ON THE EXTERNAL PHOTOELECTRIC EFFECT IN ADSORBED FILMS*

R. Suhrmann, Breslau

Contents

1. Basic concepts and laws.
2. Historical survey.
3. General physical properties of adsorbed films.
4. The external photoelectric effect in simple adsorbed films:
   a) general influence of gases; action of hydrogen and oxygen on metallic surfaces;
   b) influence of degassing on the energy distribution of photoelectrons;
   c) action of various gases and vapors on the surface of metals and carbon;
   d) adsorbed metallic films;
   e) vectorial photoeffect in metal films
5. The external photoelectric effect in complex films.
6. Influence of temperature on the photoeffect of adsorbed layers.
7. Influence of the field.
   Literature.

The modern technique of manufacturing photocells consists in applying or adsorbing the photosensitive material (for the most part alkali metals), obtained in an extremely fine dispersed state, onto a metallic surface. Such systems give a relatively large yield of electrons and are distinguished by a number of remarkable physical properties that depend on their physicochemical structure. The study of these questions has greatly promoted the deepening of our knowledge of the nature of the external photoelectric effect. With the aid of the data obtained it has become possible to explain a large number of previously obscure phenomena, such as, for example, the influence of gases and the phenomenon of fatigue. Thus, at the present time we have arrived at a broad and comprehensive understanding of the photoelectric properties of films adsorbed on a surface.

1. Basic Concepts and Laws

Before passing directly to the subject, it seems to me advisable briefly to set forth the basic photoelectric concepts and laws.

* Ergebn. d. exact. Naturwiss. Translated by S. Yu. Elovich.

If we illuminate a metal plate placed in a vacuum with light of a definite wavelength, the resulting photocurrent is proportional to the intensity of the light. It is assumed here that the action of the light is exerted on one definite point on the cathode, for the photosensitivity of the photocathode, generally speaking, is different in different places.

If, on a diagram, we plot the values of the photocurrent calculated per unit intensity of the light as a function of the wavelength of this light, we obtain a curve of the spectral distribution of the sensitivity, which merges with the axis of abscissas on the side of long waves. The position of this boundary varies with temperature owing to the additional energy of the electrons. Therefore the boundary cannot be determined directly from the curve, and it has to be found by an indirect method, which we shall discuss below. In this way we obtain the boundary on the long-wave side \(\lambda'\), into which the thermal energy of the electrons no longer enters. \(\lambda'\), or the limiting frequency \(\nu'\), is connected with the work function \(\psi\) by Einstein’s equation

\[ h\nu' = e_0\psi, \tag{1} \]

where \(h\) is Planck’s constant (equal to \(6.55\cdot 10^{-27}\ \text{erg}\cdot\text{sec}\)), and \(e_0\) is the elementary charge (equal to \(1.591\cdot 10^{-19}\ \text{coul}\)).

If a black body is irradiated with an undecomposed beam of light, then the photocurrent obtained depends in the same way on the temperature of the radiator as the number of electrons emitted by an incandescent cathode depends on its temperature. Namely:

\[ I = MT^r e^{-\frac{e_0}{kT}\psi}, \tag{2} \]

where \(M\) (referred to unit area) is a certain universal constant, different from the thermionic constant \(A\) (equal to 60.2 or \(120.4\ \text{A}/\text{cm}^2\ \text{deg}^2\)), \(k\) is Boltzmann’s constant, equal to \(1.374\cdot 10^{-16}\ \text{erg}/\text{deg}\), and \(r\) is a number whose value is of order 2. The photocurrent \(I\) is called the “total photoelectric emissivity.” By means of equation (2) one can immediately obtain, for aluminum, the value of the electron work function from the photoelectric effect and, consequently, the red limit of the photoeffect.

If we connect the anode of the photocell to ground and the cathode to an electrometer, then under illumination with light of a definite wavelength the latter will become charged to a definite potential \(V_m\), which does not depend on the wavelength and depends on the frequency of the light, according to the relation

\[ h\nu = e_0 V_m + e_0\psi. \]

If \(\nu=\nu'\), then, as follows from equation (1), \(V_m=0\). Equation (3) assumes a somewhat different form when the work functions of the cathode \(\psi_K\) and of the anode \(\psi_A\) differ from one another. In this case, between the two electrodes there arises a certain contact potential difference

\[ V_{K,A}=-(\psi_K-\psi_A), \tag{4} \]

as a result of which the electrons emitted by the cathode acquire acceleration. Therefore

\[ h\nu = e_0 V'_m + e_0\psi_K + e_0 V_{K,A} \]

or

\[ V'_m=\frac{h\nu}{e_0}-\psi_A. \tag{5} \]

Thus, the measured maximum potential \(V'_m\) does not depend on the properties of the cathode and depends only on the work function of the anode.

As can be seen from the dependence of the current strength on the voltage in a field retarding the electrons, under irradiation with light of a given wavelength

it is also the case that electrons are obtained whose velocity (in volts) is less than the maximum potential \(V'_m\) of equation (5). Hence it must be concluded that there exists some distribution of electrons over energies. The energy-distribution curves obtained at the same saturation current with the same maximum value depend little on the frequency of the incident light. Whereas the majority of authors had previously assumed that the curve of photoelectric photosensitivity intersects in a quite definite way the abscissa axis, on which wavelengths are marked, some other authors expressed the supposition that this intersection occurs at an insufficiently definite angle. Millikan and his students believe that the uncertainty found by a number of authors for pure metals is explained by insufficient spectral purity of the light sources; Zhurman^47 established that this uncertainty is explained by the temperature dependence of the red limit of the photoeffect. The additional thermal energy makes possible the emission of electrons that have received a quantum of energy smaller than that required to overcome the work function. Zhurman substantiates his point of view by the fact that the calculation of the red limit (equation 2), which theoretically gives the limit at absolute zero, always gives values somewhat smaller than those directly observed.

In recent years Fowler^145, using similar assumptions, developed, on the basis of the ideas of Fermi–Dirac–Sommerfeld, a theory of the temperature dependence of the external photoeffect in an electron gas. The experiments of Morris^154, Winch^167, Dobbridge^176, and others experimentally established the existence of this dependence; with the aid of the theory developed by Fowler it proved possible to calculate the true work function, and at the same time the red limit, taking into account the influence of the thermal energy of the electron gas.

For the photoelectric sensitivity \(i\) (in coulombs per quantum of light of frequency \(\nu\)) at temperature \(T\), the theory gives

1) for \(\delta \leqslant 0;\quad h\nu \leqslant e_0\psi\)

\[ i=\operatorname{const}(kT)^2(c-h\nu)^{-\frac12} \left\{e^\delta-\frac{e^{2\delta}}{2^2}+\frac{e^{3\delta}}{3^2}\mp\ldots\right\}; \tag{6} \]

2) for \(\delta \geqslant 0;\quad h\nu \geqslant e_0\psi\)

\[ i=\operatorname{const}(kT)^2(c-h\nu)^{-\frac12} \left\{\frac{\pi^2}{6}+\frac{\delta^2}{2} -\left(e^{-\delta}-\frac{e^{-2\delta}}{2}\pm\ldots\right)\right\} \tag{6a} \]

where, for brevity, we have taken

\[ -\frac{e_0\psi-h\nu}{kT}=\delta . \]

Equation (6) refers to the region immediately adjoining the red limit, i.e. to the part of the photocurrent curve determined by the electrons’ own energy.

Let us transform both equations and at the same time include the expression

\[ (c-h\nu)^{-\frac12} \]

in the constant, which is quite permissible near the red limit; we obtain

1) for \(\delta \leqslant 0;\)

\[ \lg\frac{i}{T^2} =\lg\left\{e^\delta-\frac{e^{2\delta}}{2^2} +\frac{e^{3\delta}}{3^2}\mp\ldots\right\}+\operatorname{const}; \tag{7} \]

2) for \(\delta \geqslant 0;\)

\[ \lg\frac{i}{T^2} =\lg\left\{\frac{\pi^2}{6}+\frac{\delta^2}{2} -\left[e^{-\delta}-e^{-\delta}+\frac{e^{-2\delta}}{2^2}\pm\ldots\right]\right\} +\operatorname{const}, \tag{7a} \]

With the aid of equations (7) and (7a), one can determine \(\psi\) from the photocurrent curve taken at a given temperature. From the equations one calculates the values of \(\lg \dfrac{i}{T^2}\) (to within an additive constant) for any positive and negative values of \(\delta\), and the numbers obtained are plotted as a function of \(\delta\). After this the measured values of \(\lg \dfrac{i}{T^2}\) are plotted as a function of \(\dfrac{h\nu}{kT}\) in the same coordinate system. The latter curve is shifted relative to the former and can be made to coincide with it by horizontal and vertical displacements. The horizontal shift, owing to the choice of scale for the abscissa axis \(\left(\text{first } \dfrac{h\nu}{kT} - \dfrac{e_0\psi}{kT}, \text{ then } \dfrac{h\nu}{kT}\right)\), is equal to \(\dfrac{e_0\psi}{kT}\), whence \(\psi\) is easily determined.

2. Historical survey

As early as 20 years ago it was noticed that the photoelectric sensitivity of a fresh metallic surface in contact with air decreases with time. This phenomenon, called “fatigue,” was investigated in various directions by Galvaksom\(^{1,2}\) and his pupils.

Other investigators, such as, for example, Werner\(^{16}\), Wulf\(^{8}\), Krisler\(^{10}\), and Greinacher\(^{13}\), established that a metallic surface prepared by cathode sputtering of the metal in hydrogen, or subjected to the action of a glow discharge, exhibits a considerably greater emission than under similar treatment in an atmosphere of air or nitrogen.

At the same time, Pohl and Pringsheim observed a change in the spectral distribution of the sensitivity of aluminum and magnesium evaporated and deposited in vacuum; they explained this phenomenon by a chemical change of the surface owing to the liberation of residual gases\(^{12}\). In exactly the same way, the sensitivity of the surface of a calcium amalgam in vacuum changed as a result of renewing the surface by pouring off the amalgam\(^{14}\). Unfortunately, many of the investigations carried out earlier were performed with nonmonochromatic light and under vacuum conditions which can no longer be considered sufficient. Likewise, the gases used by the investigators were prepared in far from irreproachable fashion. Therefore all these experiments cannot be used for the purpose of establishing the role and influence of gases.

With the improvement of vacuum technique as a result of the introduction of diffusion pumps, and also the use of new methods for separating gases and freezing out vapors of stopcock grease and mercury with the aid of solid carbon dioxide and liquid air, the results of investigations became more distinct. Galvaks and Widmann were able to greatly reduce the sensitivity of potassium by repeated distillation in vacuum\(^{17}\); Zende and Simon\(^{28}\) established that the sensitivity of platinum passes through a maximum when it is degassed by heating in vacuum; here, evidently, two opposite processes occur, which may be reduced to the action of ad- and adsorbed-

of the gas: the adsorbed gas causes a decrease in sensitivity, while the absorbed gas causes an increase in sensitivity.

The influence of gases on the sensitivity of a platinum surface was studied by Zuurman^30. He found that gases present on the surface shift the curve of spectral sensitivity toward short wavelengths, whereas the content of gases inside the metal leads to a shift of the curve toward long wavelengths. These results were later confirmed by DuBridge^55 and other authors^37, ^48, ^56, ^60, ^54, ^141, ^179, and DuBridge was able to carry the degassing even further than Zuurman.

At the same time Elster and Geitel^29 investigated the increase in sensitivity to visible light that occurs after the heating of a platinum wire placed in a potassium photoelement is stopped. Since platinum is not sensitive in the region of visible light, it is evidently necessary to conclude that a layer of potassium, invisible to the naked eye and highly dispersed, was adsorbed on the wire, producing the increase in sensitivity. Elster and Geitel indicated that experiments should be carried out on the action of polarized light on a platinum mirror coated with atomic potassium, in order to establish whether a selective vector effect exists.

Such systematic investigations on different metals were first carried out by Ives^40, ^41, ^51, who, however, limited his experiments to the visible region of the spectrum. Zuurman and Teichmann^83, ^101 continued the study of thin potassium layers further into the ultraviolet part of the spectrum, where they determined the photosensitivity in coulombs per calorie. They discovered certain remarkable properties of adsorbed layers. In connection with the discovered properties of adsorbed layers, Zuurman was also able to explain the influence of gases. To explain the regularities obtained, use was made both of theoretical propositions advanced by Schottky^7 and of experimental data on the study of phenomena occurring on a wire heated in vacuum (Langmuir and Kingdon^35, ^34, ^42, ^43; Becker^49). As a result, an affinity was established between the photoeffect and thermionic emission^36, ^69, ^62. Since an adsorbed layer of potassium acquires, under certain conditions, exceptional photosensitivity even in the long-wavelength part of the visible spectrum, thin layers of alkali metals acquire special importance in the technique of manufacturing photoelements. Therefore, in recent years very many investigations of adsorbed films have been carried out precisely from the standpoint of increasing the sensitivity of photoelements; these investigations have supplemented and deepened the study of the external photoeffect. Bainbridge, Zuurman, Campbell, Olpin, Koller, de Boer, Teves, Kluge, and others* took part in these works. A number of investigations whose aim was the study of the external

* The results are set forth for the most part in patent applications; applications concerning especially important photocathodes, in which the potassium layer is deposited on a layer of metal oxide, were made almost simultaneously at the end of 1927 and the beginning of 1928 by Bainbridge, Zuurman, and Campbell.

photoeffect, was carried out by Ives, Zurman, Lukirskii and his collaborators. Thanks to all these works, as well as to the work of Fleischmann and the theoretical work of Fowler and Campbell, the explanation of the phenomena of the spectral and vector selective photoeffect has recently advanced considerably.

3. General physico-chemical properties of adsorbed films

To understand the photoelectric properties of adsorbed films, it is expedient first to dwell on the physico-chemical features of adsorbed atoms, otherwise called “adatoms.”^85 We may confine ourselves to considering the influence of adatoms on metallic surfaces.

If a little cesium is introduced into an electron tube equipped with a tungsten filament, the vapor pressure of which at room temperature is equal to \(10^{-6}\) Hg, then already at \(300^\circ\) C, when a clean tungsten filament certainly does not yet give any noticeable emission, we nevertheless observe a considerable, quite measurable emission. With increasing cathode temperature the electron current grows, reaches a maximum, and at a temperature of \(1000^\circ\) assumes the same value as that given by a clean tungsten filament in vacuum. At a sufficiently high temperature (about \(1500^\circ\)) one can observe that, when the electrodes are switched, a current of positive cesium ions comes from the wire.

Cesium atoms at not very high temperatures are retained on the tungsten surface by adsorption forces and lower the work function of the electrons. The heat of evaporation of cesium adatoms from a clean tungsten surface, according to Taylor and Langmuir,^211 is equal to 65,100 cal (2.83 eV) and is many times greater than that for pure cesium (18 thousand cal). Since cesium atoms at high temperature leave the tungsten surface in the form of positive ions, it follows that the adatoms adhering to the surface undergo a profound change already at lower temperatures; it may be assumed that they are polarized and that their valence electron is drawn into the surface of the metal. The polarized atoms form a double layer with a positive charge directed outward, and therefore lower the work function.^7

The degree of polarization evidently depends on the magnitude of the electron affinity of the metal—the carrier of the adatoms, i.e. on its work function—and on the magnitude of the ionization potential, i.e. on the electron affinity of the adatoms. Potassium and cesium atoms adsorbed on one and the same metal do not lower the work function to the same extent. At equal surface coverage the action of cesium is considerably greater.

Strongly electronegative atoms, such as, for example, oxygen atoms, increase the work function as a result of the formation of an outwardly directed electronegative layer. According to the most recent work of Zurman and Spesh^219 concerning hydrogen on platinum, silver, thallium, and aluminum, the decrease in the work function proceeds linearly with the fall—

...by the work function of the carrier metal. Weak oxidation of a tungsten or silver surface, which strongly increases the work function, causes polarization of the adsorbed alkali atoms that is stronger than on a clean metal surface.

The above-mentioned emission maximum, which occurs when the temperature of the wire is raised, can be explained as follows. At a given cesium vapor pressure, the number of cesium atoms \(n\) falling per second on \(1\ \mathrm{cm}^2\) is constant. The number \(n'\) of atoms evaporating back is a function of the cathode temperature; the higher this temperature, the faster the quantity \(n'\) approaches the value \(n\). At a lower temperature \(n'\), of course, is less than \(n\). If \(n' = n\), and the temperature is so high that every incident cesium atom leaves the surface already in the form of an ion, then the number of ions leaving the cathode surface (when the poles are reversed) can serve as a measure of \(n\).

If \(n' < n\), then the surface concentration increases with time, and the work function \(\psi\) at first decreases, and moreover approximately in proportion to the coverage, so long as the cesium dipoles are still at such large distances from one another that they do not interact with one another. For the value of \(\psi\) what is important is not only a single place on the cathode surface, but a whole series of such places. In the case of a not quite uniform distribution of dipoles, the proposition remains true that the change \(\Delta \psi\) of the work function is proportional to the moment \(pN_0\theta\) of the double layer. Here \(p\) is the magnitude of the dipole, \(N_0\) is the number of dipoles at \(\theta = 1\). With increasing coverage this is no longer true, for the dipoles acting on one another mutually weaken each other. The work function falls less than is required by direct proportionality. Finally, the mutual weakening predominates, and \(\psi\) increases again, especially when the dipoles overlap, after the formation of a monomolecular layer. There exists, therefore, a most favorable coverage \(\theta_0\), at which the work function has, at a given temperature, the smallest value, and the electronic emission the greatest.

With increasing coverage of the metal, the heat of evaporation of the adatoms also decreases. In the experiments of Taylor and Langmuir \({}^{211}\) it is, for \(\theta = 0.67\), \(44\,500\ \mathrm{cal}\) (\(1.93\ \mathrm{V}\)), and for \(\theta \simeq 1\), \(40\,800\ \mathrm{cal}\) (\(1.77\ \mathrm{eV}\)). Therefore the number of evaporating atoms \(n'\) must increase both with time and with increasing \(\theta\), until, when equilibrium is reached, \(n'\) becomes equal to \(n\). The lower the temperature of the incandescent filament, the larger the value of the equilibrium coverage \(\theta_g\). At room temperature it may amount to several atomic layers; with increasing temperature the coverage decreases and finally, at a sufficiently high temperature, reaches zero.

The decrease of the coverage \(\theta_g\) with increasing filament temperature and the dependence of the work function on \(\theta\) make it possible to understand how the emission maximum is attained when the temperature is raised; for, as the temperature is raised, \(\theta_g\) passes through the optimum value \(\theta_0\). In addition to the temperature maximum, at low temperatures ...

is observed; even before the establishment of the equilibrium coverage, there is a temporary maximum of emission at the value \(\theta=\theta_0\). At high temperatures, when \(\theta_g\) lies below \(\theta_0\), instead of a temporary maximum there occurs a certain limiting value of the emission.

To what coverage does \(\theta_0\) correspond? It may be assumed that it is either equal to unity or close to unity; it is precisely then that the work function has its smallest value. Becker’s experiments\(^{49}\) confirmed this supposition. Becker calculated, on the one hand, the number of molecules in a monomolecular layer on the metal; on the other hand, he determined from the number \(n\) of molecules falling per unit time on \(1\ \mathrm{cm}^2\), and from the time elapsed before the attainment of the temporary maximum, the number \(N_0\) of molecules evaporated before the occurrence of this maximum. For \(N\) he obtained the value \(3.8\text{--}4.0\cdot 10^{16}\) atoms per \(1\ \mathrm{cm}^2\), for \(N_0\), \(3.7\cdot 10^{16}\) atoms per \(1\ \mathrm{cm}^2\); hence

\[ \theta_0=\frac{N_0}{N}=0.97\text{--}0.92. \]

In Becker’s method for determining \(\theta_0\) it is assumed in advance that, at the temperature chosen for the measurements, every cesium atom that falls adheres to the surface. If this is not the case at higher temperatures, then the temporary maximum must be reached later, and \(N_0\) will therefore be too large. At lower temperatures the emission is too small to be measured accurately. It would be very worthwhile to repeat Becker’s experiments at very low temperatures, measuring the electron emission by the photoelectric effect.

Another route for determining \(\theta_0\) was used by Taylor and Langmuir\(^{211}\). They introduced into the electron-emission tube two parallel stretched filaments, the distance between which is precisely known. Filament \(A\), whose filling is to be determined, is rapidly heated in a field that removes positive ions, so that all adsorbed atoms leave the filament.

A known fraction falls on filament \(B\), heated to \(1500^\circ\mathrm{C}\); the atoms falling on \(B\) are removed already in the form of ions, which can be measured by a galvanometer. In this method it is assumed that on the surface of filament \(A\) all atoms are distributed completely uniformly and evaporate in all directions in exactly the same way. In this manner the equilibrium filling \(\theta_g\) is first reached at so high a temperature that \(\theta_0\) is not yet reached. When the temperature is lowered, \(\theta_g\) increases until the limiting value \(\theta_m\), occurring at \(325^\circ\mathrm{K}\), is obtained. A further increase of \(\theta\) can occur only when we cool filament \(A\) below the temperature at which the apparatus is kept. As long as the temperature of \(A\) is above the bath temperature, the value \(\theta_m\) is restored again. The authors consider that \(\theta_m\) corresponds to monatomic filling. \(\theta_m\) is equal to \(4.8\cdot 10^{14}\) atoms per \(1\ \mathrm{cm}^2\); calculation of this quantity from the magnitude of the wire surface gives \(3.6\cdot 10^{14}\) atoms per \(1\ \mathrm{cm}^2\). They attribute this discrepancy to the roughness coefficient of the surface of the tungsten filament.

If \(\theta_m\) corresponds to a monoatomic filling of the surface greater by a factor of \(4.8:3.6\) than is calculated from the dimensions,

On the External Photoeffect on Adsorbed Films

of thread diameters, then the optimal emission is attained at \(\theta_0 = 0.67\). It is precisely at this filling that the dipoles begin to interact so strongly that the further addition of adatoms no longer decreases, but increases, the work function. Further, if the temperature of the tungsten filament is only slightly higher than the temperature of the bath, a multilayer deposit of cesium atoms can no longer form on the surface. This may seem unconvincing, since investigations in the field of the photoeffect make the formation of multiatomic layers already at room temperature very probable. In addition, it should be noted that the ratio \(4.8 : 3.6 = 1.33\), which we have for estimating the surface of the filament as compared with the geometrical one, is very close to the value obtained by Becker, Taylor, and Langmuir for the optimal filling of the surface. This value fluctuated, for these authors, between 1.45 and 1.37. Therefore we cannot consider the question of the magnitude of the optimal filling to be unambiguously resolved. In any case it is certain that the optimal filling \(\theta_0\) lies between 0.67 and unity.

We have hitherto used the assumption that a clean metallic surface is completely homogeneous in all its parts with respect to all adsorbed atoms. From this point of view the heat of evaporation of atoms at all points of the surface must be the same and may depend only on the concentration of the latter in the region under consideration. However, all this is already incorrect if we consider a very small region of the surface. Taylor’s and Langmuir’s experiments \(^{211}\) showed that the fraction of active sites on the surface of a tungsten filament amounts to only \(0.5\%\) of its whole surface. On this part of the surface the adatoms are bound to the surface atoms especially strongly, so that the heat of evaporation of cesium atoms at these active sites is \(37\%\) greater than on the remaining parts of the surface. The proportionality between the filling of the surface and the lowering of the work function indicates the circumstance that the atoms sitting on the surface do not interact with one another. Hence it may also be concluded that the active sites are distributed uniformly over the entire surface.

If the surface, apart from the active sites, is regarded as not fully homogeneous, then, obviously, in this region too the value of the heat of evaporation \(l'\) will differ from the mean value \(l\). If \(l' > l\), then in this region there will be more adatoms; if \(l' < l\), then, conversely, there will be fewer adatoms.

Under certain conditions atoms adsorbed on a solid surface exhibit a certain mobility. Thus, for example, the experiments of Becker and Brattain \(^{85,109,195}\) showed that thorium adsorbed at \(T = 1500^\circ\mathrm{K}\) on one side of a tungsten strip is found also on its other side. At a temperature of \(1655^\circ\mathrm{K}\) it begins to evaporate noticeably from it. The diffusion rate of adsorbed atoms depends on the degree of filling. The temperature dependence of the diffusion rate makes it possible to calculate the heat of this process; it proves to be equal

110,000 cal. When coating a filament with barium, it has to be heated to \(900^\circ\) K in order to obtain a uniform coating on both sides of the wire \(^{109}\). The magnitude of the activation energy for the surface diffusion of cesium on tungsten was determined by Langmuir and Taylor \(^{185}\) to be 14,000 cal. The presence of an activation energy for surface diffusion indicates that atoms adsorbed on a crystalline surface do not occupy arbitrary positions, but at a sufficiently high temperature are compelled to execute vibrations between definite adsorption sites.

Atoms forming the second adsorbed layer exhibit still greater mobility. In the experiments of Taylor and Langmuir the condensation coefficient of incident cesium atoms was taken to be unity when \(\theta = 0.98\). Atoms forming the first layer when falling onto a clean surface must move over it until they find a free place for themselves. Since the number of atoms in the second layer is much smaller than the number of atoms in the first adsorbed layer, their mobility at relatively lower temperatures must also be considerably smaller. Thus, on a surface that is nonuniform with respect to the heat of evaporation, there may be both condensation and empty regions; this should be especially apparent if we carry out evaporation of the metal while cooling the adsorbing surface.

4. The external photoelectric effect on simple adsorbed films

a) The general influence of gases on the photoelectric effect; the action of hydrogen and oxygen on the photoelectric effect of metallic films

A number of investigators, such as Kober, Zende, and Simon \(^{18,28}\), showed that the photosensitivity of heated platinum or palladium, obtained under illumination by a quartz mercury lamp and recorded with heating interrupted every 5 sec., at first rapidly increases to a maximum and then decreases to a few percent of this latter value (Fig. 1). This course of the sensitivity was connected by Suhrmann \(^{30}\) with a shift of the long-wavelength limit, first toward longer wavelengths up to 350 m\(\mu\), and then into the short-wavelength region. DuBridge \(^{55}\) repeated Suhrmann’s experiments, using a more perfect photocell, and found that by increasing the degassing it is possible to obtain a shift to 194 m\(\mu\). The photocell used by DuBridge (Fig. 2) had no grease at all; the photocell contains electrically heated platinum foil \(Pt\), the cylinder \(C_2\), serving as the anode, two protective cylinders \(C_1\) and \(C_3\), and a soldered-in quartz window. A piece of metallic magnesium \(Mg\), after degassing, is evaporated and serves as a getter.

The course of the photosensitivity curve shown in Fig. 1 was explained by Galvaks in the following way: as the temperature is raised, the amount of gases adsorbed on the surface and interfering with electron emission decreases; as a result, the photosensitivity increases. With further heating of the filament, a new release of gases from the filament already leads to a decrease in photosensitivity \(^{24}\). Thus, the presence of gases adsorbed on the surface hinders the photoeffect, whereas the content of gases inside the filament is a necessary condition for the photoeffect. The question of which gas is acting remains open.

Fig. 1. Photoemission of platinum foil under illumination by a quartz lamp as a function of the number of heatings in vacuum

To answer this last question, German \(^{45}\) performed the following experiment. He degassed platinum foil and then heated it in hydrogen and in oxygen. Whereas heating in an atmosphere of hydrogen strongly increased the photosensitivity, heating in oxygen produced the opposite effect. Frese \(^{26}\) noted similar effects of evolved hydrogen, on the one hand, and nitric acid and evolved oxygen, on the other. Some authors \(^{38,20}\) reported a positive influence of oxygen on photosensitivity, but in these cases the influence of oxygen must be attributed to special factors. In general, oxygen reduces the photosensitivity even on a very clean surface of mercury \(^{72}\).

Fig. 2. Photoelement used for the investigation of metallic foil.

The positive influence of hydrogen was always observed. Even in the case of silver–palladium alloys of various compositions, which dissolve hydrogen at high temperatures, Kroler and Escher found a significant photoeffect also at low temperatures \(^{33,21}\). Further investigations of this phenomenon for silver–palladium and gold–palladium alloys were carried out by Schneider \(^{192}\). Unfortunately, the spectral photosensitivity curves were not taken \(^{33,21,192}\).

In contrast to Galvaks’s point of view concerning the influence of gases inside the metal, Klumb \(^{76}\) showed that surface coverage by gases of tantalum, tungsten, molybdenum, nickel, and palladium causes an increase in photosensitivity; at the same time,

it was shown that the presence of gases inside the metal is immaterial for the photoeffect. As for hydrogen, Klumb attributes to it a specific action that intensifies the photoeffect. Bomke \(^{130}\) arrived at the same point of view in experiments with a cadmium photoelement.

Fig. 3. Photoelement used for studying the photosensitivity of a metal that is preliminarily heated and bombarded with gas ions

Grenacher \(^{13}\), in his earlier investigations, also points to the effect of coating the metal surface with hydrogen; he observed an increase in photosensitivity to undispersed light for various metals subjected to the action of a glow discharge. Termination of the discharge again reduces the photoeffect, which may be attributed to the increase of the vacuum.

Zhurman’s works \(^{102,103}\) clarified the causes of the action of gases on the photoeffect. Zhurman showed that the increase in photosensitivity is caused by hydrogen atoms adsorbed on the surface.

Fig. 4. Influence of a glow discharge in \(\mathrm{H}_2\) and electron bombardment on the photosensitivity of silver. \(I\)—after the discharge, \(II\)—after 2 hours, \(III\)—after electron bombardment

In Fig. 3 is shown a photoelement provided with a quartz appendage \(Q\) and a quartz window \(F\); the metal plate \(K\) serves as the cathode, while the anode is the silvered inner section \(E\) with the corresponding lead. The joint is lubricated on the outside with sealing compound, previously degassed in vacuum and having a very low vapor pressure. By heating the tungsten filament \(W\) in hydrogen, a certain quantity of ions can be obtained, and, by selecting suitable electric fields, these ions can be deposited on the metal foil; moreover, \(K\) can be bombarded with electrons from \(W\).

A glow discharge in a hydrogen atmosphere very strongly increases the sensitivity of silver and gold; even after several hours of standing in va-

in a vacuum the photosensitivity remains almost constant. Only bombardment with electrons from W reduces the photosensitivity to its initial value (Fig. 4). That the action of the glow discharge amounts to covering the surface with H ions is evident from Fig. 5. If an electric field is applied that hinders the adsorption of hydrogen ions on the metal, the photosensitivity increases only very slightly. If, however, the applied field promotes adsorption of ions on the metal surface, the photosensitivity increases strongly.

Fig. 5. Effect of bombardment with H\(^+\) ions on the photosensitivity of silver. \(I\) — after electron bombardment, \(II\) — after ionization by a field of the opposite direction, \(III\) — after treatment with H\(^+\), \(IV\) — after further treatment with H\(^+\)

From all the foregoing it should be concluded that H atoms reaching the surface form H\(_2\) molecules only after some definite concentration has been attained. This follows from the fact that H\(_2\) does not increase the photosensitivity, but either has no effect at all\({}^{56}\) on the photoeffect or even lowers it\({}^{23, 9, 136}\). Hydrogen atoms, however, forming dipoles, reduce the work function.

The influence of heating the metal can now be reduced to the deposition of diffusing atomic hydrogen on the surface and to the consequent lowering of the work function.

Electron bombardment removes H atoms and reduces the photosensitivity, as can be seen from Figs. 6 and 7\({}^{103}\).

With increasing bombardment the photosensitivity passes through a maximum and then falls to a very small value. Paral-

...corresponding to this there occurs a shift of the long-wavelength boundary toward longer waves and a lowering of the curve toward shorter waves.

Fig. 6. Change in the photosensitivity of platinum (for \(\lambda = 265.5\,m\mu\)) freshly annealed in vacuum by means of electron bombardment.

The initially high photosensitivity can be restored by repeated short-time heating and then completely eliminated by electron bombardment. This can continue until all the hydrogen contained in the metal has been exhausted.*

The maximum of the curve in Fig. 6 must be attributed to H atoms; H atoms may be obtained as a result of electron bombardment of \(H_2\) molecules and then may already cause an increase in photosensitivity.

Langmuir\(^{22}\) assumes the existence of two forms of hydrogen on the surface of a tungsten filament. He is led to this conclusion by data on the mechanism of dissociation and on the thermal coefficient of accommodation of \(H_2\) on tungsten. If the equilibrium between these forms of hydrogen at the temperature of incandescence of the filament favors the form responsible for a decrease in the work of electron emission, then cooling the filament to room temperature may freeze this equilibrium and consequently preserve it also for low temperatures. It may be assumed that the form of hydrogen,

* R. Stuart (Phys. Rev. 45, 488, 1934) observed the formation of adsorbed layers of vapors of organic compounds on electrodes under the influence of electron bombardment in discharge tubes in which the concentration of vapors of organic substances did not exceed \(10^{-5}\) mm Hg. The author supposes that such layers were formed also in all the experiments mentioned above by researchers; however, in his experiments bombardment by hydrogen ions always led to an increase in photosensitivity, which contradicts the consequences that can be drawn from his own supposition. Therefore Stuart’s experiments cannot help to explain the facts set forth above.

Fig. 7. Spectral photosensitivity of platinum. The upper curve—platinum coated with hydrogen after heating for 10 sec. in vacuum at \(1000^\circ\)C. The remaining curves—decrease of the photocurrent owing to diminution of the surface coverage by hydrogen as a result of electron bombardment.

favorable to the photoeffect and adsorbed on the surface is atomic.

The most recent experiments of Baethe^194 show that molecular hydrogen does not affect electron emission. Likewise, diffusion of hydrogen through a heated platinum tube does not affect the photoeffect; the same occurs if the metal is heated in hydrogen previously passed over heated platinum.

On the contrary, the author observed an increase in photosensitivity if a glow discharge was produced in this hydrogen or if the heating of the metal was carried out in hydrogen obtained electrolytically or taken from a bomb; the same occurred if the hydrogen was passed through platinized asbestos or subjected to treatment with liquid air.

Baethe assumes that the last methods of purifying hydrogen are insufficient to free it completely from very small amounts of impurities (water vapor or oxygen), which catalytically accelerate the dissociation of molecular hydrogen into atoms. Unfortunately, these interesting experiments are not beyond reproach, since picein was used in them (for grinding joints and amber insulators), which even at room temperature gives a noticeable vapor pressure.

Fig. 8. Change in the work function \(\Delta\psi\) on the surfaces of different metals covered with atomic hydrogen.

If we bring atomic hydrogen into contact with a metallic surface, we shall obtain the same course of the photosensitivity curve as in the case of the glow discharge. Suhrmann and Scesh carried out such experiments on platinum, silver, aluminum, and thallium^219. The surfaces of these metals were prepared by evaporation in vacuum. Whereas the photosensitivity of platinum and silver increased with increasing wavelength, the photosensitivity of thallium decreased somewhat; the sensitivity of aluminum fell very strongly.

The curve of spectral photosensitivity is well expressed by Fowler’s equation (7) and (7a); therefore it is possible to calculate the work function for a clean surface and for a surface covered with atomic hydrogen, and, consequently, to determine the change in the work function \(\psi\) resulting from adsorption of atomic hydrogen. Fig. 8 shows that \(\Delta\psi\) depends linearly on \(\psi\).

If we assume that \(\Delta\psi\) is a measure of the polarization of the adsorbed hydrogen atoms, then the higher the work function, the stronger the polarization.

According to Bomke^136, insignificant amounts of water vapor on the surface of cadmium intensify the photoeffect. This should be attributed to the atomic hydrogen formed from the water vapor. In addition, it is known that dry argon decreases the photoeffect (Fig. 9), in contrast to moist argon, which intensifies it (Fig. 10, curve 2). However, large amounts of moisture already act by reducing the photoeffect (Fig. 10, curve 3). It is quite possible,

that the first molecules of water, falling on the clean surface of the metal, are catalytically decomposed, and atomic hydrogen is formed here.^5

The facts described compel one to acknowledge that Hallwachs’ point of view on the influence of gases on the photoeffect can be regarded as correct only in part.

If a metallic surface is “cleaned” with alcohol or nitric acid, then such a surface is very little photosensitive;

Fig. 9. Effect of dry argon on the photosensitivity of a clean cadmium surface evaporated in vacuum: curve 1—clean surface; curve 2—after treatment with Ar (pressure 0.03 mm); curve 3—treatment with Ar at a pressure of 1 mm; curve 4—treatment with Ar at a pressure of 4 mm.

Fig. 10. Effect of moist Ar on the photosensitivity of a clean cadmium surface obtained by evaporation in vacuum. Curve 1—clean surface. Curve 2—effect of 0.1 mm of moist Ar. Curve 3—effect of 0.5 mm of moist Ar.

the cause of this phenomenon may be assumed to lie in the adsorption of electronegative particles, such as, for example, oxygen atoms.

The increase in photosensitivity after brief heating in vacuum must be attributed to the adsorption of molecules forming dipoles with their positive poles directed outward; such, for example, are hydrogen atoms. The effect of degassing consists, on the one hand, in removing electronegative molecules from the surface and, on the other, in removing adsorbed hydrogen, which is a source of atoms that lower the work function. Ultimately there remains a clean metallic surface with the usual high work function.

It is now clear that metals must behave differently when heated in vacuum. If, for example, a metal contains little or no atomic hydrogen, but bears on its surface

itself an electronegative adsorbed layer, then degassing will lead to an increase of the photoeffect up to a certain constant value95, 199.

If, however, from the very beginning an electropositive layer predominates on the surface, then the photoeffect will decrease until it reaches a constant final value. The degassing temperature is very important68, since the amount of the form favorable to the photoeffect probably depends on it. At one and the same temperature the optimal ratios of both these forms for different metals must be different.

b) The influence of degassing on the energy distribution of photoelectrons

According to the point of view developed above, the photoelectrons leaving the surface belong to the metal, and not at all to the adsorbed molecules. Adsorbed molecules decrease or increase the work function. If the adsorbed molecules themselves are sources of electrons, then metal surfaces covered with gas should give an exclusively universal velocity distribution, with a sharply expressed preference for the maximum velocity (in volts), calculated from equation (5); this was in fact observed by Lukirskii and Prilezhaev78 in the photoeffect of thin metallic layers.

Indeed, the influence of adsorbed gases on the velocity distribution can be fully established, but only in the sense of their influence on the lowering of the work function.

If, during degassing, electronegative gas molecules are removed, this must be connected with an increase in the number of slow electrons, since the latter will be retained to a greater extent by the double layer, whose negative charge is directed outward. Therefore, in the case of fatigue or, as before, in the case of formation of a double layer, the number of electrons with low velocities should decrease considerably more strongly. This was also shown by Klemperer32.

For the state with the most favorable gas adsorption, when the work function is strongly lowered, one should expect an increase in the fraction of slow electrons to a greater extent than occurs during degassing. The further degassing proceeds, the more the fraction of fast electrons increases. Lang206 verified these conclusions on the photoeffect of a massive platinum layer irradiated with light \(\lambda = 254\ \mathrm{m\mu}\). He studied the velocity distribution before degassing both in the region of the maximum of photosensitivity and beyond it; Lang used a retarding central electric field and obtained the dependence between current strength and voltage.

The platinum layer was obtained by him by cathode sputtering onto a quartz surface. Degassing of the platinum deposit was achieved by heating the quartz plate with radiation from a tungsten filament heated specially for this purpose.

The curves giving the dependence between current intensity and voltage are shown in Fig. 11. It must be remembered that the energy distribution curve is obtained by differentiating the current–voltage curve, and that the slope of the latter is proportional to the number of electrons with the corresponding value of the velocity.

Fig. 11. Current–voltage curve in a central retarding field. The platinum layer was taken at different stages of degassing. Curve 1—at the beginning, curve 2—at maximum photosensitivity, curve 3—beyond the maximum.

From curves 1 and 2 (Fig. 11) one can see that the initial heating makes the potential more positive (in the sense of lowering the work function of the cathode). The maximum potential \(V_m\) remains constant within the limits of experimental error, as is required by equation (5). After a transition through a maximum, the value of the final potential takes on a negative value, so that the work function of the cathode increases. However, in the initial value the work function was not reached in these experiments, for degassing was far from complete.

In Fig. 12 the curves are shown with a correction for the contact potential. The maximum potential \(V_m\) [equation (5)] increases strongly from curve 1 to curve 2 and then (curve 3) falls.

All this can be understood from the point of view of an initial decrease and subsequent increase of the work function of the cathode; here one may also expect a shift of the red limit. The distribution curve is plotted for the initial state of the surface over a smaller energy interval than for the case of considerable photosensitivity. For the region of greater photosensitivity the distribution curve becomes wider. With a decrease in the coverage of the surface by gases the distribution curve becomes narrower.

Fig. 12. The curves of Fig. 11 with correction for the contact potential.

Similar results were obtained by Kluge\(^{60}\), Benevitt\(^{54}\), and Gerold\(^{78}\). They investigated the influence of degassing on the velocity distribution of photoelectrons emitted from platinum, aluminium, and palladium.

c) The action of various gases and vapors on the surface of metals and carbon

The effect on the photoelectric effect of various chemically inactive gases, apart from the hydrogen and oxygen already mentioned by us, has so far been studied very little; this is all the more regrettable because, with the present means for studying this field, one might hope to obtain very interesting data on changes of the potential $\Delta \psi$ at the surface and, in this way, to judge the changes taking place on the surface[^102][^165].

The first data in this field were obtained by Bröer[^175], who determined the influence, at various temperatures and pressures, of adsorbed gases—hydrogen, nitrogen, ammonia, helium, and oxygen—on iron and platinum.

Bröer’s data on spectral photosensitivity, unfortunately, were calculated for only a single intensity; this circumstance does not make it possible directly to calculate the work function and its changes under the influence of gas adsorption. One may disregard errors arising from neglect of impact ionization, since the anode potential reaches only 9 V; however, some complication of the results occurs as a consequence of the partial reflection of photoelectrons from gas molecules and their return to the cathode[^214]. Nevertheless, from the experiments mentioned one can extract many interesting conclusions.

Adsorption of ammonia on iron (1.6 mm) very strongly increases the emission in comparison with the action of nitrogen and hydrogen. With increasing temperature this effect decreases and, at temperatures above $200^\circ$, disappears completely.

The temperature dependence may be attributed to the decrease in the quantity of adsorbed gases with increasing temperature. The long-wavelength limit shifts (at low temperatures) from $257\,\mathrm{m}\mu$ for pure iron to $400\,\mathrm{m}\mu$ when coated with ammonia. The approach of the photocurrent curve to the wavelength axis is very gradual, so that here, evidently, the surface is nonuniform and has separate regions with small work function.

Nitrogen and hydrogen do not change the emission up to a temperature of $600^\circ$. Traces of oxygen very strongly reduce the photoemission—almost to zero. The emission of platinum is increased by ammonia to a considerably smaller degree than was the case for iron. With temperature it likewise decreases, in contrast to the case of hydrogen, where it increases. Nitrogen changes the emission little when the temperature is raised.

In helium the emission decreases, which should be attributed to degassing. The action of oxygen amounts to the following: at room temperature the emission in oxygen does not differ from the emission in nitrogen. When the temperature is raised, oxygen lowers the photosensitivity of a surface previously coated with hydrogen, which should be attributed to the formation of water.

Particularly noteworthy are the data concerning the action of ammonia; since these data can be compared with those of Hallwachs \(^{24}\) and Leopold \(^{61}\), all the facts are in agreement with one another. As for Bröser’s investigations, since they have not yet been completed, it is still premature to compare them.

In this connection one should also mention the old works of Krüger and Teget \(^{19}\), in which the influence of catalytic poisons on the photoeffect of platinum illuminated by a quartz lamp was investigated. The authors found a decrease in photosensitivity under the action of hydrogen sulfide, hydrocyanic acid, and carbon monoxide.

Abendroth \(^{193}\) studied the influence of water, propionic acid, and benzene on the photoeffect of platinum; however, the experiments were carried out with undecomposed light, and, moreover, the anode potential was so high (180 V) that impact ionization phenomena must have had a strong effect. The latter also applies to Kollmer’s work \(^{79}\), which studied the action of toluene on the photoeffect of the surface of mercury. It is even less possible to mention the influence of adsorption on the photoeffect of carbon, which remains obscure to this day. The investigations carried out up to the present time shed no light on this question, for they were made under illumination with undecomposed light. According to Auwer’s \(^{65}\) works, a number of secondary effects occur here, since after prolonged heating the photoeffect changes with time and depends on the application of an accelerating voltage. From all these experiments, however, it is impossible to draw an unambiguous conclusion.

d) Adsorbed metallic films

Elster and Geitel \(^{29}\) believed that the presence on a metallic surface of adsorbed atoms of alkali metals leads, with increasing light intensity, to a kind of saturation; they assumed that alkali-metal atoms account for the enhanced photoeffect of such a surface. One may also expect that, with insufficiently thorough degassing of the substrate, the same selective photoeffect may be observed in the region of visible or ultraviolet light as occurs in the case of alkali metals covered with gases. The first observations made by Ives \(^{40}\) on the photoeffect of adsorbed films of alkali metals when illuminated by spectrally resolved light show that up to \(\lambda = 380\ \mathrm{m\mu}\) there is a normal rise (Figs. 16 and 17), similar to what occurs, for example, for pure platinum and iron. This does not exclude the possibility of a selective photoeffect in the ultraviolet region.

The experiments of Suhrmann and Teissing showed that a platinum surface containing only a small amount of adsorbed alkali-metal atoms does not exhibit a selective photoeffect in the ultraviolet region. The photosensitivity in this region fully corresponds to the emission of the substrate metal, so that at very small coverages (below the optimum) the electrons arise exclusively from the metal of the adsorbate, while the alkali atoms adsorbed on the metal only lower the work function. Suhrmann and Teissing carried out their experiments with potassium adsorbed on platinum black and on pure platinum (Fig. 13).

A glow discharge in hydrogen changes the photosensitivity very little and gives no spectral maximum; at the same time, for the surface of a massive layer of potassium there is a strong selective maximum.

Fig. 13. Spectral photosensitivity of the surface of platinum black before and after coating with atomic potassium. The coating is less than optimal; I—platinum black, II—atomic layer of potassium on platinum black, III—atomic layer of potassium on platinum black (magnified 100 times)

The red limit for platinum black coated with potassium atoms (Fig. 13) is shifted far into the region of visible light. However, as can be seen from the curve, there exists only a small number of special points with an exceptionally small work function; this is indicated by a very extended portion of the curve with very small electron emission.

The influence of the substrate is reflected especially strongly in the shape of the curve in those cases where the substrate metal exhibits optical anomalies in the spectral region under investigation; this applies, for example, to silver. In this case the normal rise is interrupted at 317 mµ by a saddle; at the same wavelength there is a minimum of reflection for the surface of silver; at a wavelength of 317 mµ the light penetrates so deeply into the metal surface that the electrons that have received the energy of the light cannot reach the surface. Figs. 14, according to the data of Suhrmann and Schalamach[^190], and 15, according to Ives and Briggs, illustrate all that has been said above.

Fig. 14. Spectral photosensitivity of silver prepared by evaporation in vacuum; before (I) and after (II) deposition of atomic potassium (the potassium coating is less than optimal)

In Section 3 we established that, with an increase in the surface concentration of adsorbed electropositive dipoles, the work function first falls and, after the attainment of the optimum,

values of \(\theta_0\) begins to increase. The red boundary \(\lambda'\) shifts toward longer wavelengths; when the coverage reaches \(\theta_0\), \(\lambda'\) reaches the value \(\lambda'_0\), and then the shift already begins toward shorter wavelengths.

The position of \(\lambda'\) is a criterion of whether the coverage \(\theta_0\) has been reached or not. If \(\theta < \theta_0\), then with increasing condensation of electropositive atoms \(\lambda'\) increases.

Fig. 15. Spectral photosensitivity of a silver mirror coated with atomic sodium (\(\theta < \theta_0\)); light is incident perpendicularly (after Ives)

Fig. 16. Spectral photosensitivity of a platinum mirror, measured during the condensation of atomic sodium. Curve 1—after 1.5 hours, curve 2—after 3 hours, curve 3—after 19.4 hours, curve 4—after 75 hours. (after Ives)

If \(\theta > \theta_0\), then \(\lambda'\) decreases. Such a shift can be seen in Fig. 16\(^{40}\). Fig. 17 shows the same for evaporation of a cesium layer\(^{40*}\).

The position of \(\lambda'\), corresponding to the optimum surface coverage and, consequently, to the smallest work function, as in the case of coverage with atomic hydrogen, probably also depends on the work function of the substrate metal.

Ives and Olpin\(^{92}\) studied the photoeffect of sodium, potassium, rubidium, and cesium on platinum and silver, of barium on silver, and of lithium on platinum and tungsten; they found that \(\lambda'_0\) in all these systems coincides with the corresponding resonance line of the alkali metals, and they concluded, therefore, that the electrons arise on the alkali metals. However, it must be assumed that the coincidence of the value \(\lambda'_0\) with the resonance line is accidental; indeed, it is known that the work functions of substrate metals differ little from one another

* In addition to the alkaline-earth metals, the same relations as for alkali metals were observed for platinum, silver, and gold. However, Kollier, Jordan, and Woodward in their work with mercury films (Kollier and Woodridge, Phys. Rev. 45, 119, 1934) did not find on cadmium films a transition of the maximum to the long-wavelength boundary; Kollier and Woodridge explain this by the fact that the condensing cadmium accumulates on the surface before atomic layers several atoms thick are formed.

... one another, and, moreover, Campbell^88 observed for potassium and cesium on gold a photoeffect also on the far side of the resonance line (Nottingham^127). Generally speaking, investigations of the photoeffect and of the optimal operating yields for alkali metals deposited on various metals are highly desirable. It is only necessary to take all precautions in order to obtain as clean a metal substrate as possible; for example, it should be obtained by evaporation in vacuum.

Fig. 17. Spectral photosensitivity of a platinum mirror, measured during evaporation of cesium. Curve 1—before heating the platinum; curves 2, 3, and 4—with increasing duration of heating

While platinum coated with potassium gives the normal picture of an increase of photosensitivity at values of $\theta$ smaller than $\theta_0$, Suhrmann and Teissing^101 found in the ultraviolet region a selective maximum (curve III in Fig. 18) at $\theta$ somewhat greater than $\theta_0$; in unpolarized light the maximum has a value equal to $3.4 \cdot 10^{-}$ quanta/cal. The same value was found by Elster and Geitel for

Fig. 18. Spectral photosensitivity of platinum covered with an invisible layer of potassium. Curve I—coverage less than optimal; curve II—coverage nearly optimal; curves III and IV—coverage greater than optimal; curve V—potassium becomes visible

the maximum photosensitivity of potassium covered with hydrogen. With increasing coverage this maximum decreases, and with

filling, which is already beginning to be noticeable to the eye, the maximum shifts to 410 mμ (curve V)—to a value that was observed for compact potassium.

Consideration of curve II, taken on a metal with an almost optimal surface filling and the largest shift of the red boundary, gives no indication of the existence of a spectral maximum of curve III.

A similar maximum in the ultraviolet region was observed for potassium on platinum[^164], sodium on platinum black[^191], and potassium on silver[^190,^151,^209]. In Fig. 19 one can see a maximum

Fig. 19. Photoelectric sensitivity to traces invisible to the eye of a potassium layer on a silver mirror. Curve I—filling less than optimal, curve IV—filling greater than optimal, curves II and III—intermediate filling

of photosensitivity (343 mμ) for a potassium layer on silver, prepared by evaporation in vacuum; the surface filling of the silver here is somewhat above optimal[^190].

We have not yet clarified the question of the magnitude of the optimal filling corresponding to the spectral maximum in the ultraviolet region. According to Becker’s data (section 3), \(\theta_0\) is close to 1, and according to Taylor and Langmuir it is slightly below 1. Since the spectral maximum appears after the optimal filling of the surface has been reached, the polarizing action of the substrate on the adsorbed atoms is by this time already largely screened. The spectral maximum is, generally speaking, associated with selective absorption of light by definite centers. In our case these particles are atoms of the alkali metal, partially polarized by atoms located above the first layer.

We may assume that the optimal filling of the surface with alkali atoms is close to monatomic. Below the optimal

of coverage \(\theta_0\), the absorption of light energy, in all probability, occurs in the metal substrate (the normal photo-sensitivity curve); above \(\theta_0\)—by the screened alkali atoms (selective photoeffect). The emission of electrons for the spectral maximum need not have its source in the alkali metal; atoms of the alkali metals can transfer their energy to the substrate, and the electrons can then fly out of the adsorbate.

Bredig\(^{174}\) attempted to measure the optimum density of the adsorbed layer and the optimum coverages at maximum emission (in the case of a selective maximum).

In Fig. 20 the photocell used by him is shown. The alkali metal is heated in the electric furnace \(K\); the metal atoms enter through the heated opening \(C\) onto \(A\), which is illuminated by a beam of light incident through \(I\). \(H\) serves as the anode. A silver spiral is wound on the tungsten filaments \(F\) and \(G\). By evaporating it, a metallic substrate for \(A\) is obtained; \(M\) is an outlet from the cathode. During evaporation of the metal, liquid air is poured into \(B\). Liquid air is likewise poured into the tube whose end constitutes \(A\). In this way, as far as possible, multiple reflection of alkali atoms in the supply tube is reduced; the diffusion of atoms along \(A\) is likewise reduced.

Fig. 20. Photocell used by Bredig for direct determination of the quantity of adsorbed alkali metals

The experiments were conducted by Bredig as follows: first the furnace \(K\) was heated to a definite temperature and moved along the tube \(D\). This item included the uncertainty of the time needed to heat the alkali metal—the principal errors of the experiments. The thickness of the layer obtained was determined from the vapor pressure of the metal at the given temperature, the time of evaporation, and the dimensions of the apparatus.

The data obtained by Bredig for silver as substrate and potassium are as follows: the optimum coverage corresponds to three atomic layers, and the emission maximum to 12.4 atomic layers. The optimum emission for rubidium occurs at \(\theta_0 = 1.5\), and the maximum emission at \(\theta = 5.0\). The values of \(\theta_0\) are large, especially for potassium; the values determined by methods of thermionic emission (Section 3) are considerably smaller. On the other hand, from the experimental data it may be concluded that, when the maximum emission (selective maximum) is reached, the adsorbed films attain several monoatomic layers.

e) Vector photoeffect on metal films

In the first experiments with films of alkali metals, Ives\(^{40}\) discovered a remarkable phenomenon. If the substrates were platinum, tantalum, nickel, tungsten, and copper, then at a certain thic

layer thickness the phenomenon of the vector effect was observed, discovered for the first time by Elster and Geitel for a liquid alloy of potassium with sodium. When a beam of polarized light falls obliquely, the magnitude of the emission depends on the direction of the electric vector of the light. When the latter is directed parallel to the plane of incidence of the light beam \(E_{\parallel}\), the emission increases by a factor of 10 in comparison with the emission for the perpendicular direction. Further investigation shows that the vector effect of the alkali film is expressed most sharply when the surface coverage is greater than the optimum. If platinum serves as the substrate, then the magnitude of the photocurrents for different directions of the light vector changes by a factor of 30.

Fig. 21. Spectral photosensitivity of a thin layer of potassium deposited on a platinum mirror under illumination by an obliquely incident beam of polarized light. Angle of incidence \(\varphi = 60\text{--}70^\circ\)

At first it may seem that the vector effect, in contrast to what occurs for the sodium–potassium alloy, is not confined to a definite spectral region; however, Zuurman and Teissing\({}^{101}\) showed that for potassium on platinum the vector effect is associated with a selective maximum (Fig. 21).

Fig. 22. Spectral photosensitivity of a platinum mirror on which a layer of atomic potassium has been deposited, with less than optimum coverage. Illumination by a beam of polarized light at an angle of \(60\text{--}70^\circ\)

If the mirror surface of the metal is covered with potassium atoms only to a small extent, then the photosensitivity curve has an entirely normal form, while the vector ratio has a value,

which follows from the optical properties of the surface (Fig. 22). When the coverage is made greater, approaching the optimum, a weak spectral maximum begins to appear at \(E_{\parallel}\)

Fig. 23. Spectral photosensitivity of a platinum mirror coated with potassium, when illuminated by polarized light incident at an angle of \(60\text{–}70^\circ\); the surface coverage is close to the optimum

(Fig. 23); when the surface coverage exceeds the optimum, the maximum becomes very sharp (Fig. 24). The emission here exceeds that which was observed on a potassium surface.

Fig. 24. Same as in Fig. 23. Coverage greater than the optimum

The analogy between the phenomena of the vector effect and the effect studied by Elster and Geitel, and also by Pohl and Pringsheim on a liquid alloy of potassium with sodium, compels one to suppose that the causes of both these effects are the same; namely, that the cause of the vector effect we are studying lies in the inhomogeneity of the upper layer of atoms. The dependence between the vector effect and the composition of an alloy of alkali metals was studied by Ives, Jonesrud, and Stilwell; however, they were unable to establish this connection in a sufficiently definite and clear form \(^{41,59}\).

The study of the vector effect of alkali-metal films on metallic mirrors led Ives, Briggs, and Fry to interesting conclusions[^150][^151][^184][^57][^76][^182]. Thus, Ives believes that, for electron emission from a selective-photo-sensitive surface, the presence of standing light waves arising as a result of incident and reflected light waves is of very great importance; such standing waves were observed by Wiener on a thin metallic mirror covered with a photographic layer. When the thickness of such a layer is of the order of the wavelength, the distance of the particles from the metal that determine the selective effect is equal to several atomic distances. For the magnitude of the photoeffect, what is important is not the intensity in the antinodes of standing waves situated at comparatively large distances from the surface of the metal, but the intensity of the light directly at the surface itself.

Fig. 25. Solid curve: calculated values for the intensity of light directly at the surface of the metal as a function of the angle of incidence. Dashed curve: observed photocurrent for \(E_{\parallel}\) and \(E_{\perp}\), reduced for the potassium layer on platinum.

At the surface of an ideal metallic mirror there are nodes of the standing waves; however, for a metal with finite electrical conductivity this is not the case, since, owing to the phase shift of the reflected waves, their complete cancellation does not occur. Therefore the standing waves at the surface of the metal possess a definite finite intensity, depending both on the angle of incidence and on the wavelength of the incident light; this intensity can be calculated, knowing the optical constants of the metallic mirror, for both directions of the electric vector.

In Fig. 25 are shown the curves observed for a potassium layer on a platinum mirror; on the same figure are plotted the points obtained by calculation[^150].

The intensity of the light, as is seen in Fig. 25, depends on the angle of incidence in the same way as the photocurrent; between \(70\) and \(80^\circ\) the value of the light intensity reaches an especially large magnitude for \(E_{\parallel}\) in comparison with \(E_{\perp}\). Thus we can reduce the vector relation of the photocurrents to the vector relation of the light intensities.

For a complete calculation of the spectral photosensitivity, the optical data mentioned above are still insufficient. It is also necessary to know the constants of the alkali metals. In some cases, however, one can determine the form of the photosensitivity curve in a limited spectral region. In Fig. 26[^184] are given the values of the absorption of light by a silver mirror (a), the absorption of light at the surface (b), and the intensity of light directly at the surface—

tivity (c). Fig. 27, on the other hand, indicates that the maximum of the photosensitivity for perpendicularly incident light coincides with the corresponding maximum in Fig. 26c, and at the same time it is evident that this maximum has nothing in common with curves 26a and 26b. The selective maximum of curve IV in Fig. 19 is close to this as well; it can likewise be explained by means of the intensity curve for perpendicularly incident

Fig. 26. Optical properties of a silver mirror, obtained by calculation for $\lambda = 300—360\,\mathrm{m}\mu$ for linearly polarized light incident at an angle of $60^\circ$, for $E_{\parallel}$ and $E_{\perp}$;

a) total absorptivity,
b) absorption at the surface,
c) intensity of the light directly at the surface

Fig. 27. Photosensitivity of a sodium layer on a silver mirror at $\lambda = 300—360\,\mathrm{m}\mu$; the light is incident perpendicularly; the filling is greater than optimal (according to Ives and Briggs)

light (26c) much better than by means of curve 26a and 26b.

Fig. 28^184 shows that, with the aid of 26c, one can well understand the appearance both of a minimum for $E_{\parallel}$ on a sodium layer deposited on silver, and of a maximum for $E_{\perp}$.

Thus the theory of Ives gives a good account of a number of features of the photoelectric properties of alkali films on metals. However, Ives’s theory still cannot give a complete explanation of the selective maximum. This is recognized rather clearly when comparing the photosensitivity curve of potassium on platinum at a filling greater than optimal (Fig. 24^101) with the curve in Fig. 29^184.

Thus, for $E_{\parallel}$ in unpolarized light the spectral maximum occurs at a wavelength of $350\,\mathrm{m}\mu$; a layer of potassium on silver gives a maximum almost at the same place $(343\,\mathrm{m}\mu)$. The intensity curve же inten-

dependence (Fig. 29c) reveals no maximum between 300 and 360 mµ; there is only a slight rise for \(E_{\perp}\) at longer wavelengths. On the contrary, for the intensity ratio at \(E_{\parallel}\) and \(E_{\perp}\) there is agreement with the vector ratios of the photocurrents, which cannot be associated with the light-absorption curves in Figs. 29a and b.

Fleischman\(^{142}\), in his measurements of light absorption carried out on thin potassium layers deposited on quartz, showed that the optical properties of the metallic mirror of the substrate and of the alkali metal are still insufficient for a complete explanation of the spectral maximum. Fleischman found that

Fig. 29. Optical properties of a platinum mirror for wavelengths \(\lambda\) between 300 and 360 mµ under a normally incident \((\perp)\) beam and under an oblique (60°) beam of polarized light for \(E_{\parallel}\) and \(E_{\perp}\):

a) total absorption of light,
b) absorption of light at the surface,
c) intensity of light directly at the surface (after Ives and Briggs)

Fig. 28. Same as in Fig. 27 for an obliquely incident (60°) linearly polarized light beam; coverage greater than optimal (after Ives and Briggs)

thin layers of alkali-metal atoms, barely perceptible under perpendicular viewing, appear distinctly colored when viewed in an oblique direction. If a beam of polarized light falls obliquely, the coloration appears rather strongly and, moreover, is dichroic; in other words, the color is seen only when the electric vector oscillates parallel to the plane of incidence.

Fig. 30 gives the curve of the dependence of light absorption; the course of the curve is analogous to the curve of the selective photoeffect (for example, Fig. 21). The curve in Fig. 30 was taken for a potassium layer on a quartz plate, whereas Fig. 21 shows the course of the photoeffect of potassium on platinum; therefore the small difference in the positions of the maxima is quite natural.

Potassium atoms covering the surface are completely not ad-

are adsorbed by the perpendicular component of the electric vector. The view that optical properties are the most essential for explaining the selective photoeffect is refuted not only by Fleischmann’s investigations, but also by the work of Zuurman and Teising on the photoeffect for sodium on platinum black (maximum at 317 mµ^191).

Still, it must be said that the spectral maximum must be attributed primarily to the selective absorption of light by metal atoms not directly adjacent to the metallic substrate*; this absorption of light determines only the position and form of the maximum.

When the coverage does not reach the optimum, the optical properties completely determine the course of the photosensitivity curve; especially important is the magnitude of the absorption of light at the surface of the metal, as is seen from comparing Figs. 14 and 16 with Fig. 26b.

Fig. 30. Absorption of light in a thin layer of potassium for \(E_{\parallel}\) at an angle of incidence of \(55^\circ\) (after Fleischmann)

For the photocurrent obtained from the selective layer at \(E_{\perp}\), the optical properties retain their significance, despite the fact that the film of the alkali metal does not absorb this vector; according to Ives, the photosensitivity curves for layers on various substrates are arranged in the order of reflectivity, the electron emission for \(E_{\perp}\) being the greater, the greater the reflectivity of the given metal. Poor reflectivity leads to such deep penetration of light that the emission of electrons becomes very difficult. Conversely, good reflectivity makes possible a significant emission of electrons from the outer layer of the metal.

The centers of light absorption for the selective effect (at coverage greater than the optimum) for \(E_{\parallel}\) are located in the film of the alkali metal. For \(E_{\perp}\), just as for both vectors in the normal photoeffect, the emission centers are located in the upper layer of the metal substrate, and, consequently, the photoelectrons in both these cases are emitted by the substrate. If the electrons were emitted by the layer of adsorbed alkali metal, we would obtain very steep curves of the distribution of electrons by energy. As Lukirskii and Prilezhaev^78 showed, the distribution curve—

* Hluchá (Z. Physik 81, 66, 516, 521, 1933) observed a (0.5 µ) spectral (selective) maximum on thin layers of aniline dyes; the position of this maximum almost coincided with the maximum of the absorptive capacity of the dyes corresponding to their intrinsic oscillations.

determination of the energy, taken on a thin layer of silver, proceeds considerably more steeply than the curve obtained on massive silver.

The first portion of electrons emitted by a selective layer of alkali metal deposited on a metallic mirror must be the same both for \(E_{\parallel}\) and for \(E_{\perp}\). Ives, Olpin, and Djonsrud \(^{75}\) observed almost coincident distribution curves for a liquid alloy of potassium with sodium at a vector ratio of \(12:1\) (\(\lambda = 436\ m\mu\)); however, this seems very difficult to understand.

Fig. 31. Apparatus used for determining the energy distribution of photoelectrons emitted by a liquid alloy of potassium with sodium (for \(E_{\parallel}\) and \(E_{\perp}\), according to Ives et al.)

Figure 31 shows the arrangement of the apparatus used to investigate the distribution of electrons by energies for liquid alloys of alkali metals. The metal sphere \(A\) serves as the anode. The liquid alloy of potassium with sodium is in the glass vessel \(C\), coated on the outside with potassium. The incoming alloy in \(C\) is produced from \(B\) through the connecting tube \(D\). Light from \(G\), through the aperture \(H\), falls on \(C\) and is reflected through \(I\) and \(J\), without striking the anode. The electrodes \(E\), placed near \(C\), serve to measure the angular distribution.

Figure 32 gives the curve of the dependence of current strength on voltage for the saturation current at \(E_{\parallel}\) and \(E_{\perp}\). The two curves coincide; this means that the energy distribution for both vectors is the same. The maximum velocities are also the same, which is in full agreement with the experiments of Wolf \(^{63}\) and Teichmann \(^{104}\) on the photosensitivity of selective potassium surfaces.

For alkali layers deposited on platinum, the same results were obtained as for alloys of potassium with sodium. The assumption that electrons are emitted by the surface of the substrate metal agrees well not only with the facts relating to the distrib-

division of the electron velocities for \(E_{\parallel}\) and \(E_{\perp}\), but also with the angular distribution of the electrons. A verification of the latter was carried out by Ives, Olpin, and Johnsurd\(^{75}\).

To study the angular distribution of photoelectrons one may use the photocell shown in Fig. 31. One may use a device in which, instead of a cup, a platinum foil with a layer of alkali metal deposited on it is placed in the spherical anode.

Fig. 33 illustrates the angular distribution of electrons. The surface of a liquid alloy of potassium and sodium was taken as the surface; according to Lambert’s law the distribution curve should be a circle tangent to the surface, which is fully confirmed (Fig. 33). It may therefore be assumed that electrons emerging from an element of volume undergo scattering on their way, so that the number of electrons passing through an element of surface in any direction is proportional to the projection of this element onto a plane perpendicular to that direction.

Fig. 32. Curves for a liquid alloy of potassium with sodium for \(E_{\parallel}\) and \(E_{\perp}\); \(\lambda = 436\ \mathrm{m}\mu\); angle of incidence equal to \(60^\circ\) (after Ives et al.)

Even at oblique incidence this law is satisfied, as is seen from Fig. 34; the vector \(E_{\parallel}\) gives no preferential direction of emission in comparison with \(E_{\perp}\). For both vectors the direction of emission does not depend on the angle of incidence—it is normal to the surface.

Consideration of the results of experiments at a field equal to zero (with allowance for the contact potential) shows that Lambert’s law, in the case of oblique incidence, is obeyed better for \(E_{\perp}\) than for \(E_{\parallel}\). Fig. 34 further shows that the curve for \(E_{\parallel}\) in the direction of the normal extends farther than the curve for perpendicular incidence \(E_{\perp}\); the latter, in turn, extends farther than the curve for \(E_{\parallel}\); these deviations from Lambert’s law increase from \(E_{\perp}\) to \(E_{\parallel}\) and \(E_{\parallel}\). All that has been set forth above also occurs at very high vector ratios, for example at a ratio of the photocurrents for \(E_{\parallel}\) and \(E_{\perp}\) reaching 23.

The applicability of Lambert’s law, on the one hand, and the independence of the direction of emission from the angle of incidence of the light and from its polarization, on the other, can best be explained by the fact that, for the selective absorption of each vector, there exist special centers of its own, whereas for the electronic emission in both cases one and the same surface of the metal substrate is operative. Therefore only the absorption of light depends on the polarization and on the angle of incidence of the light beam; the direction of emission and its total distribution are not influenced in any way by either of these latter factors.

Fig. 33. Angular distribution of electrons; potassium alloy with sodium; the light falls perpendicularly; illumination by a mercury lamp (after Ives et al.)

Fig. 34. Angular distribution of electrons; potassium alloy with sodium. The angle of incidence of the light beam is 90 and 60°; \(\lambda = 436\); the light is polarized. The effective cathode potential is 0.1 V (after Ives et al.)

If we assume that the electrons are scattered on their way to the surface, then the above-mentioned deformation of the distribution curves can also be explained. The deeper the emitting region lies, the greater the scattering, and conversely.

We know that the energy of the vector \(E_{\perp}\) is absorbed by the metal substrate and that, consequently, the emitting centers lie relatively deep; therefore in this case one must expect scattering here, affecting the fulfillment of Lambert’s law. The energy of the vector \(E_{\parallel}\), on the contrary, is absorbed by the atoms of the alkali layer, which transmit this energy to neighboring electrons of the carrier metal. Therefore, for \(E_{\parallel}\), the electrons of the emitting centers lie closer to the surface, and the scattering of electrons is much weaker.

5. External Photoelectric Effect on Complex Films

In the preceding section we established that the photosensitivity of a pure metallic surface covered with an adsorbed layer, when the coverage exceeds the optimum, passes through a maximum. In this chapter we shall study cases in which the layer of atoms absorbing light does not lie directly on the metallic substrate, but is separated from it by one more layer of atoms, which weakens the action of the polarizing forces of the metal substrate. The intermediate layer of atoms must in this case bind the layer of alkali atoms located on it either by adsorption forces or by valence forces, i.e. purely chemically. Since the intermediate layer is penetrated by the emitted electrons, its electrical conductivity must be sufficiently high.

In the case of poor electrical conductivity it is necessary to make this intermediate layer of very small thickness. On complex surfaces a spectral selective maximum is observed, in which, to a certain extent, the absorption bands of the alkali metal are reflected.

The ideas and schemes developed in the preceding chapters cannot always be applied completely and without reservations to complex layers. Thus, for example, in interpreting the experiments of Elster and Geitel, who were the first to investigate the influence of a glow discharge in hydrogen on a potassium surface, one has to assume that not only is potassium adsorbed on potassium hydride, but that potassium particles are also present in the hydride layer.

For sufficiently small regions the old scheme can also be applied. Gudden and Pohl52, 44 were the first to express the idea that the study of the selective maximum gives us a method of measuring the optical absorption spectrum by electrical means. The correctness of the above-stated view of the structure of a selective surface was confirmed by a number of investigators, such as: Suhrmann161, 162, 163, Lukirsky and Ryzhanov159, 186, by the systematic preparation of various complex surfaces on massive layers of alkali metals. Investigations of the same kind on other substrates (chiefly on silver) were carried out by Kollmer121, Suhrmann162, 163, de Boer and Teves108, 177, 178, and Kluge203, 204, 205, 202.

In order to show that there is a real connection between the presence on the surface of a layer of alkali metal separated from the substrate by an intermediate atomic layer and the spectral maxima in the long-wave ultraviolet for Na and in the visible region for K, Rb, and Cs, Suhrmann162 deposited on the surface of freshly evaporated potassium a certain very small quantity of naphthalene, which reacts only weakly with potassium. Fig. 35 clearly shows that the photosensitivity in the visible region falls sharply. However, even traces of potassium vapor brought into contact with the potassium-naphthalene surface lead to the formation of an intense spectral maximum at 420 mμ.

In the same way one can obtain a spectral maximum for a hydrided potassium surface^162; the intermediate layer here is potassium hydride. The position of the maximum is almost the same as in the preceding example. The hydride layer is obtained by the action of atomic hydrogen*. Since much heat is liberated during the formation of the hydride, at the very beginning of hydriding a certain amount of potassium evaporates; then, condensing, the potassium covers the hydride layer.

Fig. 35. Spectral distribution of photosensitivity for a potassium surface prepared by evaporation in vacuum (I); (II)—onto a surface coated with a layer of naphthalene, (III)—after further condensation of potassium atoms

It is not improbable that some potassium atoms acquire greater mobility as a result of the liberation of the heat of hydride formation, move over the surface, and in this process become situated on the hydride layer. The spectral maximum is observed immediately after the action of atomic hydrogen. Prolonged action of atomic hydrogen leads to the disappearance of atomic potassium and to a decrease of the photosensitivity to a certain final value. However, the spectral maximum appears again as soon as traces of potassium vapor are introduced.

Fig. 36. Relative photosensitivity \(\left(\dfrac{i}{i_0}\right)\) of the potassium surface under interaction with atomic hydrogen \((N_{\mathrm{H}}\)—the absorbed amount of hydrogen) as a function of the duration of exposure to hydrogen. The experiments were carried out at room temperature. \(T\)—temperature of the tungsten filament serving to obtain atomic hydrogen was \(1320^\circ\mathrm{K}\), \(p = 0.09\ \mu\mathrm{m}\), \(S = 115.0\ \mathrm{cm}^2\) (after Ryzhanov).

Lukirskii and Ryzhanov^159,186 repeated these experiments, working with undecomposed light but determining the amount of hydrogen bound with potassium. The results obtained by them are illustrated in Fig. 36. The amount of absorbed hydrogen is 30–40 times greater than the amount required for a monomolecular layer; one may assume that at the very beginning of the reaction the potassium surface is loosened, and that KH covers this surface with layers lying one upon another

* Molecular hydrogen has absolutely no effect on the photosensitivity of a potassium surface^83. The data of Fleischer and Teichmann^116, which contradict this, should be attributed to hydrogen impurities.

atomic layers; one may suppose that within these layers there is also free, chemically unbound hydrogen. The latter is confirmed by the fact that partial evolution of hydrogen begins already at \(80^\circ\text{C}\), ending at \(250^\circ\text{C}\)—at the temperature of decomposition of KH. If hydriding is carried out at the temperature of liquid air (Fig. 37), then less hydrogen is absorbed, and the photosensitivity remains low (curve \(B\)). At the temperature of liquid air the thermal conductivity is so great that the heat is already insufficient for potassium evaporation and, consequently, a potassium layer is not formed on the hydride. When the temperature is raised to room temperature, the diffusion rate of the adsorbed, chemically unbound hydrogen increases, and some of the potassium atoms react with hydrogen. In this case the photosensitivity increases. If the hydrogen is made to react with potassium again at low temperature, then hydride is formed, but the potassium no longer evaporates, and the photosensitivity decreases.

Obtaining more precise data on the distribution of alkali-metal atoms in the intermediate layer is rather difficult. From the coloration of hydrided layers of photocells, Ostwald\(^{132}\) concluded that the alkali metal is in a highly dispersed dissolved or adsorbed state*.

Fig. 37. Relative photosensitivity \((BD)\) as a function of the time of action of hydrogen \((N_{\mathrm{H}}\), curve \(AC)\) at the temperature of liquid air (after Lukirskii and Ryzhanov)

Oxygen acts on the surface of a massive alkali metal similarly to hydrogen in the atomic state. According to Pol and Pringsheim\(^{15}\), quite minute traces of oxygen are sufficient for a spectral maximum at \(400\,\mathrm{m}\mu\) to appear for the photoeffect on a potassium surface.

Koller\(^{121}\) investigated the influence of oxygen on cesium when undecomposed light acted upon it. In this case the first portions of oxygen had almost no effect on the photosensitivity; only when the amount of absorbed oxygen reached one third of the amount needed for complete oxidation of cesium to CsO did the photosensitivity begin to rise.

It should be assumed that the first portions of oxygen, or of the oxide formed, dissolve in the alkali metal and therefore leave the surface unchanged. With further admission of oxygen an intermediate layer is formed, and the photosensitivity increases.

Further oxidation leads to complete transformation of cesium into oxide and to a fall in photosensitivity. Lowering the temperature of the layer to the temperature of liquid air strongly reduces the amount of oxygen required to attain the maximum photosensitivity; the reason for this, presumably, is the decrease in diffusion over the surface.

* Kluge and Rupp\(^{153}\) electronographically examined a hydrided potassium surface by means of slow electrons and concluded, from the form of the maxima for KH and K, that the particles located on the surface are very small.

In addition to oxygen, one can name a number of substances that cause the appearance of a spectral maximum. Kluge \(^{152}\) found a selective maximum for a potassium surface that was in contact with vapors of sulfur, tellurium, or selenium; Olpin investigated the action of a large number of organic and inorganic substances on massive layers of alkali metals; in all cases an increase in photosensitivity was obtained (for visible light)*.

What properties must the intermediate layer possess in order to be capable of producing a selective maximum? Campbell \(^{111}\) believes that the intermediate layer should consist of an electronegative substance; as an example he cites photocathodes obtained by depositing cesium on slightly oxidized silver.

Fig. 38. Photosensitivity of a potassium layer on slightly oxidized silver. Curve \(I\)—nearly optimal filling, curve \(II\)—filling greater than optimal, curve \(III\)—compact potassium in the form of a matte layer on silver.

From his observations, Zurman concludes that the intermediate layer must possess the ability to chemically bind the alkali metal or at least adsorb it. Thus \(^{161, 162, 163}\) paraffin freed from unsaturated compounds and fatty acids, when used as an intermediate layer, gives no selective maximum; the reason for this is that it does not adsorb potassium metal on its surface. Evidently, a thin distribution of the metal on the surface can be retained for a long time only in the presence of adsorption forces. This relation also holds when the intermediate layer is not an alkali metal but some other metal.

In technical photocathodes this metal is usually silver. The intermediate layer is prepared by treating silver with a glow discharge in oxygen at a pressure of the order of tenths of a millimeter. Then the alkali metal is deposited on the oxide layer obtained. As Fig. 38 shows, such a layer first gives a maximum at a wavelength of \(350\ \text{m}\mu\), which is explained in the same way as the maximum (\(343\ \text{m}\mu\)) for potassium deposited on pure silver—

* The action of various gases and vapors on different metals leads to the formation of layers exhibiting a spectral maximum, usually located in the invisible region; thus, for example, Pohl and Pringsheim observed a spectral maximum on Al and Mg \(^{12, 196, 112}\).

…surface. With stronger coverage the maximum shifts toward shorter waves, to 400 mμ; at the same time it becomes broader and more elongated.

The position and height of the maximum depend strongly on the method of preparation. Thus, for example, Koller[^121] showed that for a Cs—O—Ag cathode the maximum is observed only in the case where the layer of cesium deposited on the silver oxide was heated during its preparation for some time at 250°.

The investigations of de Boer and Teves[^108][^177][^178] confirm the idea of the role of adsorption forces in maintaining that dispersion of the alkali metal which is necessary for the spectral maximum. The data on the absorption of light in these layers lead the authors to the view that the salt on which the evaporated cesium is deposited induces dipoles in the latter and adsorbs it. In addition, these complex layers give a spectral maximum in the region of visible light; with increasing coverage the maximum shifts into the short-wave region. A decrease in coverage (evaporation of excess atoms) again leads to the former maximum.

Fig. 39. Distribution of potential on a selectively photosensitive surface when the intermediate layer consists of an electronegative substance.

What is the mechanism of this spectral maximum?

Campbell and Fowler[^111][^117] assume that, for selective surfaces, the transmission coefficient* for electrons whose velocities lie within a definite range is very large. The transmission coefficient is a measure of the probability of the passage of electrons with a given energy through the surface to the outside; this is fulfilled when the potential at the surface (Fig. 39) reaches the corresponding value.

This takes place if an electronegative substance serves as the intermediate layer. On the basis of this assumption Olpin[^157] calculated the positions of selective maxima for various surfaces; in the calculation he assumed that the intermediate layer has a crystalline structure. The potential relief shown in Fig. 39, consisting of alternating potential barriers and wells, naturally follows from the crystalline structure itself of the intermediate layer. However, calculation of the maxima for hydrides gave data in poor agreement with the most recent X-ray determination of the lattice structure (Zachariasen[^171]). Therefore these calculations cannot serve as confirmation of the Campbell–Fowler theory. De Boer and Teves evidently modified the Campbell and Fowler theory; they consider that in the electron emission of photolayers of complex composition the primary proces—

is the ionization of adsorbed alkali atoms. Electrons emitted from the metallic substrate are selectively transmitted through the intermediate layer and, as a result, a spectral maximum is obtained.

It seems to us that the selective photoeffect of a layer of complex composition can be explained in the same way as in the case of layers of simple composition, namely, as the result of selective absorption of light and the emission of electrons by centers absorbing the light energy. The differing electrical conductivity of the intermediate layers leads to the selective effect.

This conception had already been formulated earlier by Wolff[^105], though in a narrower sense. He believed that, in the photoeffect, conduction electrons are emitted, to which, as a result of collisions of the second kind, the energy of optically excited atoms is transferred.

Fig. 40. Experiments of Ives and Fry on the photoeffect of a cesium layer on a quartz wedge. The dotted curves are the intensity of light on the layer (calculated); the other curves are the magnitudes of the photocurrent obtained from experiment. Along the abscissa axis are plotted the positions of the light spots along the length of the plate; \(\lambda = 546.1\).

The study of the energy distribution of photoelectrons makes it possible to determine whether the role of the alkali metal is limited to the absorption of light, or whether the latter itself also participates in the electron emission. In the case of photoionization, the electrons must arise from layers of extremely small thickness; the energy-distribution curve is very steep, and the maximum must correspond to the maximum energy.

The experiments of Ives and Fry[^201], who studied standing waves on photocathodes of complex composition, show that the role of the selective layer consists not in selective transmission of electrons, but in the selective absorption of the incident light. The authors obtained radiation reflected by a quartz wedge on a platinum mirror; on the quartz wedge, whose thickness was of the order of the wavelength, a layer of cesium was deposited. When such a photocathode is illuminated with spectrally resolved polarized light, the photocurrent gives a series of minima or maxima depending on whether nodes or antinodes of standing waves are formed on the surface of the cesium (Fig. 40). This experiment is an analogue of Wiener’s well-known experiment on a photographic plate.

Especially noteworthy is the circumstance that, at different places of the quartz wedge, Ives and Fry obtained different curves for the photocurrent; the latter indicates that, for different light vectors, the spectral maxima lie in different regions of the spectrum.

Thus it turns out that the form of the photosensitivity curve depends strongly on the thickness of the intermediate layer deposited on the metal. In one case it proved possible to explain the form of the photosensitivity curve for \(E_{\parallel}\) and \(E_{\perp}\); owing to the dependence of the photocurrent on wavelength it was also possible to study the intensity of light at different places in the cesium film. Ives and Fry generalized this case and explained, for example, the purely optical spectral selectivity of a potassium layer on an intermediate naphthalene film. We cannot take this extreme point of view, because of the excellent reproducibility of most selective maxima occurring at a definite, specified thickness of the intermediate layer. Nevertheless, the optical properties of the metallic substrate and of the intermediate layer often influence the position of the maximum, and in other cases cause insufficient reproducibility.

The passage of electrons through the intermediate layer depends on its conductivity (dark or photoelectric).

Since the material usually chosen for the intermediate layer is one with good insulating ability, in contrast to pure metals the electrical conductivity of such a material is increased when foreign atoms are introduced into it[^3]; it is possible that this partly explains the increase in emission obtained by Zeleny[^99], Campbell[^139], Asao[^172], Zworykin[^189], de Boer and Teves[^178][^197], Gerlich[^200], Fleischer and Gerlich, and brought by de Boer and Teves to a quantum yield of 200% at the spectral maximum. The increase in conductivity at the same time gives better saturation[^197].

The introduction of atoms of substances with a different crystal lattice is most effective, since such atoms disturb the lattice of the intermediate layer most strongly. De Boer and Teves[^178] observed, on an intermediate layer consisting of cesium oxide, a number of remarkable secondary phenomena. They established that part of the light increases the conductivity of the oxide as a result of an internal photoelectric effect. This phenomenon is of special importance for cathodes with thick oxide layers, for which, even with introduced foreign atoms, the current–voltage curves reach saturation only with difficulty. If the current is drawn from such a cathode,* the photosensitivity also falls, and the more strongly the greater the photocurrent; at low intensities this decrease in sensitivity is thus less than at high ones (Fig. 41).

From Fig. 42 it is seen that increasing the voltage increases the fatigue effect. De Boer and Teves explained this by the fact that cesium ions formed as a result of photoionization at the surface are drawn inward by the action of the electric field; as a result the surface becomes poorer in cesium ions. Strangely, the influence of voltage at \(0^\circ\)C is greater than at room temperature. At the temperature of liquid nitrogen the fatigue is still considerable, while the influence of voltage, however, is small; here, probably, the phenomenon is involved which we shall discuss in Section 6.

The dependence of the fatigue effect on voltage gives us grounds to reduce it to the residual positive charges remaining in the intermediate layer after the emission of electrons as a result of the internal photoeffect. Therefore the work function for an electron in the fatigued layer is greater than in the initial one; the photosensitivity in this case falls in the region of long waves (Fig. 43).

Fig. 41. Dependence of fatigue, with decreasing current, on the intensity of light for a photocell of cesium on cesium oxide (illumination with white light)

Fig. 42. Dependence of fatigue on voltage at different temperatures

Fig. 43. Spectral photosensitivity of a cesium cathode with an intermediate layer of cesium oxide: B) after fatigue, C) after rest; course of the curve at the red boundary

The greater the intensity of light, the more electrons are liberated and the more of them are pulled outward by the electric field. With an increase in voltage, the electrons emitted as a result of the internal photoeffect enter the surface layer more quickly; the positive charge in the layer they have left increases and, consequently, the force opposing their emission grows. The dependence on temperature can evidently be explained from the point of view of the slowing down or acceleration of the neutralization of residual positive charges when the temperature is raised or lowered. Similar to an increase in temperature is irradiation with red light, which accelerates the equalization of charges in the layer mentioned.

The investigations of de Boer and Teves on the dependence between the degree of fatigue and the wavelength are illustrated by Fig. 44, where it is seen that the smaller the wavelength, the greater the influence of illumination during the passage of current.

Complex photocathodes usually give two or even more spectral maxima, the positions of which often are not especially

well determined. Thus, for example, the shape of the maximum in the ultraviolet region, obtained by Colwell[^121] for cesium on silver oxide, depends on the absorption of light in the glass window of the photocell. This same circumstance evidently affected the results of Kluge[^203], who obtained, for various layers of alkali elements on silver oxide, almost one and the same position of the maximum, namely 400 mµ. The second maximum, established by Kluge, is more reliable and is reproduced rather well; for a cesium layer the maximum lies between 730 and 850 mµ, for rubidium—between 620 and 680 mµ, for potassium—between 460 and 520 mµ. The maximum for the photoeffect on a sodium layer probably lies at the boundary of visible light. If, instead of a mirror-polished silver backing, we use a rough silver surface, then the maximum for cesium will already lie at 860 mµ[^205].

Fig. 44. Dependence of the fatigue effect on wavelength. Change of the photocurrent with time under irradiation by 1) ultraviolet light, 2) red light, 3) green light, 4) blue light

The maximum in the long-wavelength region is reproduced well only when the intermediate layer is prepared each time in one and the same way. Thus, for example, Zurgman found a maximum for a strongly oxidized silver layer at 460 mµ, whereas in the case of a weakly oxidized layer the maximum was already at 400 mµ (Fig. 38). If the silver-oxide layer is coated with potassium, a short-wavelength maximum at 350 mµ is observed; Young and Pierce[^170] observed a similar short-wavelength maximum, so that on the photosensitivity curve there proved to be two maxima (Fig. 45)—a long-wavelength one at 700 mµ and a short-wavelength one at 370 mµ. But, besides this, as is clear from the figure, there is also a third maximum at 290 mµ. According to Zurgman, the short-wavelength maxima are due to the appearance of optical absorption by inclusions of the metallic backing, which are distributed in the cesium oxide in the finest form (or as atoms). The maximum at 350–370 mµ is fairly constant, and we could partly attribute it to the optical properties of the metallic backing. The long-wavelength maximum in the visible part of the spectrum is the result of the presence of a layer of alkali metal deposited on the intermediate layer[^162].

Fig. 45. Photosensitivity curve of a photocell with a cesium layer on silver oxide (after Young and Pierce).

Kluge observed a third maximum at 290 mµ, which perhaps may be attributed to the interaction of oxygen atoms contained in the intermediate layer. In favor of the latter supposition are the observations of Fleischer and Gerlach[^215], who used a photocell with a quartz-

one whose cathode consisted of a layer of silver oxide and cesium, in which silver atoms were present. It turned out that, in addition to the long-wave maximum between 720 and 850 mµ and the second maximum at 375 mµ, there is also a third maximum at 280 mµ.

Zurman and Dempster ^216, working with a potassium photoelement that had an intermediate naphthalene layer, found, alongside the long-wave maximum at 420–480 mµ, a second spectral maximum at 290 mµ. Whereas the cause of the appearance of the long-wave maximum is potassium located ^161 on the intermediate layer, the short-wave maximum may be attributed to atoms that have diffused into the intermediate layer.

Fig. 46. Change in the sensitivity of a compound layer potassium—naphthalene—potassium.

A freshly prepared surface gives both maxima, with the short-wave one located lower (Fig. 46, curve 15/XII 1933). With time, the long-wave maximum decreases, while the short-wave one becomes higher (curves 13/II 1934 and 27/III 1934).

In presenting the state of the question concerning surfaces of complex composition, we have used chiefly those works which could be useful for studying selectively emitting surfaces or for understanding the mechanism of emission itself.

It is necessary, however, to mention that a number of investigators, such as, for example, Ives ^91, Campbell ^67, Zeleny ^99, ^135, Case ^112, Zworykin ^106, ^107, and others ^113, ^114, ^122, ^126, ^140, ^166, ^188, ^179, ^198, ^207, dealt chiefly with questions of the practical fabrication of highly sensitive photocathodes; however, because of lack of space we cannot dwell in detail on the results of their work.

6. INFLUENCE OF TEMPERATURE ON THE PHOTOEFFECT OF ADSORBED LAYERS

If the composition of the adsorbed layer does not depend on temperature (there is no condensation or evaporation), then the temperature dependence of the photoeffect is determined by the following factors:

1. Additional thermal energy of the electrons.
2. Change in the optical properties of the surface of the metal substrate.
3. Change in the lattice constants, which may cause a change in the work function of the metal substrate.*

* K. F. Herzfeld, Phys. Rev., 35, 248, 1930.

4. Influence of adsorbed foreign atoms.

Thermal oscillations of electrons are revealed in the change of the sensitivity of a clean metallic surface in the region immediately adjacent to the long-wavelength boundary. The influence of temperature on the optical properties is already manifested at a considerably greater distance from this boundary $\lambda'$. The photocurrent-sensitivity curves obtained at two different temperatures mostly intersect at a fairly considerable distance from $\lambda'$. The curve obtained at the lower temperature has a less steep course$^{154,167,176,217}$. This can be explained by a change in the optical properties of the surface alone. A change in lattice spacings apparently causes no change in the work function, since the shift of the red boundary can be explained solely by the addition of the energy of the thermal motion of the electrons.

The influence of an increase in temperature on dipole-adsorbed foreign atoms should be manifested in a decrease of the total dipole moment. At coverages smaller than the optimum $(\theta_0)$, the decrease of the work function $\Delta\psi$ with increasing temperature diminishes, while the total work function of the surface increases. One may suppose that at higher temperatures this effect will outweigh both of those mentioned earlier.

At low temperatures the effect of a change in the thermal motion of the dipoles may prove more significant than the other influences.

Investigations of the photoeffect at high temperatures were carried out chiefly on continuous surfaces, where secondary phenomena play an especially significant role. Thus, for example, Keys$^{25}$, Coppius$^{27}$, Grew$^{50}$, Newbury$^{96}$, Berger$^{86}$, Newbury and Lemery$^{155}$ studied the surface of platinum and tungsten covered with oxides of alkaline-earth metals. When the temperature was raised from $20^\circ$ C to red heat, a thousandfold increase of the photocurrent was observed; this may be attributed partly to an increase in electrical conductivity and partly to structural changes in the oxide layer.

If the thermionic emission is already considerable, then the increase of emission under illumination is also caused by space charges. Such a view is held, for example, by Bodeman$^{87}$. Ramadanov$^{158}$ avoided the formation of space charges by using light of variable intensity and amplifying the alternating photocurrent. He found that, under illumination with a quartz lamp, the photoemission also increases with rising temperature; the cause of this increase he considered to be changes in the structure of the barium–oxygen layer.

The temperature dependence of the photoeffect on layers of simple composition, namely on atomically distributed barium and nickel, was studied by Zhurman and Depont$^{210}$. They determined the work function from the total photoemission by means of equation (2). Since $\psi$ is a function of the coverage $\theta$ and of the temperature $\vartheta$, the equation takes the following form:

$$ I = MT^2 e^{-(e_0/kT)\cdot\psi(\theta,\vartheta)}; \tag{2a} $$

taking logarithms gives

\[ \lg I - 2 \lg T = -\frac{1}{T}\frac{e_0}{k}\psi(\theta,\vartheta)\lg e+\lg M. \tag{2b} \]

Plotting \(\lg I - 2\lg T\) on the ordinate axis, and \(\frac{1}{T}\) on the abscissa axis, we obtain a straight line, the slope of which gives the value of \(\psi(\theta,\vartheta)\), while the intercept on the ordinate axis gives the value of \(M\).

Fig. 47. Logarithmic curve for the photoeffect on a massive layer of barium at different temperatures \(\vartheta\)

To determine the dependence of the total emission \(I\) on temperature, the authors \(^{210}\) used the null method. Two photocells were connected in parallel; both were illuminated by different light sources. The intensity of the light source incident on the photocell being compared with the photocell under investigation was varied until the photocurrents were made unequal, which was established by the absence of current in the corresponding compensation circuit.

Fig. 48. Logarithmic curve for atomically distributed barium \((\theta < \theta_0)\) on nickel at different temperatures

From Figs. 47 and 48 it is seen that the dependence of the photoeffect on a massive layer of barium in the temperature interval \(100\text{–}400^\circ\mathrm{C}\) depends very little on temperature; on the contrary, for a layer of atomic barium on nickel, increasing the cathode temperature lowers the photocurrent.

A more detailed examination leads to the conclusion that the decrease in photocurrent occurs not as a consequence of a decrease in the value of \(M\), but as a result of an increase in the work function. In the case of constant \(M\), \(\psi(\theta,\vartheta)\), beginning from \(350^\circ\mathrm{K}\), decreases linearly with temperature; this means that

\[ \psi(\theta,\vartheta)=\psi_u+\theta(\alpha_0+\beta\vartheta); \tag{2c} \]

\(\psi_u\) is the work function of pure nickel, \(\alpha_0\) is a constant referred to a definite coverage (the value of \(\alpha\) is negative),

\[ \beta=\frac{1}{\theta}\frac{d\psi(\theta,\vartheta)}{d\vartheta}, \tag{2d} \]

\(\beta\) is a temperature-independent quantity (Fig. 40), and depends relatively little also on the coverage.

Equation (2c) requires that the constant \(A\), determined by the thermionic method, decrease with increasing coverage. For the emission at red heat \(I_T\), for which \(\vartheta \equiv T\), the relation* holds:

\[ \begin{aligned} I_T&=A_0T^2 e^{-(e_0/kT)\cdot\psi(\theta,\vartheta)} =A_0T^2 e^{-\frac{e_0}{kT}(\psi_u+\alpha_0\theta+\theta\beta T)} \\ &=A_0 e^{-\frac{e_0}{k}\theta\beta}\cdot T^2\cdot e^{-\frac{e_0}{kT}(\psi_u+\alpha_0\theta)} =AT^2 e^{-\frac{e_0}{kT}(\psi_u+\alpha_0\theta)}. \end{aligned} \tag{2e} \]

Thus

\[ A=A_0 e^{-\frac{e_0}{k}\theta\beta}, \tag{2f} \]

where \(A_0\) is a universal constant.

From equation (2f) it is seen that, since \(\beta\) is positive, \(A\) decreases with increasing \(\theta\), which can occur only for \(\beta>0\). Since an increase of coverage (in the region below \(\theta_0\)) causes a decrease of the work function \(\psi(\theta,\vartheta)\) and a shift of the red boundary (toward longer waves), the increase of photosensitivity observed in this case (at room temperature) may be associated with a decrease of the constant \(A\). The latter was in fact observed by Campbell \(^{139}\). It was also noted by him that the value of the work function for a layer consisting of cesium deposited on silver oxide depends on the method of its determination. If the work function is determined from thermionic emission, then this value is smaller than in the case when it is established from the photoeffect. This difference can be satisfactorily explained by the temperature dependence of the work function. Indeed, determination of the value of the work function from photoelectric data gives \(\psi_u+\alpha_0\theta+\theta\beta T\), whereas the thermionic method according to equation (2e) gives \(\psi_u+\alpha_0\theta\). Since \(\beta\) is positive, this latter quantity is, of course, smaller. The last considerations indicate

* See Schottky and Rothe; further see the theoretical work of Becker and Brattain, Phys. Rev. 45, 694, 1934.

that the conclusions drawn by Campbell on the basis of the difference he observed in the work function are probably incorrect.

The temperature coefficient of the work function decreases with decreasing temperature (Fig. 49). It may be accepted that at low temperatures the work function, both of pure metals and of metals with the inclusion of foreign atoms, does not depend on temperature. In this case the temperature dependence of the photosensitivity is reduced mainly to a change in the optical properties of the substrate metal and of the layer adsorbed on it.

Fig. 49.

The study of the photoeffect at low temperatures presents special experimental difficulties, since foreign gases, alkali metals, etc., condense on the surface of the cathode, which obscures the temperature effect. Ives and Johnsrud\({}^{46}\) wished to avoid these difficulties by using the alkali metal under study not only as the cathode but also for the anode; in doing so they made the cathode small in comparison with the anode.

The cathode and anode were cooled, which led to the condensation of foreign gases and vapors mainly on the surface of the anode. Schallamach\({}^{209,217}\) immersed the entire photocell (Fig. 50) in a thermostat, so that the quartz window \(A\), the electrode, and the leads to them \(C\) and \(F\) were cooled to the required temperature. It goes without saying that the photocell was constructed without ground joints and without cement.

Fig. 50. Photocell adapted for investigating the photoeffect at low temperatures

To avoid condensation of air on \(A'\) during cooling to the temperature of liquid hydrogen (\(20^\circ\)K), the space between the quartz windows \(A\) and \(A'\) was evacuated. In \(G\) there was activated charcoal. The sealed-in platinum foil \(B\) constitutes the lead to the cathode

mirror, obtained by evaporation of \(E\). The tungsten filament \(E\) was also used as an anode.

Fig. 51 gives the temperature dependence of the photosensitivity of a layer consisting of potassium \((\theta > \theta_0)\), deposited on silver oxide.\(^{209}\) As is seen from Fig. 51, near the red limit the influence of temperature in the range \(293\text{--}83^\circ\mathrm{K}\) considerably exceeds the influence of cooling in the region from 83 to \(20^\circ\mathrm{K}\). The additional energy of the electrons does not fall linearly, as classical theory requires, but the less so the lower the temperature, as follows from the Fermi–Dirac–Sommerfeld theory. The spectral maximum at \(343\,\mathrm{m}\mu\) becomes, as a result of cooling, more gently sloping, while its position changes almost not at all in the temperature interval from 83 to \(20^\circ\mathrm{K}\). We cannot yet say with certainty what is the cause of the latter phenomenon: the disappearance of the absorption capacity of the alkali layer, or the independence of the optical properties of the silver mirror at low temperatures from temperature.

Fig. 51. Atomic potassium deposited by repeated evaporation in a high-vacuum silver mirror (coverage greater than optimum). The photoeffect was studied at 20, 83, and \(293^\circ\mathrm{K}\).

Fig. 52. Photocurrent curve for atomic sodium deposited on a smooth plate, at 83 and \(293^\circ\mathrm{K}\); coverage greater than optimum.

Zurman and Teising \(^{191}\) observed the appearance of a very sharp spectral maximum upon cooling of a continuous layer consisting of sodium \((\theta > \theta_0)\) on platinum black, from room temperature to the temperature of liquid air. If, however, the coverage was made less than optimal (in this case sodium gives a normal photocurrent curve), then cooling led to the opposite deviation—

tion of this curve near the red boundary and to the intersection of both curves in the region of short waves (Fig. 52). These phenomena were also observed in the case where, instead of platinum black, smooth platinum was used.

Fig. 53. Decrease of the photosensitivity of a cooled (83° K) hydrogenated potassium layer for \(\lambda = 405\ \mathrm{m\mu}\). The sensitivity was measured at 80 V; irradiation was carried out between experiments, the anode being either insulated or, conversely, a potential of 80 V being applied to it.

With proper experimental technique, quite reproducible results can be obtained on cooling and heating photoelements consisting of simple layers of alkali metals deposited on a metallic substrate.

For complex photolayers, however, this is not the case. Zhurman and Dempster \(^{216,218}\), using the photoelement shown in Fig. 50, showed that the sensitivity of complex layers consisting of potassium on naphthalene or a hydrogenated potassium surface decreases upon cooling. Heating to room temperature again restores the former sensitivity, which then no longer changes even under long exposure.

One might suppose that here, as was the case in the already cited work of de Boer and Teves, photoelectric excitation of the intermediate layer is essential. This, however, is contradicted by the following facts. The decrease in sensitivity does not depend on the voltage (Fig. 53) or on the strength of the current flowing and, on the contrary, depends on the intensity of the light. A decrease in photosensitivity can also be observed even in the complete absence of a potential difference between cathode and anode. Further, the drop in photocurrent does not become stronger on passing to shorter waves; the strongest decrease in sensitivity occurs when illuminated by light of wavelengths corresponding to the selective maximum (Fig. 46).

As Fig. 54 shows, when illuminated by light corresponding to the minima (334 and 254 \(\mathrm{m\mu}\)), the sensitivity changes hardly at all. Illumination by light corresponding to the second maximum (297 \(\mathrm{m\mu}\)) leads to a decrease

Fig. 54. Decrease of photosensitivity in separate spectral regions for a complex potassium—naphthalene—potassium surface under irradiation with different wavelengths.

of the first maximum (404 mµ); the short-wave maximum (297 mµ), however, does not change upon illumination with light of the wavelength corresponding to the maximum at 404 mµ (the slight decrease visible in Fig. 54 may be attributed to observational errors).

All the results we have presented can be explained if it is assumed that the absorbing centers of the photosensitive layer are excited; when the temperature is lowered, their lifetime increases greatly, just as occurs in phosphorescence phenomena. The potassium—naphthalene—potassium surface has two excitation levels corresponding to the two selective maxima. Since the centers of the long-wave maximum are excited by the frequencies of the short-wave maximum, these centers must be connected with one another.

Fig. 55. Photocurrent obtained from a potassium–naphthalene–potassium surface upon irradiation with light corresponding to the ultraviolet maximum, and then with red light; I — before illumination with ultraviolet light, II — after illumination with ultraviolet light.

At room temperature the excitation energy is instantaneously transferred to a free electron. At a low temperature, however, irradiation with red light is required for this. Fig. 55 shows that if, after excitation by light corresponding to the short-wave

Fig. 56. Sensitivity of the potassium—naphthalene—potassium surface at 83° K in the excited and unexcited state. ○—○—○ — after preliminary illumination with light of \(\lambda = 297\) mµ. ×—×—× — between separate measurements upon illumination with long-wave light.

to the maximum, illumination with red light should be applied, then the photocurrent decreases as a result of a reduction in the number of excited centers.

Since the excited states are quenched upon irradiation with red light, photosensitivity can also be observed at the temperature of liquid air, if illumination with red light is used. In Fig. 56 a curve obtained in this way is shown. A comparison of the curves of the complex layer potassium—naphthalene—potassium at room temperature and at 83° K (Fig. 57) shows that at low

Fig. 57. Photocurrent curves of the potassium—naphthalene—potassium layer at room temperature and in the unexcited state at a temperature of 83° K (between the individual measurements, illumination with long-wavelength light)
—○—○— \(T = 293^\circ\) K. ×—×—×— \(T = 83^\circ\) K

Fig. 58. Photocurrent curves of a naphthalene—potassium layer at room temperature and in the unexcited state at a temperature of 83° K

temperature both maxima have a flatter form. Whereas the long-wavelength maximum shifts, when the temperature is lowered, toward shorter waves by 10 mμ, the short-wavelength maximum remains in its former place. On a naphthalene–potassium surface, having only one maximum (see the preceding paragraph), lowering the temperature leads to an increase of the maximum, but not to its shift (Fig. 58).

Investigations of the photoeffect of cooled layers of complex composition lead to the conclusion that, in the regions corresponding to the spectral maximum, the mechanism of the photoeffect cannot consist in the ionization of absorbing centers. Evidently, what occurs is that the absorbing centers transfer light energy to free electrons, which, in leaving the surface, must overcome the corresponding potential. The lifetime of the excited state is very long; however, it can be considerably shortened as a result of sufficiently

of an intense impact of a molecule or as a result of the action of long-wave radiation. At room temperature the latter case, obviously, is of no significance.

The mechanism of electron emission described above makes it understandable why no deviations from Einstein’s law^3 were found in Olpin’s^128 study of the selective photoeffect on composite surfaces. Olpin established in this case complete equality between the incident quantum of energy and the energy of the emitted electron, so that the energy-distribution curves had no disturbances in the region of the spectral maximum.

7. Influence of the Field

In the well-developed Schottky theory the electric field acting on an electron at the surface of the cathode may be represented as consisting of two quantities: the external field \(E_a\), arising from the potential difference between cathode and anode, and the field \(E_i\), which is the result of the action of the so-called electric image force. In Fig. 59 it is shown that the potential \(V\) passes through a maximum at a certain distance \(x_m\) from the surface.

Fig. 59. Scheme of the fields acting on an electron at the surface of the cathode

\[ \psi=-\int_{0}^{x_m} [E_i(x)+E_a]\,dx = -\int_{0}^{\infty} E_i(x)\,dx+ \]

\[ +\int_{x_m}^{\infty} E_i(x)\,dx-E_a x_m; \]

\[ \psi=\psi_0+\int_{x_m}^{\infty} E_i(x)\,dx-E_a x_m . \tag{8} \]

For \(x_m\) and \(\psi\) the theory gives

\[ x_m=\frac{1}{2}\sqrt{\frac{e_0}{4\pi e}}\cdot \frac{1}{\sqrt{E_a}}, \tag{9} \]

\[ \psi=\psi_0-\sqrt{\frac{e_0}{4\pi e}}\cdot \sqrt{E_a}. \tag{10} \]

According to equation (10), \(\psi\) is a linear function of \(\sqrt{E_a}\); this conclusion has been repeatedly confirmed by experiments with pure metals

at high temperatures. However, Becker and Möller\(^{66}\), in their experiments with metallic surfaces on which foreign atoms were adsorbed, observed deviations from the Schottky straight line. They believe the reason for the deviation from linearity to lie in the presence of additional local electric fields on the surface. Equation (8), however, is also valid in the case when \(E_i(x)\) denotes the total field; differentiating equation (8), they obtain:

\[ \frac{d\psi}{dE_a}=-E_i(x_m)\frac{dx_m}{dE_a}-x_m-E_a\frac{dx_m}{dE_a}. \]

If one sets \(-E_i(x_m)=E_a\), then

\[ \frac{d\psi}{dE_a}=-x_m. \tag{11} \]

If the function \(\psi(E_a)\) is determined, then the negative tangent at the point \(E_a\) gives the distance \(x_m\) at which \(E_i=E_a\), and, consequently, in this way we can find \(E_i(x)\). Subtracting from the total value of the field \(E_i(x)\), we obtain that magnitude of the field which owes its existence to the adsorption of foreign atoms.

To determine the surface field in this way, one uses either \(\psi(E_a)\), or at least \(\dfrac{d\psi}{dE_a}\) as a function of \(E_a\). If the thermionic emission \(I_T\) is known to us as a function of the cathode temperature \(T\), then we can apply equation (2e)

\[ I_T=AT^2 e^{-\frac{e_0}{kT}(\psi_u+\alpha_0')} \tag{2e} \]

where

\[ A=A_0 e^{-\frac{\varepsilon}{k}\beta\theta}. \tag{2f} \]

Moreover,

\[ \ln I_T=\ln A+2\ln T-\frac{e_0}{kT}(\psi_u+\alpha_0'\theta). \]

Hence

\[ \frac{d\ln I_T}{dE_a} = \frac{d\ln A}{dE_a} - \frac{e_0}{kT} \frac{d(\psi_u+\alpha_0'\theta)}{dE_a}. \]

Becker and Möller assume that \(A\) does not depend on \(E_a\), and calculate \(\dfrac{d(\psi_u+\alpha_0'\theta)}{dE_a}\) from

\[ \frac{d\ln I_T}{dE_a} = -\frac{e_0}{kT} \frac{d(\psi_u+\alpha_0'\theta)}{dE_a}. \]

However, it is quite possible that the temperature coefficient \(\beta\theta\) of the work function [equation (2c)] depends on the external field, and then \(A\), according to equation (2f), is a function of \(E_a\). In this case the definition of the surface-field magnitude given above cannot lead to any result. It is necessary to use me-

ON THE EXTERNAL PHOTOEFFECT ON ADSORBED FILMS

…method independent of the cathode temperature; namely, to determine the total emission at different temperatures or the red limit from the photocurrent curve as a function of the external field.

The dependence of the work function on the external field applied to the surface of a pure metal has hardly been studied. On the contrary, for metallic surfaces on which atoms are adsorbed, this dependence has been investigated many times by a number of authors. The first works were carried out by Ives,^40 who studied thin layers of alkali metals deposited on a mirror-like metallic surface. However, in later studies,^75 he himself cast doubt on his first observations because of insufficient vacuum. Indeed, the photoeffect in his case must be very small unless special measures are taken (especially when thin wires are used as the cathode) so that the potential gradient at the cathode surface is sufficiently large. Thus, Suhrmann,^81,164 working with potassium and sodium salts on platinum black, owing to the slight curvature of the surface and, consequently, the small macroscopic potential gradient at small external fields, obtained, however, a very considerable effect because of the large microgradient. As is seen from Fig. 60, the maximum influence of the field is observed near the red limit. This effect is much weaker if the adsorbed layer is located as a thin layer on a smooth substrate surface (Fig. 61). At very small coverages (less than optimal) and a small potential gradient, the photoeffect disappears (Fig. 62).

Fig. 60. Current–voltage curves obtained when platinum black coated with atomic sodium is irradiated with light of various wavelengths. The coating is less than optimal. The cathode is a sphere 2 cm in diameter, situated at the center of a spherical anode 20 cm in diameter.

Since the influence of the field is manifested most strongly at the red limit \(\nu'\), it may be thought that the action of the field is connected with a shift of the \(d\nu'\) boundary toward longer waves. Lawrence and Linford^124 did indeed show that the photosensitivity curve shifts parallel to the applied field (Fig. 63). The cathode used was a tungsten filament \(0.023\) mm thick, surrounded by an anode cylin-

... 5.8 mm in diameter. The fields applied were very large, and therefore a significant photoeffect could be observed even on a smooth platinum plate.

Fig. 61. Current–voltage curves under illumination with light of various wavelengths and the spectral sensitivity curve of a photo-layer consisting of sodium on a smooth platinum plate. Coverage greater than the optimum.

From the shift of the red boundary the authors determined the surface field as a function of the distance \(x\), substituting into equation (11)

\[ e_0\psi = h\nu'; \qquad d\psi = \frac{h}{e_0}\,d\nu' \]

the expression

\[ \frac{d\nu'}{dE_a}=\frac{e_0}{h}\,x_m . \]

Just as in the considerations of Becker and Mueller, here we obtain the distance \(x_m\) at which the surface field is equal to \(E_a\). In a similar way the curves \(C\) in Fig. 64 were obtained.

The surface field follows the field of the electrical image up to a distance of \(1.2\cdot 10^{-6}\) cm. At greater distances the deviations of the surface field are smaller than those observed by Becker and Mueller\(^{66}\), and also by Reynolds\(^{134}\), in thermoelectronic measurements. This may be explained by the fact that the temperature dependence of the work function makes the determination of \(\psi\) by the thermionic method unreliable.

Gexford\(^{148,149}\) investigated the influence of an external field for oxide cathodes. He found that the red boundary depends on the magnitude of the field in the following way:

\[ \nu'=\nu_0' - b\sqrt{E_a}, \tag{10a} \]

where \(\nu_0'\) denotes the limiting frequency at \(E_a=0\); \(\nu'\) is obtained as a result of extrapolating the photoeffect curve to its intersection with the axis of abscissas.

Fig. 62. Current–voltage curve under illumination of a sodium layer on a smooth platinum plate with light of various wavelengths, and the spectral-sensitivity curve. Coverage less than the optimum.

Zurman and Eichborn\(^{180}\)

established the validity of equation (10a) for atomic potassium on tungsten. They determined the work function \(\psi\) from the dependence of the total emission on the temperature of the light source at different external fields;

Fig. 63. Shift of the photocurrent curve toward longer wavelengths with increasing potential gradient at the cathode surface; cathode—a tungsten wire coated with potassium. Potential gradient in V/cm for curves: \(A\)—63100; \(B\)—36200; \(C\)—26100; \(D\)—15800; \(E\)—9000; \(F\)—3100; \(G\)—1000; \(H\)—260; \(I\)—0 (after Lawrence)

they obtained a linear dependence between \(\psi\) and \(\sqrt{E_a}\) *

\[ \psi=\psi_0-b'\sqrt{E_a}. \tag{10b} \]

Equation (10b), up to the constant \(b'\), coincides with Schottky’s equation (10), derived for the ordinary field of an electric image, given by the equation

\[ E_i(x)=-\frac{1}{4\pi\varepsilon}\cdot \frac{e_0}{4x^2}. \]

For the ordinary field of an electric image

\[ b=\sqrt{\frac{e_0}{4\pi\varepsilon}}. \]

* To obtain equation (10b), one must substitute into equation (10a)

\[ \psi=\frac{h\nu'}{e_0}. \]

In the case considered by us, however, this quantity has a value two or 3 times greater. Therefore one may assume that, in the adsorption layers studied, \(E_i\) also decreases proportionally to \(\frac{1}{x^2}\), but that the constant of proportionality is greater than \(\frac{1}{4\pi\varepsilon}\cdot\frac{e_0}{4}\). This means that an electron with charge \(e_0\) receding from the surface can be returned by the action of a positive charge whose magnitude must be greater than the charge induced by \(e_0\) at a distance \(2x\).

Fig. 64. Curves A and B: shift of the red limit as a function of the applied field. Curve C—the surface field calculated from here as a function of distance from the surface. The solid (calculated) curves correspond to the case when the external field is absent. Cathode—a tungsten wire with a deposit of potassium (after Lawrence).

Up to this point we have considered the influence of a large potential gradient on photoelectric emission; let us call this phenomenon the external field effect. In addition, Suhrmann \(^{164}\) observed yet another effect at very small external fields, which he called the internal field effect. The latter is expressed in a shift of the saturation current and occurs for layers consisting of alkali-metal atoms on a smooth metallic substrate, when nearly optimal filling is reached. On passing beyond the optimal filling the effect disappears.

Nottingham \(^{187}\) studied in more detail the influence of weak fields on the sodium–nickel photosurface. He determined the red limit (or, what is the same thing, the work function) as a function of the anode potential. At potentials greater than 3 V he extrapolated the photocurrent curve to its intersection with the wavelength axis; at small potentials he measured the photocurrent as a function of potential at different light frequencies.

As the limiting frequency he took that which corresponded to a photocurrent of \(10^{-14}\) A. Fig. 65 is a graphical representation of the results of Nottingham’s experiments; in the graph the limiting frequencies (expressed in volts) are given as a function of the anode potentials.

The curves measured for a thick layer of sodium consist of a horizontal part and an inclined linear segment with an angular coefficient equal to 1. At the point of transition from the horizontal part to the inclined one, the anode potential is equal to the contact potential

\(V_{K,A}\), below which all electrons reach the anode; the red limit, therefore, does not change. For the inclined linear part, Einstein’s equation holds:

\[ \frac{h\nu}{e_0}=-(V-V_{K,A})+\frac{h\nu_0}{e_0}. \]

Indeed, if the limiting frequency is \(\nu\) at potential \(V\), then, when illuminated with frequency \(\nu\), the emitted electrons must perform work against the potential \(V_{K,A}-V\); the work function must then be equal to \(h\nu\).

The abrupt transition from one straight line to the other occurs only for very clean surfaces. Since foreign atoms are present, the graph has the form shown in Fig. 65 (a thin metal layer). The true limiting frequency \(\nu_0\), independent of the external field, determined by the deviation of our graph from linearity, is smaller in magnitude than \(\nu\) for a pure metal substrate; but in any case this quantity is greater than that obtained for a massive foreign metal. This limiting frequency, therefore, does not pass through a minimum value, as is the case for the work function determined at higher anode potentials* (the horizontal part).

Fig. 65. Limiting frequency (expressed in volts) as a function of anode potential. Surface—sodium on nickel (according to Nottingham)

The curved portion \(bc\) owes its origin to the surface field produced by foreign atoms; the magnitude of this field can be determined from this part of the curve. However, these quantities turn out to be too large: at a distance of \(10^{-2}\) cm the field gradient is, for example, \(5\text{--}10\ \mathrm{V/cm}\). It is possible that Nottingham’s method for determining the limiting frequency is not sufficiently correct, since it does not take into account the thermal motion of the electrons.

These experiments must be continued, but in such a way that, for determining the limit, the photoelectric path is used—ordinary emission. In the latter case \(\nu\) is determined more accurately and, moreover, independently of the additional energy of the electrons.

* The most recent works of Reynolds and Nottingham (Phys. Rev. 45, 765, 1934) on the work function of electrons from heated thoriated tungsten in the absence of a field give an optimal filling equal to almost a monatomic layer.

In addition to all the influences mentioned, it is also necessary to mention the influence of a nonuniform coating, which can produce an effect similar to the superposition of electric fields of considerable magnitude.

When determining the red limit with the aid of spectrally resolved light, one always obtains the limiting frequency of the most electropositive spots of the surface. By means of the method of constructing the photoeffect line, it is also possible to take into account the electronegative spots of the surface and to obtain a result fairly close to the mean value.

LITERATURE

1. R. Pohl and P. Pringsheim, Die lichtelektrischen Erscheinungen, Braunschweig, 1914.

2. W. Hallwachs, Die Lichtelektrizität, Handbuch d. Radiogie, Leipzig, 1916.

3. B. Gudden, Lichtelektrische Erscheinungen, Berlin, 1928.

4. A. L. Hughes and L. A. Du Bridge, Photoelectric phenomena, New York and London, 1932.

5. H. Simon and R. Suhrmann, Lichtelektrische Zellen und ihre Anwendung, Berlin, 1932.

6. R. Suhrmann, in the section “Elektronenemission metallischer Leiter” of the course of physics Müller-Pouilletschen Lehrbuchs der Physik, 11th ed., Vol. 4, Part 4, Braunschweig, 1934.

7. W. Schottky and H. Rothe, in the section “Physik der Glühelektroden,” Handbuch der Experimentalphysik, Vol. 13, Part 2, Leipzig, 1928.

8. Th. Wulf, Beiträge zur Kenntnis der lichtelektrischen Wirkung, Ann. Physik (4) 9, 946, 1902.

9. W. Hallwachs, Über die lichtelektrische Ermüdung, ibid. (4) 23, 459, 1907.

10. V. L. Chrisler, Influence of absorbed hydrogen and other gases on the photoelectric properties of metals, Phys. Rev. 27, 289, 1908.

11. J. Elster and H. Geitel, Über gefärbte Hydride der Alkalimetalle und ihre photoelektrische Empfindlichkeit, Physik. Z. 11, 257, 1910.

12. R. Pohl and P. Pringsheim, Über einige lichtelektrische Beobachtungen am Al und Mg, Verh. dtsch. physik. Ges. 14, 516, 1912.

13. H. Greinacher, Über den Einfluss der Glimmentladung auf den lichtelektrischen Effekt, ibid. 15, 797, 1913.

14. R. Pohl and P. Pringsheim, Über die langwellige Grenze des normalen Photoeffektes, ibid. 15, 625, 1913.

15. R. Pohl and P. Pringsheim, Über den Einfluss des Sauerstoffs auf den selektiven Photoeffekt des Kaliums, ibid. 15, 637, 1913.

16. S. Werner, Über lichtelektrische Elektronenemission, Uppsala, 1913.

17. S. Wiedemann and W. Hallwachs, Über die Abhängigkeit der Lichtelektrizität der Metalle von Gas, insbesondere den Grund der starken Lichtelektrizität des Kaliums, Verh. dtsch. physik. Ges. 16, 107, 1914.

18. W. Hallwachs, Mitteilungen aus dem Physikalischen Institut der Technischen Hochschule Dresden, Physik. Z. 16, 95, 1915.

19. F. Krüger and E. Täge, Über den Einfluss von Katalysatorgiften auf die lichtelektrische Empfindlichkeit des Platins, Z. Elektrochem. 21, 562, 1915.

20. E. Eichler, Der Einfluss der Gase auf den lichtelektrischen Effekt, Z. wiss. Photogr. 16, 10, 60, 1916.

21. F. Krüger, Über den Parallelismus von lichtelektrischer Empfindlichkeit und Voltaeffekt, Z. Elektrochem. 22, 365, 1916.

1. I. Langmuir, The dissociation of hydrogen into atoms, III. The mechanism of the reaction, J. Am. Chem. Soc., 38, 1145, 1916.
2. W. M. Varley, Phil. Trans. Roy. Soc., Lond., 202, 365, 1916.
3. W. Hallwachs, Lichtelektrizität als Funktion des Gasgehaltes, Physik. Z., 21, 561, 1920.
4. T. W. Case, New strontium and barium photoelectric cells, Phys. Rev., 17, 398, 1921.
5. W. Frese, Über Passivität und Lichtelektrizität, Z. wiss. Photogr., 21, 37, 1921.
6. O. Koppius, A comparison of the thermionic and photoelectric work function for platinum, Phys. Rev., 18, 443, 1921.
7. M. Sende and H. Simon, Lichtelektrizität als Funktion des Gasgehaltes, Ann. Physik, 65, 697, 1921.
8. H. Geitel, Die Proportionalität von Photostrom und Beleuchtung an sehr dünnen Kaliumschichten, Ebenda 67, 420, 1922.
9. R. Suhrmann, Rote Grenze und spektrale Verteilung der Lichtelektrizität des Platins in ihrer Abhängigkeit vom Gasgehalt. Ebenda 67, 43, 1922.
10. G. Wiedmann, Über die Entwicklung der normalen und selektiven Lichtelektrizität seit 1914, Jb. Rad. u. Elekt. 19, 112, 1922.
11. O. Klemperer, Über die lichtelektrische Geschwindigkeitsverteilung, Z. Physik, 16, 280, 1923.
12. F. Krüger and A. Ehmer, Über die lichtelektrische Empfindlichkeit von wasserstoffbeladenen Palladium-Silberlegierungen, Ebenda 14, 1, 1923.
13. J. Langmuir and K. H. Kingdon, Thermionic phenomena due to alkali vapors, Part I. Experimental, Phys. Rev., 21, 380, 1923.
14. J. Langmuir, The electron emission from thoriated tungsten filaments, Ebenda 22, 357, 1923.
15. R. Suhrmann, Der Einfluss der Gasbeladung auf die thermische und lichtelektrische Elektronenemission des Pt und Ta, Z. Physik, 13, 17, 1923.
16. F. G. Tucker, Effect of heat treatment on the photoelectric emission from platinum, Phys. Rev., 22, 574, 1923.
17. R. Dümpelmann and W. Hein, Beiträge zur Kenntnis des lichtelektrischen Effektes, Z. Physik, 22, 368, 1924.
18. B. Gudden, Elektrizitätsleitung in kristallisierten Stoffen unter Ausschluss der Metalle, Erg. exakt. Naturwiss., 3, 116, 1924.
19. H. E. Ives, Photoelectric properties of thin films of alkali metals, Astrophysic. J., 60, 209, 1924.
20. H. F. Ives and A. D. Johnsrud, The normal and selective photoelectric effect in the alkali metals and their alloys. Ebenda 60, 231, 1924.
21. K. H. Kingdon and I. Langmuir, Electron emission from caesium covered filaments, Phys. Rev., 23, 112, 1924.
22. K. H. Kingdon, Electron emission from adsorbed films on tungsten, Ebenda 24, 510, 1924.
23. B. Gudden and R. W. Pohl, Zum Nachweis des selektiven Photoeffektes, Z. Physik, 34, 245, 1925.
24. K. Herrmann, Über den Einfluss der Gasbeladung auf die lichtelektrische Elektronenemission und die elektrische Leitfähigkeit des Platins. Ann. Physik, 77, 503, 1925.
25. H. E. Ives and A. L. Johnsrud, The influence of temperature on the photoelectric effect of the alkali metals. J. opt. Soc. Am., 11, 565, 1925.
26. R. Suhrmann, Rote Grenze und Austrittsarbeit lichtelektrischer Elektronen Z. Physik, 33, 61, 1925.
27. A. E. Woodruff, The variation in the photoelectric emission from platinum, Phys. Rev., 26, 665, 1925.
28. J. A. Becker, Thermionic and adsorption characteristics of caesium on tungsten and oxidized tungsten Ebenda 28, 341, 1926.
29. W. H. Grew, An effect of light on the electron emission from hot filaments, Ebenda 28, 1365, 1926.

1. H. E. Ives, Photoelectric properties of thin films of alkali metal. II. Phenomena at high temperatures, Astrophysic J., 64, 128, 1926.

2. R. Pohl, Zur quantenhaften Lichtabsorption in festen Körpern, Naturwiss., 14, 214, 1926.

3. A. Prewoditelew and G. Joffe, Der Einfluss des adsorbierten Gases auf die Grösse des photoelektrischen Effektes, Z. Physik, 38, 280, 1926.

4. W. Bennewitz, Die Variation der Geschwindigkeitsverteilung lichtelektrischer Elektronen beim Entgasungs- und Gasbeladungsprozess an Palladium und Platin, Ann. Physik, 83, 913, 1927.

5. L. A. Du Bridge, The photoelectric properties of thoroughly outgassed platinum, Phys. Rev., 29, 451, 1926.

6. H. K. Dunn, Changes in the photoelectric threshold of mercury, Ebenda, 29, 693, 1927.

7. Th. C. Fry, Plane waves of light: I. Electromagnetic behavior, J. opt. Soc. Am., 15, 137, 1927.

8. H. E. Ives and A. L. Johnsrud, The thickness of spontaneously deposited photoelectrically active rubidium films measured optically, Ebenda, 15, 374, 1927.

9. H. E. Ives and G. R. Stilwell, Photoelectric emission as a function of composition in sodium-potassium alloys, Phys. Rev., 29, 252, 1927.

10. J. Kluge, Einfluss des Gasgehaltes auf die Geschwindigkeitsverteilung lichtelektrischer Elektronen bei Platin, Aluminium und Palladium, Ann. Physik, 82, 432, 1927.

11. Leupold H., Lichtelektrische Untersuchungen an ammoniakbeladenem Platin, Ebenda 82, 841, 1927.

12. A. H. Warner, A comparison of the thermionic and photoelectric work functions for clean tungsten, Proc. nat. Acad. Sci. U. S. A., 13, 56, 1927.

13. F. Wolf, Über die Elektronengeschwindigkeiten beim normalen und selektiven lichtelektrischen Effekt, Ann. Physik, 83, 1001, 1927.

14. Th. Zimmer, Über den lichtelektrischen Effekt an Schichten, die durch Naturmatomstrahlen hergestellt sind, Diss. Univ. Frankfurt a. M., 1927.

15. O. Vauwers, Über den Einfluss der Gasbeladung auf die lichtelektrische Empfindlichkeit von Kohle, Z. Physik, 51, 618, 1928.

16. J. A. Becker and D. W. Mueller, On electrical fields near metallic surfaces, Phys. Rev., 31, 431, 1928.

17. N. R. Campbell, The photo-electric properties of thin films of the alkali metals, Phil. Mag., 6, 633, 1928.

18. A. B. Gardwell, The photoelectric properties of thin films of the alkali metals P.

19. L. A. du Bridge, The photoelectric and the thermionic work functions of outgassed platinum, Phys. Rev., 31, 236, 1928.

20. Th. C. Fry, Plane waves of light. II. Reflection and refraction, J. opt. Soc. Am., 16, 1, 1928.

21. B. Gudden, Bemerkung zur Mitteilung von Rud. Suhrmann: Über eine neuartig lichtelektrische Beobachtung an dünnen Alkalimetallschichten, Naturwiss., 16, 547, 1928.

22. W. B. Hales, Optical photoelectric potential of clean mercury and the influence of gases and of the circulation of the mercury upon it, Phys. Rev., 32, 950, 1928.

23. F. Herold, Über die Geschwindigkeitsverteilung lichtelektrischer Elektronen, Ann. Physik, 85, 887, 1928.

24. H. E. Ives and A. L. Johnsrud, The thickness of spontaneously deposited photoelectrically active rubidium films, measured optically, J. opt. Soc. Amer., 16, 121, 1928.

25. H. E. Ives, A. R. Olpin and A. L. Johnsrud, The distribution in direction of photoelectrons from alkali metal surfaces, Phys. Rev., 32, 57, 1928.

26. H. Klumb, Über den Einfluss der Gasbeladung auf die lichtelektrische Empfindlichkeit der Metalle, Z. Physik, 47, 652, 1928.

27. J. C. McLennan, L. A. Matheson and C. D. Niken, The photoelectric

ON THE EXTERNAL PHOTOEFFECT ON ADSORBED FILMS

effect at low temperatures, Trans. Roy. Soc., Canada, Sect. 111(3), 22, 279, 1928.

1. R. Lukursky u. S. Prilezaev, Über den normalen Photoeffekt, Z. Physik, 49, 236, 1928.

2. D. Roller, Effect of toluene on the photoelectric behavior of mercury, Phys. Rev., 32, 323, 1928.

3. Sander u. Nitsche, Die lichtelektrische langwellige Empfindlichkeitsgrenze des Ammoniumamalgams, Z. Elektrochem., 34, 244, 1928.

4. R. Suhrmann, Über eine neuartige lichtelektrische Erscheinung an dünnen Alkalimetallschichten, Naturwiss, 16, 336, 1928.

5. R. Suhrmann, Wasserstoffionen als Ursache für das Auftreten der lichtelektrischen spektralen Selektivität des Kaliums, Physik Z., 29, 811, 1928.

6. R. Suhrmann u. H. Theissing, Über den Einfluss des Wasserstoffs auf die lichtelektrische Elektronenemission des Kaliums, Z. Physik, 52, 453, 1928.

7. G. B. Welch, Photoelectric thresholds and fatigue, Phys. Rev., 32, 657, 1928.

8. J. A. Becker, The life history of adsorbed atoms and ions, Trans Am Elektrochem. Soc., 60, 153, 1929.

9. C. E. Berger, The effect of light on the electron emission from cerium dioxide, Phys. Rev., 34, 1566, 1929.

10. E. Bodemann, Über eine Steuerung des glühelektrischen Stromes oxydbedeckter Metallfolien durch Bestrahlung mit ultraviolettem Licht, Ann. Physik, 3, 614, 1929.

11. N. R. Campbell, Photoelectric thresholds of the alkali metals, Phil. Mag., 8, 667, 1929.

12. N. R. Campbell, Wasserstoff und die photoelektrische Emission aus Kalium, Physik. Z., 30, 537, 1929.

13. R. Fleischer, Beitrag zu R. Suhrmann: Wasserstoffionen als Ursache für das Auftreten der lichtelektrischen spektralen Selektivität des Kaliums, Ebenda 30, 320, 1929.

14. H. E. Ives, The preparation of photoelectric cells with thin films of lithium as the photoactive material, Phys. Rev., 33, 1081, 1929.

15. H. E. Ives a. A. R. Olpin, Maximum excursion of the photoelectric long wave limit of the alkali metals, Ebenda 34, 117, 1929.

16. L. R. Koller, Thermionic and photoelectric emission from caesium at low temperatures, Ebenda 33, 1082, 1929.

17. E. O. Lawrence and L. B. Linford, The effect of intense electric fields on the photoelectric behavior of alkali films on tungsten, Ebenda 34, 1492, 1929.

18. M. J. Martin, The photoelectric and thermionic properties of molybdenum Phys. Rev., 33, 991, 1929.

19. K. Newbury, The effect of light on the electron emission from oxide coated filaments, Ebenda 34, 1418, 1929.

20. A. R. Olpin, Use of dielectrics to sensitize alkali metal photoelectric cells to red and infrared light, Ebenda 33, 1081, 1929.

21. A. R. Olpin, Apparent modulation of light by films of dielectrics on catodes of alkali metal photoelectric cells, Ebenda 34, 544, 1929.

22. P. Selenyi, Über rotempfindliche Natrium-Photokathoden, Physik Z., 30, 933, 1929.

23. A. Smits u. H. Gerding, Über den photoelektrischen Effekt des Aluminiums und seiner Amalgame, I, Ebenda 30, 322, 1929.

24. R. Suhrmann u. H. Theissing, Versuche zur Klärung der selektiven äusseren lichtelektrischen Wirkung. III. Untersuchungen über den selektiven lichtelektrischen Effekt an dünnen auf einem Platinspiegel adsorbierten Kaliumhäuten, Z. Physik, 55, 701, 1929.

25. R. Suhrmann, Über die Änderung des elektrischen Zustandes von Metalloberflächen durch Beladen mit H-Ionen und durch Elektronenbombardement, Z. Elektrochem., 35, 681, 1929.

26. R. Suhrmann, Beziehungen zwischen dem normalen lichtelektrischen

effect and electrical surface properties of various metals, Physik Z., 39, 939, 1929.

1. R. Teichmann, On the maximum velocity of photoelectric electrons in the selective sensitivity region of potassium, Ann. Physik (5), 1, 1069, 1929.

2. H. Th. Wolff, On the theory of the photoelectric effect, Z. Physik, 52, 158, 1929.

3. V. Zworykin and E. D. Wilson, The caesium-magnesium photocell, Phys. Rev., 33, 633, 1929.

4. V. Zworykin and E. D. Wilson, The caesium-magnesium photocell, J. opt. Soc., Am., 19, 81, 1929.

5. J. H. de Boer and M. C. Teves, The influence of the photoelectric properties of caesium by adsorption on salt layers, Z. Physik, 65, 489, 1930.

6. W. H. Brattain, Effect of adsorbed thorium on the thermionic emission from tungsten, Phys. Rev., 35, 1431, 1930.

7. A. K. Brewer, The photoelectric and thermionic properties of platinum-coated glass filaments, Ebenda 35, 1360, 1930.

8. N. R. Campbell, A theory of selective photoelectric emission (photoelectric cells and their applications, London 1930).

9. Th. W. Case, Barium photoelectric cells (Photoelectric cells and their applications, London 1930).

10. L. J. Davies and H. R. Ruff, The manufacture and use of the thin-film caesium cell for sound reproduction (Photoelectric cells and their applications, London 1930).

11. G. Déjardin, Photoelectric cells with alkali metals on a magnesium support, J. Physique Radium (7) 1, 66, 1930.

12. R. Fleischer and H. Teichmann, On the influence of silicon oxide on the photoelectric sensitivity of potassium, Z. Physik, 60, 317, 1930.

13. R. Fleischer and H. Teichmann, The increase of the photoelectric effect of potassium by hydrogen, Ebenda 61, 227, 1930.

14. R. H. Fowler, A possible explanation of the selective photoelectric effect, Proc. Roy. Soc. Lond., (A), 128, 123, 1930.

15. H. Fröhlich, On the photoeffect in metals, Ann. Physik, and 7, 103, 1930.

16. H. Gerding, The photoelectric effect of aluminum and of alloys of aluminum with mercury, Z. physik. Chem., 2, 1, 1930.

17. R. F. Hanstock, Effect of systematic surface treatment on the photoelectric emission from metals, Phil. Mag., 10, 937, 1930.

18. L. R. Koller, Photoelectric emission from thin films of caesium, Phys. Rev., 36, 1539, 1930.

19. J. Kunz, Photoelectric cells and their applications, London 1930.

20. E. O. Lawrence and L. B. Linford, The effect of intense electric fields on the photoelectric behavior of thin potassium films, Phys. Rev., 35, 1421, 1930.

21. E. O. Lawrence and L. B. Linford, The effect of intense electric fields on the photoelectric properties of metals, Ebenda 36, 482, 1930.

22. L. B. Linford, Electrostatic surface fields near thoriated tungsten filaments by a photoelectric method, Ebenda 36, 1100, 1930.

23. D. H. Loughridge, The manufacture of photoelectric cells and their use in sound reproduction (Photoelectric cells and their applications, London 1930).

24. W. B. Nottingham, Influence of accelerating fields on the photoelectric and thermionic work function of composite surfaces, Phys. Rev., 35, 1128, 1930.

25. A. R. Olpin, Validity of Einstein’s photoelectric equation for red-sensitive sodium compounds, Ebenda 35, 171, 1930.

26. A. R. Olpin, Selective maxima in the spectral response curves of light-sensitive compounds as a function of valence, Ebenda 35, 561, 1930.

1. A. R. Olpin, Method of enhancing the sensitiveness of alkali metal photoelectric cells, Ebenda 36, 251, 1930.

2. A. R. Olpin, Inhibition of photoelectric emission by near infrared light, Ebenda 36, 376, 1930.

3. Wo. Ostwald, Kolloidwissenschaft, Elektrotechnik und Catalyse, Dresden u. Leipzig 1930.

4. D. Ramanadoff, A method of studying the effect of temperature on photoelectric currents, Rev. Sci. Instr., 1, 768, 1930.

5. N. B. Reynolds, Schottky effect and contact potential measurements on thoriated tungsten filaments, Phys. Rev., 35, 158, 1930.

6. P. Selenyi, The manufacture, properties and use of sodium photoelectric cells. (Photoelectric cells and their applications, London 1930).

7. H. Bomke, Über die lichtelektrischen Eigenschaften des Kadmium, insbesondere den Einfluss von Gasen auf dieselben, Ann. Physik (5), 10, 579, 1931.

8. A. K. Brewer, Photoelectric properties of ammonic catalysts, J. Am. Chem. Soc., 53, 74, 1931.

9. A. K. Brewer, The effect of adsorbed K-ions on the photoelectric treshold of iron, Phys. Rev., 38, 401, 1931.

10. N. R. Campbell, The photoelectric emission of thin films, Phil. Mag., 12, 173, 1931.

11. G. Déjardin, R. Schwegler et M. Warin, Sur les propriétés photoélectriques des couches minces des métaux alcalins, J. Physique Radium, 2, 88, 1931.

12. E. H. Dixon, Some photoelectric and thermionic properties of rhodium. Phys Rev., 37, 60, 1931.

13. R. Felischer, Die lichtelektrische Elektronenemission an dünnen Kalium-und Cäsiumschichten, Physik Z., 32, 217, 1931.

14. R. Fleischer u. H. Teichmann, Über den Zusammenhang zwischen dem Einfluss von Stickstoff-Sauerstoffverbindungen und dem ihrer Komponenten auf die lichtelektrische Empfindlichkeit des Kaliums. Z. Physik, 67, 184, 1931.

15. R. Fleischmann, Eine selektive Lichtabsorption in dünnen Alkalimetallschichten, Nachr Ges. Wiss. Göttingen, Math-physik. Kl., 1931, 252.

16. R. H. Fowler, The analysis of photoelectric sensitivity curves for clean metals at various temperatures, Phys. Rev., 38, 45, 1931.

17. J. Frenkel, Some remarks on the theory of the photoelectric effect, Ebenda 38, 309, 1931.

18. H. Göthel, Über den äusseren lichtelektrischen Effekt an Phosphoren und seine Abhängigkeit vom Erregungszustand, Physik Z., 32, 218, 1931.

19. W. S. Huxford, Photoelectric properties of oxide cathodes, Phys. Rev., 37, 102, 1931.

20. W. S. Huxford, Effect of electric fields on the emission of photoelectrons from oxide cathodes, Ebenda 38, 379, 1931.

21. H. E. Ives, The vectorial photoelectric effect in thin films of alkali metals, Ebenda 38, 1209, 1931.

22. H. E. Ives a. H. B. Briggs, The photoelectric effect from thin films of alkali metal on silver, Ebenda 38, 1477, 1931.

23. W. Kluge, Über die photoelektrische Sensibilisierung von Kalium mittels Schwefel, Selen und Tellur, Z. Physik, 67, 497, 1931.

24. W. Kluge u. E. Rupp, Über lichtelektrische Wirkung und Elektronenbeugung an hydrierten Kaliumoberflächen, Physik Z., 32, 163, 1931.

25. L. W. Morris, Certain photoelectric properties of gold, Phys. Rev., 37, 1263, 1931.

26. K. Newbury a. F. Lemery, The photoelectric effect from a barium oxide coated platinum filament, J. opt. Soc. Am., 21, 276, 1931.

27. A. R. Olpin, Correlating the selective photoelectric effect with the selective transmission of electrons through a cathode surface, Phys. Rev., 37, 464, 1931.

1. A. R. Olpin, An interpretation of the selective photoelectric effect from two component cathodes, Ebenda 38, 1745, 1931.

2. D. Ramadanoff, Photoelectric properties of composite surfaces at various temperatures and potentials, Ebenda 37, 884, 1931.

3. S. Rijanoff, Photoelectric properties of the potassium surface modified by the action of hydrogen atoms, Z. Physik 71, 325, 1931.

4. D. Roller, W. H. Jordan and C. S. Woodward, Some photoelectric properties of mercury films, Phys. Rev. 38, 396, 1931.

5. R. Suhrmann, On the structure of the emitting surface in the selective photoelectric effect, Physik. Z. 32, 216, 1931.

6. R. Suhrmann, The physico-chemical constitution of the metal surface in the selective photoelectric electron emission of the alkali metals, Z. Elektrochem. 37, 678, 1931.

7. R. Suhrmann, Causes of the occurrence of the selective photoelectric effect, Z. wiss. Photogr. 30, 161, 1931.

8. R. Suhrmann, New observations on field and photoeffects at outer boundary surfaces, Physik. Z. 32, 929, 1931.

9. R. Suhrmann, On chemical and electrical processes at gas-laden metal surfaces, Z. anorg. allg. Chem. 203, 235, 1931.

10. M. C. Teves, On highly sensitive vacuum photoelectric cells, Z. techn. Physik 12, 556, 1931.

11. E. P. Winch, The photoelectric properties of silver, Phys. Rev. 37, 1269, 1931.

12. E. P. Winch, Photoelectric properties of thin unbacked gold films, Ebenda 38, 321, 1931.

13. A. A. Young and N. H. Frank, Temperature dependence of photoelectric effect in metals, Ebenda 38, 838, 1931.

14. T. F. Young and W. C. Pierce, The wave-length-sensitivity curve of a caesium oxide photocell, J. opt. Soc. Am. 21, 497, 1931.

15. W. H. Zachariasen, On the interpretation of the selective photoelectric effect from two-component-cathodes, Phys. Rev. 38, 2290, 1931.

16. S. Asao, Physics 2, 12, 1932.

17. K. B. Blodgett and J. Langmuir, Accommodation coefficient of hydrogen; a sensitive detector of surface films, Phys. Rev. 40, 78, 1932.

18. J. J. Brady, The photoelectric properties of alcali metal films as a function of their thickness, Ebenda 41, 613, 1932.

19. A. K. Brewer, The effect of adsorbed gases on the photoelectric emissivity of iron and platinum, J. Am. chem. Soc. 54, 1888, 1932.

20. L. A. du Bridge, A further experimental test of Fowler’s theory of photoelectric emission, Phys. Rev. 39, 108, 1932.

21. J. H. de Boer and M. C. Teves, The red shift of the photoionization of alkali atoms through adsorption on negative salt surfaces, Z. Physik 73, 192, 1932.

22. J. H. de Boer and M. C. Teves, Secondary phenomena following the primary photoelectric effect in caesium atoms adsorbed on salt layers, Z. Physik 74, 604, 1932.

23. L. A. du Bridge and W. W. Roehr, Photoelectric and thermionic properties of palladium, Phys. Rev. 39, 99, 1932.

24. J. L. V. Eichborn, Dependence of the electron work function on the external field in atomic and compact potassium layers deposited on tungsten, dissertation, Breslau, 1932.

25. J. Erenkel, Note on the catalytic action in the photoelectric effect, Physik. Z. Sowjetunion 2, 243, 1932.

26. Th. C. Fry, Plane waves of light. III Absorption by metals, J. opt. Soc. Am. 22, 307, 1932.

27. H. Gerding and R. Gerding-Kroon, Photoelectric effect on magnesium surfaces, Rec. Trav. chim. Pays-Bas (4), 13, 51, 612, 1932.

28. H. E. Ives and H. B. Briggs, The depth of origin of photoelectrons, Phys. Rev. 40, 802, 1932.

1. J. Langmuir and J. B. Taylor, The mobility of caesium atoms adsorbed on tungsten, Ebenda 40, 463, 1932.

2. P. J. Lukirsky and S. G. Rjanoff, Abhängigkeit der lichtelektrischen Emission des Kaliums von der Anordnung von atomaren Wasserstoff-und Kaliumschichten auf ihrer Oberfläche, Z. Physik, 75, 249, 1932.

3. W. B. Nottingham, Photoelectric and thermionic emission from composite surfaces, Phys. Rev., 41, 793, 1932.

4. C. H. Prescott jr. and M. J. Kelly, The caesium-oxygen-silver photoelectric cell, Ebenda 41, 395, 1932.

5. R. Sewig, Lichtelektrische Zellen mit dünnschichtigen Alkalikathoden, Z. Physik, 76, 91, 1932.

6. R. Suhrmann and A. Schallamach, Über das Zustandekommen des spektralen selektiven Photoeffektes an dünnen Alkalimetallhäuten, Ebenda 79, 153, 1932.

7. R. Suhrmann and H. Theissing, Spektrale lichtelektrische Empfindlichkeit dünner Alkalimetallhäute bei Zimmertemperatur und bei der Temperatur der flüssigen Luft, Z. Physik, 73, 709, 1932.

8. J. Schmiegiermann, Lichtelektrischer und thermoelektrischer Effekt wasserstoffbeladener Palladium—Silber- und Palladium-Goldlegierungen, Ann. Phys. (5), 13, 761, 1932.

9. B. Abendroth, Über den Einfluss der adsorbierten Gasschicht auf die lichtelektrische Empfindlichkeit, Z. Physik, 85, 530, 1933.

10. G. Bethe, Über die chemischen und physikalischen Bedingungen der lichtelektrisch wirksamen Wasserstoffbeladung des Platins und Palladiums, Ebenda 80, 701, 1933.

11. W. H. Brattain and J. A. Becker, Thermionic and adsorption characteristics of thorium on tungsten, Phys. Rev., 43, 328, 1933.

12. R. J. Cashman and W. S. Huxford, Photoelectric sensitivity of magnesium, Ebenda 43, 881, 1933.

13. J. H. de Boer and M. C. Teves, Thermische und photoelektrische Emission von Cäsium-Cäsiumoxyd-Kathoden und deren Beeinflussung durch Einbau von Cäsiumatomen in das Dielektrikum, Z. Physik, 83, 521, 1933.

14. Déjardin, G. and R. Latarjet, Colorations superficielles des cathodes photoelectriques à l’oxyde de caesium, J. Phys. Radium, 4, 85, 1933.

15. A. Engelmann, Über Bestimmung der lichtelektrischen Grenzwellenlänge am Rhenium. Ann. Physik (5), 17, 185, 1933.

16. P. Görlich, Bemerkung zu J. H. de Boer and M. C. Teves: Thermische und photoelektrosche Emission von Cäsium—Cäsiumoxyd—Kathoden und deren Beeinflussung durch Einbau von Cäsiumatomen in das Dielektrikum, Z. Physik, 85, 128, 1933.

17. H. E. Ives and Th. C. Fry, Standing light waves, repetition of an experiment by Wiener, using a photoelectric probe surface, J. opt. Soc. Am., 23, 73, 1933.

18. W. Kluge, Empfindliche lichtelektrische Zellen für Rot und Ultrarot, Z. wiss. Photogr., 32, 142, 1933.

19. W. Kluge, Über das Auftreten und die Deutung der selektiven lichtelektrischen Elektronenemission an zusammengesetzten Alkalikathoden, Physik, Z., 34, 155, 1933.

20. W. Kluge, Die optische Absorption an lichtelektrisch zweifach selektiv wirkenden Oberflächenschichten, Ebenda 34, 465, 1933.

21. W. Kluge, Zur lichtelektrischen Sensibilisierung des Cäsiums, Ebenda 34, 844, 1933.

22. G. Lang, Der Einfluss der Materie auf langsamste Elektronen nach lichtelektrischen Untersuchungen. Ann. Physik (5), 16, 781, 1933.

23. R. Latarjet, Instabilité des cellules photoelectiques à l’oxyde de caesium, J, Physique Radium, 4, 82, 1933.

24. P. J. Lukirsky, Über die Austrittsarbeit der Elektronen und die photoelektrischen Eigenschaften der Metalle, Physic. Z. Sowjetunion 2, 212, 1933.

1. R. Suhrmann, “On the external photoelectric effect at low temperatures,” Physik. Z., 34, 877, 1933.

2. R. Suhrmann and R. Deponte, “Photoelectric investigation of the temperature dependence of the electron work function on a nickel surface covered with atomic barium,” Z. Physik, 86, 615, 1933.

3. J. B. Taylor and J. Langmiur, “The evaporation of atoms, ions and electrons from caesium films on tungsten,” Phys. Rev., 44, 423—458, 1933.

4. P. W. Timofeew and W. W. Nalimow, “Influence of oxygen and sulfur on the photoelectric effect of alkalis (K and Na),” Z. Physik, 81, 687, 1933.

5. J. P. Widmer, “Long-wave sensitivity limit of the cesium-oxide photocell,” Helv. phys. Acta, 6, 269, 1933.

6. A. M. Cravath, “The motion of electrons near a plane photoelectrode in the presence of a gas,” Phys. Rev., 45, 138, 1934.

7. R. Fleischer and P. Görlich, “On composite photocathodes,” Physik Z., 35, 289, 1934.

8. R. Suhrmann and D. Dempster, “New observations on the selective photoeffect at low temperatures,” Ebenda 35, 148, 1934.

9. R. Suhrmann and A. Schallamach, “Temperature dependence of the photoeffect of pure metal surfaces and metal surfaces covered with foreign atoms at low temperatures,” Z. Physik, 1934 (in press).

10. R. Suhrmann and D. Dempster, “The selective photoelectric effect of composite surface layers at low temperatures,” Ebenda 1934 (in press).

11. R. Suhrmann and H. Gesch, “The electrical polarization of hydrogen adsorbed on pure metal surfaces and its influence on hydrogen recombination,” Z. physik, Chem., 1934 (in press).

1. Müller—Pouillet, Lehrbuch d. Physik, vol. 4, part 4, p. 228. ↩