Ionization of Gases by Positive Bombardment Citation Sutton, Richard Manliffe (1929) Ionization of Gases by Positive Bombardment. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/rw38-2t80. https://resolver.caltech.edu/CaltechTHESIS:08142024-200938006 Abstract No abstract. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Physics) Degree Grantor: California Institute of Technology Division: Physics, Mathematics and Astronomy Major Option: Physics Thesis Availability: Public (worldwide access) Research Advisor(s): Millikan, Robert Andrews Thesis Committee: Unknown, Unknown Defense Date: 10 May 1929 Record Number: CaltechTHESIS:08142024-200938006 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:08142024-200938006 DOI: 10.7907/rw38-2t80 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16638 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 15 Aug 2024 21:04 Last Modified: 15 Aug 2024 21:04 Thesis Files PDF - Final Version See Usage Policy. 10MB Repository Staff Only: item control page

IONIZATION OF GASES :BY POSITIVE ION BOM BARDMENT. THESIS by Richard Manliffe St.itton In Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY. California Institute of Technology Pasadena, California 1929 INTRODUCTION. The experimental work reported in this paper deals with a study of the problem of ionizing gases by the impact of slow positive ions using accelerating potentials less than 1000 volts. Definite evidence of ionization distinguished from other effects produced by the positive ions is submitted. The conditions of voltage, gas pres- sure, and collecting potentials are herein recorded, together with some estimate of the efficiency with which the positive ions ionize the gases used. In all of the present work a source of singly-charged potassium ions was employed to bombard successively hydrogen, nitrogen, argon, neon, and air. While the experimental conditions involve cer- tain com9lications of quantitative interpretation which render impossible an accurate calculation of efficiencies from the results obtained with the tube utilized, yet the qualitative evidence of ionization is unambiguous; and under certain conditions of collecting potentials to be described later, the calculation of efficiencies should be reliable within approximately ten per cent. An enormous amount of experimental work has been done upon the problem of ionization of gases by electrons since the early work of Lenard in 1913. This field has been so fruitful in yielding re- sults of value in the development of spectroscopic theory that relatively little attention has been paid to the similar evidence of electric fracture of atoms by positive ion impact. -1- The reasons for this condition of our present knowledge are several; among them are, i) the difficulty of getting an effective source of positive ions of sufficient intensity; ii) the relatively small efficiency of the ionization produced by the positives as compared to electrons; iii) the recognition of secondary effects which vitiate the early results obtained; iv) the difficulty of interpreting results, due chiefly to the large mass of the positive ions compared with electrons and their complex structure which together render the impact between+ ion and gas molecule an exceedingly complicated occurrence. When electrons ionize gas molecules and atoms, the chief consideration is one of energy transfer, and the simple equation of evergy loss in terms of radiation (the inverse photoelectric effect) is applicable. Ve= ½mv2 = hv. The question of the conservation of momentum is of minor importance in this case since the mass of the electron is so small in comparison with that of the atom struck. The electron can transfer virtually all of its kinetic energy (and momentum) to the ionized atom in an "inelastic collision." :SU.t when the impacting particle is an ion of mass comparable with, or greater than that of the atom struck, even though it may receive the same energy from the accelerating field as does the electron, it is evident from simple considerations of the conservation of momentum and of energy that the quantity of energy which it may transfer to an atom upon impact is, at best, only a fraction of its total kinetic energy. This fraction depends upon the nature of the collision in a calculable -2- manner, but we have no way of knowing what the nature of the collisions is. The best we can do is to set a minimum value for the accelerating potential necessary to produce ionization by positives, with no assurance that we will be able to detect the few collisions which satisfy this minimum condition. A second consideration which appears to reduce the probability of ionization by positives is that the kinetic energy transfer upon impact mu.st in some way be localized in a single electron of the atom whi ch is ionized. This may be an event of relatively small probability in comparison with the number of times the positive ion transfers kinetic energy to the impacted atom as a whole. Furthermore, in order to ionize an atom by+ ion bombardment, the electron which is liberated mu.st escape from a greatly increased positive field due to the presence of the positive charge of the bombarding ion and to the field of attraction of its own parent atom • .All three of these last considerations point to i) the appearance of ionization by positives at higher accelerating potentials than for electron ionization; ii) the lack of a clearly defined "ionizing potential 11 for positive ions; iii) relatively small efficiency of ionization by positives. Nowhere, in the present work was there found any evidence of the 11 breaks 11 due to radiation or double ionization such as are common in experiments on electron ionization. In no case was there found any certain evidence for ioni- zation below 100 volts, which is several times the potential required for a=: electron ionization of the sa.~e gases. However, it is conceiv- able that some ionization below 100 volts mi ght be detectable by the -3- use of more sensitive measuring devices which would record the less p robable impacts resulting in fracture of the atom. The theory of t his type of ionization is still in a primitive state, dealing chiefly. with considerations of eonsiie atioas of conservation of energy and momentum and the (unknown) conditions of force fields about the colliding particles. 1,2,3,4.) Mention should be made of the previous work in this field with some estimate of the results obtained. 1) Bahr and Franck5)(1914) using Lenard's method which had just been developed found what was supposed to be ionization~- 121! the potentials for electrons. This work has been generally dis- counted on the basis of secondary effects. 2) Pawlow6) (1914) used the same method in hydrogen and found ionization below 10 volts. This also has not been substantiated by later repetitions. 3) Horton and Davies7)(1919) attempted to ionize helium by positive ions from glowing tantalum and found no evidence for ionization up to 200 volts. They were the first to recognize the importance of the secondary effects, such as radiation and the secondary emission from metals. 4) Saxton8)(1922) believed he found ionization in H2 by hy- drogen positives below 18 volts. This has not been supported by later work (Hooper). 5) Dem:p ster9)(1916) reports the excitation of light by posi- tive rays a s low as 5 volts! In some later work of his ownlO) he finds -4- no evidence of ionization by 900 volt protons in helimn or hydrogen. 6) Tatell)(l924) finds spectroscopic evidence of the ionization of mercury va-por by sodium positives between 40 end 70 vol ts. Mercury was ionized by Hg+ at 70 volts. 7) Jones 12 )(1927) excited mercury vapor by 1200 volt potassium positives. (Spectroscopic) 8) Hooperl 3 )(1925) carried on extensive experiments on hydrogen using sodium ~ositives up to 925 volts without evidence of ionization. 9) Gurneyl4)(1928) attempted to ionize hydrogen with 7000 volt potassium positives. The experiment led to inconclusive results. For a swmnary and discussion of previous work in this field, see L.B. Loeb 1 s paper in Science.15) It would appear that, with the exception of the spectroscopic evidence, little definite knowledge of ionization by positive ions has been found previously which did not have the taint of "secondary effects." None of these references deal with the range of potentials usually referred to as the 11 0anal Ray region" in which the positives certainly produce ionization since they have energy thousands of times the ionization potential values. We are concerned more particularly with the range below 1000 volts. The effect of the impact of positive ions upon metal surfaces causing the emission of electrons has been studied by Klein16 ), Jacksonl7), Hyattl8), Oliphant19 ) and others. .Any attempt to detect ionization in gases by positives must carefully exclude this effect -5- or adequately acco1mt for it. It is believed that the present work separates ionization from secondary emission from metals in an unmistakabl e manner. A lesser effect, reflection of positives, has been investigated by Read20), Gurney21 ), and Oliphant19 ). Evidence of its presence in this work is practically lacking and it may be considered throughout as negligible in amount. It is essential to adjust experimental conditions so that any possible photoelectric effect due to the presence of excited atoms is ruled out. tentials. This may be done by the proper choice of collecting po- Furthermore, the collecting potentials should not be so great as to introduce the possibility of ionization by electrons. With a recognition of these secondary phenomena which may affect the results of experiment, the controlled conditions utilized throughout the major portion of this work were such as to exclude these effects, or at least to make them negligible in amount compared with ionization within the volume of the gas. -6- APP.A.RATUS. The tube used in these experiments was built in the Norman Brio.ge Labora.tory after a design worked out by Mr. Max Wehrli and Dr. R. A. Millikan in 1924. It is schematically pictured together with its connecting system in Fig. l. Several changes in the original electrode system have been made by the author as the necessities of the problem suggested. The source of positive ions was a platinum strip 1.2 x .3 cm resting in a cylindrical porcelain trough held rigidly between copper terminals. The platinum was coated with a few milligrams of oxid.e catalyst developed by C. H. Kunsman • .All of the present work was done with the potassium source, the emission from which has been shown by Barton, Harnwell, and Kunsman 22) to be practically pure potassium. After reducing the oxide in an atmosphere of hydrogen to obtain a good coating, the filament and cathode system was sealed to the rest of the tube. Not wishing to r...ave the contamination of organic vapors enter into the tube, it was deemed best to seal the tube together in this manner without using a wax-joint although this made the necessary changing of filaments inconvenient. The filament was ke?t at fairly constant temperature by a heavy duty six-volt storage battery, the average current requirement being about 6.5 amperes to secure su.fficient positive emission. A well coated fila..'llent was usually good for about 50 hours of operation, barring accidents, before the emission fell to an inconveniently low value. -7- Unfortunately it re- quires a good many minutes for the emission current to become constant after closing the heating circni t; but it was found tha.t, with the method of handling resuJ. ts utilized, wide variations in the filament emission did not appreciably affect the ratio of currents to the recording galvonometers. Emission currents of the order of one microampere were used in most cases; the positive current entering the ionization region in the upper part of the tube was of the order of 10-? amperes, or less. The entire filament system was enclosed in a hollow cylindrical cathode turned from a single piece of steel with a 2 mm hole in the top for the ion stream to enter the upper region of the tube under the accelerating potential, Va• mately 2 mm from this hole. The source was placed approxi- A 5 cm diameter nickel plate located just above the cathode served as the collector for the products of ionization. The ions were prevented from striking the collector, S, directly from the primary beam by means of a projecting ridge on the cathode which protruded through a hole in the collector. Situated approximately one centimeter above the collector plate was a widemeshed grid of nickel wire inserted to control the potential within the space arm.md it without offering much cross-section for the stoppage of ions and electrons. Above the grid av-shaped nickel plate served to collect the positive ion stream; its shape was supposed to prevent the reflection of the positive ions. This element of design is rather unfortuiiate because reflection of positives is of minor importance, whereas the unsymmetrical field produced by the plate-8- grid system and t he indeterminable volume between G and P from which the products of ionization are collected make the accurate calculation of the efficiency of ionization impossible. It should be bene- ficial to change the plate to a flat disc parallel to t he grid and as close to it a s convenient so as to reduce and make calculable the corrections necessary to account for ionization between grid and plate. The electric connections are shovm in Figure 2. It will be seen that the accelerating potential, Va, which was supplied by a bank of small lead accumulators, could be varied continuously from zero to the highest values used, - about 1100 volts. ra was applied to the filament through the midpoint of a 30 ohm resistance. The total current in the space between F and C was measured by galvanometer Gi_. This current, IE• consists of the positive ion emission in- creased by any secondary emission of electrons from the cathode and by any ionization within the furnace. In general, less than 10% of the total emission from F passed through the hole in the cathode. The collecting potentials on Sand P could also be varied continuously and independently from zero to ± 150 volts, although these potentials, Vs and Vp• were in general maintained at values below 15 volts to prevent electron ionization from entering int~ the observ~d effects. It is important to note that nowhere in the re- gion above the cathode are there any potentials applied which can give to electrons energy sufficient to ionize the gas, except in -9- those cases where Vs and Vp were purposely raised above the ionizing potential of the gas being used. The currents to collector and plate, I 6 and IP' were measured by the two galvanometers, Gz and G3 , of sensitivity approximately 10-lO rurrp/rmn. By carefully working out the system of grounds and insulating the collecting potential batteries, it was possible to make wide variations of voltage on all three batteries without drift of the galvanometers. Whenever zero-drift was suspected due to change of potentials, a careful check of the zero readings was made immediately. The gas-handling system used for neon and argon is shown in Fig. 1. The gases were obtained commercially in reasonable pure condition and were placed successively in the reservoir, R. A small amount of gas introduced into the bulb M was glowed for hours between magnesium electrodes attached to a 15,000 volt transformer; sputtering was continued until a good spectrum of the gas appeared. From M the gas was then passed into the experimental tube through Stopcock §4 was of special design suggested by Mr. CJ. ancy for the introduction of small amounts of gas. The 11 pocket 11 within the head allowed the transfer of small quantities of gas with consequent facility in controlling pressure. c 1 A mercury cut-off, T, of the 11 suck-down 11 type was used to seal off the tube from the -pumps when desired. cock, 00 • This was operated by the two-way stop- Gas pressures were read on a MacLeod gauge whose connec- -10- tion entered the system on the pump side of the liquid air trap. Either liquid air or solid CO 2 were used on the trap during pUlnping and in the course of rillls. The whole tube was baked out occasional- ly, particularly between runs with different gases. The experiments with hydrogen and oxygen were carried out with commercial gases collected over P2o5 to insure dryness. Nitro- gen was generated from sodium azide, NaN , after the method described 3 by Tiede.23) A few grams of recrystallized sodium azide were placed in a side tube sealed to the reservoir system and after careful evacuation the azide was heated in a sandbath to 330°0. At this tempera- ture decomposition begins with the emission of nitrogen in pure form; sodium is left deposited within the tube. The nitrogen was collected over P o and admitted to the apparatus as needed. 2 5 -11- t Figure 1. 40 Cl'L Experimental tube and its connections. C F s G tl I I -p I I I I I I I .J J. + 1 1 - - - - + , . - - - - + - - - - - V, Figure 2. El.ectrical Connections. -12- METHOD OF HAlIDLI NG OBSERVATIONS. There are three variable features in the operation of this apparatus for any one gas which may be utilized: I) Variation of collecting potentials, Vs and Vp. II) Change of Pressure within the tube. III) Alteration of Accelerating Potential, Va. This suggests three types of runs which are significant in studying gases with this tube; i.e., variation of each of the previous factors while other conditions are maintained as constant as possible. All three modifica tions were used, al though by far the greater part of the work was done with the third factor, Va, as variable after the general characteristics of the phenomena of positive ion bombardment were 1mderstood from a study of I) and II). In any case, the observations consist in recording the currents received upon collector and plate, I 5 and Ip, together w~th the gas pressure, accelerating potential, collecting potentials, and filament current at the time. In the latter part of the work the total emission current, IE, was al so re corded by galvanometer G1 in order to study more carefully the conditions within the furnace. It was suspected that a glow discharge might be occurring unobserved within the furnace at the pressures and potentials used, affecting the accuracy of results and causing suspicion of a strong photoelectric effect which might prevail. This suspicion has been shown to be gro"Wldless, although the emission from the filament is undoubtedly affected by the presence of gas in a mai.,ner which needs careful study. -13- It is likely that the observations of Barton, Harnwell, and Kunsman22) regarding the emission from catalyst-coated filaments are not perfectly valid when gas is present around the filament. It is essential to analyze the currents arriving upon S and P for any particular set of conditions of collecting potentials and gas pressure, since variations in the ratio of these currents occur with changes in the values of Vs and VP; and inasnn.ich as the evidence for ionization is based upon these ratios, we should be able to interpret any change which occurs due to change of conditions in the tube. This has not always been possible to do in a quantita- tive manner because of the design of the tube, but it is comparatively easy to make a qualitative interpretation of results. I. V.ARIATI ONS IN COLLECTING POTENTIALS. .A. .ACCELERATING POTEllJTI.AL CONST • Plate Positive. l. Collector Positive: Vs > VP This proved to be the most easily interpreted condition and it is the one which was used in most of the work under heading III. In order to isolate the phenomenon of ionization, due account must be ta.~en of the secondary effects of electron emission from metals due to positive ion bombardment, photoelectric emission due to the presence of excited atoms, and reflection of positives. In order to eliminate these effects as far as possible, the plate was usually maintained slightly positive with respect to the grid (ground), and the collector was kept more positive than the plate. -14- Thus, electrons could not escape from the plate nor from the region bet\'!een the plate a."ld grid to the collector. The slightly greater positive po- tential of S kept any positive ions from reaching S by reflection from the plate or by stoppage of the incident ions from the filament between G and P due to collisions with gas atoms. Hence S collected only such electrons as were liberated from the gas between Sand G, together with any liberated from the grid by seconda:ry emission. The latter were small in amount, as may be seen from data taken at low pressures where collisions between gas atoms and positive ions were ra:re. (See Fig. 6-9) The ratio of the electron current, I 6 , to the positive ion current, Ip, was taken as a close approximation to the number of ions formed per positive ion within the region S-G at any particular pressure and accelerating potential. subject to corrections. Actually, both Is and Ip a:re As already mentioned, Is consists of elec- trons from the gas a.~d secondary electrons from the grid. 1 io;ization = Is (observed) - r;ec. emission. (1) The last term can be found from runs taken at low pressure ( < .0001 mm). The correction to Ip is not so simple, and the design of the tube does not perrr~t of an exact correction being made. Ip consists of initial positive ions arriving on the plate from the fila.'Ilent, diminished by electrons liberated from the gas between G and P, and also diminished by any reflected positives leaving the plate to discharge through the grid to ground. Furthermore, some of the initial positive ions may lose too mu.ch energy in passage through the gas to -15- reach the plate against the reta.ra_ing potential, VP+. The number of ions stopped within the gas cannot be exactly calculated unless an assumption is made that the sa~e per cent of the total emission passes through the hole from the filament at high pressures as at low. Iinitial ~ Ip(obs) + IIonization + I~efl. + rtcattered (2) The first correction term is approximately calculable on the assumption that the positive ions a.re as effective ionizers above the grid as below. Then, Iionization = kIS, ffi8,tely 3 for this apparatus. where k - ~~I: ~a' approxi- The next term seems to be negligible in amount, as may be seen by making Vs negative and atteErpting to collect reflected positives at low pressure. The last term is not accurately known, but judging from the ratio of Ip/ Ig at various p ressures, it is not a very· large factor until pressures above • 7 mm are reached. It then mounts rapidly in importance as the gas becomes more and more 11 opaque 11 to the positive ions. Neglecting the last two terms of Eq. 2, we can get a fairly close approximation to the ratio r-/ r+ in terms of observable galvanometer currents: I- = Is(obs) - Isec. emiss. :+ ____- ~ - - - - ~ - - - , approximately: I ( Is(obs) ) +kl ( -) I Ip(obs) + kis(Obs) p obs s Obs 2. VP > Vs ) O. (3) Collector still positive. Where Vs is less than VP, the ratio of I 8 / Ip changes rapidly. The calculation of corrections is complicated by the fact -16- that an unpredictable fraction of the positive ions formed or scattered between G and P succeed in reaching the collector S. An approximation utilizing the rel a tion between VP and V8 gives only partial success in interpreting results . .1..:. = Is(obs){ l+k(l-Vs/Vp) - Isec. emissJ r+ Ip(obs) + k Is(obs) . 3. (4) Collector Potential Negative. When Vs is negative, VP positive, the interpr.etation of the ratio Is/Ip is not so clear as in Case 1). tive ions. Snow collects posi- All positives formed by ionization throu~hout the whole volume S-P, plus positives from the initial beam which have been retarded too much to reach P, plus reflected positi ves,--all reach the collector. In now consists of those positives from the original beam . which succeed in reaching P, diminished by reflected positives and by all electrons released from the gas by ionization. The ratio Is/Ip consequently increases materially and appears to be very sensitive to slight changes in gas pressure. This increase may be seen in Fig. 3 in the case of air at .65 mm pressure, Vp + 10 volts, and Vs ranging from +50 to -50 volts. To obtain the true I1onization I +.1D1't'al l when Sis negative, we analyze I 6 and Ip as follows: Is(obs) =lionization+ Itcattered + I+refl. (5) I p (obs) ~ Itnitial - 1Ionization - r;ec. em. from grid - I~efl. (6) Neglecting scattering from the primar y beam and the minor items of secondary emission from the grid and reflection of positives from -17- the plate, we have a fair approximation: 1ionization - (7) r+initial This ratio should be roughly four times as great as the ratio r-;r+ found for Vs ) Vp• since the ~,hole volume S-P is approximately four times the partial vol1m~, S-G from which electrons were collected in the previous case. B. Plate Potential Negative. 1. Vs Positive. When the plate potential is made negative, secondary emission from the plate enters as an important correction to the current received on the collector. With no gas present, one collects secondary electrons on S so long as Vs is positive. I p consists in the initial primary beam of positives, increased by the amount of secondary emission. Under these conditions a rough check can be made on the experiments on secondary emission at low pressure and the efficiency for this plate can be calc-ctl.at ed. f = I-/r+ = Is(obs) 1p (obs) - 1 s(Obs) ( 8) When gas is admitted, Snow collects the electrons liberated from the whole volume of gas between Sand Pin addition to the secondary electrons from grid and plate. I p consists in the primary beam, in- creased by the secondary emission and by the positive ions liberated from the gas. lionization= Is(obs) - r;ec. em. -18- (9) The latter term is calculable from low pressure observations= f Ip+. Iinitial = Ip(obs) - Isec. em. - !Ionization. (10) These last two correction terms are, however, nothing but the observed Is· Hence, to good approximation the correct ratio for ionization is: = 1 s(obs) - f Ip(obs) 1p(obs) - 1 s(obs) (11) in which 11 f 11 is the secondary emission factor determined at low pressure. 2. O>V 8 >Vp• ~ Vs and V-o negative. This is a transitional region in which the ratio Is/Ip changes rapidly. With the present tube it is difficult to interpret the ratio quantitatively with accuracy. The ratio decreases since the collector now begins to receive positive ions from the space S-G and also retards some of the electrons from above the grid. Finally Vs becomes sufficiently negative so that the positive ion current pre- dominates and the ratio Is/Ip changes sign. 3. \V sl > \Vpl Both negative. Vs now prevents any electrons liberated above G from reach- ing S. I 8 becomes pure positive ion current due to positives liberated between Sand G, plus any from the primary beam which are sufficiently retarded by collision to be affected by the field of the collector. I ) re-presents those positives which get through to the plate from the 1 filament (practically all of them, since VP help s them along), plus positive ions formed above the grid by ionization, plus secondary emission from the plate, the current from which must go out on the grid. -19- Iionization = 1 s(obs) - r;cattered 1tnitial (Collected from region S-G) = I p (obs) - k I!onization - r;ec. em. The last term of the previous equation is ap~roxirnately calculable as This correction is good for reasonabl y low gas pressures when Ip is composed chiefly of initial ions from the filament. A fair approach to the correct ratio for ionization in this case is then: . Iionization Itni tial = (l + f) 1 s(obs) (12) No specific mention has been made regarding photo-electric effect upon the metal parts. This correction may be included in most cases under the heading of "secondary emission," except that it may be dependent upon gas pressure whereas the secondary emission was assumed not to be. The foregoing analysis of currents and their ratios for various collecting potentials has been given considerably in detail in order to show what the characteristics of the apparatus are. Figs. 3 and 4 show how the ratios vary for a particular case when air was used under an accelerating potential of 550 volts and .65 mm pressure so that am9le ionization would be detectable. It would be a tedious process to repeat these curves for every gas pressure and accelerating potential desired, since a large number of observations are represented in each case. Consequently, most of the work was done with constant collecting potentials, V6 ) Vp, with both positive as analyzed in Case I A 1. -20- II. VARIATION OF PRESSURE. COLLECTING .A:ND ACCELERATI NG POT 1 L CONST. A few runs of this type were taken in order to determine at what pressure the effect of ionization set in. No appreciable ionization was found below .1 m.~ in general, a fact which is remarkable when one turns to mean free path considerations. This point will be discussed at length in a later section of this paper. Fig. 5 shows the characteristic observations taken in this type of run. The chief information to be gained concerns the range of pressure through which the effect is present. A great many curves of this t y-pe could be conrpoun.ded from chosen data in a large number of runs taken under III. III. VARIATION OF ACCELERATING POTENTIAL. Vs, VP, .AND PRESSURE CONST. The most systematic method of measuring ionization was to set the collecting potentials at chosen values and vary the accelerating potential, Va• from zero to the maximum value which was usually 750 volts; in a few cases, potentials to 1100 volts were used. Care was taken in adjusting the zero readings of the galvanometers and efforts were made to detect the first trace of ionization. Since the currents to the collector were feeble, it was not always possible to locate the exact inception of ionization although it never appeared below 100 volts. also. As Va was increased above 100 volts, I 8 increased In the previous analysis when V6 and VP were both positive, the currents to the collector were shovm to be as free as possible from secondary phenomena. Hence most of the results for the gases studied were based upon this system of -potentials, with Vs > VP. -21- After changing Va by steps from zero to 750 volts, the pressure was altered and the process repeated. Several curves were plotted on this basis for each gas at different pressures, as in Figs. 6-9, clearly showing how the ratio of 1-/1+ increases with increasing pressure and Va· The lowest curves are those taken when the pressure was too low for ionization to play any important part; hence they represent the part played by secondary emission from the grid. These curves have been corrected according to Eq. 3. In order to reduce these various curves to a comparable basis, the method suggested by Compton and Van Voorhis24) was utilized. The values of the ratio 1-/1+ for each curve were divided by the pressure in millimeters of mercury and by the distance in centimeters over which the products of ionization were collected, thus giving the number, N, of fresh ions formed per initial positive ion per centimeter path at one millimeter pressure. N = lionization p X L X 1 tnitial (13) For the case where the ions were collected only from the region between collector and grid, L was approximately one centimeter. number, N, may be called the 11 efficiency of ionization. 11 This The values calcuJ.ated in t hi s way m._9,y be several per cent hig11; they are no:t likely to be too low. The reason for this lies in the fact that the true value of l!nitial which produces t he ionization cannot be accuratel y determined on account of scattering, is probably too low, making N too large. and the observed val 1.1e Another correction term of unlrnown value is that of ionization produced by neutral atoms which -22- started as K+ ions and were neutralized after havin~ fallen through all or part of Va. Theoretically, such fast moving atoms would be better ionizers than positive ions moving with the same velocit y, but they would not be recorded in Ip. A few additional remarks may clear uP certain points of procedure: 1. The grid and cathode were always maintained at zero 2. .All calculations of efficiencies have been corre c ted -potential. for the difference in gal vanometer sensitivities on G2 and G3 . 3. • variations in Vs b oth posiSo long as Vs + > v+ :P , wide tive and negative appear to have no effect upon the positive ion current to the plate when no gas is present. Evidently the + ion stream is not accelerated or retarded by Vs, due no doubt to an equipotential 11 core 11 between the ridge on the cathode and the grid through whi ch the ion beam passes. W'.o.en gas is present there is sufficient scattering and retarding of the positive ions to throw some of t hem into control of Vs· 4. So long as Va+>±Vp, the current to Pis unaffected by wide vari ations in VP' both positive and negative, wi th no gas present. 5. The total fila..'1lent emission appears to be reduced a few per cent by the presence of gas, after a slight increase when the gas is first admitted. This reduction is perhaps due to a slight temperature decrease caused by conduction of heat from the filament by the gas . The decrease is contrary to what would be expected if ionization within the furnace were occurring to any -23- appreciable extent. 6. The ratio I sf Ip for V/ > VP-+ is essentially independent of the value of Ip: an increase in Ip of one-hundred fold changes the ratio by only 3%. Hence small changes in the emission do not affect the results materially. 7. The practical limit of sensitivity of the apparatus as used was approximately .001; i.e., currents to S less than .001 those to P could not be detected . .An. electrometer in place of G3 would in- crease this sensitivity. -24- RESULTS. I. VARIATION OF COLLECTING POTENTIALS. To illustrate the analysis undertaken on pp. 14- 7.0 , curves are given in Fig s. 3 and 4 showing the measured Is/Ip for Air at .65 mm pressure with Va cons·~e.:nt at 550 volts, together with curves taken wider the same conditions of potential but with the gas pressure below .0001 mm. The corrections indicated in the previous pages have been calculated for this ca se and are recorded in the following tables. Greater reliance is placed upon the a ccuracy of the experiment when Vp is positive tha.n when it is negative; the r atio in the latter case is undoubtedly too l ar ge. With a different placement of the Plate, as mentioned previ ously, better corrections to the vp- case could be made. It will be seen that the lat i tude of variation in the ratio Is/Ip is greatly reduced in the corrected ratios; and although not all that might be desired, the corrections applied appear to work consistently in the ri ght direction. The small rise in I-/I+ at ±50 vol ts may be due to ionization by electrons. TABLE I. VP = +10 v. Va= 550 v. Pr. ,., . 65 mm Air. Correction Is/Ip corrected applied. Eq. 3 .096 II .080 II .080 II .076 .141 Eq . 4 .080 Eq. 7 II .099 JI .102 vs 1:si Is Is/Ip 50 25 15 10 10.5 5.9 10.2 6.5 5.9 14.9 1.6 .7 1.2 .7 .4 -7.0 -5.55 - 6.65 .152 .119 .118 .108 . 068 .47 .66 .69 5 - 5 -25 - 50 8.4 9.6 -25- TABLE IL . VP : -10 v. Vs Ip 50 25 15 10 -15 23.4 14.6 20.2 16.8 16.3 14. 7 20 .2 -25 -50 II. Va= 550 v. Pr. = .65 mm Air. Is Is/Ip Is/Ip corrected Correction applied. 6.3 .269 .226 .223 .208 .098 .095 .118 .228 Eq. 11 3.3 4.5 3.5 -1.6 -1.4 -2.4 .174 II .168 .148 .153 .147 II II Eq. 12 .203 II II VARIATION OF PRESSURE. In Fi g. 5 are some collected results for argon, neon, nitrogen and air, showing the ionization as a function of gas pressure with Va chosen at 750 volts so as to insure sufficient ionization. As will be seen from the curve, little measurable ef- fect appears for pressures lower t han .1 mm, a result which is striking from the viewpoint of kinetic theory since the expected mean free path of the ions is so short at these pressures where the effect of ioniza tion ~ppears. This will be discussed in the next section. Pressures were used in a few instances above 1 mm, but most of the work was carried out between .1 and .7 m.~. Above .7 rrnn the gas seems to scatter the initial beam very strongly and Ip falls off rapidly a s may be shown by the decrease in the ratio Ip/I g; Furthermore, at the higher pressures there was greater danger of arcing at the pote ntials used, with consequent hazard to the sensitive instruments. A few mi shaps occurred on this account. - 26- III. VARIATION OF ACCELERATI N~ POTENTIAL, Va. The collected results of a large number of runs in various gases are shown in Figs. 6, ?, 8, 9. The corrected ratio Is/Ip is plotted as a function of Va for a number of pressures in each gas. All of these results were obtained with VP~ +10 volts, Va• +15 volts; the correction applied is the one derived on Page 13, Eq. 3. There is a great similarity between the results for different gases as far as general characteristics of the effect are concerned; the chief difference lies in the numerical values of Is/Ip for the various gases used when compared a t the same pressure and accelerating potential. The curves for .Argon indicate tha t the maximum efficiency is not far above 750 vol ts. HYDROGEN. Hooperl3), Gurneyl4), and some of the previous workers attempted to ionize Hydrogen by positive ions without success. They all worked at far lower pressures tha.,.~ those used in these experiments, so it was of interest to see whether Hz could be ionized in this tube under the conditions which yielded unmistakable ionization in Neon and .Argon. The results were negative, and although pressures as high as 1.2 mm were used and accelerating potentials up to the arcing voltage were a~plied , no ionization could be distinguished from secondary emission fro m the grid. The collected results of the efficiency of ionization, N, caused by K+ ions in t hose gases studied are shown in Fig . 10 These curves are calcula ted from the curves shown in Figs. 6-9 according -27- to Eq. 13, dul y averaged. Comparing these values of N with those obtained by Compton and Van Voorhis 24 ), we see that ionization by positives is far less efficient than by electrons. None of the values of 11 N' 1 in Fig . 10. have reached a maximum at 750 vol ts, whereas electron ionization pa sses t hrough its most efficient values at far lower voltage. In the following table are given the values for N~. for electron bombardment at the potential for the maximum, together with the value of N for K+ bombardment in the same gases at 750 volts. The relative efficiencies of the two processes will readily be seen, that for 750 volt K+ ions being in every case less than 4% of that for electron ionization at the respective maxima for the various gases. TABLE III. Gas N:"'laX. Vma.x. N750 (K+) 1'1750/N~. Argon Neon Nitrogen Hydrogen 10.33 3.22 9.96 3.55 140 340 175 145 .288 .112 .124 None .028 .035 .013 -28- 0 ,., . .' ,._ ,--~, -, - - I I .. ) - ' · - r,-----,-r - I Iv';,_ +.s I ,_o; __ A, •. -~ --~7 . . •I I I I I _ _:_o __ ...., __ f"oool >,un. Vp = + 10 volts. ,. - ,_ I C I I ~ it"""' Ill Ys - -- ff- -.10 - - - ~ 0 - - ., I ,__. - I_ .1 -- : -1 -I I Variation i n I /I s ' with change in potential of the p Vp= +10 v. Va= 550 v. Pressure= .65 mm Air. - -- I __ .:1 I - - 0 - ---o- ____(. ,z ooaJ 1nm. ~=-1/0~}+s. I •I I ' Vs I 1, I ~ ½~,. ' -iS>--e..-,+- .bStnmA11' -S( .J ·~ ~ , I 1- ,_ -1.0 collector. J ""'. - ,_ Figure 3. I : ,./' , -~ / I ; +~ 0 'is ' .. ..... - - - lt5 r I t...--~- - - - .. -- I -- ~~ ,__..J___ - - - o - - - ---- ---- J -I t - - ./ ,..-_.t- 0 -25 ZS i +5 ' .... I . = 0Vs . .... - .I ' - I ... I . l ~ -,3 I Figure 4. -Variation in I 8 /Ip with change in potential of the collector. Pressure = .65 mm Air. - .7 I .6 • - '- - .5 I I ~l- I I . I I I ~ .1 I . I .05 I - I I I 'T' I I I/ I r t v 'i .. .2 I ' I I V • • • -- -- • .1 :Pres.sur>e (mm fig.) •• Ar-gon 1 I I · -,,. 1/j I I l I NihoSeri I!> I / I I I I ~Y // ✓• Neon .- • X I .5 I I 1 Figure 5 . Variation of ionization in neon, argon, and nitrogen plotted against the log of gas pressure. I /I has not been cor rected in this case for ionization s p A similar graph for air would appear very occurring above the grid. close to that for nitrogen. Is/ tfp corrected .OD JOO 0 t oo 3 00 -too Joo 700 800 At'celerdting potential (volts) Figure 6. Ionization in argon at several selected pressures, plotted against the accelerating potential. I /I s p has been corrected in each case shown in the curves of Figs. 6-9 according to Eq. ~. .00 L---lilllf:L..---c::::::lC:..___.c,1,c:==.::J,--,,.!-----,-L----,,,i..-~ JOO too 300 +oo Joo 1,00 7 o Accelera lD otental ( 1~ Figure 7. Ionization in neon at several selected pressures, plotted against the accelerating potential. Nitro<3en .a t - - - - , - - - - - 1 - - - - + - - - - 1 - -- - + - - - - + - .lOt----f---------l----+---+----+---+ Is fr r .0 8 t - - ~ t - - - , - - - - + ---t-----+-,,__-+- + c:orredecl. .oo L - ~ ~ _ L_ _j_ _L:i:::::::±~7 o 100 ioo ~ 300 400 500 Accderahn~ potential (volts) Figure 8. Ionization in nitrogen at several selected press~res, plotted against the accelerating potential. " .IZ. r - - - - - - - - - . - - - - r - - - - - - - . - - - . - - - - - i - - ----,.----, Air .10 t---+---+--+---+---+---+---+----l ,0001 ,u, i-00 0 5 00 600 100 eao in~ potmt.ial (vC1lts) Fig...1.re 9. Ionization in air nt several selected pressures, plotted against accelerating potential. .30 At>gon' .20 ,-----t------j---+---f-1---+---+----l-------- N .15 r------t------t--------,!f--------t---+----+---+--~ z:oo 300 -1-00 > Acceleratin* paten ol (vo III Figure 10. Plot of N, the number of ions formed per initial positive ion per centimeter path, rsduced to 1 millimeter preaayre for argon, neon, nitrogen, and air. These curves are calculated from the curves of Flga. 6-9 together with a number of similar curves according to Eq. 1;, VALIDITY OF RESULTS . . In the section on Method of Handling observations, a considerable amount has been said concerning the treatment of secondary effects. One further curve is given in Fig. 11 showing the amount of secondary effect present in the ratio Is/Ip at low pressure under three conditions: 1) plate negative, collector positive, so as to collect secondary electrons from grid and plate on S; 2) plate posi- tive, collector negative, so that collector receives only reflected ( positive ions (less than 1% so long as Va>Vp); 3) both plate and collector positive, as actually used in the entire portion of the work shown in Figs. 6-9. The dotted line (3) in Fig. 12 shows the part played by secondary effects,- almost a negligible amount. The reasonable assumption is made that this secondary emission (with the exception of that due to photoelectric effect) is a maximum at low pressures. There are three other items which deserve consideration in this critique of results: 1) Collisions of the Second Kind; 2) Photo- electric Effect; 3) Metastable atoms. The first has been studied for positive ion collision by Harnwe11 2•); it is the effect observed in a mixture of gases of different ionization potentials wherein a metastable atom may ionize another atom whose ionization energy lies below that of the energy of excitation of the metastable atom. The ef- fect may be ruled out so long as the gases dealt with are pure. Also, the energy liberated upon neutralization of a K+ ion is not sufficient to ionize any of the gases used in these experiments. Second, any I' PPe.ssurQ <.0001 mm I I .SecondaPy ami:ssion \ (1) 1 7% P-9v. C+13.5v. • I ·Yi P Q ~--,--~~:::;:;;,"="""'_ _ _ _ _ _;;,,;;;;_=_=_;;;.._=.::_:.;;:_..:_:.:_:.:_:.:-0~(3:j) ( E) 'Re2fkction <1% P +9v. C - 13.5v. 0 Figure 11. 100 300 400 500 Va (volts) 600 100 600 The ratio I 8 /Ip as a function of Va for different arrangement& of the collecting potentials at preesure less than .0001 millimeter of mercury, showing effect of secondary emission. Curve (1) Secondary electrons from grid and plate collected on the collector, S. Curve (2) Reflected and scattered positive ions collected on the collector, s. curve(~) Amount of aecondary effect actually observed in rune where V8 VP, with both potentials positive. photoelectric effect upon me tais is probably dependent upon gas pressure since any radiation with energy sufficient to liberate electrons must come from highly excited gas atoms. Hence this may be a serious i mpediment to success, as all other secondary effects we re assumed to be maximum at low pressure. However, by keeping VP= 10 volts positive, no photoelectrons would be liberated with energy sufficient to escape to the collector. It is conceivable that the photoelectric currents from the pla.te in the case where Vn ~ was negative (compare Tables I and II, p. ZS- 2B) account for the larger efficiency of ionization in this case. In fact, changing the potential of P shoul d. , with suff icient refinement of measurement, serve to distinguish 9hotoelectric effect from ionization and secondary emission phenomena. Third, Brode r aises the objection that ion- ization may be p roduced by collisions between t wo metastable atoms and that such collisions might occur with great probability at the high pressures used. No check of this point has yet been de vised, and it stands as a valid criticism of interpret a tion. However, if a positive ion can impart energy sufficient to raise an atom to a metastable state upon collision, it would seem reasonable to expect some impacts to result in complete ionization although, admittedly, the ratio Energy required to Ionize Energy required to Excite is gr eater for positive ions than for electron s, due to the intense positive field from which the electron must escape i n t he former case. It has been questioned by some writers whether positive ions can ionize gas wi t hout acquiring the actual velocity obtained by electrons which ionize,--without passing through thousands of volts accelerating potential. A 100 volt K+ ion has the same kinetic energy as a 100 volt electron, but its velocity is that of a 0.00135 volt electron,--far below ordinary ionizing potentials. These experi- ments and those on secondary emission show without a doubt that positive ions of 100 vol ts a.rid over are able to affect both gases and metals, causing the liberation of electrons from them. This represents some five or ten times the energy necessary for electrons to produce similar results, and the efficienc~· of the ionization is far smaller than for electrons as shown in Table III. Nothing can be said at the present time regarding the probability of an impact between a certain velocity positive ion and a gas atom resulting in ionization, for we do not know yet what the mean free path of these positives is in the gas. We can calculate the efficiency of ionization expressed as the number of ions formed per positive ion per centimeter ·path at one millimeter pressure without knowing the mean free path of the ions. The abnormally high pressures at which the effect of ionization occurs, roughly ten times the pressure expected on kinetic theory calculations, is striking evidence that the kinetic theory mean free path calculat4ii~hi~\ for these ·positive ions. They apparently exhibit a remarkable pro-pensity for getting through the 5 cm path from filament to plate, even against a 10 volt retarding potential in the latter half of their path, without heavy scattering in Sl') ite of suffering 50-100 (kinetic theory) collisions. We must conclude that they lose only a small fraction of their forward momentum, on the average, upon an encounter with a gas atom. This abnormally high mean. free path for positive ions has been observed by other workers whose conclusions agree at least qualitatively with the observations of this experiment . .Among those who have dealt with this a spect of positive ion kinetics are DempsterlO), Kennard26), Du.rbin27 ), and Harnwe11 28 ). Similar studies for slow electrons (one paper on positive ion mean free paths) have been carried on by Ramsauer29). The first three observers worked with magnetic analysis apparatus which is better suited for analyzing the mean free path than is the one used in these experiments. They give mean free paths up to ten or fifteen times the kinetic theory val11.es for slower positive ions t han those used in this experiwent. -40- CONCLUSION. The success of this experiment in detecting ionization by ~ositive ions where other workers failed may be attributed largely to: 1) Consistent efforts to account for and to eliminate secondary effects. 2) A convenient hot-anode source of positive ions. 3) Extension of observat ions into higher pressures. The phenomenon of positive ion ionization has been shown to exist in several gases for K+ ions above 100 volts, and the efficiency of the effect has been calculated to be less than 4% of tha t for electron ionization in the gases studied. There still re- main for future work the development of greater refinement and variety of methods, the extension of observation to a larger number of gases and to a variety of sources of positive ions; and finally, adequate theoretical interpretation of results. It is a pleasure to t he author to acknowledge his indebtedness to Dr. R. A., Millikan for his interest and hel pful discussion of the problem; to Mr. Willy Uyterhoeven and Dr. L.B. Loeb for their constructive criticisms; and to Mr. William Clancy for his assistance in the glass blowing; and to Mr. J. C. Mouzon for his help in the latter part of the experimental work. Pasadena, California, May 10, 1929. REFERENCES. 1) J. Franck and Cario, Zeits.f. Physik 25, 312, 1924 2) Joos and Kulenkampf, Phys. Zeits. 25, 257, 1924 3) C. Eckart, Science 62, 265, 1925 4) J. J. Thomson, Phil. Mag. 48, 1, 1924 5) Bahr and Franck, Verh. d.D. Phys. Ges. 16, 57, 1914 6) W. J. Pawl ow, Proc. Roy. Soc. 90A, 348, 1914 7) Horton and Davies, 8) A. J. Saxton, Phil. Mag. 44, 809, 1922 9) A. J. Dem9 ster, Proc. Nat. Acad. Sci. , 2, 374, 1916 II II 95A, 333, 1919 Phys. Rev. 29, 108, 1926 Phil. Mag . 3, 115, 1927 II 10) II 11) J. T. Tate, Phys. Rev. 23, 293, 1924 12) E. J. Jones, Phys. Rev. 31, 611, 1927 13) Jr. Franklin Inst. 201, 311, 1926 14) w. J. Hooper, R. w. Gurney, 15) L. B. Loeb, Sc ience 66, 627, 1927 16) A. L. Klein, Phys. Rev. 26, 800, 1925 (or C.I.T. thesis) 17) w. J. Jackson, Phys. Rev. 28, 524, 1926; 30, 473, 1927 18) J. 19) M. L. Oliphant, Proc. Carob. Phil. Soc. 24, 451, 1928 20) G. E. Read, Phys. Rev. 31, 629, 1928 21) R. 22) Barton, Harnwell, Kunsman, 23) Phys. Rev. 32, 1:fl5,. 1928 M. Hyatt, Phys. Rev. 32, 922, 1928 w. Gurney, Phys. Rev. 32, 461', 1928 Phys. Rev. 27, 739, 1926 Chem. Ber. 49, 1742, 1916 Tiede, 24) K.T.Comp ton & Van Voorhis, Phys. Rev. 26, 436, 1925 25) G. P. Harnwell, Phys. Rev. 29, 583, 1927 26) R. B. Kennard, Phys. Rev. 31, 423, 1928 27) F. M. Durbin, Phys. Rev. 30, 844, 1927 28) 29) G. P. Harnwell, C. Ramsauer, (Abstract) Phys. Rev. 31, 634, 1928 -42- Phys. Zei ts. 28, 858, 1927