TECHNIQUE FOR MANUFACTURING PHOTOCELLS*

Georges Déjardin

The manufacture of photocells is at present concentrated in only a few laboratories possessing the necessary equipment. This production entails a number of very serious difficulties—glassmaking, vacuum technique, work with pure alkali metals, pure gases, and so on—which make manufacture lengthy and complicated.

Because of these difficulties, in addition to equipment the laboratory must have staff members who have attained a high qualification in this production; only then can one be confident of its success.

Nevertheless, the ever-expanding field of application of photocells is compelling physicists and manufacturers to improve the methods of their production and to seek simpler and faster ways of making them. All the more so since work has now begun on designing photocells of certain definite types for use in such important matters as objective photometry, television, sound cinema, and many others. We therefore have every reason to expect that these methods will soon become industrial, just as happened recently with the construction of multi-electrode cathode tubes.

* Report at a public meeting of the Optical Institute in Paris, June 13, 1930. Rev. d’Optique; translated by V. L. Ginzburg.

GEORGES DÉJARDIN

I. MANUFACTURE OF THE BULB

The first difficulties encountered in the production of photocells consist in the manufacture of the glass bulb. Quartz is attacked by alkali metals and therefore cannot be used for this purpose. Ordinary glass has high conductivity and is easily covered with a film that conducts moisture.

Borosilicates are usually used, such as, for example, Pyrex. When working with this type of glass, the metal leads to the electrodes can be made very easily: tungsten wire (thoroughly cleaned) is taken and, at the place of sealing, is first coated with sealing glass. Other hard glasses (Schott, Philips, Photos, etc.) seal equally well to tungsten, molybdenum, and ferrochrome. In some cases it is recommended to coat the tungsten with a layer of cobalt.

There also exist so-called “intermediate” glasses (Schott, Philips, etc.), which make it possible to fuse quartz with relatively low-melting glasses, such as, for example, Thuringian glass. Thus, by using a bulb made of low-melting glass, one can at the same time avoid the difficulties of sealing in the electrodes. The use of Thuringian glass greatly simplifies the construction of photocells, and at present it is being used successfully. The leads in this case are made of platinite coated with a drop of glass.

The lead-in conductors may be mounted in a “stem,” just as is done in lighting and cathode lamps.

The metallic parts must first be degassed by prolonged heating in vacuum, and the interior of the bulb must be cleaned in the most thorough manner (with a solution of chromic acid in sulfuric acid, alcohol, etc.). It is better to seal the bulb on the pump, so that it is contaminated as little as possible. Contact with the cathode base is made by means of a platinum wire pressed from the inside against the glass and brought out by its two ends.

In some cases the cathode is deposited not on the inner walls of the bulb, but on a metal plate (nickel, silver, etc.) placed in the required position inside the bulb.

To eliminate insufficient insulation of the surface layer of the glass (the conducting film on the inner surface is extremely thin), two protective rings (inner and outer), formed by silver deposition and connected to ground, are often arranged. This device eliminates the dark current, but it is not used in photocells of widespread use.

The shape of the bulbs is made very different, depending on the purpose of the photocell: spherical, cylindrical, tubular with a flat wall, with a side tube for sealing in a quartz window. The spherical form is one of the oldest and most widespread forms. In it the window is often made round and small in comparison with the spherical part of the bulb. Owing to this, a series of repeated reflections from the surface of the cathode is obtained, which do not pass outward and which increase the photocurrent. In reality, for the most commonly used cathodes (for example, potassium treated with hydrogen), this “black-body” effect does not give a noticeable increase in sensitivity, since the most actinic radiation is strongly absorbed by the surface of the cathode, and the multiply reflected light therefore produces only a very small increase in photocurrent. However, in those cases where the cathode consists of films of alkali metal deposited on magnesium, the influence of repeated reflections may become significant.

Photocells with a glass bulb cannot be used in the ultraviolet region, except for the part close to the visible, for which the window still has sufficient transmission. But, on the other hand, bulbs made of quartz are very expensive. Therefore designs of photocells with a glass bulb, but with a sealed-in quartz window, deserve much attention.* This is extremely convenient

* See, for example, E. O. Hulburt. Astr. Journ. 41, 400, 1915.

method for obtaining simple and cheap photoelements (sodium and cadmium), sensitive to ultraviolet.

We said above that some hard glasses, such as Pyrex, possess considerable transmission, exceeding that of crystals and optical glasses, in the near ultraviolet. With a thickness of 2 mm the transmission limit for Pyrex lies at about 3200 Å. The variation of this boundary with thickness was studied with a quartz spectrograph using a mercury lamp as the source.

With a glass thickness of 0.25 mm, the resonance line—2537 Å—is readily obtained on the spectrogram. By reducing the thickness of the plate to several hundredths of a millimeter, one can obtain the spectrum of the mercury lamp down to 2200 Å.

Other glasses of this type (Sibor glass, Thuringian glass), when measured under the same conditions, gave entirely analogous results.

The great transmission of glasses at very small thicknesses was used by Davis for work with X-rays, and by Slack, for obtaining Lenard radiation.* On the other hand, Sonkin** showed that it is possible to make a mercury lamp of Pyrex with very thin walls, giving the same radiation as a lamp in a quartz bulb. The examples cited show that it is possible to make a photoelement sensitive to ultraviolet in a glass bulb (Pyrex or Thuringian glass), but with a very thin window.

The transmission of the window can be checked on a spectrograph before the final mounting of the photoelement. Very thin glass walls, necessary for good transmission of ultraviolet radiation, nevertheless offer sufficiently great resistance to atmospheric pressure and withstand, without special precautions, more—

* Davis and Slack. Journ. Opt. Soc. Amer. 19, 65, 1929.
** S. Sonkin. Journ. Opt. Soc. Amer. 19, 65, 1929.

heating. To protect against impacts, it is very convenient to arrange the window as shown in Fig. 1, but other designs may be used with equal success.

II. PREPARATION OF THE CATHODE LAYER

The walls that are to be coated with the cathode layer are first covered with a film of metal, which provides electrical contact with the cathode and promotes better deposition of the alkali metal.

For this purpose, the required portion of the inner walls is silvered, taking precautions not to coat the remaining parts of the bulb (the neck is made of a special shape, and the parts that must remain clean as a result are covered with natural wax, which is then dissolved in gasoline).

Fig. 1.

The silvering must be such that it withstands, without deterioration, strong heating (during the annealing of the photocell on the pump). In this respect, Martin’s method* gives very good results.

Instead of silvering by a chemical method, the method of subliming silver in vacuum may be used. For this, a thin silver wire is wound around a tungsten incandescent filament.**

In other cases, a layer of copper obtained by electrolysis is coated with silver. Sometimes platinum is used (but this has no practical significance) as an underlayer for the cathode layer. Finally, in especially sensitive photocells, calcium fluoride obtained by sublimation in vacuum serves as the underlayer for the alkali metal (cesium).*

* Martin. See A. Danjon. L’Astronomie. 255, 1924.
* English patent No. 303476, 1928, and French patent No. 668997, 1927.
* English patent No. 809224, 1928.
*** French patent No. 682254, 1929.

The metal most easily deposited on glass is undoubtedly magnesium. Indeed, it is enough to point to its use in “electron tubes,” where it plays the role of a “getter.” A piece of magnesium, when heated, readily evaporates in vacuum and gives a shiny deposit on the cold walls of the bulb. For sputtering, a strip of magnesium is introduced inside a heated tungsten spiral. In the same way, by attaching a piece of magnesium to the nickel anode of a photoelectric cell, it can be heated in a high-frequency furnace. The window and the neck of the bulb must be shielded from deposition of magnesium. Small electric furnaces can also be made and used to heat separate parts of the bulb, thereby preventing deposition of magnesium on them.

In some cases, in order to obtain the necessary spectral sensitivity of the photoelectric cell, the metallic film serving as the underlayer for the cathode (silver, copper, magnesium) is slightly oxidized. For this purpose, oxygen is admitted into the bulb at a pressure of the order of 1–10 mm, and for several hours a discharge from an induction coil is passed through it. Oxidation is obtained even more simply by heating the bulb, filled with oxygen at low pressure, to a temperature of about 350° C.

The photoelectric cell is baked out on the pump (up to 400° C for about an hour or two) before sputtering the metal for the underlayer, which in turn immediately precedes the introduction of the alkali metal.

The last operation is extremely simple when potassium or sodium is used for the cathode layer. The metal is taken in its usual form and carefully cleaned of the kerosene in which it is ordinarily kept; the surface crust is removed, and it is repeatedly washed in petroleum ether. After cleaning, the required amount of sodium is placed in a Pyrex ampoule and distilled in vacuum. After distillation the ampoule is sealed. The residue from the first distillation is kept.

The photoelectric cell is placed on the pump, and through a narrow tube a balloon is attached to it, into which a previously opened Pyrex ampoule with sodium is placed.

Heated by an electric furnace, sodium sublimes and passes into the colder bulb of the photoelement. If certain precautions are taken, sodium vapors settle in the form of a thin film on the metallic underlayer. With an underlayer of magnesium and a layer of sodium, everything turns out very well and quickly. If the proper voltage is applied to the photoelement (illuminated), a galvanometer connected in series with the photoelement shows a current as soon as the evaporated metal enters the bulb. The operation can also be monitored by the attainment of maximum sensitivity. When the process is going well, no shiny droplets of metal should form on the surface of the cathode. It is possible, however, to remove excess metal by carefully heating the photoelement.

If a large quantity of output is needed, then of course several photoelements can be installed at once on a single vacuum apparatus.

Sodium can be obtained in a very pure state by direct decomposition (at a temperature of about 275° C.) of sodium azide, \(N_3Na\). To obtain a sodium deposit one may also use the exceedingly ingenious method indicated by Warburg in 1884, which is based on the electrolytic conductivity of ordinary glass.

In Fig. 2 there is schematically shown an apparatus for depositing sodium, which can easily be reproduced.* An electric lamp, an evacuated one of the economical type, into the glass of whose bulb soda enters, is fastened in such a way that the lower part of the bulb is immersed in molten sodium nitrate (\(NaNO_3\), at a temperature of about 312° C.). When the lamp filament is heated, electrolysis takes place, and sodium ions from the melt penetrate through the walls into the bulb. Inside, they are neutralized by electrons emitted by the incandescent filament and form metallic \(Na\), which settles on the walls in the upper cold part of the bulb.

* R. C. Burt, Journ. Opt. Soc. Amer. 11, 87, 1925; Phil. Mag. 49, 1168, 1925. See also “Physics, Chemistry, Mathematics, Technology in the Labor School” 4, 79, 1930. (Translator’s note.)

Glass, however, undergoes no decomposition in this process.

With an appropriate arrangement, somewhat different from that shown in Fig. 2, it is possible to obtain, with a 60-watt lamp at 110 volts, a magnesium deposit of 300 mg per hour. This method is also used in the industrial production of photoelements. “Telegraf” photoelements are made precisely by this method. These elements are vacuum types, of low sensitivity, and are suitable only in certain cases.*

Fig. 2.

The same method may also be applied to potassium, but the bulb must be made of a special grade of glass. The introduction of lithium by this method presents a number of difficulties.** For cesium and rubidium another method is usually used. The metal is prepared in a glass ampoule connected with the bulb, or directly in the photoelement itself. This method, also very convenient, has been described by Boer, Clausing, and Zecher.***

A mixture of solutions of barium azide and cesium chloride, taken in definite proportions, is rapidly evaporated in a vacuum. A glass vessel with this mixture is soldered to the opening in the bulb of the photoelement and slowly heated to a maximum temperature of 350° C. Barium azide decomposes at 120° C, and at a higher temperature the liberated barium decomposes the chloride, itself taking the place of the cesium that is released.

The indicated method is very convenient for introducing cesium

* Zworykin. Phys. Rev. 27, 613, 1926.
** See, for example: Miss Seiler. Astr. Jour. 52, 129, 1920.
*** Boer, Clausing and Zecher. Zeitschr. anorg. allg. Chemie 160, 1928, 1927.

TECHNIQUE FOR MANUFACTURING PHOTOCELLS

into the bulb of the photocell. This method gives excellent results and makes it possible, in particular, to prepare cesium photocells easily, following the technique indicated by Zworykin and Wilson.* Cesium can likewise be prepared inside the bulb itself, by treating a cesium salt with a suitable substituting metal.** The mixture is placed in a capsule made of iron or nickel, which is heated by a high-frequency furnace. These methods are also successfully applied to photocells in which the cathode consists of a plate of oxidized silver.

It should finally be noted that other metals as well (Cd, Zn, Mg) can be used for constructing the cathodes of photocells sensitive in the ultraviolet part of the spectrum.

In the case of sodium, the layer (thick or very thin) of alkali metal is subjected to treatment in order to increase the overall sensitivity, as was indicated by Elster and Geitel as early as 1911. This treatment also promotes a shift of the photoelectric threshold toward the red and affects the position of the maximum corresponding to the selective photoeffect.

Hydrogen, thoroughly dried and purified, is introduced into the bulb under a pressure of about one millimeter of mercury. Then a discharge (luminous) is passed through the gas from a small induction coil or from storage batteries of several hundred volts. Excellent results were obtained with batteries of 400 V. The voltage is applied to the electrodes of the photocell in the proper way (the positive pole to the anode).

As a precaution, so that the current should not exceed a few milliamperes, an additional resistance is included in the circuit. The discharge spreads inside throughout the entire volume of the bulb. After a certain interval of time, ranging from a few seconds to a minute, the surface of the cathode changes; correspondingly, the photoelectric—

* Zworykin and Wilson, Journ. Opt. Soc. Amer. 19, 81, 1929.
** See, for example (use of zirconium), Boel, Brovs, Emmens, Zeits. anorg. allg. Chemie 191, 118, 1930.

tric sensitivity. This change is controlled with extraordinary precision by the change in the color of the surface. The surface takes on a color depending on the nature of the deposit and on the discharge conditions (blue—steel-colored, blue-violet, light blue, greenish-blue, also grayish, when the droplets are very small or the metal film is very thin). The increase in sensitivity is followed by means of a galvanometer, and the operation is stopped when the maximum sensitivity has been reached.

The cause of the change in the photoelectric effect is, for the most part, a change in the structure of the surface film treated with gas, and in the optical absorption as a result of the change in color. In most applications it is preferred to increase the sensitivity at the expense of the stability of the photoelement.

The photocurrent spreads inside the bulb of the photoelement, making use of the well-known phenomenon of ionization by impacts. An inert gas, chiefly argon, is introduced into the bulb under a pressure of the order of ten millimeters of mercury (varying somewhat depending on the intensity of the light for which the photoelement is intended).

The argon must be well purified and dry; the presence of traces of nitrogen in it is not permissible.

In some cases, in order to eliminate the decrease of sensitivity in the red part, it is preferred to introduce hydrogen into the photoelement. It is very convenient to place inside the photoelement a palladium osmoregulator.*

III. SPECTRAL SENSITIVITY OF DIFFERENT TYPES OF PHOTOELEMENTS

The current produced by a photoelement obviously depends both on the intensity and on the spectral composition of the incident light. In the case of monochromatic illumination, the emission ratio (expressed, for example, in amperes per watt) is the ratio of the primary current (the saturation current of a vacuum photoelement) to the flux incident on the cathode.

* Astr. Journ. 52, 129, 1920.

In a first approximation, neglecting the influence of different angles of incidence, we regard emission as a quantity characterizing the cathode, independent of its geometrical form and of the manner in which the radiation enters the element. The curves representing the variation of sensitivity with wavelength (the curves of spectral sensitivity at equal energies of the incident radiation) are “bell-shaped” curves, having a maximum that shifts toward longer wavelengths as the atomic weight of the cathode increases. For pure metals, the wavelengths corresponding to the maximum, according to Zeiler’s measurements,* are:

The study of monochromatic sensitivity was carried out by M. J. Tovpert with alkali photocells having a magnesium underlayer, prepared by the research laboratory of the “Fotos” lamp company.

As light sources there were used a standard lamp with a carbon filament (2,000° K) and a 100-watt gas-filled lamp with a temperature of about 2,600° K. The beam of light from one of these sources was focused on the slit of a Hilger monochromator. In another series of measurements this monochromator was replaced by a Leiss monochromator with quartz optics for work in the ultraviolet. The photocurrent was amplified by a two-tube amplifier, which made it possible to use a microammeter for direct readings. The sensitivity of photocells of one and the same type may vary within fairly wide limits. Nevertheless, photocells with a thin layer on a magnesium underlayer are considerably more uniform than photocells with a thick layer treated with hydrogen.**

A large number of elements measured accurately under identical conditions make it possible to form an exact conception of the specific features of each type studied. The greater part of the photocells investigated were [[unclear: word continues on next page]]

* G. Rougier. Revue d’Optique 138, 1928.
** Likewise, they are also more stable.

field, but all the cells were measured under a voltage of 80 V, sufficiently far from the discharge potential. Under these conditions the stability of the phenomena seems beyond reproach, and the confidence in the measurements made is very high. However, the ordinates of the curves obtained (representing the “spectral sensitivity,” under the conditions just enumerated) are quantities of the same order as the emission values, as we defined it earlier. The absolute values of the ordinates were obtained by comparison of the values with a single photoelement with a Cs–Mg cathode, which has high sensitivity over a wide range of wavelengths.

We construct the first part of the relative-emission curve (with a definite light source, for example with a carbon incandescent lamp), introducing corrections for prismatic dispersion, i.e. we take into account the effective width of the slit. Denoting by \(S_\lambda\) the “spectral sensitivity,” we thereby specify the current in amperes per watt, for a certain wavelength referred to a wavelength interval equal to unity. The total current of the photoelement is proportional to the area bounded by the curve, i.e. to the integral

\[ I = \int S_\lambda E_\lambda\, d_\lambda , \]

where by \(E_\lambda\) we denote the spectral energy density (referred to a black body).

The flux incident on the photoelement is proportional to the integral

\[ F = \int K_\lambda E_\lambda\, d_\lambda , \]

(where \(K_\lambda\) is the luminous efficiency in lumens per watt).

Let us also introduce:

\[ I_\lambda = \frac{S_\lambda E_\lambda}{I}; \qquad F_\lambda = \frac{K_\lambda E_\lambda}{F}, \]

which characterize, if one may say so, the relative photoeffect and the relative illumination. These two quantities include only the ratios \(S_\lambda\), \(K_\lambda\), and \(E_\lambda\). They can therefore be easily obtained from the photocurrent curve (corrected

…for dispersion), from the curve of sensitivity of the eye and the curve of black-body radiation at the temperature of the light source.

The ratio \(I/F\) is obtained directly from experiment. In front of the photoelement window a diaphragm is placed, to limit the beam and eliminate stray light, and a lens focused exactly on the cathode surface. The entire photocurrent measured under these conditions is produced only by the given source, placed at some definite distance from the photoelement.

The magnitude of the incident flux, in lumens, is obtained by a simple calculation. The spectral sensitivity \(S_\lambda\) is found from the expression:

\[ S_\lambda = K_\lambda \frac{I_\lambda}{F_\lambda}\frac{I}{F}. \]

For \(\lambda = 5500\,\text{\AA}\) one may take \(K_\lambda = 621\) lumens per watt, and the value of \(S_\lambda\) (in amperes per watt) is obtained correspondingly for the given wavelength. From the calculation of the radiation curve (of a black body) we can construct curves of spectral sensitivity in equal energies for all the elements studied (the relative values of the ordinates may be reduced to one, if all the photoelements were tested under identical conditions). The preceding calculation is made with respect to the ordinates of the curve for the element with the cathode Cs—Mg, and from this all the remaining curves are obtained by simple subtraction.

For thin layers on a magnesium underlayer the sensitivity increases continuously toward the shorter waves throughout the entire visible spectrum (Fig. 3). The emission maximum for heavy metals, treated and untreated with hydrogen (Pohl, Pringsheim, and Miss Seiler), is not observed under these conditions. Measurements at the beginning of the ultraviolet give something like a maximum in the vicinity of wavelength \(3600\,\text{\AA}\), which is fully explained by the nature of the noble metals. The K—Mg photoelements, however, possess considerable sensitivity in the region of shorter waves than the visible spectrum. They can very successfully replace the commonly used photoelements (thick

of hydrated K) in most cases of practical application. The Cs—Mg photocell is very interesting; its curve has a slow, greatly extended decline over a broad region from 4000 to 5500 Å; even at about 6000 its sensitivity retains so considerable a value that the photocell can be used for measurements in this region as well. But the maximum sensitivity in the vicinity of 4850 Å found for it by Zworykin and Wilson* was not observed.

Fig. 3.

The preceding results should be compared with the measurements of Ives,** made by him with thin films of alkali metals (Na, K, Rb, Cs) deposited on a metallic surface (Au, Ag, Cu, etc.), carefully cleaned and polished, in particular on platinum. The photoelectric sensitivity of these films has no maximum in the visible region. But, on the other hand—

* Zworykin and Wilson. Journ. Opt. Soc. Amer. 19, 81, 1929.
** Ives. Ast. Journ. 60, 209, 1924.

...rons, Aives regards as possible the existence of a maximum in the near ultraviolet. The study of the change in spectral sensitivity with the thickness of the film and, in particular, the displacement of the photoelectric threshold has not been carried out for the case of a magnesium underlayer. It is possible that, in photocells prepared in precisely this way (with an underlayer of Mg), the layer of alkali metal corresponds to a thinner film than Aives studied.

Spherical photocells with a thin layer deposited on magnesium, as has already been said, are a kind of “black body,” owing to the influence of repeated reflections on the metallic layer, and therefore differ from photocells with a cathode consisting of a thick deposit treated with hydrogen. Here, for example, are the results obtained with a Cs—Mg photocell (A) and a potassium hydride photocell (B) as the light source; in both cases a 100-watt gas-filled lamp was used.

The extension of sensitivity into the region of red rays is at present the most urgent technical problem, to which very many attempts have already been devoted.

Photocells with a thin layer of potassium on a magnesium underlayer are notable for their great stability and simplicity of manufacture. It has therefore become of great interest, by subjecting their photosensitive layer to certain modifications, to increase thereby their sensitivity in the region of long waves. Experience has shown that the following two operations lead to good results.

1°. The potassium layer is treated with a luminous discharge in hydrogen, in exactly the same way as is done for thick...

layers of metal. Thus the overall sensitivity of the photoelement also increases (but much less than in the case of a thick potassium layer), and it is especially sensitized in the region around 6000 Å. In this respect the photoelement proves analogous, and often even superior, to Rb—Mg photoelements.

Fig. 4.

2°. The magnesium underlayer is first subjected to slight surface oxidation. In addition, after the introduction of the alkali metal the cathode is partially heated (to the required temperature), which helps to remove the excess metal. Under these conditions a photoelement is obtained that is sensitive to orange and red (Fig. 4). Such photoelements considerably surpass rubidium ones and are almost not inferior to cesium ones in the visible region. The use of an oxidized metallic underlayer (Ag, Cu, Ni, W, Ba, and Mg) is mentioned in various patents, but only for the purpose of increasing the overall sensitivity and stability, independently of the spectral sensitivity-

TECHNIQUE OF MANUFACTURING PHOTOCELLS

However, Campbell* showed that in the case of thin layers of metals the nature of the metallic underlayer (Ag, Pt, Cu) has a great influence on the sensitivity (global and spectral) of the photocell. Greater oxidation of the surface contributes to an increase in sensitivity in the region of long waves. According to Campbell, high sensitivity in the red region is obtained by using oxidized copper as the underlayer.** A cathode prepared in this way has two maxima of sensitivity at about 4200 Å and 6000 Å. Very high total sensitivity, advancing the boundary into the region of long waves up to the beginning of the infrared, to 1 μ, is obtained by using as the cathode a thin layer of cesium deposited on an underlayer of oxidized silver.***

The surface of all special cathodes probably consists of one continuous, monomolecular layer, a layer (K or Cs, O) analogous to the photocells studied by Langmuir and Kingdon (Cs, O, W) from the point of view of the appearance of thermionic emission.****

The photoelectric effect is mainly an action of the surface and is only slightly determined by the nature of the metal composing the bulk of the cathode.

The presence of a gas-treated film and of a monomolecular layer can produce a significant change in emission and a redistribution of the spectral sensitivity.

In the same way we may picture the process of sensitization by hydrogen—the principal role of the discharge probably consists in the formation of hydrogen ions.

The high sensitivity of potassium-hydride photocells in blue rays is of interest in a considerable number of applications.

It is possible, while preserving high total sensitivity, at the same time to increase it in the red region; this is achieved by coating the potassium deposit with a very thin layer of cesium.

* Campbell. Phil. 6, 683, 1928.
** Campbell and Ritchie.—Photoelectric cells. 37 and 47, 1929.
*** “Osram” C. M. photocell; see Campbell and Ritchie, p. 36.
**** Proc. Roy. Soc. A 107, 61, 1925.

Such a mixed cathode is treated in the usual manner: by a discharge in hydrogen. Photocells prepared in this way have considerable sensitivity and can be used over a considerably wider region than potassium photocells (measurement of absorption, etc.).

Sensitivity curves reduced to equal energy for a number of photocells do not make it possible to judge the relation between them, because of the complex radiation incident upon them.

Fig. 5.

However, in the case of solar radiation (a black body at a temperature of \(6000^\circ\mathrm{K}\)) the change of intensity over the entire visible region is sufficiently small; therefore the ordinates of the sensitivity curve may be identified with the relative values of the current for each wavelength. This cannot be done for a gas-filled lamp. In Fig. 5 are given data on the relative values of the photocurrent for photocells of various types, under illumination by a 100-watt gas-filled lamp (filament temperature \(2600^\circ\mathrm{K}\)).

In photocells with a very thin window transmitting the ultraviolet region, it is evidently also necessary to use a cathode sensitive to ultraviolet (Na, Cd, Zn). One of the most convenient methods consists in repeated deposition of sodium on a magnesium substrate. The spectral sensitivity of photocells prepared in this way was studied with a Leiss monochromator, using a quartz mercury lamp.

The relative values of the photocurrent thus obtained, corresponding to the various lines, are collected in the following table.

The values given are undoubtedly somewhat inaccurate because of the continuous spectrum accompanying the line spectrum of the mercury lamp. The distribution of energy in the spectrum of the mercury lamp was recently studied by E. O. Hulburt*—absolute values of the energy (ergs per atom per second) for a number of lines, under a definite operating regime of the lamp.

Using these intensities and the relative values of the photocurrent, one may try to represent the general form of the sensitivity curve of this photocell.

The curve obtained, in its general features, has the form of a bell and gives a fairly sharp maximum in the region between 3 400 and 3 800 Å.

* Hulburt. Phys Rev. 32, 593, 1928.