of precise lamp installation), which fully justify efforts to replace carbon arcs in projectors with SVD lamps.

c) The use of SVD lamps in many optical instruments is very varied, where the determining factors are the small dimensions of the light source itself and of its luminous field. In these cases SVD lamps are indispensable, since their dimensions make it possible to place the lamps very close to the optics—in the limiting case, when quartz optics are present, up to contact with the condenser, thereby making use of the greatest luminous flux. The small dimensions of the luminous field of the light source make it possible to focus the lamp with great precision.

Fig. 14. Inspection of hollow narrow bodies with the aid of an SVD lamp and an ocular mirror.

From all the varied applications of SVD lamps in optical instruments in industry, technology, and scientific research practice, the following examples may be cited:

1) inspection of the internal surface of narrow hollow bodies (Fig. 14);

2) projection onto a screen of an image produced by a microscope;

3) study of the processes of air flow and vortex formation (in aerodynamics);

4) use of SVD lamps in loop oscillographs.

B. The high intensity of radiation in the blue and violet regions of the spectrum makes the use of mercury SHP lamps suitable for photochemical processes, such as, for example:

1) in microphotography;
2) in studio motion-picture filming *);
3) in photosynthesis.

C. The high content of ultraviolet radiation, especially of the lines with wavelengths 3650 Å and 4047 Å (see Fig. 9), which produce luminescence, makes mercury SHP lamps an ideal light source in all cases where the phenomenon of luminescence is used **).

These include, for example:

1) Luminescence microscopy, where only with the use of SHP lamps was it possible to observe living objects under the microscope without specially staining them. As for specimens stained in the ordinary way, with SHP lamps one can obtain luminescent images so bright that they can be projected onto a small screen.

2) The use of luminous paints (chiefly in theatrical productions), where projectors with SHP lamps covered by a “black” filter are used.

II. SHP LAMPS FILLED WITH INERT GASES

8. FEATURES OF SHP GAS LAMPS

The advantages and disadvantages of mercury SHP lamps also determine the limits of their application. It is quite clear that practical needs for light sources possessing increased economy (in comparison with incandescent lamps), high brightness, and desired spectral characteristics cannot be exhausted by the availability of mercury SHP lamps alone. The question arises: can vapors of other metals or gases be used for SHP lamps?

Vapors of sodium, cadmium, zinc, tellurium, cesium, and other metals have been used and are being used in low- and high-pressure discharge lamps for experimental and technical purposes. However, the dependence of the vapor density in the lamp on the bulb temperature, which for almost all metals leads to the need for extremely high temperatures in order to attain the required pressures, sets a natural limit on the use of most metals for SHP lamps.

*) In motion-picture filming with SHP lamps, the stroboscopic effect, which interferes with filming, is eliminated by switching the lamps into different phases of the mains.

**) The visible radiation is in this case blocked by a special “black” filter placed in front of the lamp.

In individual cases, as for example for cesium, an obstacle is the chemical action of the metal vapor on the lamp glass. This action increases strongly with an increase in the vapor temperature.

But, in addition, and this is especially important, a common shortcoming of almost all light sources using a discharge in metal vapors is the spectral composition of the radiation, usually represented by a few individual lines.

As for gases, the practical application of an electric discharge in them has existed for a comparatively long time (for example, advertising tubes with argon, neon, helium); however, the use of inert gases has hitherto proceeded along the line of applying low pressures, low current densities, and gases of small atomic weight. This resulted in a comparatively low efficiency and low brightness of the discharge in inert gases.

Quite recently,^16,17,18 with regard to heavy inert gases (argon, krypton, xenon), use has been made of the previously found, for metal vapors, dependence between the efficiency of the discharge and its brightness—on the one hand—and the vapor density (pressure) and current density—on the other hand. The use of this dependence in the case of inert gases is attractive already because, in the most general case, the gas density in the lamp is not connected with the temperature of the coldest part in it, as occurs in lamps with metal vapors. Hence follow the fundamental advantages of gas HID lamps in comparison with mercury lamps: the possibility of considerably lightening the thermal load on the lamp bulb (relatively low wall temperature), and the independence of the luminous and electrical characteristics of the lamps from the temperature conditions of the external environment over very wide limits (the lamps do not require thermal insulation). Very significant and valuable in operation is the absence of a process of evaporation of a liquid phase and, consequently, the absence of the warm-up process characteristic of lamps with a discharge in metal vapors.

But in addition to these differences, HID lamps filled with heavy inert gases possess other features as well, the chief of which is the good color of their radiation.

9. DESIGN AND TECHNOLOGY

The constructional design of gas HID lamps may, in the main, be the same as that of mercury lamps. However, taking into account that in gas lamps there is no dependence between the volume and surface area of the bulb and the lamp power, on the one hand, and the principal lamp parameters (warm-up time, brightness, luminous efficacy), on the other hand, the task of the designer of gas lamps is greatly facilitated. It is facilitated to the same degree also because, unlike mercury lamps, for gas HID lamps the presence of “pockets,” i.e., cold excess volumes, not only

capillary ones, but also sufficiently large ones, is absolutely harmless.

A practical consequence of this circumstance is the possibility of sealing the lamp leads into the bulb in such a way that there is a small gap between them and the wall of the bulb. Such sealing of the leads, in which the pressure on them from the outside is equal to the pressure experienced by the sealed-in electrode and the molybdenum foil from the inside, prevents delamination of the foil welded into the quartz or rupture of the leads, which occurs quite often in mercury UHP lamps.

Owing to the very high breakdown potential, gas UHP lamps must, as a rule, be provided with a third—ignition—electrode. Fig. 15 shows an example of the construction of a lamp.

With respect to manufacturing technology, gas UHP lamps are simpler than mercury ones for two reasons:

1) There is no need for very precise dosing of the mercury, which is replaced by the simpler operation of dosing the amount of gas in the lamp. Since the magnitude of the voltage gradient in an arc in mercury vapor is approximately 5 times greater than in heavy inert gases (see below), and, moreover, for mercury this quantity varies more strongly with pressure, equal percentage errors in dosing have an incomparably weaker effect on the magnitude of the voltage gradient on the lamp in the case of gas lamps. This greatly reduces rejects due to improper dosing.

Fig. 15. UHP lamp filled with an inert gas (scale in cm).

2) The presence of a high gas density at the moment of discharge formation during ignition of gas UHP lamps and, thus, the absence of the enormous change in pressure between the moment of ignition and the moment of establishment of the operating regime, which occurs in mercury UHP lamps (a change by hundreds and even a thousand times!), makes it possible to dispense with the use of oxide or other cathodes with a reduced electron work function. The principal role of the oxide in mercury lamps (apart from lowering the ignition potential) consists in reducing the cathode potential drop in the phase of the glow discharge preceding the occurrence of the arc during ignition and, consequently, in reducing cathode sputtering.

Since in gas lamps the gas density at ignition is three orders of magnitude higher than in mercury lamps, this circumstance alone is sufficient for the sputtering effect to prove ...

negligible. In this connection it is possible to dispense completely with the use of oxide cathodes.

The use of pure tungsten electrodes instead of oxide ones greatly simplifies and cheapens the production technology.

10. SWITCHING CIRCUITS, IGNITION AND RUN-UP OF LAMPS

The switching circuits of HPMV gas lamps are analogous to those for mercury lamps. Ignition is carried out with the aid of an igniting device connected to the third, auxiliary, electrode of the lamp. Repeated ignitions of a burning and switched-off lamp take place in the same way as the first. The run-up time is practically absent. It is true that, because in an operating lamp a definite thermal regime is established, under which the density of the gas inside the arc cord decreases in comparison with that which existed at room temperature, its luminous characteristics change somewhat during several minutes after switching on. But this is not of such substantial importance and does not constitute such an inconvenience as the run-up of HPMV mercury lamps (Fig. 16).

Fig. 16. Run-up characteristics of HPMV mercury and gas lamps:
1 — luminous intensity, 2 — current through the lamp, I — HPMV gas lamp, II — HPMV mercury lamp.

11. ARC FORM; STABILIZATION OF THE DISCHARGE

Owing to the great lightness of argon, krypton, and xenon in comparison with mercury vapor, convection in the discharge in the indicated gases is extremely strong, increasing as the atomic weight of the gas used decreases. For this reason, stabilization of the arc discharge in gases at superhigh pressure is determined chiefly by convection. This type of stabilization, analogous to the case of a carbon arc in air, is the least favorable of all the three described above. It determines the external form of the discharge: a relatively broad bright arc, slightly asymmetric with respect to the electrodes (expanding toward the upper electrode). The presence of strong convective gas currents leads to more or less significant displacements of the arc over the lamp electrodes. The shape of the electrodes, as well as the distance between them, can change the external appearance of the arc. Thus, in the presence

cone-shaped electrodes with a small (2–3 mm) distance between them, the stabilization of the arc is of a mixed type—electrode-convective. In this case the arc becomes narrower (and correspondingly brighter), and the stabilization of the discharge is considerably improved.

Convective gas flows lead to the fact that, when the arc deviates from the vertical, the dimensions and shape of the arc change greatly. Carried along by these convective flows, the arc bends and, in limiting cases, may reach the wall of the bulb. Changes in the linear dimensions of the arc lead to a significant change in the electrical and luminous characteristics of the discharge. Thus, for the normal burning of HID gas lamps it is necessary to place them in a vertical position or with only a slight deviation from it. This is their disadvantage. At very small distances between the electrodes, when stabilization of the arc by convection is entirely replaced by stabilization by the electrodes, the position of the lamp, as in mercury lamps, becomes immaterial.

The arc in inert gases is very sensitive to magnetic and electric fields. This feature can easily be used for forced stabilization of the discharge, as is practiced for carbon arcs. When “magnetic blowing” is used, the position of the arc likewise plays no role.

Finally, another distinctive feature of the discharge in heavy inert gases is the dependence of the diameter of the luminous channel (cord) of the arc on the current. In mercury HID lamps the diameter of the arc changes almost not at all as the current increases, and the growth of radiation, therefore, occurs almost exclusively as a result of an increase in the temperature of the arc¹⁹. As for gas HID lamps, they are characterized by an increase in the width of the arc simultaneously with an increase in current.

Measurements show that the cross section of the arc in inert gases is proportional to the current, and its diameter to the square root of the current.

Consequently, the current density in the arc remains practically almost constant as the current increases. The increase in radiation occurs at the expense of an increase in the cross section of the discharge. It follows from this that the temperature of the arc discharge in inert gases and the concentration of electrons in it, with increasing current, change almost not at all.

12. ELECTRICAL CHARACTERISTICS

Table III gives the results of measurements of the voltage gradient in the arc and the total voltage drop at the electrodes for HID lamps filled with different gases at the same pressure¹⁸.

For comparison, analogous figures are also given for mercury SVD lamps.

Table III

Kind of gas or vapor	Pressure (atm)	Voltage gradient in the arc (V/cm)	Sum of the anode and cathode voltage drops (V)	Magnitude of the ionization potential (eV)
Argon	35	26	16	15.7
Krypton	35	30	12	14
Xenon	35	38	11	12
Mercury	35	130	10–12	10.4

As can be seen, the magnitude of the voltage gradient in gas SVD lamps, in contrast to mercury lamps, is very small. The explanation of this fact lies in the high concentration of electrons in the discharge plasma and their very great mobility. The latter is due to the fact that, at a discharge-column temperature of the order of 8000–10 000° K, which corresponds to a mean electron energy of about 1 eV, the effective transverse cross section of atoms of the heavy inert gases for these electron velocities has a sharply pronounced minimum (the Ramsauer effect).

The low gradient is a disadvantage of the lamps, since it leads to small voltages on the arc. Consequently, if one wishes to increase the power of the lamp, the increase in power is practically possible only by increasing the current. Apart from the fact that, as has already been pointed out, the problem of designing hermetic seals into quartz for large currents has not yet been finally solved, the use of large currents has a second negative aspect. The point is that the low voltage on the lamp (20–30 volts), with a relatively high supply voltage (150–200 volts), makes it necessary to extinguish the excess voltage in a ballast resistance connected in series with the lamp, and this greatly reduces the economy of the entire installation. It is true that, provided special supply circuits or special lamp designs are used (for example, with one movable electrode, when ignition is effected by contact between the electrodes), a considerable increase in economy is possible while simultaneously using lower supply voltages. But this is associated with certain complications in the supply circuit and in the design of the lamp.

The magnitude of the total voltage drop at the electrodes in gas SVD lamps is of the same order as in mercury lamps. It is interesting that

This quantity almost coincides with the ionization potential of the gas with which the lamp is filled. Since the magnitude of the ionization potential decreases from argon to xenon, the sum of the cathode and anode voltage drops is correspondingly smallest in xenon lamps.

On the other hand, the voltage gradient increases from the gas with the smaller atomic weight—argon—to the heaviest—xenon.

Both of these circumstances make xenon the most advantageous gas for use in ultra-high-pressure lamps.

13. SPECTRAL AND LIGHT CHARACTERISTICS

The most interesting and valuable distinction between gas ultra-high-pressure lamps and mercury lamps is their spectral characteristics. If in a mercury discharge, as we have seen, the radiation is mainly concentrated in intense lines falling in the visible and near ultraviolet regions of the spectrum, then for a discharge in inert gases (especially in xenon) at high current densities there is characteristically an almost complete absence of intense lines in the visible part of the spectrum and their absolute absence in the ultraviolet. Line radiation occurs mainly in the near infrared region. As for the ultraviolet and visible regions, all the radiation in them is due to an intense continuous spectrum.

The lower boundary of the output of ultraviolet radiation is determined by the transparency of the quartz bulb of the lamp, i.e., practically it falls at a wavelength of about 2000 Å.

The continuous spectrum also extends into the infrared region, but there its intensity is considerably less than the intensity of the line radiation. The intensity maxima of these lines for different gases are connected with the atomic number of the gas—the larger it is, the more the intensity maxima shift toward longer wavelengths.

In Fig. 17 are shown the energy distributions in the visible and near infrared regions of the radiation spectrum of gas ultra-high-pressure lamps (the curves are given on different scales). In Fig. III (see insert) a photograph of the spectrum in the visible and ultraviolet regions is given. The radiation intensity depends strongly on the current. In a first approximation it is proportional to the square of the current.

It may be considered that, over wide ranges of variation of current and pressure, the character of the energy distribution over the spectrum remains unchanged. This is also supported by the fact that krypton and xenon lamps recently designed for powers up to 10 kW with low gas pressure (about 700 mm Hg), but with high current density, give a spectrum that does not differ from the spectrum of lamps with pressures of tens of atmospheres2.

The luminous efficacy of the discharge in heavy inert gases, other conditions being equal, increases with increasing atomic weight of the gas. Its greatest values, therefore, were obtained for a discharge in xenon and, in absolute magnitude (at a pressure of about 35 atmospheres and a current of 30 amperes), are equal to 30 lm/W. The brightness of the discharge arc in this case has a value of the order of 15–25 thousand stilbs.

Fig. 17. Energy distribution in the visible and near-infrared regions of the spectrum of SVD gas lamps: a — krypton SVD lamp, b — xenon SVD lamp.

The combination of high brightness, with relatively good efficiency, and very favorable spectral characteristics of the discharge, in which its radiation differs little from solar radiation, may make gas lamps a competitor not only to mercury SVD lamps, but also to carbon arcs in many areas of their application.

14. ORIGIN OF THE CONTINUOUS SPECTRUM IN THE RADIATION OF SVD GAS LAMPS

It was indicated above that a continuous background of radiation is also observed in the spectrum of a mercury discharge when the mercury-vapor pressure is increased. However, in the spectrum of a mercury SVD discharge the intensity of the continuous background is much less than the intensity of the individual broadened lines, whereas in SVD lamps with inert gases most of the lines disappear against the background of the continuous spectrum.

As a result, the color of the radiation of SHP gas lamps is very close to white. The shades of color depend on the type of filling gas. Thus, argon SHP lamps are characterized by a pleasant bluish tint, krypton lamps by a pinkish one, and xenon lamps by a yellowish one.

The question of the origin of the continuous spectrum in the glow of gases arose much earlier than the sealed-off electric discharge in inert gases at high and superhigh pressures was realized. Incandescent gaseous atmospheres of the sun and of many stars possess a continuous spectrum, in contrast to the line spectrum of gaseous nebulae. A continuous spectrum is also observed in the radiation of the channel of a spark discharge.21—29

It is impossible to explain the transformation of a line atomic spectrum into a continuous one by the influence of atomic fields on the energy levels when gas atoms pass at close distances from one another, since the broadening and splitting of spectral lines caused by this influence is clearly insufficient for such an effect.

The origin of stellar and gas-discharge continuous spectra has been subjected to detailed theoretical consideration, followed by experimental verification of the conclusions obtained.30, 31, 32

One of the proposed explanations assumes that in a continuous gas spectrum we are dealing with spectral bands of a molecular spectrum, composed of different lines, under conditions analogous to those that occur in so-called predissociation.33, 34 However, in all cases in which continuous spectra of gases are observed (stellar atmospheres, the column of an arc discharge, the channel of a spark), the temperature of the gas is so high that prolonged existence of undissociated gas molecules is improbable. Still less probable is the formation of stable molecules when neutral and excited atoms of a monatomic inert gas meet. Only the formation of “quasimolecules” that immediately decompose is possible. But even in this case calculation shows that the intensity of the continuous spectrum of such “quasimolecules” must be extremely small.21 Moreover, the regions occupied by the continuous spectrum would have to correspond to the positions of the individual line bands of the molecular spectrum, and the continuous spectrum could not extend continuously through all regions, from the ultraviolet to the infrared.

In the radiation spectrum of a low-pressure gas discharge, in some cases (for example, in the head of a glow discharge) regions of continuous spectrum are observed adjoining the limits of individual spectral series on the short-wave side. The presence of regions of continuous spectrum is explained by the process of recombination of free electrons of the most varied energies with positive ions. In this process transitions of electrons take place from nondiscrete (free) energy levels to discrete

atomic levels. For the “recombination glow” to have an appreciable intensity, a considerable concentration of free electrons and positive ions is necessary, together with a comparatively small velocity of their relative motion.

In the isothermal plasma of a channel of an ultra-high-pressure discharge (as also in the isothermal plasma of stellar atmospheres) these conditions are present. The very process of equilibrium thermal ionization presupposes not only the constant detachment of electrons from atoms, but also their constant recombination. At the same time, the influence of the electric fields of some atoms on others leads not only to a broadening of spectral lines, but also to a lowering of the ionization potential of the atom or, in other words, to a lowering of the potential barrier at the edges of the atom’s “potential well.” When the number of atoms is large, the ionization boundary is, as it were, smeared out toward the lower-lying and, in turn, likewise smeared-out excitation levels, and merges with them. As a result, a considerable fraction of the atoms which under other conditions would only be excited proves to be ionized.

In exactly the same way, in place of some of the excited ions there appear doubly ionized atoms. All this leads to an increase in the number of recombination events, and consequently also to an increase in the intensity of the recombination spectrum.

Nevertheless, the radiation of light quanta in transitions of free electrons to discrete, though broadened, levels is still insufficient to explain the continuous background in all regions of the spectrum, since the energy given up by an electron settling onto a discrete energy level is at least equal to the energy required to detach the electron from this level, and therefore the recombination spectrum ought to have a more or less sharp boundary on the long-wavelength side.

However, quanta of electromagnetic radiation arise not only when electrons pass to discrete atomic levels, but also when electrons are decelerated in the field of ions.

An example of such radiation under deceleration is found in the generation of “white” X-ray radiation when electrons strike the anticathode of an X-ray tube. Radiation under deceleration is a consequence of the transition of an electron from one nondiscrete energy level to another, likewise nondiscrete, level. An electron moving in the chaos of the microfields of a plasma may fail to enter a decelerating field. This will occur, for example, in the case when the electron describes some curve around a positive ion, or shoots past the ion without remaining in the latter’s system. The inevitable deceleration in such cases over a certain segment of the electron’s trajectory will also cause the inevitable radiation resulting from a transition from one energy level of the electron to another level. In this case neither the one nor the other

level is not discrete—they belong to levels capable of changing continuously (the so-called “free” levels, corresponding to a continuous series of values of the energy of a free electron). The values of the difference of the energies of both free levels, and consequently the energy of the emitted quantum, may be arbitrary.

The continuous emission spectrum may in this case extend arbitrarily far toward both long and short waves. Thus, in brief, it may be said that the continuous spectrum of the arc-discharge column at very high pressure in inert gases and the spectrum of the channel of a spark discharge are at present explained by “discrete-free” and “free-free” electron transitions, i.e. by recombination radiation and by the bremsstrahlung of electrons in an isothermal plasma.^17

A calculation of the intensity, based on integration over all frequencies of the intensity of bremsstrahlung and of the intensity of separate continuous bands corresponding to broadened energy levels, led to Unsöld’s formula:

\[ 4\pi \varepsilon_\nu = \gamma \frac{128\pi^3}{3\sqrt{3}} \left(\frac{e^2}{hc}\right)^3 Z_{\mathrm{eff}} p\, e^{\pm \frac{(U_i-\Delta U)e}{kTp_g}}, \tag{5} \]

(where \(Z_{\mathrm{eff}}\) is the effective nuclear charge, which may be taken equal to unity, \(\gamma\) is the statistical weight of the ground state of the given atom, \(p\) is the gas pressure, \(U_i-\Delta U\) is the effective ionization potential, equal to \(U_{i\,\mathrm{eff}}\)), according to which the intensity of the continuous background, referred to unit frequency (\(1\ \mathrm{cm}^{-1}\)), should not depend on the frequency of the emitted light.

Experiment shows that this relation holds only approximately. Later investigations explain the discrepancy between theory and experiment by the fact that Unsöld did not quite legitimately replace the summation of the different recombination levels by integration.^31,^32

In order to understand why the ratio of the brightness of the continuous spectrum to the brightness of individual lines is much greater in the case of a very-high-pressure discharge in monatomic inert gases than in the case of the same discharge in mercury vapor, it is necessary to compare the arrangement of the electron energy levels in a mercury atom and in the atoms of monatomic inert gases.

The arrangement of these levels is shown schematically in Fig. 18. The diagram in Fig. 18 shows that the ratio of the ionization potential to the first excitation potential is much greater for a mercury atom than the same ratio for atoms of inert gases. In the case of monatomic

for inert gases the average distance between the excited levels and the ionization level is much smaller. Therefore, when the broadened ionization levels merge in strong atomic electric fields with the likewise broadened excitation levels, a much larger number of lower-lying levels is captured than in the case of mercury vapor, and the probability of recombination, compared with the conditions occurring in mercury vapor, increases sharply.

Fig. 18. Arrangement of the excitation and ionization levels in atoms of neon, argon, krypton, xenon, and mercury.

In addition, proceeding from the Boltzmann and Saha equations (1), (2), it is not difficult to show that the intensity of radiation of individual spectral lines \(I_{\text{line}}\) is proportional to the first power of the concentration of gas atoms \(N\) and to the factor \(e^{-\frac{U_a e}{kT}}\), where \(U_a\) is the value of the excitation potential of the upper level of the given line, i.e.

\[ I_{\text{line}}=\mathrm{const}\,N\cdot e^{-\frac{U_a e}{kT}}, \tag{6} \]

ULTRA-HIGH-PRESSURE LAMPS

whereas the intensity of the recombination glow \(I_{\text{rec}}\) is proportional to the square \(N\) and to the factor \(e^{-\frac{U_i e}{kT}}\), where \(U_i\) is the effective, i.e. reduced by the atomic electric fields, ionization potential. Hence:

\[ I_{\text{rec}}=\operatorname{const} N^2 e^{-\frac{U_i e}{kT}} . \tag{7} \]

Therefore, with increasing atom concentration \(I_{\text{rec}}\) grows much faster than \(I_{\text{lin}}\).

Further, the temperature of the discharge channel is determined by the energy balance of the latter. In this balance the expenditure of energy is mainly for radiation. Since \(U_m\) (the averaged value of the excitation potential) in the case of inert gases is higher than in the case of mercury vapor, then, at one and the same temperature, the expenditure of energy on the radiation of the super-high-pressure arc in inert gases is less than in mercury vapor. This leads to an increase in the arc temperature in the case of inert gases as compared with mercury vapor. In turn, the higher values of the arc temperature \(T\) compensate in expression (7) for the brightness of the recombination glow the higher initial values of the ionization potentials of the inert gases. For inert monatomic gases the ratio \(\frac{I_{\text{rec}}}{I_{\text{lin}}}\), and consequently also the ratio of the brightness of the continuous background to the brightness of individual lines, increase with increasing current density and gas pressure more rapidly than for mercury vapor, since, for the reasons indicated above, in the case of inert gases the atomic electric fields, when the gas pressure is increased, lead to a greater lowering of \(U_{i\,\text{eff}}\) than is observed for mercury vapor*).

In addition, the expanded Saha equation

\[ \frac{\alpha^2}{1-\alpha^3} = \frac{\sigma_i \sigma_e}{\sigma_0}\cdot 4.73\cdot 10^3 \frac{1}{p} \left(\frac{kT}{e}\right)^{\frac{5}{2}} e^{-\frac{eU_i}{kT}} \tag{8} \]

contains the factor \(\frac{\sigma_i \sigma_e}{\sigma_0}\), where \(\sigma_i\), \(\sigma_e\), and \(\sigma_0\) are, respectively, the sums of states of the ion, electron, and neutral atom, practically equal to the statistical weights of the fundamental levels. For a free electron the statistical weight \(\sigma_e\) is always equal to 2 (the two possible directions of the electron spin vector). For mercury \(\sigma_0=1\); \(\sigma_i=2\) (i.e.

* To this it may be added that for mercury there is at least a diminished, but still definite, value of \(U_{i\,\text{eff}}\) (9.7 volts as against 10.4 volts for free atoms),\(^{35}\) whereas for inert gases, owing to the structure of their atoms, the value \(U_{i\,\text{eff}}\) must continuously decrease with increasing \(T\). On the contrary, the excitation potential of the lines does not change with temperature.

for the state of the valence electrons of the \( \mathrm{Hg}^+ \) ion there are two possibilities). For Ne, Ar, Kr, and Xe \( \sigma_0 = 1 \); \( \sigma_i = 6 \) (six different possible combinations of the quantum numbers of the valence electrons at the smallest value of the principal quantum number). Therefore the factor \( \dfrac{\sigma_i \sigma_e}{\sigma_0} \) in equation (8) for a discharge in inert gases is 3 times greater than for a mercury discharge, and the degree of ionization at the same value of \(T\) is at least \( \sqrt{3} \) times greater. A higher degree of ionization leads to a greater electron concentration and, further, to an even greater lowering of \(U_{i\,\mathrm{eff}}\) by atomic fields than follows from all the considerations set forth above.

Thus, the experimental fact that the ratio of the brightness of the continuous spectrum to the brightness of individual discrete spectral lines in the case of a SHP discharge in inert gases is much greater than in the case of a discharge in mercury vapor is explained by the whole complex of considerations presented above.

Fig. 19. Distribution of the radiation energy of SHP gas lamps in the ultraviolet region of the spectrum.

In the ultraviolet region, the radiation of SHP lamps with inert gases, in contrast to mercury SHP lamps, which have an intensity gap at \( \lambda = 2537\,\text{\AA} \), embracing a rather wide interval of wavelengths and explained by reversal of resonance lines, is continuous, without gaps and peaks (Fig. 19 and Fig. IV). Such a gap is not observed because the resonance lines of inert monatomic gases, owing to the comparatively large value of the excitation potential \(U_a\), lie in the far ultraviolet region, which escapes observation because of its absorption by the quartz envelope of the lamp, by air, and by the spectral apparatus.

15. POSSIBLE FIELDS OF APPLICATION OF GAS SHP LAMPS

As with mercury SHP lamps, the possible fields of application of SHP lamps filled with heavy monatomic gases are determined by their luminous, spectral, and electrical characteristics.

It may be asserted that in a large number of cases where mercury SHP lamps are used, gas lamps may be used with still greater success. At the same time, thanks to their inherent features, gas SHP lamps may find specific fields of application distinct from those of mercury lamps. We shall cite some of them:

a) Owing to the presence of intense continuous radiation in the ultraviolet and visible parts of the spectrum, SHP gas lamps, even of low power, are successfully used in the field of spectral analysis in the study of absorption spectra, for the investigation of photochemical processes, and for other similar purposes. The advantage of SHP gas lamps in comparison with low-pressure hydrogen lamps is the high intensity of the radiation, which makes it possible greatly to accelerate research work in the indicated fields.

b) The exceptionally favorable spectral distribution of the radiation energy in the visible region, in combination with considerable brightness and small dimensions of the luminous field, makes SHP gas lamps indispensable for illumination and projection purposes in all cases where the fullest possible preservation of the color shades of objects is required. These include, for example, color photography, the shooting of color films, the projection of these films, etc.

c) The presence of a continuous spectrum in the ultraviolet region makes it possible to use the lamps in biology and biochemistry.

d) In those cases where a source of illumination is required that approximates solar illumination as closely as possible—for example, for mines, for dwellings under conditions of the polar night, for winter sports halls or swimming pools—SHP gas lamps, by virtue of their spectral characteristics, are almost ideal.

e) A number of applications of the lamps, both for technical and for scientific-research work, is possible in connection with the presence of selective radiation from the lamps in the near infrared region.

The enumerated fields of application, of course, do not exhaust all possible ones. Further research work in the field of discharge in gases at high pressures, in particular a detailed study of the electrical and electro-optical characteristics of the discharge, will be able considerably to expand the range of application of SHP gas lamps.

Light sources based on the radiation of a discharge in gases and vapors, including the SHP gas lamp, are the result of the development of one of the branches of new technology. In turn, this branch owes its successes to numerous investigations in the field of the physics of gas discharge. Thus, just as in other fields, the connection between scientific research and the development of technology is an indispensable condition for the further progress of both science and technology.

References Cited

1. V. V. Petrov, News of Galvanic-Voltaic Experiments..., St. Petersburg, 1803. (Republished in 1934 by Energoizdat.)
2. W. Lochte-Holtgreven and Maecker, Zeits. f. Physik, 105, 1 (1937).
3. V. A. Fabrikant, UFN, 305 (1938).
4. V. Fabrikant and V. Pul’ver, DAN, 3, 435 (1934).
5. C. Boj, Ingenieur (Utrecht) 50 (1935) E 91–92.
6. Heller, Physics, 6, 389 (1935).
7. Uyterhoeven, Elektrische Gasentladungslampen (1938).
8. R. Rompe, W. Thouret, W. Weizel, Zeits. f. Physik, 122, 1 (1944).
9. W. Weizel, R. Rompe, Schön, Zeits. f. Physik, 115, 179 (1940).
10. R. Rompe, W. Weizel, Zeits. f. Physik, 119, 366 (1942).
11. W. Shottky, Phys. Zeits., 25, 342 (1924).
12. N. A. Kaptsov, Electrical Phenomena in Gases and Vacuum, 1947, p. 479.
13. G. N. Rokhlin, DAN, 60, No. 5 (1947).
14. Mankopf, Zeits. f. Physik, 76, 396 (1932); Elenbaas, Physica, 1, 211 (1934); Becken and Seelieger, Ann. d. Physik., 24, 609 (1935); Rompe, Zeits. f. techn. Phys., 17, 377 (1936); Seelieger, Phys. Zeits., 42, 69 (1941).
15. Das Licht, No. 7/8 and 9/10, 1 (1944).
16. C. G. Suits, J. Appl. Phys., 10, 370 (1939).
17. P. Schulz, Ann. d. Physik, 1, 95, 107 (1947).
18. D. A. Gokhberg, Izv. AN SSSR, OTN, No. 9, 1277 (1949).
19. R. Rompe, P. Schulz, Zeits. f. Physik, 112, 691 (1939).
20. Photogr. Jorn. 89, 58 (1949).
21. Finkelnburg, Kontinuierliche Spektren, 1938, Ch. IV.
22. McCallum, Nature, 142, 614 (1938).
23. F. Mohler, J. Opt. Soc. Am., 29, 152 (1939).
24. Dejardin, J. de phys. et rad., 9, 132 (1938).
25. Vogel, Ann. d. Physik, 41, 196 (1942).
26. K. S. Vul’fson, Elektrichestvo, No. 11, 1946.
27. Hahn and Finkelnburg, Zeits. f. Physik, 122, 36 (1944).
28. Maecker, Zeits. f. Physik, 114, 500 (1940).
29. Maecker, Zeits. f. Physik, 116, 257 (1941).
30. A. Unsöld, Ann. d. Physik, 33, 607 (1937), and also his Physics of Stellar Atmospheres, IL, 1949; Schubert, Ann. d. Physik, 39, 295 (1941).
31. P. Schulz, Zeits. f. Physik, 119, 167 (1942).
32. Rösler, Zeits. f. Physik, 110, 352 (1938); 112, 667 (1939); 122, 285 (1944).
33. A. N. Terenin, Introduction to Spectroscopy, 1933.
34. M. I. Batuev, Izv. AN SSSR, Ser. Fiz., 14, 429 (1950).
35. R. Rompe, P. Schulz, Zeits. f. Physik, 110, 223 (1938).

N. A. Kaptsov and D. A. Gaukhberg, Superhigh-Pressure Lamps.

Fig. I. Improvement of stabilization and narrowing of the discharge arc with decreasing distance between the electrodes.

Fig. II. Arc of a superhigh-pressure mercury lamp, stabilized by electrodes:
a — burning vertically; b — burning horizontally; c — burning horizontally but rotating at a speed of 600 rpm; d and e — the same as a, b, and c, but photographed through a red filter that blocks the radiation of the arc torch.

a  b  c

d  e

Fig. III. Photograph of the spectrum of a krypton SVL lamp in the ultraviolet and visible regions of the spectrum (for comparison, the spectrum of a high-pressure mercury lamp is also given).

a

b

Fig. IV. a—arc of a high-pressure mercury lamp immediately after switching on; b—the same after warm-up.

1. In addition, under these conditions the shielding of the luminous flux by the lamp electrodes begins to have an effect. ↩

2. 20. ↩