November 17, 192 1] 



NATURE 



375 



the foregoing argument. We do not need to hesi- 

 tate between these alternatives, and we need only 

 look to see how the alleged behaviour of A and B 

 will need to be modified in order that no useful 

 work may appear. There are two alternatives. 

 Either A. and B necessarily emit equal numbers of 

 electrons at all temperatures, or the charges which 

 develop owing to the unequal rate of emission are 

 not discharged, even to the slightest degree, when 

 the two bodies are placed in contact. 



The first alternative is definitely excluded by 

 the experimental evidence, so I shall proceed to 

 interpret the second. It means that bodies have 

 natural states of electrification whereby they 

 become charged to definite potential differences 

 whose magnitudes are independent of their rela- 

 tive positions. There is an intrinsic potential 

 difference between A and B, which is the same, 

 at a given temperature, whether they are at a 

 distance apart or in contact. 



Admitting that the intrinsic potentials exist, a 

 straightforward calculation shows that they are 

 intimately connected with the magnitudes of the 

 thermionic emission at a given temperature. The 

 relation is. in fact, governed by the following 

 equation. If A and B denote the saturation ther- 

 mionic currents per unit area of the bodies A and 

 B respectively, and V is the contact potential 

 difference between them at the absolute tempera- 

 ture T, then V = feT/e log A/B, where fe is the gas 

 constant calculated for a single molecule (Boltz- 

 mann's constant), and e is the electronic charge. 



I have recently, with the help of Mr. F. S. 

 Robertson, obtained a good deal of new informa- 

 tion on this question from the experimental side. 

 We have made measurements of the contact poten- 

 tial difference between heated filaments and a sur- 

 rounding metallic cylinder, both under the high- 

 vacuum and gas-free conditions which are now 

 attainable in such apparatus, and also when small 

 known pressures of pure hydrogen are present. 

 As is well known, both contact potentials and 

 thermionic emission are very susceptible to minute 

 traces of gas, but we find that under the best con- 

 ditions as to freedom from gas there is a contact 

 potential of the order of one volt between a pure 

 tungsten filament and a thoriated filament. We 

 also find that changes of a similar magnitude in 

 the contact potential difference between a thoriated 

 tungsten filament and a copper anode take place 

 when the filament is heated. These changes are 

 accompanied by simultaneous changes in the 

 thermionic currents from the filament, and w^e 

 find that the change in the contact potential cal- 

 culated from the change in the currents with the 

 help of the foregoing equation is within about 

 20 per cent, of the measured value. Considering 

 the experimental difficulties, this is a very sub- 

 stantial agreement. Whilst the evidence is not 

 yet as complete as I hope to make it. it goes a 

 long way towards disproving the chemical view 

 of the origin of contact potential difference. 



From Avhat has been said you will realise that 

 the connection between contact potentials and 

 NO. 2716, VOL. 108] 



thermionic emissions is a very close one. I would, 

 however, like to spend a moment in developing it 

 from another angle. To account for the facts of 

 thermionic emission it is necessary to assume that 

 the potential energy of an electron in the space 

 just outside the emitter is greater than that inside 

 by a definite amount, which we may call iv. The 

 existence of this w, which measures the work done 

 when an electron escapes from the emitter, is 

 required by the electron-atomic structure of matter 

 and of electricity. Its value can be deduced from 

 the temperature variation of thermionic emission, 

 and, more directly, from the latent heats absorbed 

 or generated when electrons flow out of or into 

 matter. These three methods give values of 10 

 which, allowing for the somewhat considerable 

 experimental difficulties, are in fair agreement for 

 any particular emitter. The data also show that 

 in general different substances have different 

 values of ihk This being so, it is clear that when 

 uncharged bodies are placed in contact the poten- 

 tial energies of the electrons in one will in general 

 be different from those of the electrons in the 

 other. If, as in the case of the metals, the elec- 

 trons are able to move freely they will so move 

 until an electric field is set up which equihbrates 

 this difference of potential energy. There will 

 thus be an intrinsic or contact difference of poten- 

 tial between metals which is equivalent to the 

 difference in the values of w and is equal to the 

 difference in lu divided by the electronic charge.^ 



Photo-electric Action. ' 



We have seen that there is a connection on 

 broad lines between thermionic emission and both 

 contact potentials on one hand and photo- 

 electric emission on the other. The three groups 

 of phenomena are also related in detail and to an 

 extent which up to the present has not been com- 

 pletely explored. In order to understand the 

 present position, let us review briefly some of the 

 laws of photo-electric action as they have revealed 

 themselves by experiments on the electrons 

 emitted from metals when illuminated by visible 

 and ultra-violet light. 



Perhaps the most striking feature of photoelec- 

 tric action is the existence of what has been called 

 the threshold frequency. For each metal whose 

 surface is in a definite state there is a definite 

 frequency n^, which may be said to determine the 

 entire photo-electric behaviour of the metal. The 

 basic property of the threshold frequency n^ is 

 this : When the metal is illuminated by light of 

 frequency less than tIq no electrons are emitted, 

 no matter how intense the light may be. On 

 the other hand, illumination by the most feeble 

 light of frequency greater than n^ causes some 

 emission. The frequency «„ signalises a sharp 

 and absolute discontinuity in the phenomena. 



Now let us inquire as to the kinetic energy of 

 the electrons which are emitted by a metal \vhen 

 illuminated by monochromatic light of frequency, 



' This statement is only approximately true. In order to condense the 

 argument certain small effects connected with the Peltier effect at the 

 junction between the metals have been left out of consideration. 



