Positive Ionisation Produced by Platinum tvhen Heated .
117 and without immediate change of spectrum , when the current ceases .
Sometimes , however , a change of spectrum eventually results from unequal decay of the lines .
These effects , which last less than 1/ 1000 of a second , are distinct , not only in degree , but in kind from the afterglows in nitrogen , in gaseous mixtures containing oxygen , and in gaseous mixtures containing hydrogen , previously investigated .
These have durations extending in some cases to several minutes , and their spectra are from the first radically different from those of the exciting discharges .
As previously shown , they are due to secondary chemical actions of substances produced by the discharge .
The Positive Ionisation Produced by Platinum and by Certain Salts when Heated .
By Frank Horton , M.A. , D.Sc .
, Fellow of St. John 's College , Cambridge .
( Communicated by Prof. Sir J. J. Thomson , O.M. , F.R.S. Received November 28 , 1912 , \#151 ; Read January 16 , 1913 .
) In the 'Proceedings of the Royal Society ' for 1910 , * there is an account of a spectroscopic investigation of the nature of the carriers of positive electricity when an electric current is sent from a glowing platinum strip covered with aluminium phosphate to a surrounding platinum electrode , the whole being contained in a highly evacuated vessel .
It was found that aluminium phosphate heated under these conditions evolved carbon monoxide gas , and as a molecule of this gas carrying a single electronic charge gives a value for ejm which agrees fairly well with the mean value found for the carriers of positive electricity from heated metals , it was concluded that the positive ions are charged molecules of carbon monoxide .
Hydrogen was also detected in the gas evolved by the hot electrode in this experiment , and it now seems probable that atoms of hydrogen also take part in carrying the current , for Garrett has found that about 10 per cent , of the positive ions present when aluminium phosphate is heated on a platinum strip in a vacuum have a mass corresponding to that of the hydrogen atom.f A different view of the nature of the positive ions from glowing metals and metallic salts is held by Prof. O. W. Richardson .
As the result of * Series A , vol. 84 , p. 433 .
t A. E. Garrett , ' Phil. Mag.,5 1910 , VI , vol. 20 , p. 582 .
118 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , numerous experiments by himself and his pupils , * Richardson has come to the conclusion that the positive ions emitted by heated salts are charged atoms of the metallic constituent of the salt under test , and that in the case of incandescent metals the carriers of positive electricity are charged atoms of sodium or potassium which are present as impurities in the form of salts .
It was in order to obtain further evidence of the origin of the positive ionisation from incandescent bodies that the experiments described in the present paper were undertaken .
The paper contains the results of an investigation of the effect of the surrounding gas upon the thermionic currents from different substances .
The general method of experiment was as follows:\#151 ; The positive leak from a strip of platinum foil was first carefully investigated .
Observations were made of its rate of decay with time , of its variation with the potential difference used , and with the gas pressure in the apparatus .
The strip was then covered with the salt to be tested , and the observations were repeated .
When the experiments with this salt were complete , the strip of platinum was carefully cleaned and another salt substituted .
In this way observations were made of the thermionic emissions under exactly similar conditions from platinum , from four samples of aluminium phosphate , from two samples of sodium ortho-phosphate , and from one sample of sodium pyro-phosphate .
Comparisons wTere made , in particular , at the two temperatures 1080 ' and 1190 ' C. In an earlier paper , f in which the discharge of positive electricity from sodium phosphate heated in different gases was examined , 800 ' C. was used as the temperature of comparison , and reasons were given why it is best to work at as low a temperature as possible .
In the present experiments , however , it was found in some cases that even at 1080 ' C. the thermionic current became so small on long-continued heating as to be hardly measurable with the galvanometer used , and it was on this account that the comparisons were finally made at 1190 ' C. These earlier experiments with phosphates of sodiumf and lithium^ have shown that the nature of the surrounding gas has a considerable effect on the magnitude of the thermionic current .
In the present experiments the gas in the apparatus was^usually air , but in one instance , with a sample of pure aluminium phosphate , some experiments were made in carbon monoxide .
The positive leak in this gas was found to be larger than that under the same conditions in air , but in other respects * O. W. Richardson , 4 Phil. Trans. , ' 1906 , A , vol. 207 , p. 1 ; 4 Phil. Mag. , ' 1908 , VI , vol. 16 , p. 740 ; O. W. Richardson and E. R. Hulbirt , 'Phil .
Mag. , ' 1910 , VI , vol. 20 , p. 545 ; O. W. Richardson , 'Phil .
Mag. , ' 1910 , VI , vol. 20 , p. 981 ; 'Phil .
Mag.,5 1911 , VI , vol. 22 , p. 669 ; C. J. Davisson , 'Phil .
Mag. , ' 1912 , VI , vol. 23 , p. 121 .
t 'Camb .
Phil. Soc. Proc.,5 1911 , vol. 16 , p. 89 .
J 'Camb .
Phil. Soc. Proc. , ' 1911 , vol. 16 , p. 318 .
1912 .
] by Platinum and by Certain Salts ivlien Heated .
119 similar results were obtained with the two gases .
It was therefore thought undesirable to prolong the research by repeating experiments with different gases in the case of the other substances .
A description of the apparatus used and an account of the results obtained will now be given .
Description of the Apparatus .
The discharge tube used in these experiments is represented in fig. 1 .
It is made of glass , and consists of two parts which fit together by a ground joint with a mercury seal .
The platinum strip A can thus easily be removed for covering with the salt to be experimented on .
When in position it is situated vertically in the centre of the tube between two parallel platinum plates P , P , which together form the negative electrode .
The strip A is of thin foil 2 cm .
long and 2 mm. wide .
It can be heated electrically by means of the thick platinum leads B and C , the current being supplied from eight accumulator cells , and regulated by wire resistances .
The temperature of the central portion of the strip is obtained by a thermocouple consisting of two fine wires , one of pure platinum and the other of platinum with 10 per cent , of rhodium .
These wires are 0*0025 cm .
in diameter , and are fused together at one end , the junction being lightly welded on to the centre of one side of the platinum strip as indicated in the diagram .
The wires are insulated for the greater part of their length by fine glass tubes , and are sealed through the lower portion of the discharge tube at E. The thermo-electric current is measured by a d'Arsonval galvanometer , and the deflections are standardised from observations of the melting point of pure potassium sulphate , making use of the curve given by Callendar* for this purpose .
The welding of the thermojunction on to the platinum strip is an operation requiring great care .
A very small junction has to be made , and this must be welded to the strip * Callendar , ' Phil. Mag. , ' vol. 48 , p. 519 .
Fig. 1 .
A , platinum strip forming the anode ; B , C , leads of heating circuit ; E , wires of thermocouple ; P , P , platinum plates forming the cathode .
120 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , without any part of the fine wires becoming affixed , otherwise they may tap off a portion of the E.M.F. of the heating circuit and thus invalidate the temperature observations .
Before standardising the thermo-couple used in these experiments , it was ascertained that the galvanometer used to indicate the temperature gave the same deflection when the direction of the heating current was reversed .
The two platinum plates P , P , are made of stout metal , and are 2*5 cm .
long by 1 cm .
wide .
They are connected together by a wire outside the discharge tube .
The plates were about 1 cm .
apart , and were at equal distances from the anode A. The thick platinum wires B and C of the heating circuit were covered with glass in order to make them more rigid and to keep them cool .
The whole apparatus was very carefully cleaned out with hot chromic acid and with boiling nitric acid , and then washed with distilled water and dried before being fixed on to the mercury pump .
A drying tube with phosphorus pentoxide was between the discharge tube and the pump , and the latter was also connected to a McLeod gauge for registering the gas pressure .
In order to facilitate the gradual alteration of the pressure in the apparatus , there was in connection with it a large vertical glass tube 5 cm .
wide and 50 cm .
long , which could be filled with mercury by raising a cistern of mercury attached by indiarubber tubing to its lower end .
For obtaining a thermionic current the platinum strip A was connected to the positive pole of a set of small accumulator cells .
The platinum plates P were connected to the negative pole of the battery through a fine tinfoil fuse and a sensitive d'Arsonval galvanometer .
The galvanometer was well insulated on paraffin blocks ; it gave a deflection of one scale division for a current of 3*68 x 10"10 ampere , and could be shunted for measuring larger currents .
The Experiments with Platinum .
Before being finally fitted up for experimenting , the platinum strip was boiled in strong nitric acid and then carefully washed with hot water .
The first observation of the thermionic current was made with a potential difference of 40 volts , and an air pressure of 3*42 mm. in the apparatus .
As the temperature of the platinum strip was gradually raised a current was first detected at 700 ' C. ; at higher temperatures the current was larger , but it decreased very rapidly with time in the manner which has already been described by Richardson.* A comparison of the rates of decay of the positive emission from platinum , and from the salts experimented on , will be given in a later part of this paper , but it will be convenient to point out here that * O. W. Richardson , ' Phil. Mag. , ' 1903 , VI , vol. 6 , p. 80 .
1912 .
] by Platinum and by Certain Salts when Heated .
121 although the current under given conditions of voltage and gas pressure became roughly constant after about five or six hours ' heating , it continued to decrease slowly throughout the series of experiments with the platinum anode , and if the heating had been continued for some hours longer 1 believe the positive leak would have been too small to measure on the galvanometer used , It should also be mentioned that after leaving the platinum strip cold overnight , a much larger current was obtained when first testing on the following morning .
This effect was most noticeable when the gas pressure in the apparatus was very low , and I thought it was possibly due to phosphorus vapour coming over from the phophorus pentoxide drying tube , for commercial phosphorus pentoxide often contains free phosphorus .
I investigated this point by having a mercury barometer tube so arranged that by raising the cistern I could cut off the connection between the discharge tube and the rest of the apparatus .
Doing this , I found that the largely increased leak was still obtained on allowing the strip to remain cold for several hours , although there was no possibility of it having come in contact with phosphorus vapour .
This effect is exactly similar to that which was obtained when investigating the negative leak from platinum at higher temperatures , * and I believe it is due to the mercury vapour from the pump forming some compound with the platinum .
The abnormally high value of the thermionic current only lasted for a short time ; in a few minutes the leak had decreased to a steady value , usually somewhat smaller than it had on the previous evening .
This decrease in the leak was accompanied by a slight increase in the electrical resistance of the strip , which suggests that some volatile product was subliming from its surface .
Even at the highest temperature of these experiments no negative leak could be detected while this process was going on .
When the thermionic current had been reduced to its nearly constant value , a series of observations of the relation between the current and the applied E.M.F. were made .
A typical curve showing this relation is given in fig. 2 .
This was obtained with the anode at 1190 ' C. , and with a pressure of 0*01 mm. * 4 Phil. Trans.,5 1907 , A , vol. 207 , p. 149 .
122 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , of air in the apparatus .
The values of the current obtained when the potential difference was being gradually increased are marked \#169 ; , and those obtained on decreasing the voltage are marked x .
These fall slightly below the first set\#151 ; a result which was obtained with each of the anodes used in these experiments .
The difference in the two readings is not much in the case of platinum , but when this is covered with aluminium phosphate a larger hysteresis effect is obtained .
This will be referred to again when the experiments with aluminium phosphate are being considered .
The shape of the curve in fig. 2 is similar to that given by Eichardson* for platinum at 976 ' C. in air at atmospheric pressure .
It will be seen that up to 320 volts the current is never completely saturated , but that after about 120 volts it increases approximately proportionally to the voltage .
At higher pressures the linear part of the curve did not begin until a somewhat larger E.M.F. was reached , indicating that it was more difficult to saturate the current under these conditions .
In my earlier experiments with sodium phosphate , which were performed with an apparatus of similar construction , and of about the same dimensions , as that used in the present research , fairly complete saturation was obtained with 40 volts potential difference , even with a pressure of 2 cm .
of air in the discharge tube .
This difference is no doubt due to the different temperatures used in the two cases .
The sodium phosphate anode in the earlier research was only heated to 800 ' C. , whereas the curve of fig. 2 corresponds to a temperature of 1190 ' C. In his experiments with platinum , already referred to , Eichardson also found that it was more difficult to saturate the current at high temperatures , and he states that this may possibly be due to the relatively greater magnitude of the negative ionisation , which would make recombination a factor to be reckoned with .
In the present experiments I could detect no negative emission at 1190 ' C. , and I think the difficulty in saturating the current may be due to the presence of slowly-moving ions at this high temperature .
Eutherford , f when measuring the velocity of the positive ions from a sheet of platinum heated electrically in air at atmospheric pressure , found that slowly-moving ions were formed in increasing numbers as the temperature of the platinum was raised .
These ions probably owe their origin to the dust which is produced by the disintegration of hot platinum , and may be formed by gas ions sticking to the dust particles .
With each of the anodes used in the present apparatus it was found that when the applied potential difference had been raised to 200 volts , the thermionic current was well within the stage when the increase with * O. W. Eichardson , 'Phil .
Trans./ 1906 , A , vol. 207 , p. 40 .
t E. Butherford , ' Pliys .
Eev .
, ' 1901 , vol. 13 , p. 321 .
1912 .
] by Platinum and by Certain Salts when Heated .
increasing voltage is small and uniform , and this voltage was used for the purpose of comparing the various thermionic emissions .
In the experiments with different gas-pressures in the apparatus , it was found that the form of the current-pressure curve was the same whether a potential difference of 40 volts or of 200 volts was used .
This was tested in several cases in order to make quite sure that the applied voltage was not an important factor in the results obtained .
Before the series of observations of the variation of the positive emission from the hot platinum strip with the pressure of the residual air in the apparatus were begun , a test was made to see whether the same value of the current could be obtained if the platinum strip was allowed to cool down for a few minutes , and was then reheated to the original temperature .
It was found that after reheating for a few minutes the original steady value of the current was obtained .
In taking a series of observations at different pressures it was usual to begin at the highest pressure , and gradually pump the air out of the apparatus .
After each observation of the thermionic current , while the pressure was being reduced , the heating current was cut off , and the platinum strip was allowed to cool down .
On reheating at the lower pressure , a few minutes were allowed for the current to become steady before the new reading of the galvanometer deflection was taken .
By cutting off the heating current during the alteration of pressure one is able to guard against accidentally overheating the anode\#151 ; a most important precaution , for if the temperature is raised too high , the thermionic current at the lower temperature is at first larger than the normal value , and it takes some time before the steady state is obtained .
In this way observations of the thermionic current under different potential differences were made at pressures between about 60 mm. and a good vacuum , the temperature of the anode being in the first experiments 1080 ' C. , and in the later ones 1190 ' C. It was found that there was very little difference in the magnitude of the current at pressures between 60 mm. and 30 mm. The relation between the current and the pressure at lower pressures wilWoe seen from the curve of fig. 3 , which was obtained with the anode at 1190 ' C. , and with a potential difference of 160 volts .
Pressure : mm. Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , The current decreases steadily with the pressure until a pressure of about 1 mm. is reached , after which there is a sharp increase in the current as the pressure is still further reduced .
At pressures below 0*03 mm. the current was always rather irregular , but usually tended to decrease as the pressure was lowered .
These irregularities are possibly due to an inaccurate knowledge of the pressure in the discharge tube containing the glowing platinum , which is continually evolving gas .
At low pressures this evolution may conceivably cause the gas-pressure in the discharge tube to be much greater than that registered on the McLeod gauge .
Another difficulty in making observations at very low pressures is the mercury vapour which comes over from the pump , and which at the ordinary temperature of the laboratory has a pressure of about 0*001 mm. , * so that observations cannot be taken at pressures lower than this unless arrangements are made to prevent mercury vapour from entering the discharge tube .
It seems not improbable that the rapid increase in the positive leak with diminishing pressures below about 0*5 mm. is due to chemical action between the mercury vapour and the hot platinum , and it would be interesting to repeat the observations , taking care to exclude mercury from the discharge tube .
It should be mentioned that observations of the thermionic current made at gradually increasing pressures gave curves which were similar to the curve of fig. 3 .
The Experiments with Aluminium Phosphate .
Sir J. J. Thomson found that aluminium phosphate heated in air at atmospheric pressure gave rise to a much larger emission of positive ions than any other salt of the large number he tested under similar conditions.* ) In the present research four samples of aluminium phosphate wrere tested , and for convenience these will be numbered in the order in which they were used .
Sample No. II was from the same bottle as that tested by Sir J. J. Thomson ; it was the ordinary commercial pure salt .
Sample IY was also sold as pure by the chemist from whom it was obtained .
The spectroscopic test showed that each of these specimens contained sodium .
Richardson holds that the positive emission from commercial aluminium phosphate\#151 ; which he did not find to be remarkably large\#151 ; is due to the presence of alkaline impurities .
In the 'Philosophical Magazine ' for November , 1911 , he has described some experiments made with aluminium phosphate prepared from materials all of which had undergone distillation , hoping in this way to obtain the salt quite free from the alkali metals .
It * T. H. Laby , ' Phil. Mag./ 1908 , VI , vol. 16 , p. 789 .
t J. J. Thomson , ' Camb .
Phil. Soc. Proc.,5 1907 , vol. 14 , p. 105 .
1912 .
] by Platinum and by Certain Salts when Heated .
125 was found that with this pure phosphate the positive emission , even at 1050 ' C. , was very small , and Richardson therefore attributes the large emission from the commercial phosphate to the presence of alkaline impurities .
Sir J. J. Thomson found that of all the classes of salts he experimented with the phosphates gave by far the largest emission of positive ions , which suggests that the emission is increased by the presence of phosphorus in the salt , and does not depend mainly upon the metal which happens to be combined with it .
The importance of this point is so great that I decided to test some aluminium phosphate prepared in the same way as that used by Richardson .
I am indebted to Mr. F. R. Ennos , B.A. , and to Mr. W. W. P. Pittom , B.A. , who prepared the salt for me in the Chemical Laboratory of this University .
The precipitated phosphate , after washing with boiling distilled water , was divided into two parts , one of which was kept in a moist condition and the other was dried in an air oven .
The former of these was the first sample of aluminium phosphate tested in these experiments , the dried portion was Sample No. III .
It was found that each of these samples of pure aluminium phosphate gave an initial positive emission , which was considerably smaller than that given by either of the impure specimens .
This confirms the observations of Richardson .
It was also found , however , that after long-continued heating there was not much difference between the emissions from the different specimens .
In preparing the first aluminium phosphate anode tested , the platinum strip was covered on both sides with some of the pure sample which had been preserved in a moist condition .
The strip was then gently heated in the air by means of an electric current , and the phosphate was thus dried and formed a thin uniform layer over the surface of the platinum .
Anodes of the other samples were prepared in a similar manner ; the platinum strip was first cleaned by careful scraping and washing , it was then covered with an emulsion made by mixing some of the new specimen with distilled water , and dried by gentle heating .
In all cases when a new anode was fitted into the discharge tube , the air-pressure was reduced to a few millimetres , and the apparatus was left overnight in order that any moisture present might be absorbed by the drying tube .
With each of the aluminium phosphate anodes the first observations which were made were those of the rate of decay of the thermionic emission with time .
The temperature of the anode was usually 1080 ' C. , the potential difference was either 40 or 200 volts , the gas pressure in the apparatus was about 16 mm. With Samples I , III , and IY the current decreased continuously from the commencement of the experiment , but with 126 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , Sample II\#151 ; the impure aluminium phosphate used by Sir J. J. Thomson in his experiments\#151 ; the current steadily increased during the first half hour .
It then gradually decreased for over an hour , when some sudden disturbance seemed to take place and in the next 20 minutes the current rose rapidly to its highest value , after which it decreased continuously to a fairly steady value in about three and a half hours .
The rise in the current on first heating and the subsequent decay to a steady value is in agreement with the results obtained by Garrett , but this effect was not obtained with the other impure specimen tested in these experiments ( Sample IY ) .
This point will be referred to again later in the paper .
In all cases after the first few hours ' heating there was a gradual diminution in the emission throughout the experiments , but on leaving the apparatus overnight the current on the following morning was usually larger than when last tested .
If the air pressure were several millimetres , the current at first would be about two or three times the normal value , and it would decrease to a steady value in about an hour .
If an air pressure of only a small fraction of a millimetre had been left in the apparatus overnight , the effect already referred to with platinum was obtained , and the initial current on the following morning was perhaps 100 times the normal value .
This large current always decreased very rapidly during the first few seconds of heating , afterwards falling off gradually to a steady value as in the case when the pressure was higher .
It must be understood that the " steady " values were never quite constant ; the current was always slowly decreasing as time went on , but when a " steady " value had been reached there was no appreciable alteration in the galvanometer deflection in the course of 5 or 10 minutes .
With both of the specimens of impure aluminium phosphate a curious effect was observed on first testing the emission after the anode had been left for some time in air at a pressure of several millimetres .
Under these circumstances it was found that the initial current gradually increased for a few minutes before the usual decay to a steady value began .
This effect was only noticed on one or two occasions when the pure salt was being tested ; but , as will be seen later , it was also found to occur when a sodium phosphate anode was being used .
When , after reheating an anode , a steady state of the emission had been obtained , the thermionic current was always slightly less than the steady value when last tested .
It was therefore decided to test whether the current would decrease to a further extent if the anode were allowed to remain cold for several days .
This was tested with Samples III and IV , and it was found that a very considerable diminution in the steady emission was produced by leaving the anode cold for two or three weeks with the air in the apparatus at a few millimetres pressure .
It thus 1912 .
] by Platinum and by Certain Salts when Heated .
appears that the change which causes the decay of the positive leak with time goes on even in the cold , though at a very much slower rate than when the anode is hot .
It seems probable that this change is the evolution of gas by the anode , which one would expect to take place most rapidly at high temperatures .
With all four samples of aluminium phosphate the variation of the thermionic current with the potential difference was studied at different temperatures and with different gas pressures in the apparatus .
The curves given in fig. 4 will serve to illustrate the results obtained .
The points marked \#169 ; were obtained when the voltage was being gradually increased , those marked x were obtained afterwards , on decreasing the voltage .
It will be seen that the currents under an increasing voltage were rather larger than those obtained under the corresponding decreasing voltage .
There are two causes which tend to produce this effect .
The first of these is the slow but gradual diminution of the emission with time , but that this does not account for the whole of the effect was proved by commencing the series with the highest E.M.F. , gradually reducing it to zero , and then gradually raising it again to the initial value .
Taking the observations in this order , the values of the current at increasing voltages were still much larger than the corresponding values with decreasing voltages , although the decay of the leak with time tends to equalise the values in this case .
The second cause , and the one which seems to be mainly responsible for the hysteresis , is the fact that a sudden increase in the applied potential difference gives rise to a current which is at first larger than the normal value for that voltage , and which gradually decreases to the normal value in the course of 5 or 10 minutes .
If the voltage is suddenly decreased the opposite effect takes place , the current to begin with is too small and gradually increases .
A hysteresis effect similar to this was obtained by Pdchardson in his experiments with platinum* In making the observations for a current-E.M.F .
curve the readings have to be taken without much delay , firstly on account of the time-lag effect , secondly , because of the difficulty of maintaining the * O. W. Eichardson , 'Phil .
Trans. , ' 1906 , A , vol. 207 , p. 10 .
128 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , temperature constant for a long time .
At each change of the applied potential difference the new voltage was left on for about 2 minutes before the corresponding reading of the current was taken .
The indications of the galvanometer showed that in this time a great part of the effect due to the sudden change of voltage had subsided , but it is probable that the current had not quite settled down to the steady value , and this causes each of the readings at increasing voltages to be slightly too high , and those at decreasing voltages to be slightly too low , thus accounting for the difference in the curves obtained under the two conditions .
The curves in fig. 4 were obtained after the anode ( Sample I ) had been heated for several days at 1190 ' C. , and the emission had become very small .
The air pressure in the apparatus was 0*035 mm. ; at higher pressures the curves rose more gradually , showing that a greater E.M.F. was required to produce approximate saturation of the current .
The relation between the thermionic current from aluminium phosphate and the air pressure in the apparatus has been investigated by Garrett , * who found that as the pressure was lowered from about 150 mm. the positive emission increased to a maximum value at somewhere between 15 mm. and 5 mm. , according to the temperature of the phosphate , the maximum being at a lower pressure the higher the temperature .
After passing through the maximum value the current rapidly decreased with the pressure and became very small in a good vacuum .
In the present experiments a result similar to this was obtained with both specimens of impure aluminium phosphate , but with the pure salt the curves obtained were very similar to those given by platinum , and between 100 mm. pressure and a good vacuum there was no sign of the current attaining a maximum value .
With all the samples experiments were made at 1080 ' C. and at 1190 ' C. At each temperature a series of readings was taken as soon as the positive leak became fairly steady , and other series were taken after the anode had been heated for many hours so that the emission at the particular temperature had been reduced to a low value .
With Samples I and III of pure phosphate it was found that the current decreased gradually as the air pressure was reduced , from its highest value ; there was , however , little change in the emission at pressures above about 30 mm. At pressures lower than this there was a steady fall in the current to a minimum value between 1 mm. and 2 mm. pressure , followed by a slight increase as the pressure was still further reduced .
When the observations were made at gradually increasing pressures precisely similar current-pressure curves were obtained .
That the applied potential difference makes no appreciable difference to the general form of the current* A. E. Garrett , ' Phil. Mag. , ' 1910 , YI , vol. 20 , p. 573 .
* 1912 .
] by Platinum and by Certain Salts when Heated .
129 pressure curves will be seen from the observations recorded in fig. 5 .
Here the temperature of the anode was maintained at 1190 ' C. and the air pressure in the apparatus was gradually decreased from 30 mm. to a low value .
At each pressure observations were taken first with 40 volts and then Pressure : mm. with 200 volts applied potential difference .
It will be seen that under both conditions the current gradually decreased to a minimum value at about 1 mm. pressure , after which it increased slightly as the pressure was still further reduced .
When Samples II and IY of impure aluminium phosphate were used as anodes the current pressure curves resembled those obtained by Garrett , but at low pressures the current increased with diminishing pressure .
This is well shown in fig. 6 , which was plotted from observations at decreasing pressures made with the anode ( Sample II ) at 1190 ' C. , and with an applied potential difference of 200 volts .
On pumping down from a pressure of 83 mm. the current slowly increased to a maximum value at about 19 mm. pressure , after which it decreased rapidly with diminishing pressure , attaining a minimum value at a pressure of about 1 mm. , and then increasing somewhat on further pumping .
It was found that after an impure aluminium phosphate anode had been heated for a long time , the prominence of the maximum in the current-pressure curve became much less marked , and in the final set of readings , when the current had become very small , it increased continuously as the 130 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , air pressure in the apparatus was increased from about 1 mm. to 100 mm. At pressures below 1 mm. the curves were always similar to the corresponding portion of the curve in fig. 6 .
It seems , therefore , that the cause of the maximum current at a certain pressure gradually disappears on long-continued heating , whereas the cause of the increased leak at very low pressures still remains .
This latter effect is the same as that recorded in the case of a platinum anode , an explanation of which has already been suggested .
A pressure of maximum thermionic emission had been obtained in these experiments only in the earlier stages of heating the impure aluminium phosphate ; it was not observed in the experiments with the platinum anode , nor with the pure phosphate .
It had , however , appeared to a very marked extent in some earlier experiments which had been made by the author with sodium phosphate , and thus it seemed possible that the maximum emission at a certain pressure was connected with the presence of sodium in the anode .
In this case the gradual disappearance of the maximum on continued heating of the impure phosphate , ought to be accompanied by a disappearance of the sodium which is known to be present as an impurity in the fresh sample .
In order to test this , at the conclusion of the series of current-pressure observations with Sample IY of impure aluminium phosphate , the apparatus was taken down , and the platinum strip , which was still covered with a white powder , was introduced into a non-luminous Bunsen flame .
The flame did not give any evidence of the presence of sodium , but when a few grains of the original dry powder were placed on the strip , a yellow sodium flame was at once obtained .
It thus appears that in the course of the experiments the sodium leaves the impure phosphate , an effect which may be due simply to volatilisation , or it may be an essential part of the mechanism of the large positive emission from the freshly heated anode .
After this flame test the anode was refitted into the apparatus , which was evacuated to a pressure of about 6 mm. , and the thermionic current at 1190 ' C. was again tested by applying a potential difference of 200 volts .
The initial current was 1*26 xlO-6 ampere ; this decreased in the manner which is characteristic of the positive leak from freshly heated substances , and after about 90 minutes ' heating was only 6*3 x 10"8 ampere , a value which was still somewhat greater than the last reading at this pressure before the flame test was made , namely , 2*2 x 10"8 ampere .
The largely increased emission on first heating might possibly have been due to some of the powdered phosphate which had been heated in the Bunsen flame on the strip of platinum adhering to it and thus introducing a fresh supply of sodium .
I did not consider this likely to be the cause , on account of the 1912 .
] by Platinum and by Certain Salts when Heated .
131 small amount of fresh phosphate which was placed on the strip , but it seemed worth while to test the effect of heating in the Bunsen flame without risking contamination with sodium .
Accordingly , on the following day , the anode was again heated to 1190 ' C. , and was maintained at that temperature for nearly five hours , during which time the thermionic current decreased]to 2*5 x 10"8 ampere .
The apparatus was then taken down and the platinum strip was quickly and carefully introduced into a non-luminous Bunsen flame .
Again there was no sign of the presence of sodium .
After a few minutes ' heating in the flame the anode was placed in position in the apparatus , and after pumping down the emission was tested under the same conditions as before .
The initial current was greater than in the last experiment , being of the order of 10-5 ampere , but it decreased more rapidly , and in half an hour was no larger than it had been after 90 minutes ' heating on the former occasion .
On leaving the anode overnight , the emission was found to be less on the following morning , showing that , as usual , the change had gone on in the cold , and after heating for about an hour longer the current was reduced to a nearly steady value , 29 x 10-8 ampere , which is about the same as that obtained before repeating the flame test .
We thus see that simply heating the platinum strip in the Bunsen flame had the effect of increasing the positive emission to a value which is about that of the freshly heated anode .
It is well known that when platinum ] is heated in a flame it absorbs some of the flame gases , notably hydrogen , and there can be no doubt that it is the emission of these gases on subsequently heating in a vacuum which is responsible for the large current observed .
This view is , at all events , in agreement with the rapid decay of this induced activity on continued heating , and with the fact that the decay is more rapid on repeating the experiment , for it would be expected that continually Ailing and emptying the platinum would have the effect of rendering the escape of the occluded gas more easy .
That the sodium disappears from the anode during the experiments is extremely interesting , although it has been shown that the presence of this metal is not necessary in order to obtain a large positive emission , for there seems undoubtedly to be a connection between the presence of sodium in the anode and the variation of the thermionic current with the gas pressure in the apparatus .
At all events , as far as these experiments go , on reducing the pressure of the air from about 60 mm. , the thermionic current always increased to a maximum value at a certain pressure , depending on the temperature of the anode , when the latter contained sodium , and it did not do so in the absence of this metal .
VOL. LXXXVIII.\#151 ; A. L 132 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , The Experiments with Sodium Phosphate .
Experiments were made with two samples of Kahlbaum 's sodium phosphate , and with one sample of Kahlbaum 's sodium pyrophosphate .
A long series of observations was only made in the case of one of the samples of sodium phosphate , for the tests made with the other salts showed that there was practically no difference in the thermionic emissions from the three samples .
The results of a series of experiments with sodium phosphate at a temperature of 800 ' C. have already been published.* For purposes of comparison with the results for aluminium phosphate it was necessary to repeat the observations with sodium phosphate in air , at the higher temperatures 1080 ' C. and 1190 ' C. The platinum strip after being carefully cleaned was covered with sodium phosphate by evaporating a water solution of the salt upon it .
It was heated electrically during this process and the temperature was taken to a bright red heat to make sure that the salt was firmly attached to the platinum before the anode was placed in position in the apparatus .
The thermionic current at 1080 ' C. under a potential difference of 200 volts was tested at a pressure of 55 mm. after an interval of 24 hours for the absorption of moisture by the phosphorus pentoxide drying tube .
The initial current was 1*7 x 10-7 ampere , and this gradually increased during the first 40 minutes to 8*4 x 10"6 afflpere , after which it decreased in the usual manner .
The alteration of the initial current with time is thus similar to that which was found with one sample of the impure aluminium phosphate ; it will be referred to again in the next part of this paper .
Throughout the experiments it was found that after leaving the anode cold for some time with a considerable air pressure in the apparatus the effect which has been mentioned in connection with impure aluminium phosphate occurred , the current on retesting increased for the first few minutes and then the decay continued .
The maximum current was usually smaller than the last reading of the previous evening , showing that the change which causes the decay of the emission had gone on in the cold .
The proportional amount by which the activity of the anode was reduced by allowing it to remain cold was somewhat variable , but it seemed to depend in the main upon two things , firstly , upon the total length of time during which the heating had been kept up , and , secondly , upon the interval since the last observation was made .
The experimental results were in agreement with the assumptions that the freshly heated anode evolves a gas which takes part in carrying the current , and that the gas is also evolved , though at a very much slower rate , when a new anode is allowed to remain cold in a good vacuum .
* ' Camb .
Phil. Soc. Proc.,5 1911 , vol. 16 , p. 89 .
1912 .
] by Platinum and by Certain Salts when Heated .
133 On those occasions when the apparatus was left unused for some time with the gas pressure very low , the initial current was usually very large but decreased in a few seconds to a much smaller value , which gradually increased and then decreased in the manner described above .
This large leak at the start is the effect which was nearly always obtained on leaving any anode cold for some time in a good vacuum , and which I have suggested is due to the action of mercury vapour upon the platinum strip .
The relation between the current and the applied potential difference was investigated with different gas pressures in the apparatus , at both 1080 ' C. and 1190 ' C. The general character of the results obtained was precisely similar to that of the results recorded in the account of the experiments with aluminium phosphate .
Measurements of the positive emission under a constant potential difference of 200 volts were made with both increasing and diminishing pressures of air up to about 60 mm. The first series of observations was made at 1080 ' C. after about 10 hours ' heating at that temperature , and a maximum value of the thermionic current occurred at a pressure of 12 mm. when the pressure was being gradually reduced , and at about 15 mm. with increasing pressures .
Similar results were obtained at the higher temperature , 1190 ' C. , but here the pressure of maximum emission was not so well marked , probably because the anode had been heated for a long time before observations at this temperature were made .
A similar effect was noticed in the experiments with aluminium phosphate containing sodium , but in that case the pressure of maximum emission could not be detected after long-continued heating , whereas in the present case the thermionic current always had a maximum value at a pressure of about 10 mm. when the temperature of the anode was 1190 ' C. That the variation of the current with the gas pressure became much reduced by continued heating at 800 ' C. was pointed out in the account of my earlier experiments with sodium phosphate .
In the case of impure aluminium phosphate , the removal of the pressure of maximum emission by continued heating was accompanied by the disappearance of sodium from the anode .
In the present experiments , although the amount of sodium phosphate on the anode was less at the conclusion of the experiments than it had been at the beginning , it did not completely disappear\#151 ; a fact which was also obvious from the magnitude of the positive emission .
The thermionic current from impure aluminium phosphate is at first larger than that from sodium phosphate under similar conditions , but on continuing the heating for some time , the current from aluminium phosphate falls off to a much lower value than that from sodium phosphate .
At the end of the experiments with the latter salt the current at 1190 ' C , in an air pressure of 10 mm. was about 450 times the final value from impure aluminium L 2 134 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , phosphate .
The existence of this large emission from an anode which has been heated for so long that most of the occluded gas must have been driven out of it , is evidence in favour of the view that at this stage the current is largely carried by sodium atoms .
At low pressures the thermionic current from sodium phosphate always decreased continuously as the pressure was reduced .
There was never any sign of the increasing current obtained in other cases .
This is accounted for by the greater emission in the case of this anode , which would completely mask an increase of the magnitude usually observed .
The Decay of the Positive Emission from .
Freshly Heated Anodes .
The thermionic emissions from the various anodes , when freshly heated , cannot in all cases be compared , because the conditions under which the initial tests were made were not always the same .
In the case of platinum itself , and of the first samples of aluminium phosphate , observations were made at the lowest temperature at which a conveniently measurable current was obtained , whereas with the later salts the observations were commenced at 1080 ' C. With a view to comparing the initial currents from the different anodes , and the rates of decay under similar conditions , a fresh series of observations was made at the highest temperatures used in these experiments , 1190 ' C. , the air pressure being in all cases 15 mm. , and the potential difference 200 volts .
The platinum strip was well washed to free it from sodium phosphate .
It was then dried , and the absence of sodium was proved by heating it in a Bunsen flame .
The anode was replaced in the apparatus and the decay of the initial current from platinum under the standard conditions was investigated .
The values of the currents during the first two hours are plotted in Curve I , fig. 7 , and in the same diagram the corresponding curve for pure aluminium phosphate is given ( Curve II ) .
In this latter case , the current during the first few minutes was too large to be included in the figure , but the emission decreased continuously from the start , very rapidly at first , and the curve when plotted to a smaller scale was exactly similar to the curve for platinum .
The decay curves for the emission from sodium phosphate , from sodium pyro-phosphate , and from impure aluminium phosphate are given in fig. 8 , the curves being numbered in that order .
The Curves I and II for the sodium salts show the characteristics which have already been described , namely , an emission which increases to a maximum value during the first few minutes and then decays away .
The curve for impure aluminium phosphate ( Curve III ) is similar to those given in fig. 7 for the pure phosphate and for platinum ; there is an enormous decrease in the current 10~6 ampere 1912 .
] by Platinum and by Certain Salts token Heated .
135 s c3 \#163 ; 3 ' \lt ; P u Q 1\#174 ; ( \ Fig. 7 .
o v V S\#177 ; y^e_2 Curve 1 O 20 40 60 80 100 Time from beginning of heating ( minutes ) .
Curve 3 Time from beginning of heating ( minutes ) .
136 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , during the first few minutes .
The salt with which this curve was obtained was from the same bottle as Sample IV in the experiments already described , and the decay curve is similar to that which was obtained before with the anode at 1080 ' C. It will be remembered that the other specimen of impure aluminium phosphate ( Sample II ) gave a result resembling that obtained with the phosphates of sodium , the decay being preceded by an increase in the current to a maximum value .
This initial increase in the current before the decay sets in was not only obtained with these anodes when new , but also , to a smaller extent , on reheating after they had been left cold for some hours in a pressure of several millimetres of air .
A preliminary increase of the emission on reheating under these conditions was also occasionally noticed when the anode consisted of pure aluminium .
phosphate , or of platinum , although the magnitude of the effect in these cases was very small .
The aluminium phosphate used by Garrett in his experiments* also gave an initial current which increased to a maximum in about eight minutes , and then decreased rapidly .
He points out that the form of the curve seems to indicate that the first effect of the high temperature is the production on the anode of an unstable substance , which does not itself emit ions , but which gives rise to a second unstable substance which is the origin of the ionisation .
In the following table I have collected together the values of the thermionic currents from the different anodes at the commencement of the heating , and after the temperature ( 1190 ' C. ) had been maintained for 1 , 2 , 10 , 50 , and 100 minutes :\#151 ; Anode .
Current in 10 8 amperes .
At start .
1 min. 2 mins .
10 mins .
50 mins .
J 00 mins .
Platinum 183 18 0 6*9 2 5 1 *74 1-24 Pure aluminium phosphate Sodium phosphate 2040 201 87 18 *0 5*6 3-6 2550 3350 4430 5650 3750 1600 Sodium pyro-phosphate ... 3380 5220 5600 5270 2400 940 Impure aluminium phosphate 7560 3220 1450 250 59 32 The values of the current at the start must be taken as being approximate only , for the rapidity of the alteration during the first minute makes the observation very difficult .
The numbers given for platinum and for the phosphates of aluminium were obtained from the largest deflection of the galvanometer , and the rapidity with which this indicates changes in the * A. E. Garrett , 4 Phil. Mag.,5 1910 , YI , vol. 20 , p. 573 .
1912 .
] by Platinum and by Certain Salts tvhen Heated .
137 current depends on the extent to which it is shunted .
The instrument was shunted to the same extent for the measurements with platinum and with pure aluminium phosphate ; a shunt of smaller resistance was used in the cases of impure aluminium phosphate and of the phosphates of sodium .
From the numbers in the above table it will be seen that the decay of the current during the first minute of heating is much the greatest with a platinum or a pure aluminium phosphate anode ; with the aluminium phosphate containing sodium the decrease is considerably smaller , and in the case of the phosphates of sodium the current increases .
Now it is remarkable that when an impure aluminium phosphate anode which had been used for so long that all traces of sodium had disappeared from it was heated for a few minutes in a Bunsen flame , the emission on retesting was found to be of the same magnitude , and to decay during the first few minutes in the same rapid manner , as the initial discharge from the freshly heated pure salt .
The increase of the current after heating in the Bunsen flame is , no doubt , due to the emission of gases which have been absorbed from the flame , and this points to the conclusion that the large initial currents from freshly heated platinum and from pure aluminium phosphate are also due to the emission of gases by these anodes .
There is doubtless an emission of gas also in the case of the other anodes used , and this is probably responsible for carrying some of the observed current , but in these cases there is apparently some other source of ionisation which increases either in amount or in activity , or in both respects , during the first few minutes ' heating , and the most probable origin of this effect seems to be the sodium which these anodes contain .
In the case of impure aluminium phosphate this second source is not very large , and serves only to decrease the rate of decay of the total emission ; with the phosphates of sodium , however , the decay is completely overcome , and there is an increasing current during the first few minutes .
It will be remembered that an initial increase in the current was also observed with Sample II of impure aluminium phosphate .
This would seem to indicate that Sample II contained a greater amount of sodium impurity than Sample IV , which was used in obtaining the figures in the above table .
After a very few minutes ' heating the decay of the emission from the impure salt is more rapid than the decay of the activity of the pure phosphate , a state of affairs which can readily be explained by the gradual disappearance of the sodium , in addition to the falling off of the gas emission , in the case of the former anode .
The numbers recorded in the table give the relative activities of the different substances at 1190 ' C. The observations which had been made at lower temperatures gave the same general result .
The largest initial current 138 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , was always given by impure aluminium phosphate , and next in order of thermionic emission come the phosphates of sodium , pure aluminium phosphate , and , last of all , platinum .
The initial increase in the emission from sodium phosphate was relatively larger , and continued for a longer time , when the anode was at 1080 ' C. than when it was heated to 1190 ' 0 .
, a result which is probably due to the rate of evolution of gas from the heated anode increasing very rapidly with rise of temperature .
A comparison of the final values of the thermionic currents from the different anodes after heating for a long time is given in the following table .
In all cases the applied potential difference was 200 volts , and the temperature 1190 ' C.:\#151 ; Anode .
Thermionic currents ( 1 = 10 9 ampere ) at various air pressures .
0*005 mm. 1 mm. 2 mm. 5 mm. 10 mm. 20 mm. Platinum 5*2 1 *65 1 *65 2*2 2 *9 3*9 Pure aluminium phos- 0*7 0*6 0*7 0*9 1 *2 2*0 phate Impure aluminium phosphate Sodium phosphate 2*9 1*5 2*0 3 *1 41 5*0 1080 1500 1630 1750 1810 1740 The numbers given are interesting as indicating the order of magnitude of the emission from the different anodes at the conclusion of the experiments , but too much importance should not be attached to the actual figures , for it must be remembered that there was every indication that continued heating would have reduced the emission still further , although the rate of decay of the current had become so small that , in the case of the salt anodes , it might perhaps be accounted for by the phosphate gradually becoming detached from the platinum .
From the table it would appear to be doubtful whether the small emission given by pure aluminium phosphate is due to that salt at all , for it will be seen to be smaller than the emission from the platinum strip itself at the beginning of the experiments , but this would have decreased considerably during the extra heating .
At the conclusion of the experiments with the phosphates the platinum strip was cleaned and tested alone ; it was found that after heating at 1190 ' C. for several hours , the thermionic currrent was reduced to a smaller value than any recorded in the above table , and I am therefore inclined to think that the numbers given for pure aluminium phosphate are in part due to some remaining small emission from that salt .
After heating for about the same length of time , the final current measured from the impure aluminium phosphate is not much greater 1912 .
] by Platinum and by Certain Salts when Heated .
139 than that from the pure salt , although during the course of the first few minutes ' heating it was more than fifteen times as great .
It has already been stated that this seems to be due to the disappearance of the sodium which was present as an impurity in the former anode .
The outstanding feature of the above table is the relatively enormous emission from sodium phosphate , which is obtained although the anode had been heated for at least as long as either of the specimens of aluminium phosphate .
The Nature of the Carriers of the Positive Thermionic Current .
The experiments described in this paper have shown that there are considerable variations , both in the magnitude , and in the permanence , of the positive emission from different substances .
In the case of platinum , and of pure aluminium phosphate , we have an initial thermionic emission which decreases very rapidly with time , and in the course of a few hours ' heating at a high temperature becomes extremely small .
With sodium phosphate , on the other hand , the emission at first increases , and then decreases much more slowly than with the other substances tested , so that even after many hours ' heating a very considerable thermionic current can be obtained .
The impure aluminium phosphate falls between these two cases .
At the commencement of the heating the alteration in the emission with time resembles that given by phosphate of sodium , whereas the final low value more nearly corresponds to the current from the pure phosphate or from platinum .
The experiments have shown that this is accounted for by the fact that in the beginning the impure phosphate contains sodium , but that this gradually disappears from the anode in the course of the heating .
The author is of opinion that the ionisation from platinum ( and from metals generally ) is largely due to the emission of absorbed gases on heating , but that in the case of sodium phosphate a considerable part of the emission consists of positively charged sodium atoms , the final value of the current in a good vacuum after long-continued heating being almost entirely due to this cause , although at first there is no doubt also a large current due to the evolution of ionised gas , which probably conies partly from the salt and partly from the platinum upon which it is heated .
That certain salts of sodium when heated and used as anodes in vacuum tubes emit positively charged sodium atoms was demonstrated by Gehrcke and Reichenheim in the year 1906 , * but the experimental conditions were somewhat different from those described in this paper .
In their earlier experiments the salts were heated upon platinum foil , but the cathode of the discharge tube was also heated electrically , and consisted of a platinum * Gehrcke and Reichenheim , 'Deutsch .
Phys. Ges .
Verb .
, ' 1906 , p. 559 .
Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , strip covered with barium oxide .
The potential difference applied was 110 volts , and under these circumstances a bundle of yellow rays was seen proceeding from the salt on the heated anode .
These rays Gehrcke and Eeichenheim called " anode rays , " and subsequent experiments proved them to consist of positively charged sodium atoms .
In later experiments* Gehrcke and Reichenheim obtained better results by using much higher potential differences across the tube and having neither electrode heated by an external current ; the current sent through the tube by the applied E.M.F. gradually warmed up the salt anode , and when this attained a bright red heat a luminous pencil of rays could be seen proceeding from it .
The discoverers of these rays found that they were given particularly well by the haloid salts of the alkali metals .
In each case the value of the specific charge of the ions , which was measured by two different methods , agreed with the value calculated on the supposition that they consist of atoms of the metallic constituent of the salt under test , conveying a charge equal to that carried by the hydrogen ion in electrolysis .
Anode rays were also obtained from salts of the divalent elements strontium and calcium , and in these cases the carriers appeared to be doubly charged atoms of the metal used ( Sr++ or Ca++ ) .
For a full account of the methods of obtaining these anode rays reference must be made to the original papers , but it will be convenient to mention here some of the conditions which seemed to favour their production .
In the earlier experiments , in which the salt was heated upon platinum foil and a low potential difference was used , the rays lasted for a very few minutes only .
In the later form of apparatus , in which a small stick of the fused salt formed the anode , the rays could be obtained for several hours , although the activity diminished with time .
The length of the luminous ray was greater the higher the vacuum in the discharge tube , but where the path was non-luminous it could be traced by the fluorescence produced on the glass .
The most copious supply of rays was obtained when easily dissociating or volatile salts were used as anodes , and some mixtures of salts seemed to work better than either of the constituents alone , possibly because the mixture had a lower melting point .
Anode rays could not be obtained from a cold anode , and in the form of apparatus where no external heating was used it was necessary to wait some time after the high E.M.F. was applied , until the current through the tube had raised the extremity of the anode to a red heat , before the rays appeared .
It occurred to the author that the anode rays of Gehrcke and Reichenheim were essentially the same as the thermionic emission from heated salts .
Sir J. J. Thomson had found that the greatest emission was given by phos-* Gehrcke and Reichenheim , 'Ann .
der Phys. , ' 1908 , p. 861 .
1912 .
] by Platinum and by Certain Salts when Heated .
141 phates , and that aluminium phosphate was particularly active .
It was therefore attempted to obtain anode rays from this substance , but without success.* A mixture of phosphates of sodium and lithium was also tried , and in this case some curious luminous effects were obtained , but I was unable to see the fine bright pencils of light described by their discoverers as characteristic of anode rays .
It was , however , obvious that positively charged sodium was being shot off from the anode , for , although the cathode was behind the anode , so that anything shot out by the glowing phosphates was projected away from it , yet a bright yellow sodium light began to appear on the cathode as the tube was worked , appearing first on the edge nearest to the anode , and gradually spreading all over it .
The impossibility of obtaining anode rays from these salts , which were then thought to be the most copious emitters of positive electricity when heated , led the author to the conclusion that there was no close connection between the two phenomena .
Recent researches , however , have modified this opinion .
Richardson has determined the specific charge of the carriers of the thermionic current from a number of different salts when heated in vacuo , and has concluded that these carriers are atoms of the respective metallic constituents minus a single electron .
He also found that aluminium phosphate is by no means the most active salt in producing positive ionisation when heated in a vacuum , a result which is confirmed by the experiments recorded in the present paper .
In the light of these results it seems probable that the anode rays of Gehrcke and Reichenheim are essentially the same as the emission , in a good vacuum , from a positively charged salt which has been heated until the steady state is reached .
There is , however , the difference that in the anode rays the emitted atoms are travelling much faster owing to the fact that they are liberated into a much stronger electric field .
Gehrcke and Reichenheim have shown that it is necessary to have a large fall of potential at the anode in order to obtain luminous anode rays .
Under ordinary circumstances , with such gases as oxygen or nitrogen in the discharge tube , the anode fall of potential is about 20 volts , and the velocity which the ions attain is too small for luminosity to result , but if a small quantity of any halogen vapour is introduced into the tube , the anode fall increases enormously , and may become many thousands of volts .
This fact explains why anode rays are readily obtained when haloid salts are used which have the property of dissociating when strongly heated .
Now Richardson has concluded from his experimentsf that what is required for a salt to be a good emitter of positive ions " is a * ' Camb .
Phil. Soc. Proc. , ' 1909 , vol. 25 , p. 329 .
+ O. W. Richardson , ' Phil. Mag. , ' 1911 , YI , vol. 22 , p. 669 .
142 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , combination of volatility in the possible compounds formed , together with high electro-positiveness of the metallic constituent .
" The salts of the alkali metals he found to be particularly active , and if only a trace of either sodium or potassium were present as an impurity in the salt under test , the values of e/ m obtained for the positive ions always indicated that they consisted of atoms of these metals .
Volatile salts were found to be more efficient than non-volatile salts ; for instance , zinc haloids were very active , and barium sulphide was more efficient than barium sulphate .
It will thus be seen that the class of salt which is best from the point of view of thermionic emission with low potential differences is precisely that which was found by Gehrcke and Eeichenheim to be most efficient for the production of anode rays .
The carriers of electricity in the two cases are identical , the only difference seems to be in the velocity with which they travel .
It should , however , be mentioned that whereas Gehrcke and Eeichenheim found that the anode rays from salts of strontium and calcium consisted of atoms of these metals carrying double charges , Davisson* obtained a value for the specific charge of the carriers of the thermionic current from these salts corresponding to a single unit of electronic charge per atom .
In some more recent determinations of \lt ; s/ m for strontium anode rays by measurements of the Doppler effect , Eeichenheimf obtained from certain lines of the spectrum a value corresponding to Sr+ , and from other lines a value more nearly corresponding to Sr++ .
Also Eichardson has concluded that there are ions of the type Zn++ present in the thermionic emission from the zinc haloids .
There is evidence , therefore , both in the case of anode rays , and of the ordinary thermionic emission , that polyvalent atoms may carry either single or multiple charges , but further investigation of this point is required before it can be regarded as fully established .
My experiments on the discharge of positive electricity from sodium phosphate heated in different gases led me to the view that the gas present in the discharge tube takes an active part in carrying the thermionic current\#151 ; probably by diffusing into the anode and being emitted in an ionised condition .
This view has been criticised by Eichardson , J who holds that the carriers of the thermionic current are in all cases atoms of the metallic constituent of the salt under test ; a conclusion which he bases upon his determinations of the value of the specific charge of the carriers in a good vacuum .
I think there can be little doubt that when a salt anode has * C. J. Davisson , 'Phil .
Mag. , ' 1912 , VI , vol. 23 , p. 121 .
t O. Reichenheim , ' Ann. der Phys. , ' vol. 33 , p. 747 .
t '* w* Richardson , 'Phil .
Mag. , ' VI , vol. 22 , p. 669 .
1912 .
] by Platinum and by Certain Salts when Heated .
143 been heated for so long that the thermionic emission has decreased to what I have called its " steady value , " the carriers of the current in a good vacuum are mostly metallic atoms\#151 ; the ions which have been found by Gehrcke and Reichenheim to constitute the anode rays .
At higher gas-pressures there is a considerable increase in the thermionic current , corresponding presumably to an increased emission of positively charged ions from the heated anode .
On the view which is advocated by Richardson no part of the current is carried by the gas , and this increased emission consists solely of positively charged atoms of the metallic constituent of the salt , but , so far as the author is aware , no theory has been advanced which explains how the presence of the gas increases the emission of these atoms .
The theory that the ions are entirely metallic atoms seems to rest upon some experiments of Davisson , * by which he claims to have proved ( 1 ) that the gases emitted by a heated anode are uncharged , ( 2 ) that the increased thermionic emission from a salt which is caused by the presence of a gas is not due to the cause which I suggested in the paper referred to above .
The first of these conclusions is in direct opposition to the results of other observers .
In the case of aluminium phosphate heated upon platinum , Garrett , from observations of e/ m , concluded that some of the carriers were charged hydrogen atoms , and the earlier work of Richardson on the positive discharge from heated platinum led him then to the view that the gas emitted by the hot metal was ionised .
Richardson investigated the positive leak from the outer surface of a platinum tube heated electrically in air at atmospheric pressure , while hydrogen was diffusing through the walls of the tube from the inside .
The results showed that the effect of the hydrogen was " to produce an additional number of positive ions proportional to the amount of hydrogen diffusing out , " from which Richardson concluded that " the hydrogen inside the metal , which is known from other considerations to be in the atomic state , is positively charged , " although " only a small fraction of it comes out in the ionic form."f These experiments are strong evidence in favour of the view that the carriers of positive electricity from heated platinum which is evolving hydrogen consist , partly at all events , of charged atoms of that gas , and it would seem to be extremely probable that the carbon monoxide gas which is generally evolved by a heated anode is also emitted in an ionised condition .
Of course , it does not follow that since the gases emitted by a metallic anode are ionised , that those emitted by salt anodes are also ionised , though I think this is usually the case , but it must be remembered that in the usual method of testing the thermionic emission , the salt is heated * C. J. Davisson , 4 Phil. Mag. , ' VI , vol. 23 , p. 139 .
+ O. W. Richardson , ' Phil. Trans. , ' 1906 , A , vol. 207 , p. 58 .
144 Dr. F. Horton .
Positive Ionisation Produced [ Nov. 28 , upon a strip of platinum and there is doubtless a liberation of ionised gas from this .
It was probably hydrogen evolved in this way from the platinum strip which was detected in the determinations of e/ m for the ions in Garrett 's experiments .
Davisson 's second conclusion\#151 ; that the increase in the thermionic current from a salt which is caused by the presence of gas , is not due to the emission of gas atoms or molecules which have been absorbed by the salt and are emitted in a charged condition\#151 ; rests upon determinations of the specific charge of the carriers from impure aluminium phosphate , in air and in carbon monoxide at pressures of about 0*1 mm. , and in hydrogen at about 1 mm. pressure .
The temperature of the anode was 600 ' C. ; the first observed value of e/ m in a good vacuum was taken to indicate that the ions were charged potassium atoms .
After longer heating the value found was much nearer the specific charge for sodium , and the ions were taken to be atoms of this metal .
On allowing gas to enter the apparatus the difficulty of measuring the specific charge of the carriers increased , owing to the scattering which occurs on collisions with the gas molecules , and Davisson concludes that this effect is responsible for the much smaller values which are then obtained .
From the curves given it appears that , with carbon dioxide , for example , the value of e/ m decreases from about 435 in a good vacuum to 280 at a pressure of 0T3 mm. ( The value of ejm for Na+ is 421 , and for C02+ is 243 .
) If an alteration of this magnitude is produced by the interference of the gas with the free passage of the ions , it is obvious that the method employed is quite unsuitable for determining the value of the specific charge of the carriers at the higher pressures .
Moreover , previous to these experiments of Davisson , Garrett had shown that in the case of an aluminium phosphate anode at 1005 ' C. , an increase in the gas pressure of the order used by Davisson produces only a very slight increase in the thermionic emission , and the increase would be very small indeed at the much lower temperature used by Davisson .
In the experiments recorded in this paper it was found that the emission actually decreases slightly as the gas pressure is gradually raised to about 1 mm. , so that if accurate measurements of e/ m could be made at this pressure they would not be expected to indicate that the gas is carrying any larger proportion of the current than it does at lower pressures . ?
Reference has been made at the commencement of this paper to the view recently advocated by Richardson as to the nature of the positive ions from incandescent metals .
Richardson obtained values of the specific charge of the ions from different metals varying from 486 to 337 , the mean value from measurements with 13 metals giving 25*7 for the mass of the carriers 1912 .
] by Platinum and by Certain Salts when Heated .
145 as compared with the mass of the hydrogen atom , from which he concludes they are atoms of sodium or potassium.* My experiments with impure aluminium phosphate have shown that if when first heated an anode is contaminated with sodium impurities , these disappear after the thermionic emission has been going on for some time , so that it is improbable that the value of the specific charge measured under these conditions refers to sodium atoms .
As the result of a determination of e/ m for the carriers of positive electricity from a hot platinum wire , Sir J. J. Thomson some time ago came to the conclusion that the current is carried both by atoms of the metal and by atoms or molecules of the surrounding gas.f Richardson , in his experiments , did not detect any ions of the former type , but his value of ejm for the carriers is not inconsistent with their being molecules of carbon monoxide carrying single charges ( CO+ ) .
It is known that many metals evolve this gas when heated , and that it is continually being set free in an evacuated vessel which contains traces of wax or grease , so that its presence in the apparatus used by Richardson is easily accounted for .
I think there can be little doubt that the large emission on first heating a clean platinum anode is due to escaping gas which comes from inside the metal in an ionised condition .
The connection between the gas emission from heated metallic wires and the emission of positive electricity from them has been carefully studied by Klemensiewicz.| He found that wires which had been heated in a good vacuum until the thermionic emission was very small , regained their activity when they were allowed to stand for some time in a gas at a pressure of several atmospheres , and that the recovery was aided by keeping the wires at a temperature of about 200 ' C. so as to assist the diffusion of the gas into the metal .
On the other hand there was no recovery of activity when the wires were allowed to remain for a much longer time in an evacuated glass tube .
There is thus a very close connection between the positive ionisation and the emission of absorbed gas , and this connection is most simply explained by the view that gas coming from the interior of the wire is ionised .
It is difficult to see any explanation on the supposition that the ions are atoms of sodium or potassium .
If the gas actually carries the current , we see at once the cause of the rapid fall of the emission from a new wire , for when the wire is heated the gas pressure inside it is increased and the gas diffuses out until there is equilibrium between the internal and the external conditions .
When , after * O. W. Richardson , ' Phil. Mag. , ' 1910 , VI , vol. 20 , p. 545 .
t J. J. Thomson , 4 Conduction of Electricity through Gases , ' Camb .
Univ. Press , 1903 , p. 185 .
1 Z. Klemensiewicz , 4 Ann. der Phys. , ' 1911 , vol. 36 , p. 796 .
146 Positive Ionisation Produced by Platinum when Heated .
long-continued heating , equilibrium is attained , there are as many gas atoms or molecules entering the wire in a given time as there are emitted by the wire .
The latter are ionised , a certain proportion of them ( probably a very small proportion of them ) is positively charged , and it is the charges carried by these which constitute the small " steady " current which is given by a wire which has been heated for a long time .
This is the view of the emission of positive ions from heated metals which the author thinks is most in agreement with the experimental results ; at the same time , if the metal is not perfectly clean , or if there are gases or vapours present which have chemical action upon it , there is probably , a certain amount of ionisation produced in other ways .
If there is a trace of sodium or of potassium impurity on the anode , then it is probable that atoms of these metals assist in carrying the current as they do when a salt anode is used .
In the experiments with platinum described in this paper the metal was very carefully cleaned before being tested , and later in the research it was found that the usual large emission on first heating was obtained after the platinum had been heated in a Bunsen flame , although the flame gave no evidence of the presence of sodium .
I think , therefore , that the ions cannot be sodium atoms in this case .
A small amount of chemical action may give rise to considerable ionisation , and so if gases or vapours are present which have a chemical action on the anode there may be electrification produced while this action is going on .
It seems probable that some of the ionisation which is produced when certain salts are heated has its origin in chemical actions between vapours produced by these salts and the platinum on which they are heated .
This view has already been put forward by Richardson to explain certain ionisation effects which he obtained when various salts were heated to a high temperature in a platinum tube .
The author wishes to acknowledge his indebtedness to the Government Grant Committee of the Royal Society for the means of purchasing some of the apparatus used in these experiments , and also to Prof. Sir J. J. Thomson for his advice and kindly interest in the research , which was carried out in the Cavendish Laboratory .

