XL VIII. The Heat liberated during the Absorption of Electrons by Different Metals. By 0. W. Richardson, Professor of Physics, and H. L. Cooke, Assistant Professor of Physics, Princeton University [Communicated by the Authors]. IN a recent paper [citation redacted] under a similar title the authors showed that when slow moving electrons were received by a platinum strip, part of the heat developed was independent of the kinetic energy of the electrons. This production of heat was explained as the thermal equivalent of the difference in the potential energy of the electrons inside and outside the metal. The present paper describes the results of similar [header] 405 observations on strips of the following metals and alloys : — gold, nickel, copper, silver, palladium, aluminium, phosphor bronze, and iron. The method of experimenting was practically identical with that already described (loc. cit.). The only important change made was in the method of balancing the disturbing effect of the thermionic current. This was made simpler, both in theory and practice, by the following arrangement. The resistance R 3 (fig. 1, loc. cit.) was replaced by a constantan wire of resistance 4*7 ohms wound on a circular drum and placed in an oil-bath. A movable contact-maker enabled the thermionic current to be tapped off at any point of the wire. The contact-maker w T as connected directly to the point E in the figure referred to and the resistances R 4 and R 5 were done away with. From the principles of the method of balancing the thermionic current already described,, it is clear that if it is tapped off at the right point of the resistance R 3 it will produce no effect on the galvanometer G. To balance the effect of the thermionic current it was also necessary first to oppose the two batteries Ci and C 2 so as to destroy the electromotive force in the Wheatstone's bridge circuit, and rotate the contact-arm until the galvanometer spot w T as in the same position with the thermionic current " on " or " off," The reliability of the method was tested by dummy experiments, using the electromotive force from a battery in a manner precisely similar to that used in testing the original method, and it was found quite satisfactory. Effect of Pressure In the former paper it was suggested that the results- might perhaps be affected by the pressure of the residual gas present in the apparatus. At that time experiments were made to test the point. So far as they went they seemed to indicate that, if the pressures, such as occurred, exerted an influence on the results, it was an unimportant one. It was felt, however, that these experiments were not very satisfactory; so fresh experiments have been made under better conditions. In these experiments an iron grid was used, and the osmium filaments had been heated continuously for a long time. The conditions were generally very steady. On account of the continuous evolution of gas by the hot metal, it was necessary so use a continuous pump in these experiments. This made it somewhat difficult to control the pressure. The obvious way of producing a desired change in the pressure is to vary the speed of the pump. This was 406 [header] found to be no good with the Gaede pump used, as, in order to obtain a pressure appreciably higher than that given by the pump when working at its most efficient speed, it was necessary to work it so slowly that the changes in the temperature of the apparatus caused by the periodic variations of pressure entirely precluded any attempt at measuring the effect. Several methods were tried, and that finally adopted consisted in connecting the apparatus to a side tube provided with an indiarubber joint which leaked slightly. The connexion was made through a good ground-glass cock. In this way two different pressures were available, namely, the limiting values obtained when the side tube was, and was not, connected with the apparatus. These pressures were about 2 X 10~ 3 and 5 x 10 ~ 3 mm. respectively, and were quite free from the slow periodic variations which had previously been found so objectionable. With this arrangement several experiments were made, of which the following are typical examples. With 8 volts applied potential-difference, and the side tube shut off, the pressure both before and after one set of observations was 20 X 10 -4 mm. The effect in scale-divisions per unit thermionic current under these conditions was found to be 1*680. After connecting the side tube to the apparatus the pressure at the beginning of an experiment with 8 volts applied potential-difference was 56 x 10~ 4 mm., and at the end 46 x 10~ 4 mm. The effect in this case, in the same units as at the lower pressure, was 1*683. Thus changing the pressure from 20 x 10~ 4 mm. to 51 x 10 -4 mm. produces no change in the effect per unit thermionic current, The same was true at other voltages. Thus with 18 volts the pressure was 20 x 10~ 4 mm. at the beginning of an experiment with the side tube shut off and 19 x 10~ 4 mm. at the end. The effect per unit thermionic current was 3'01 scale-divisions. After connecting, the pressure was 46 x 10~ 4 mm. at the beginning and 42 X 10~ 4 mm. at the end of an experiment under the same potential-difference. In this case the effect was found to be 2*98 in the same units as before. This is identical with 3*01 within the limits of observational error. These experiments prove conclusively that the measurements are quite unaffected by small fluctuations in the pressure of the surrounding gas ; so that such irregularities as have been found must be attributed to other causes. Experiments with the Different Metals. A general review of the results which have been obtained shows that they are much more inconsistent than those yielded [header] 407 by the former experiments on platinum. We believe that we have discovered the main cause of this inconsistency but, so far, unfortunately, we have not succeeded in regulating it. The cause seems to lie in an instability in the thermionic emission of osmium itself, and we are investigating the phenomenon in detail in the hope of being able to control it. The experiments which follow were carried out, so far as we are able to judge, in the same way and under the same conditions as those previously made with platinum. In view of the considerable range of the value of the effect for each one metal, and also of the peculiar effects first observed when experimenting with iron (see below), we shall only give the final numbers obtained in each case and shall omit the details of the measurements which led to them. In every case, the strips used were of the purest specimens of the metal obtainable. The gold, silver, palladium, copper, and nickel were obtained as pure from Messrs. Johnson, Matthey & Co., London. This specimen of copper was compared with one rolled from commercial magnet wire, and did not exhibit any notable difference. The aluminium was rolled from commercial aluminium wire and then cut by hand, and the phosphor bronze was a strip such as is used for galvanometer suspensions. As a rule, a considerable number of experiments were made on each material. The final results of all those which appeared to be satisfactory are given in the following table :— Metal. Corrected Values of [phi]. Volts. Mean. Volts. Weighted mean. Volts [table redacted] 408 [header] The weighted means were judged by inspection of the experimental points. It is questionable whether they are much more reliable than the others, as it is our opinion that the experiments are affected by causes, which we are not able to control, which may remain constant during a single set of experiments. In the case of silver and aluminium the results of the experiments seemed less trustworthy than in the other cases. Often it was impossible to get the same value twice, for the same thermionic current and the same potential-difference, owing to some change with time which was going on. Moreover, in several experiments with these metals the heating effect did not turn out to be a linear function of the applied potential-difference. This was probably due to some parts of the grid being different from others, and the heating current moving from one part to the other as the potential-difference was changed. Probably most of the difference from one part of the grid to another was due to the effect on it of the heating and sputtering. This suggestion is supported by the appearance of the grids. The aluminium ones, after they had been used a few times, were quite changed in texture and were so much altered that they crumbled to pieces as they were unwound. They were also very badly discoloured. The experiments made with iron grids led to a discovery which we think likely to account for a considerable part of the irregularities which have been noted with all the metals. It seems that under the conditions of the experiments on iron, and probably on the other metals, there is a kind of instability in the thermionic emission of the osmium filaments. The nature of this instability is best described by considering the way the thermionic emission changes as the temperature of the filament is altered. Starting at a comparatively low temperature, the emission increases rapidly with increase of temperature following the usual inverse exponential law. This goes on until temperatures of a certain value are reached, when the current shows a tendency to sag off. If the temperature is now raised and maintained at a certain value, there is a sudden drop in the thermionic emission. In favourable cases the new current may be as little as one-thirtieth of what it previously was at the same temperature. When the temperature is raised further the current is found to be stable again, and increases with temperature according to an inverse exponential law. The behaviour of the emission as the temperature is lowered is the reverse of what has been described. We have first a quick but regular decrease in current following the inverse [header] 409 exponential law. At a certain stage there is a sudden increase in the emission. This is followed by a third region in which the current again diminishes with decreasing temperature according to the regular law. There are thus two ranges or! temperature in which the thermionic current is stable. These are separated by a region of instability. We shall refer to the two' stable ranges as the low-temperature range and the high-temperature range respectively. The heating effect in the case of iron has been examined, both when the osmium was on the low-temperature range and also when it was on the high-temperature range. The numbers are given in the last two rows of the table. The mean of the six values for the low-temperature range is 4*88 volts, and of the four values for the high-temperature range 6*72 volts. There is thus a difference of nearly two volts in the effect given by the two ranges. We next attempted to make experiments upon the other metals under such conditions that we knew whether the osmium was on the low- or the high-temperature range. At the outset we obtained values for platinum in the neighbourhood of 7 volts as against the value 5*5 volts obtained in our previous investigation. We believed at the time that we were working on the high-temperature range ; but that particular filament burnt out, and later on we found it impossible to get a filament which would develop the two ranges. This situation compelled us to desist from the direct line of attack for the moment ; as it is necessary to make a more thorough examination of the thermionic properties of osmium, in order to be able to control the conditions which determine its thermionic emission. We hope to be able to report on this matter at an early date. If we return for a moment to the table on page 407 it is a striking fact that four of the mean values, viz. : gold, 7'26, copper 7*1, aluminium 7 '4, and iron (high) 6*72, are equal, within the range of experimental error, to a common mean value 7*11 ; whilst the other five, viz.: nickel 5'3, phosphor bronze 5*8, palladium 5*6, silver 5*15, and iron (low) 4*9, are equal within the same limits to the common mean value 5'35. It looks as though the values for gold, copper, and aluminium had all been obtained with the osmium on the high range, and the values for nickel, phosphor bronze, palladium, and silver with the osmium on the low range. It would follow from this that the value of the heating effect is independent of the nature of the metal which receives the electrons, being determined almost entirely by the metal which emits them. [footer] 410 [header] For the reasons which have been stated we have found it impossible to test this question directly, up to the present. The present theory o£ these phenomena requires that the heating effect should be practically independent of the metal receiving the electrons, as one of the authors has already pointed out [citation redacted]. It is satisfactory to know that our experimental results, so far as they go, do not conflict with the theory. In our previous paper we omitted to take into account the possibility of the existence of an electric field between the osmium and the platinum arising from an intrinsic electromotive force ; and this omission led us to identify the heating effect with the work done when the electrons escape from hot platinum. The identification should be with the work done when the electrons escape from hot osmium. This quantity has not been measured yet, but we propose to measure it in the course of our investigation of the thermionic properties of osmium already referred to. The most important facts which we have so far established are : — (1) The heating effect due to the difference of the potential energy of an electron inside and outside of a conductor, which we previously established for platinum, occurs in the other metals. (2) The effect is of the same order of magnitude in all cases, the measured values ranging from about 4'5 to 7-5 volts. (3) The values are influenced very considerably by the nature and state of the thermionic emitter. (The experiments do not preclude the possibility that the true effect is almost independent of the metal receiving the electrons.) (4) The measured effect is not influenced by changes in the pressure of the residual gas in the apparatus, provided this be reasonably low. (5) Under certain conditions, not yet completely determined, the thermionic emission from osmium becomes unstable ; and there are two ranges of stability, one at low and the other at high temperatures. We are glad to take this opportunity of thanking Messrs. Baldwin, Carter, Critchlow, Ferger, Frederick, and Gibbs for again assisting in taking the very numerous observations. Palmer Physical Laboratory, Princeton, N.J.