The Development of a High Intensity Magnetic Lens Type of Mass Spectrometer Citation Rumbaugh, Lynn Hamilton (1932) The Development of a High Intensity Magnetic Lens Type of Mass Spectrometer. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/8s1e-8012. https://resolver.caltech.edu/CaltechTHESIS:09132024-224005516 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): Smythe, William Ralph Thesis Committee: Unknown, Unknown Defense Date: 1 January 1932 Record Number: CaltechTHESIS:09132024-224005516 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:09132024-224005516 DOI: 10.7907/8s1e-8012 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16728 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 17 Sep 2024 16:30 Last Modified: 17 Sep 2024 17:38 Thesis Files PDF - Final Version See Usage Policy. 12MB Repository Staff Only: item control page

THE DEVELOPMENT OJ!"' A HIGH INTENSITY MAGNETIC LENS TYPE OF MASS SPECTROMETER by I.ynn H. Rumbaugh A thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy California Institute of Technology 1932 I THE DEVELOFBMEN T OF A HIGH INTENSITY TuIAGNETIC LENS TYPE OF MASS S'P8C'i'ROME 1'ER. I. Introduction II. The Theory of the iviagnetic Lens III. A.. The Type I magnetic lens B. The Type II magnetic lens Factors Affecting Focal Sharpness A. Focal defects originating in the magnetic system 1. The focal defect analogous to coma in optical systems 2. Edge effects from the fringin g field at pole piece boundaries B. C. IV. 3. Irregularities in the pole pieces 4. Variation in total magnetic flux Focal defects orig inating in the ion beam 1. The thermal distribution of ion velocities 2. Magnetic effects near t he ion source 3. Collisions and ionic interactions Summary The Apparatus A. The magnetic system 1. Pole piece construction 2. Windings and the control of magnetic fields B. The vacuum system C. The ion collector D. The source block and electric field plates II V. The Construction of Extended Equipotential Ion Sources A. The elimination of magnetic effects B. Heating methods C. Mountings and construction for control of source potentials D. Source materials VI. 'l'he Control of Accelerating Potentials VII. Experimental Results A, B. Resolution 1. Studies of thermal velocities 2, Discussion of resolution curves Intensities at the collector 1, Electrostatic cross-focusing as a means of increasing intensities VIII. The space charge limitation 3, Widths of collector slits Possible Applications of the Instrwnent A. The search for faint isotopes B. 'rhe collection of isotopic samples C. IX. 2, A. Nuclear studies B. Determination of radioactive isotopes Problems requiring intense ion beams Conclusion I. INTRODUCTION The identification and study of isotopes constitute one of the interesting and important branches of contemporary physics. In the cases of the three radioactive series, nature furnishes us with pure isotopic 'species, but, for the non-radioactive elements, the isotopes occur always in a mixture of constant proportions. Many attempts have been made to separate quantities of such isotopes. The dependence of diffusion and evaporation rates upon mass has been used with partial success. By the first of these methods Aston{l) showed neon must be composed of at least two isotopes. By repeated applications of the second method Bronstad and He~esy(2) obtained samples of mercury differing from each other in weight by about O.O5 per cent. Numerous other attempts, both success f ul and unsuccessful, are reported in the various scientific journa ls. Ii. long list of ingenious but unsuccessful methods depending on centrifuging, photo-chemica l effects, ionic mi grations and the like is g iven in Aston's "Isotopes n(l). None of the above mentioned devices can give more than extremely incomplete separation. The use of positively cha r ged atoms and molecules in mass spectrometers is the only available method by which isotopes can be separa ted in the pure sta t e and their number and mass in a given chemical element determined. Various types of this instrument have been designed by J. J. Thomson, Aston, Dempster, Smythe, Bainbridge, and a number of other investigators( 3 ). All mass spectrometers developed heretofore must collimate the initial ion beam by me ans of narrow slits and in consequence the amount of isotopic materia l received at the detector is extremely minute. For 2 this reason, only measurements of mass and relative abundance are performed on individual isotopes and very faint isotopes may escape detection entirely. The obvious extension to obtain higher intensities requires a mass spectrometer of such construction that wide beams of ions may be utilized. Thip is accomplished by the magnetic lens type of instrument which will be discussed in the following pages. Two of many possible kinds of magnetic lenses have been devised by Dr. W.R. Smythe, and, for convenience in discussion, will be referred to as Type I and Type II. A Type I instrument was designed and construct- ed at the California Institute of Technology by Dr. Smythe, who used it during 1924 and 1925. He was assisted for a time by Dr. Dinsmore Alter. Dr. Smythe then began investigation of another problem and experiments with the Type I spectrometer were continued by Dr. H. M. Evjen from 1926 to 1929. He ma.de little progress beyond that attained by Dr. Smythe. The author began work with the apparatus in 1929. This thesis is devoted to the results of studies made with the Type· I instrument, and to the construction and operation of a successful spectrometer with a Type II magnetic lens. II. THE 'r:E·IBORY OF 'l'HE MAGNETIC LENS. A wide parallel beam of ions in which individual masses, m1 , have a common velocity, v1 , may be obtained by accelerating ions in a uniform electric field. Let one boundary of this field be a plane, the equipotential ion source, y -;::.V 0 • y ~ V. The other boundary may be the plane Then all ions of mass m1 and charge e, emerging at the upper boundary of the electric field have a common velocity, v1 :::. 2e(V 0 - V) X FIG.1 y FIC.2 3 and their trajectories are the straight lines x-=- a constant. Let these ions enter a uniform magnetic field whose direction is perpendicular to .the plane of the ion beam. Consider the boundaries such a field may have if all ions of mass m1 are to be brought to a focus at x~ 0, y =- C0 • Tvro of the sets of boundaries by which this may be accomplished will be considered. Type I Magnetic Lens If the plane Yl-:-:: 0 is chosen for.the first boundary of a uniform magnetic field, the equation of the second boundary surface may be written by inspection of Fig. 1: tan 9 :::. Jr 2 - Y 22 :e Co - Y 2 Y2 xz where r is the radius of the family of circles which the ions of mass m1 and velocity v1 follow in the magnetic field. If ions of mass m2 also are present in the beam, the~ follow circles of radius rz in the magnetic field and hence their paths do not pass through the principal focus, x ~ O, y ::. c 0 • The distinguishing feature of the Type I magnetic lens is that ~11 ion paths are perpendicular. to the first boundary of the magnetic field. 1'ype II Magnetic Lens Suppose the first field boundary is the cylinder, x 1 2 - 2rx 1 t y 1 2 - 20 0 y 1 -::: O, (Fig. 2). Consider an ion whose path is x ;:::-k entering the uniform magnetic field at (x1, Y1) on this boundary. In the field the ion tr~vels a circular path whose equation is 4 G' - (:iq - r)] 2 ;--- [Y' - y~2= r2 , and leaves the field at (x 1 ,y') ~ (x2iY2) • In the field-free space its path must be the straight line, y - Co::- Y2 - Cox, x2 if it is to pass through the principal focus, x ~ O, y =- C0 • These three equations may be solved for the equation of the second boundary of the field, which is This is the equation of a circular cylinder whose axis passes through the principal focus. Hence the distinguishing feature -of the Type II magnetic lens is that all ion paths convepging to the principal focus are perpendicular to the second boundary of the magnetic field. III. FAC'l'ORS .A.FFECTUJG FOCAL SHARPNESS. The above equations for the boundaries of the magnetic field were derived from a special set of assumptions: that the magnetic field is uniform and sharply bounded; that the constituents of the ion beam have parallel trajectories; that all ions of the same mass have common velocities upon entering the magnetic field. Under such conditions there exists a sharp focus - we have called it the principal focus - to which all ions of a certain mass, m1 , converge. These ideal conditions c~nnot be realized in practice and focal defects result. Since it is obvious that focal defects can originate only in the magnetic lens and in the ion beam, let us consider first those inherent in a magnetic lens. 5 FOCAL DEFECTS ORIGINATING IN THE :tr.tAGNETIC SYS'l'EM. All possible causes of focal defects which can be traced to the magnetic system may be grouped into four classes: (1) lack of focal sharpness for ion be ams converging toward regions other than the principal focus; (2) frin g ing fields outside a. gap between pole pieces; ' (3) irregularities in the pole pieces; (4) variations in the total mag- . netic flux. We shall consider these classes in the order named. Suppose there are ions of masses different from m1 present in the beam. m3 If the paths of these ions - say of masses m2 7m 1 and < m1 - are computed, they will be found to pass above and below the principal focus respectively. found to be not sharp. The focal regions for these ions are If we use the terminology ordinarily associated with optical systems, we may say that something analogous to coma exists for those beams converging off the principal axis. In the Type I magnetic lens these focal regions are diffuse and occur at distances from the lens which are considerably different. (See Fig. 1). This factor materially decreases the linear separation between the various components of the ion beam and, if the ratio m3/m1 or m2/m1 approaches unity, t he separation becomes so small that complete resolution may be accomplished only for a beam of limited breadth. Furthermore, if two or more isotopes are to be collected simultaneously, only one can be in definite focus. Should the collector not be located fortuitously on the principal axis, even this single, definite focus will never be realized. Fortunately, this defect is removed a lmost entirely in the Type II lens, (Fig . 2). The focal regions over considerable distances on H H_=_H., -.6 0 I. 2 3. 4. cm. from pole piec.e- FIG. 3 -----Yii I I I I It ~I "~ EI ectr ic ,;I • field / I FIG. 4 6 both sides of the principal focus are comparatively sharp. Moreover, these regions are approx1.mately equidistant from the lens. Consequenlt- ly, collectors located off the principal axis can be utilized and several isotopes may be collected simultaneously. It is impossible to realize a sharp boundary for the magnetic field across a . gap between pole pieces. The central distribution of field outside a gap 6 millimeters wide between plane pole pieces, as computed from a. Schwarz tre.nsformation, is shown in Fig. 3. The effect of such distributions must be investigated and an edge correction applied to the pole pieces. In the Type I lens the fringin g field sce.rcely can be calculated, for the upper pole piece boundary is not a syrnmetrical surface. The Type II lens has circular boundaries so that the field outside the gap may be represented by a r a dial function, H(r); i.e., the field distribution ce.n be a function of the perpendicular distance to the pole piece bounde.ry only. The distribution function, H(r), may be obtained in two ways: by constructing pole pieces similar to those to be used and measuring the field at many points; by computations utilizing a Schwarz transformation. The first method is expensive and cannot g ive the true field a t a particular point because measurements must be made with a test coil of finite area. '£he second method assumes the pole pieces to have infinite permeability and is rigorously applicable only to plane pole piece boundaries. In neither case does it seem possible to express H(r) in a finite number of integr able terms. solution for the path of an ion • probably impossible to obtain. :in; In consequence, a rigorous the field H(r) is extremely difficult, 7 It is possible to approximate a circular pole piece boundary by circumscribed and insribed regular polygons. The magnetic field near the pole piece boundaries now is divided into regions which are regular trapezoids. In these regions the distribution of field is given in Car.tes ian coordinates as a function of a single varia.b le, the distance from one of the sides of the polygon. 'rhis reduces the problem to the calculation of successive short ion paths f:i;-om re gion to re gion. Furthermore, by suitable orientation of the circumscribed and inscribed polygons, upper and lower limits may be computed for the circular case. Thus, suppose a particle of charge e, mass m, and velocity v 0 in an x y plane enters a magnetic field, H(x), whose direction is parallel to the z axis. Let the initial angle between the path of the particle and the x axis be r/0 when the particle is at x0 • Then the magnetic field imparts a radius of curvature to the ion path. given by r =- Fds =- sec r/ dx =- mv 0 uY1 ~ He x Hence, sin r/0 - sin r/ = _e_ mv 0 / , H(x)dx Xo Successive applice.tions of this equation may be used to calcula.te the path through the magnetic field. Sj.nce the computation depends more critically upon the relative values of the field at the boundaries of the trapezoidal regions, it seems advisable to obtain the distribution function, H(x), from a Schwarz transformation. This distribution function is corrected to approximate the circular case, and curves similar to Fig. 3 are used to evaluate the various integrals by mechanic8.l methods. Similar, but rather involved, calculations may be made for the 8 effect of the change in direct.ion of the frin ging field for ions entering near one side of a pole piece gap. The important point is that the sum of all edge effects may be reduced to such form that the corrected pole piece boundaries may be turned in a lathe. This property of the 'fype II lens is preferable to that of Type I which must be cut point by point, in a milling machine. A third possible defect in the magnetic field may arise from irre gularities in the pole pieces. It is required that the pole piece material shall be homogeneous and that the pole piece faces sha ll be plane ·and uniformly separated. The first property is possessed to a very high de gree by Armco iron. The lower limit to the second requir&- ment is set by machine shop practice. It may be assumed that no surface irre gularities greater than one thousandth of an inch will occur and these may be made smaller if necessary. These irre gularities are usually tool marks which are so many and re gular that their integrated effect is nil, even if the effect of stresses in the magnetic field in insuring further uniformity is disregarded. The mean separations • of successive small areas of the pole piece faces may be made uniform to one ten thous andth of an inch or better, which is ample. 'fhat the tota l flux in the magnetic system may be held exceedin gly constan t has been demonstrated by Aston and many other investiga tors. They obtain sharp lines in mass spectrogr ams requiring long exposures. We shall assume that varie,tions in total flux may be elimin- ated by proper experimental technique. l!, OC-1\L DEFECTS ORIG I NATING I N THE ION BEJ\M The defects in the ion beam must be associated solely ·with lack of. parallelism and lack of common tra jectory velocities for ions of 9 identical mass passing through the magnetic field. It may be assumed that a uniform electric field and an equipotential ion source are obtainable. The remaining factors to be considered are the thermal distribution of ion velocities, magnetic disturbances near the ion source, and collisions and ionic interactions. The ions in the beam will have thermal velocities given by the Maxv,ell-Boltzmann distribution law. The thermal velocity component para.llel to the e.ccelerating field may be eliminated immediately, for thermal velocity components average only a few hundredths of a volt when the source is e.s hot as 2,000° K. Such a velocity component becomes negligible when as much as 100 volts accelerating potential is used. Hence the common trajectory velocity criterion is fully satisfied. The thermal velocity components perpendicular to the electric field are serious since they materially affect the angles at which the ions enter the magnetic field. Suppose an ion has a velocity component 1, perpendicular to the accelerating field and parallel to the ion source. Its path in a uniform electric field will be a parabola and it will leave the field at an angle Ag (Fig. 4) with respect to ♦hose ~ ions for which~ is zero. The equations of motion give, tan ( 6 9) = JI/ ( ~ m,_ ::. ~ 9 2Ve Where V is the accelerating potential, e the char ge and m the mass of the ion. If V ~ 1 e.s.u. of potentia¾6 9 will be small. In the magnetic field this ion will travel along a se gment of a circle of radius rand hence will be devia ted through some angle1 o<.. Upon emer ging from the magnetic field the ion will proceed toward the focal plane, making an angle I),. 9 with the path of an ion of the same mass which has no perpendicular velocity component, 7. This latter ion 10 may be made to follow the same path from the magnetic field to the focal pla.ne if it is accelerated first through a potential V -f- -6. V while the magnetic field remains constant. In the electric field,½ mv2:::. (V+ LJ. V)e, and in the magnetic field, mv 2 (r--t.e.r) :::.He v, where His the magnetic field strength, Then ~ r -::; .6 V , for Ar is small when 6 9 is small. It is easily shown by geometry that cos2( A 9) -:: 1 - (___.!_)2 sin2 o<. . Hence 2V r ~ 9 ::: 4 r r sin oc. = 4 V 15 ~fl~ V 2r :hr-J:- 2Ve t The number of ions having velocity components between ~ and 1-t, -t- d ~ is given by the Maxwell-Boltzman law, .. The total beam arriving at the principal focus is received after its components have been deviated through all values of ~ such that d. • (L:i. Be Then <"' ,--~~,· ~ ~ -V)->:xV -(- V T Since r and 0( depend upon the construction of the maf;netic lens, /3 is a constant of the apparatus. If Let C t:.v =- k is zero, I!. Vis zero. dN dN 0- 1'hen Therefore B ::::. dN0 z. V -f3 ( Cl - V} ··•·.,.k _ e V T CAV := is a useful equation giving the shape of the curve measured ·with an electrometer at the principal focus when the accelerating potential is ~ rl 11 varied in steps of ( AV/v). isotopes of masses m1 and m2, If it is desired to resolve two given AV/v is the voltage-separation con- AV , when the magnetic field is kept stant; i.e., m2 - m1 ::: V1 - V2 V2 =- V m1 constant. Consequently, these isotopes will be resolved. only if a certain minimum value of accelerating potential, V, is exceeded. If we consider V:;:. V0 as the potential necessary to bring the most intense part of a given isotopic beam to the principal focus, it is seen that the sharpness of resolution will increase explmentially as V0 is increased, the magnetic field strength being correspondingly increased to keep the isotopic beam on the principal focus. Let us consider magnetic effects near the ion source; i.e., in the region of the accelerating electric field, as a possible factor disturbing the paralle 1 condition of the ion beam. The magnetic field in this region may have a number of causes, such as electric currents in the heater used for the ion source, remanent fringing field from the magnetic lens, and so on. We shall require that deviations in ion paths from the direction of the accelerating electric field shall be small. Hence, magnetic forces and magnetic field components parallel to the electric field may be neglected, and the equations of motion take the simple form, y e, where m and e are the charge and mass of the ion, y is the direction of the uniform electric field of strength Y, ~ a direction perpendicular to y, and H the magnetic field component perpendicular to bot:171 y and l . H will be considered constant for simplicity. From these equations it follows that an ion initially at rest 12 will emerge from the electric field ma.king an angle Ar/ with the dir- ection of y, given by H =(2 V my½~ e D ' where Vis the difference in electric potential across the distance D. Obviously, the above equation may be generalized for the case in which H is constant only along such surfaces as s.re electric equipotentials in the cross-section of the ion beam. Then, provided His nowhere greater than Hmax, H ~ (2 V m)½ ~ max e D Suppose that H = Hz is the magnetic component perpendicular to the plane of the ion beam. It may be shown that the focal regions in the neighborhood of the principal focus, C0 , will rema in comparatively sharp for the Type II lens when A r/ ~ 0 .04 radians for all ions. Thus if Vis 1,000. volts and Dis 3 cm., a potassium ion will fulfill this condition of an gle if Hz ~400. gauss. In general, this restriction on Hz is very easily satisfied. Even were Hz in exces:. of 400. gauss, it would be necessary only to rotate the direction of the electric field through an angle Llr;/ to bring the ion paths again tangential to the lines x = constant upon emerging from the electric field. (See page 3 ). In ca se H is variable in sps.ce or time, we can extend the equation for H to obtain the permissible variations. Define a certain value of Ar/ -= ( A9) 0 as the least angular sepa ration at the focal plane, under ideal conditions, between the pe.th of an ion of mass M, on the chemical see.le, and another ion of mass M "t 1. Now, (by page / o ) 13 AG ::. .tl r r sin(){ , where r is the radius of curvature of an ion path in the field of the magnetic lens and 0:-- is the angle of deviation. Then the equations of motion for an ion in the mass spectrometer give A ) ( <JG k 0 -:: ~ M2 n 0 -- K Sin (j_ o sin~ m2 To keep the resolution high, require~ that il cj ~ ( LlG)o. This 2o leads to an equation for the allowable variable magnetic field, f1v, in the electric field space: gauss radian' where~ is a constant which may be computed from the geometry of any given magnetic lens. In case the minimum value of a.. is 45 degrees, A > 103 gauss l cm. (radians)- 1 (volts)-2 . Then, if V -:: 1,000 volts, D -= 1 cm., Hv for potassium may be as large as 25 gauss. This may be interpreted as follows: under the above conditions, in the electric field space the magnetic field may be allowed to vary in any manner, so long as the difference of the absolute values of the angula.r deviatinns of any two potassium ions, in the plane of the ion beam, is not greater :t;han would be the angular deviation of a single potassium ion acted upon by a constant magnetic field, Hz, of 25 gauss in the sa.me electric field. Collisions and ionic interactions can have little effect on an ion beam if the vacuum is good. Collisions of ions with walls and gas molecules are proportional in number to the denasity of the beam, and so must give a background of about the seme relative intensity for any type 14 of mass spectrome ter, under the same vacuum conditions. Thi s back- ground can be made sma ll, in percentage of the total ion beam in a magnetic lens t ype instrument, by the same stan dard methods used successfully in other mass spectrometers. The possibility of electrical repulsions between ions in the be am a s it converges to the focal point may be ruled o~t by elementary ca lculations. These ions are not closely space d and are in the converging part of the beam for so short a time. that inter a ction is negligible. SUMMARY The object of the above discussion is to make cle ar that certain very definite conditions must be satisfied if any magnetic lens type of mass spectrometer is to operate successfully. The most rigid of these conditions is set by the Maxwellian distribution of ion velocities. Applicat ion of the equat ion for Ct::. v ( page /0) will show that isotopes differing by 5 ~~ in mass cannot be resolved unless an accelerating potential of at least 1,000 volts is used, it being assume d that the ion source is heated to about 900° K. an d that all other conditions are ideal. The Maxwe llian angular spread in the ion beam is independent of the mass of the ion and, in a given apparatus, depends only on the temperature of the ion source an d the accelera.ting potential. The uni- formity and ste ad iness of the magnetic an d electric fields are conditions only slight l y less rigid, since the angular separat ion between isotopic beams is not lar ge , even for small mass numbers. Pole piece edge cor- cections cannot be made rigorously, so far a s I am aware. Considerable absolute error is permissible in a Type II lens, which is partially self-correcting , but r e l a tive errors, in correcting from region to re gion, must be kept small. The other factors discussed are importan t .-rr-. . ,, \._~ ,.., ~,\I\ I I I I' 15 when refinements in resolution are sought. IV~ THE APPARATUS. The principal part of this apparatus is a Type II magnetic lens in which C0 : r ~ 8 inches. These dimensions were chosen because they give an acceptable value to the a_ppara.tus constant, (3 , in the Maxwellian distribution equation for C 4 v, (page to), without making the magnetic lens abnormally large in any dimension. A Schwarz trans- formation was utilized in computing the edge corrections for the lens according to the method outlined on pages'~ 0 ~. The concave pole piece boundary is increased and the convex boundary decreased in radius as a result of this correction. • J. The pole piece gap 1.s 4 inch, the pole piece faces being held apart by brass spacers, A, F'ig. 5. These spacers a.re so placed that the least permitted deviation in the magnetic lens field is 45 degrees and the greatest deviation is 90 degrees. In consequence, an ion source 20 cm~ in length is necessary if the whole working section of the magnetic lens is to be utilized. The pole piece boundaries are extended well beyond the working section to insure uniform conditions at the edges of the magnetic field. Four one half inch brass bolts pa ss through the spacers and both pole pieces, clamping the whole rigidly toe~ether. pieces, B, in the magnet yoke. There are two side They have a cross-section of 32 square inches and are fastened to the pole pieces with stud bolts. studs, C, are screwed into each side piece. Two heavy The entire appara.tus ::. is supported on these studs, which rest in V block s on a frame. The field windings, D, consist of 890 turns of Number 12 copper wire placed over a hollow iron box, E, into which a variable number of iron bars, F', may 16 be inserted. This part of the ma gnetic circuit may be s a tur a ted a t a ny desired v a lue by cha.ng,i ng the number of the ba rs, t hereby f a cilitatin g the ma intena nce of a. ste a dy magnetic field a cross the 1?0le p iece ga p. The entire magnetic circuit is made of Ar mco Ing ot Iron because of its exceptiona l homoge neity. The pole p ieces a.re e a ch 4 i n ches wide a nd are used as a ba se for a box, G, of rectangular cross-section, wh ich is f'a stened to the sides an d ends of the pole pieces 1,vith screws. This b ox fits :i.n to a brass tube, H, 24 inches long and 8 inches in interna l di ameter. end of this tube is permanently closed. One 'f he othe r has a removable circular cover, I, for which a ton gue and groove construction is used. A narrow rubber gas ket is fitted into the bottom of the groove to insure a tight joint. A wedge-s ha ped box, J, fastened to the pole pieces with screws completes the vacuum s y stem. The whole is evacua ted through a short, large tube by me ans of a t wo stage mercury diffusion pump ba cked with a Cenco oil pump. The ion collector is i n side a. bra ss cy linder, K, in one end of which the slit system is pl a ced. The wh ole is ca rried on a Sy lphon bellows, L, to f a cilita te loca tion of the best foc a l re g ion. The source block , Tur , is a l a r ge brass block con t a ining t ube s for water-coolin g . surfa ce. .!A.. semi-cy lin drica l ope nin g runs the le ngth of t he upper This opening receive s t he ion source, ca rried below the lower a ccelera tin g field pla te, N. The cooling s y stem le a ds out throu gh a tube, O, from whic h t he 'irat e r passes through a series of g lass tubes which form a resistance to ground, The source block is sup ported on three insulators, P, by leveling screws. La tera.l adjustments of the source block are made with the screws, R. The water tubes in O pass out through an insulating block, S, which is ca rried on a small Sylphon r---- 71 I) / I I / I I / I I / I I / I I / / II /I / / / / /1 / / f'/ / /t I I / / I / I/ / I I I I I / / I I I I I / I / I I I I I /t I I / I I I J I I I I I I ' I I I I I 1 I - - - - I I - - -( - - -- . J,/ 1 I I 1 / I I I I I I I I I I I I I 17 bellows for flexibility. The upper electric field plate, T, is at the potential of the case and is adjustable both laterally and vertically. Both electric field plates are extended 4 inches beyond each end of the ion source to insure a uniform accelerating field. 'l'he potential and current terminals are brought in on the lower side of the large brass tube thr9ugh a number of insulators, U. V. THE CONSl'RUC 'i.'ION OF EX'I'Jtl'J lJED EQUIPOTENT L'\L ION SOURCES. Any ion source to be used with a Type I or '.L'ype II magnetic lens should possess the follovving characteristics: ( 1) it must be an equipotential surface; ( 2) it must be straight in the direction of its length, as it is placed in one boundary of a uniform electric field; (3) it must contribute little or nothing to the magnetic field in its neighborhood; (4) it mu st furnish sufficient heat to cause the source material to emit ions freely; (5) it should be long -lived and rugge d; (6) it should be electrically insula ted, if possible, from its surrounding s. Ten different types of source construction he.ve been used at various times during the course of this research. Nine of these, which had wire heating elements., will be dismissed without description. In these nine, the objectionable fe a ture was short life, if adequ a te temperatures were maintained. Long wire elements in general must have a continuou.s support, especially if the non-magnetic chara cteristics of the source are to be satisfied. 'fhe electrically insul1J.ting materials necess a ri a lly introduced as supports tend to make the source assembly bulky, hard to out-gas, a nd wa steful of heat. In consequence, the furnace elements must be overheated and soon disinter grate. 'l'his disinte gration is hastened by a large number of ceramic materials with which metallic 18 heating elements interact. The tenth, and best, of the types of source construction tried is diagramed in Fig. 6. A long iron tube, A, has a shallow trough, B, welded or machined along its surface. At one end, two projections, C, are used to support the tube in a lavite insulating block, D. The electrical hea~ing element is a Globar, E, whose ends are inserted in iron termin9.l cups, F and G. These cups contain a bit of aluminum which melts and insures good electrical contact bet-ween cups and Globar. cup F is threaded and screws into one end of the iron tube. 'fhe The rod- like projection of F then forms the other tube support when inserted in a lavi-te insula ting block, H. The cup G slides freely in a hole in the lavite block D, and a tungsten spring, I, holds the terminal cups firmly against the ends of the Globar. The source material is placed in the trough a nd the assembly is mounted into the lower electric field plate, J, by screvvs through the l av i te blocks. nickel strip, K, welded to the iron tube. One current terminal is a The other current terminal, L, screws over the rod projecting from the termin a l cup G. Since the potential of the iron tube is held at a high constr:i.n t value, the heating current is obtained from an insulating tr an sformer. This source assembly meets a ll. of the requirements enumerated above. The potential drop a long the iron tube is onl;<l a few thousandths of a volt. Since the heating current follows the iron tube and returns through the concentric Globar, the resulta nt magnetic field is nearly zero. 'fhe lavite blocks stay compar a tively cool and are good electrical insulators. Thermal expansions in the Globar are compensated by the tungsten spring, and one support of the iron tube ca n slide in its lavite block. Lavite is the product formed by firing high grade soapstone. It s He. p , _ _ _ - - - - 1 , __ R, FIG. 7 t +-> C (1.) ~ ~ ::::s u ,. 2. FIG.8 _1__~ G 19 , may be cut to any desired s hape before firin bo- which makes it an especi a lly suitable ma terial for these blocks. strong and ri g id. to 1800° C. The Globar heating element is It is a Carborundum product s.nd may be heated s a fely Above this tempera ture its silicon evs.porates. If the tempe rature of the outside tube is to be above ab out 10000 c., the iron pa rts should b~ repl a ced with gr aph ite. The only source mater i a l used has been Kunsman catalyst (4). The catalyst is packed in the trough of the source assembly a nd reduced for about t wenty hours in a hydrogen atmosphere preparatory to use in the a ppara tus. A water-cooled tube has been built for this purpose. There is little differential thermal expansion between the iron tube and the iron <hxide base of the cata lyst. This factor is important in an extended ion source, for otherwise the mate ria l bre sk s up an d heats poorly. It has been suggested by Dr. Smythe t hat liquid tin might be used as a. bs.se containing the element to be ionized but as yet this possibility has not been tested. VI. TI-lli COHTROL OF ACC:l~ LKfu'i.TING POTEHT IALS . Obviously , the source of potential for the uniform electric i'ie ld must be ste a dy to a hi gh degree, since all ions of identical masses must have paths of the s ame curvature in the magnetic field if g ood resolution is to result. Some method of control should be intro- duced by which the accelerating potential can be varied in small steps while the magnetic field is kept constant. Of course, it is possible to hold the voltage constant and vary the magnetic field, but such proceedure is not recommen ded. Voltage measurements are simple, rapid, and accurate compared to measurements of magnetic field streng ths. Further- 20 more, the spectrometer equation for mass number, where K is an apparatus constant, makes Va more convenient variable for computations than is H. The voltage source used to date is a small 2,000 volt direct current generator, (Fig. 7). The voltage drop across the resistance R3 , in parallel with the generator, is held constant by a helium discharge tube, He, furnished through the courtesy of Mr. Erickson of the Claude Neon Company in Los Angeles. A helium or neon discharge tube of short length may have a zero resistance characteristic over a considerable current range; i.e., R ~ AE/Ai :: 0. If the discharge tube is length- ened for use at higher potentials, a larger percentage of the total voltage drop occurs in the positive column of the tube and the resistance characteristic becomes negative but remains constant, under proper tube conditions, for a certain current range. A fixed non-inductive resistance such as R2 is introduced in series with the tube to bring the resultant characteristic back to zero. generator line. A rheostat, R1, is put in series with one It is used to vary the current through the helium tube to the proper range . .A:n.y desired voltage may be tapped off on R3 . With a good gener- ator, such a system will give satisfactory voltage control. .Actually, the generator used has so large a vol tae;e ripple, probablJ due to t'tn open or shorted armature winding, that the helium tube is throvm periodically off the zero characteristic range. In consequence, a number of chokes L1 , ~2, L3, and 14, and a two microfarad condenser, C, have to be used in addition to the helitun tube. The voltage between the grounded case 20ct i +--' C Q) x 310 volts at p~aK . L L ::::, u • sis volts at peaK . FIG.9 c' 6. • 5. 4: j.. .. . .. 3. 2 )( F IG . 10 • 21 connection, G, and the source, S, is read from a voltmete r, V, used with a multi pl y ing resistance R4. A voltage tap, P, is t aken from R4 • Its use will be e x plained l a ter - page ZS. The source voltage is v a ried by shifting the grounded t a p on the rheostat R3 . This rheostat was ma de by winding severa l thous a nd feet of number .33 enameled Advance wire on t wo glass tubes 26 inches long a nd 4 inches in diamete r. the sliding ground tap. The e n amel later wa s s a nded off un der Such a rheosta t can be used s a fely for severa l thous and volts. VII. EXPER ilvil~N 'fAL RESULTS. The usefulness of t he magnetic lens t ype of mass s pe ctrometer is determined entirely by the resolution and the intensity of the ion be am at the collector. The r e sults of experhne nts on these factors will now be g iven. When this rese a rch was be gun, a 'f-Jpe I instrument a lready ha d been built. .A -,Nater-cooling sy stem was introduced a round the source. Potassiu.rn ions from Kunsman catalyst were use~ and a large number of resolution curves were plotted from da ta taken at v a rious voltages. The dependence of resolution upon the Maxwellian distribution of t hermal ion velocities soon became a ppa rent. ing voltage. The pe ak s sharpened with increas- In Fig . 9 a re s h own t wo repres en t a tive resolution curves, reduced to t he s ame rel a tive voltage sca le, in wh ich collector current intensities a r e plotted a ga i nst a cce l e r a tin g voltage. The curve t aken at 310 vol ts is wider t han the one t aken at 51 5 vol ts by about t he amount predicted by the expression for C A V, page 10. It will be noticed that on the 515 volt curve, the isotope K 41 is just be g inning 22 to appear. Its theoretica l position with respect to the center of the broad K 39 peak is marked by the dotted line. The fundamental dependence of resolution upon accelerating voltage may be ma de more plain in another way. . dN ra t 10 - - • dN0 CA v is defined as the For convenience in measuring the resolut ion curves, define 1 c':: - - -1- 1 C •V -t 1"g" The factor dNo is added to dN 19 1/19, ( 5). 1 because the r at io of K 41 to K 39 is Evidently C' so defined is the ratio of the intensity at the center of the K 39 peak to the height a t which K 41 occurs. C1 is equal to 19 for pe rfect resolution of K 39 and K 41, (C~v =-0). Theoretical values for C1 at any voltage may be computed for infinitely narrow collector slits and me a sured values may be obta ined from the resolution curves, a lthough the measured values always must be less than the computed values because of the f inite width of the collector slit. In Fig. 10 the theoretical curve for C' at 1200° k. is plotted for comparison with the measured va lues for C' taken from over twenty resolution curve s at var ious voltages. 9 are marked with crosses. The points for the curvesin Fig. During t he se runs the source t emperatures we re kept roughly constant, t hough considerable variation must have occurred, because wire wound heating elements were being used at the time and ha d to be rebuilt quite often. The 1'ype I instrument had not been edge-corrected when built and wa s subject to the coma defect mentioned on page 5 . It a lso was found that the magnetic field in one part of the Type I lens was not homogeneous, probably due to poor annealing of the iron in the pole pieces. These f a ctors, together with the finite width of the collector slit, exp l a in the .9-~s cre_p_a_n,c_~_e_~. be t ween 23 the theoretical curve and the experimental points. The theoretical curve rises rapidly when the voltage is greater than 600 and becomes asymptotic to C':::: 19 at about 2,000 volts. The magnetic field in the Type I instrument had to be saturated to bring 600 volt potassium ions to the collector, so it was impossible to extend these investigf tions to higher voltages with the original Type I lens. For a time attempts were made to obtain higher resolution by reducing the temperature of the ion source. In addition, a collimating devic8.J consisting essentially of loo , narrow parallel openings made from one mil copper was constructed. J The device was designed to remove ions having excess thermal velocity components perpendicular to the accelerating field. These methods gave some of the higher values of C', but such a large decrease in intensities resulted from their use that they were abandoned as impractical. The information gained by these investigations warranted the construction of a new apparatus. Accordingly, the Type II magnetic lens instrument already described was designed and constructed to specifications adopted after considerable theoretical investigation of edgecorrections and thermal velocities. The first runs were taken with the new apparatus in January, 1932. The t w-o following months were wasted in attempts to improve the resolution by adjustments on the apparatus. When it was discovered that an enormous voltage ripple, ·which the helium tube could not eliminate, was present in the generator used as a source of accelerating potential, a number of chokes were obtained or constructed to operate at the correct current values and were used in conjunction with the helium tube and a condenser. The resolution curve shoVITn in F'ig. 11 resulted once the accelerating voltage was steadied. This curve and others were taken at 11 37 38 39 FI G.11 40 41 42 43 24 about 1500 volts. This is the approximate upper limit set by the voltage obtainable from the generator and helium tube, minus the voltage drop in the chokes. The shape of the curve in Fig. 11 closely follows the theoretical sh!;!ipe derived from a IviaX\"lell-Bol tzmann distribution law for the thermal velocities of ~ons at the heated source, and may be taken as an independent proof that this law holds for ions emitted by a heated body. The peak at 41 is 6 .85% of the peak at 39, if the highest part of the K 39 peak is taken to be as drawn in Fig. 11. The accepted ratio of K 41 to K 39 gives 5.26%, so there must be a background of about L6~s of the K 39 peak present in the K 41 peak. chiefly to the Maxwellian spread. This background is due 'fhe theoretical calculation of this spread gives 1.4% as the background, so the agreement is quite satisfactory. It must be remembered that the higher mass numbers come to the collector at the lower voltages; i.e., the mass spectrometer equation is 2 N -::. K H , so the background from the :Maxwellian spread is somewhat V heavier on the side of higher mass numbers. It has been found that intensities at the collector may be increased by a rather simple device. The lower field plate is be-veled slightly along the sides of the slot in which the source rests, so that the source sets at the bottom of a shallow V. In consequence, the field immediately above the lower plate will be slightly convergent and the ions leaving the source will tend to be focused into the central plane of the pole piece gap. This initial focusing can be made to offset the diverging action of the electric field along the slot in the upper plate J and will result in an increased number of ions reaching the collector. The amount of this focusing action can be varied at will by applying a 25 small potential difference, VF, between the lower plate and the source. A variable tap - P, Fig. 7 - on the multiplying resistance of the voltmeter circuit may be used for this purpose, which allows VF to be made any desired fractional part of the total accelerating potential, V. Since the ratio VF/v rema ins constant for a given setting of the focusing voltage tap, P, the resolution curves will be unaffected and will represent the true distribution of intensities about the collector slit. It scarcely need be sa id t ha t this would not be the case if VF a lone were giveri a constant value. Fig. 8 shows the marked focusing effect which may be gained by proper adjustment of the ratio VF/v for a particular beveled plate. When this ratio is zero, the ion be am presumably is overfocused and only a smal l percentage of the ions re a ches the collector. A rather broad maximum occurs as the r a tio is incre a sed to the optimum value. For still higher va lues of VF/v the beam becomes underfocused,which results in it being divergent, and most of it is lost. Doubtless this effect is not entirely due to foc using , a s the s pa ce cha.r ge limitation is more favorable when the initia l field becomes more diver gent . This may explain why the focusing curve does not come down quite as sharply on the side of high values of VF/v· The s pa ce charge sets a definite limit to the current which may be drawn from a.n ion source in an electric field. li.s the space charge increases I a condition is re a ched such that the potential gr ad ient at the source becomes zero and the emission rate of the source cannot be i nc re ased further. The derivation of the current limit set by space charge in a uniform electric field is a standard one and will not be re peated here. The limitin g current, i, is given by: 26 where y is the distance between two parallel planes whose potential difference is V. The greatest current so far obtained at the collector has been 2 x 10-6 ampereJl, when a. potential difference of 1500 volts was applied to plates one inch apart. This current is of the magnitude one would expect from the space charge equation and presumably is given by it, for neither increased source temperatures nor the substitution of sources kno~~ to be more copious emitters increases the collector current. The dependence on distance also seems to be of the nature g iven by the space charge equation. Further increases in intensity evidently will require higher voltages or reconstruction of the present electric field system to obtain shorter distances between the plates. In Fig. 11 the width of the collector slits, in terms of the linear separation between K 39 and K 41, is represented by the parallel lines drawn at the lower center of the K 39 peak. This slit width was made small purposely to g ive a truer measure of the actual resolution. If isotopes are to be collected, the collector slit width should be increased to the maximum allovvable size so as to include as much of the . total ion beam as possible. VIII. POSSIBLE APPLICATIONS OF THE INSTRIBvftJ J1' 'f his type of mass spectrometer has certain features which fit 1 it for uses not possible with other instruments. A few of these will be discussed. The instrument could be used to advantage in the search for faint 27 isotopes of a tomic numbers less tha n 40. In the present apparatus an extended source 20 cm. in length a nd 6 mm. in width is used. The inte- grated emission from such sources is quite steady and many times more intense than that obtaina ble in a Dempster t ype apparatus with narrow slits. F'or sma ller mass numbers> the resolution be tween successive masses is quite complete and very f a int isotopes should be detected, even in those elements for which only poor ion sources are at present available. . A case in point is the suspected f a int isotope Be 8 for whose existence 'Natson and Parker (6) have found evidence in the ba nd spectra of Be H. Isotopes of low ma ss number are of great interest in the problems of cosmic radia tion, nuclear structure, etcetera, for it is only when the nuclear structure is simplest . 1!:;hat pack ing fractions can be measured with most accuracy and tr ansmutation experiments performed. The instrument could be used for collecting isotopic s amp les for nucle a r studies. Efficient sources should be developed for a s many as possible of the isotopic elements lithium, boron, magnesium, sulphur, and silicon, wh ich, together with potassium ca n be subjected to nuclear disinte gration a s shown by Rutherford, Chadwick, Kirsch, Petterson a nd others (7). Of these elements, good sources for lithium a nd potassium an d fair sources for magnes itun are known. The separation of the iso- topes of these elements is a problem of considerab le interest because of the information about nuclear structure and stabilities wh ich wd>ili.ld be obtained by tr an smutation ex per iments on single isoto pes. This is pa rticularly the case for magnesium,_;which has isotopes 24, 25, a nd 26. The effect of adding one electron and proton pair to the nucleus might be made cle a r if s amples of the pure isotopes were obtained f or disintegration experiments. 28 The technique of transmutation experiments has been developed to such a point thaft ,masses between 10-3 and 10-4 grams of material, usually deposited as oxides on thin films of gold or platinum, are feasible. To obtain a deposit of 10-4 grams of K 39 an exposure of about seven hours to a collector current of 10- 5 amperes would suffice. The determination of the isotope or isotopes responsible for the radioactivity of potassium has attracted mUeh interest because this element is fl'llr removed from other re,dioactive elements and falls upon a part of Aston I s curve where nuclear energy considere,tions from packing fractions seem to forbid radioactive disintegration, unless there is a discontinuity of some kind in Aston's curve at this place. In addition, potassium is the heaviest element which all investigators agree can be disintegrated by bombardment with alpha..-particles. Kossel suggested K 43, Harkins K 40 or K 41, as the ra.dioactive isotope. More recently, Beck and Fournier have predicted K 43 and K 40 respectively as the disintegrating isotope, basing their conclusions on classifications of all known isotopes for reoccurring regularities in nuclear constitution, (8). Bainbridge (9) has found that potassium isotopes other than K 39 and K 41, if they exist at all, must be extremely faint. Hevesy and Logstrup (10) and Biltz and Ziegert (11) have measured the radioactivity of potassium in which the heavier isotopes have been slightly concentrated by repeated evaporation and believe the radioactivity to be increased slightly. Obviously, this ,vork does not isolate any particular isotope as the radioactive one. If we accept Holmes and Lawson 1 s ( 12) half period of l. 5 x 10 1 2 years for potassium we must have about 3 x 10- 3 gre.ms to obtain one particle per second. p; A 200 hour exposur.e to collector currents of 10- 5 29 amperes would be necessary to obtain tha t much potassitnn. It might be possible to use considerably less potassium, however, if the isotopic samples were placed in a. Wilson cloud chamber being used simultaneously for other purposes, since a very few P-ray tracks originating on the potassium would be photographed from time to time. The instrument in modified forms could find uses outside the ' field of isotopes. For example, suppose it were desired to make spectro- scopic studies of the energy and momentum rela t ionships in i nelastic collisions between positive ions of a single element and neutral gas atoms. A modified form of this instrument could be ada pted to obta in inten se ion be ams in which only individual e l ements would be resolved. Spectroscopic observa tions could be ma de on the excitation of resonance lines in the ga s a s a function of the energy, mass, e.nd charge of the exciting ion. Other uses perha ps will be suggested from time to time. IX. CONCLUSION It is the belief of the author that the research outlined in this thesis contributes a new _instrument a pplicable to problems in isotopic suudy heretofore beyond the re a ch of ava ilable appar a tus, a.nd t hat t h is instrument may be used successfully in other fields wherever concentra ted pure ion be ams are demanded. The author wishes to express his most sincere appreciation to Dr. W. R. Smythe under 1Nhose direction this work was done. Re desires to thank Messrs. Sherburne and Brower of the Astrophysics Ma chine Shop for the1r fine workmanship on the Type II mass spectrometer. References (1) Aston - "Isotopes", Arnold, London, 1924. (2) Bronsted and Hevesy - Phil. Mag., 43, 31. (3) J. J. Thompson - "Rays of Positive Electricity", Longmans, . Green and Co., New York, 1921. Aston -Phil. Mag., 38,709, 1919. Dempster - Phy. Rev., 11, 316, 1918. Smythe - Phy. Rev., 28, 1275, 1928. (4) Kunsman, Harnwell and Barton - Phy. Rev., 27, 737, 1926. (5) Dempster - Phy. Rev., 20, 631, 1922. (6) Watson and Parker - Phy. Rev., 37, 167, 1931. (7) Rutherford, Chadwick and Ellis -''.Jadiations from Radioactive Substances'! (8) Harkins - Natl. Acad. of Sci. Proc., 11, 630, 1925. Kossel - Phys. Zeit., 20, 265, 1919. Beck - Zeit. f. Physik, 47, 407, 1928. Fournier - J. de Physique et le Radium,_!, 94, 1930. (9) Bainbridge - J. Franklin Inst., 212, 31?, 1931. (10) Hevesy and Logstrup - z. Anorg. Allgern. Chem., 171, 1928. (11) Beltz and Giegert - Phys. Zeit., 29, 197, 1928. (12) Holmes and Lawson - Phil. :Mag.,_§_, 1218, 1926.