A Modified Aston-Type Mass Spectrometer for the Investigation of Ionization Resulting from Single Impact Citation Taylor, Daniel Dwight (1933) A Modified Aston-Type Mass Spectrometer for the Investigation of Ionization Resulting from Single Impact. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/e3cp-1s48. https://resolver.caltech.edu/CaltechTHESIS:10282024-171833043 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: 1 January 1933 Record Number: CaltechTHESIS:10282024-171833043 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:10282024-171833043 DOI: 10.7907/e3cp-1s48 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16815 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 28 Oct 2024 21:13 Last Modified: 28 Oct 2024 21:23 Thesis Files PDF - Final Version See Usage Policy. 17MB Repository Staff Only: item control page

~ Mo&ifie~ ~ston type Mass ipectrometer for tbe Itivestigation of Ionization ResultinQ from iinQle Impact I1hesis ty 1 Daniel DwiQht T2;ylor In Partial Fulfillment of the Re~uirements for the DeQree of Doctor of Philosophy ·~al ifornia Inst·i tute of I'echnology 1 PasaJeha~ ~alifornia 19:33 ·s uM M ARY 1 tA mo difi cat ion of the !Aston type mas·s-·spectrometer is described for the examination of the process of ioni~at.ion. 'The chief modifications pro'vide for .ioni.za- tion by ·electron bonib:ardment 1 and for electrometr.ic reg·The ·various peaks 1_are focu·s:s:ed .in- .istration of .ions. dividually on the collector by ·variation of the magnetic fi:eld. :Details of the apparatus are given. 'The focus-sin g pro'ved to be ·very poor, ·sons for it are di·scussed. arJd the rea- tAlso a parti:_ al correction of the focus w.as accomplished and results are gi'ven taken both before and cuives are given of the ionization in monoxide, ammonia, 'Specimen after this correction. nitrogen, carbon and a mixture of three parts hydro- gen to one of nitrogen. lA pliotron circuit is described w.ith which a 'sen- ·s i tivi ty of 2x 10'-', 17 amperes per scale division is regularly ,att·ained. .A new apparatus has been designed to eliminate the 1 difficulties found in the first one, ·and certain addi- tional features ·are included which ·should permit ·a more complete interpre~ation of the results. 'The :analysis of the design of the magnetic field is given. type o·f balance is described !A new which g,ives ·a continuous indication of the ·sq-;uare of the magnetic field. par-atus has recently been built according design but i ·s not yet in operation. !An ap- to the new PLATE - . XIV 1.An extraordinarily powerful means of in'vestigation was opened to physics when F. ;w. l.Aston announced his modification of J. J. Thomson's parabola method of posi ti've ray analysis. the in·ve·stigation molecul·ar ions) Posi ti've ray of indi'vidual technique permits particles instead of re·vealing (;atomic or only the average properties of a large number of ·such particles, apparatus and technique devised by tAston irivestigation and the advanced ·such to a high degree of accuracy, permitting precise comparison of the masses of different isotopes. ·The positive ray method also lend·s ·. itself admirably to the ider1tification of ions produced in ·various ways ·and e·ven offers hope of yielding quantitati've che mical a- nalyses where ions of the unknown substance .can be con·veniently produced. It was found very early that positi've rays could be deflected by either electric or magnetic fields and an ohser·vation on either deflection will yield a ·v:alue of e/m pro·vided the energy of the particle is known. However ·sufficiently precise is hard to obtain so knowledge of this energy that a method is ·very de·sirable which will be independent of the energy. By means of an electric deflecting field it is possible to select a beam of positi've rays energy to any ·all of which reasonable will have the same degree of approximation, and by means of magnetic deflection it is possible to ·similarly ·select for uniformity of moment um, and by me arrs of a ·series of high frequency electric field·s ,a s de'vi·s- 2 ed by :w. R. 'Smythe it is po·ssi ble to select for uniformity of ·velocity but these methods all great los-s ,in available in'volve ·a ·v.ery intensity ·and consequently are justified only in work of the highest preci·sior1. ·The method both de·velop ed by t.As ton electric and magnetic consi·sts in com b.ini ng field·s so as to annul comparatively narrow range of ·velocitie·s) ·velocity. Essen tie ally this .is (within a the effect o·f accomplished by using an electric field to spread the collimated beam ,into an energy spectrum and then picking out a narrow region of this ·spectrum erture. with a ·stationary slit of adjustable !ap- Inasmuch ias t.Aston uses •an .induction co.il both to produce and accelerate the ions, present there w.ill be ions corresponding to :any desired within a wide range different energy ·selection and since photographically all the kinds of iorrs ·are recorded ·s.imul taneously, the energy to be selected can be cho·sen almost arb.i trarily. Howe·ver if electrometric recording is to be the choice only one kind of ion can be recorded ,at one ·time and the energy region tuations ·selected should be free of fluc- which might selectively affect o-ertain kinds of ions, or comparative readings will be ·very difficult to obtain. It is worthy of mention also that the use of an induction coil introduces between the energy producing possessed by the ion an implicit relation ionization and the energy in its passage through the mass- spectrome ter. The tAston type of mas·s-spectrometer was chosen for this re·search because it avoids the assumption which ,is inherent in some other types that all ions enter the analyser ·after having fallen through precisely the ·same potential difference. ·This assumption can be closely approximated under some conditions but it was undesirable to be restricted to these conditions. Greater pre- cision is attainable by the use of :a ·veloc.i ty filter as in ·smythe • ·s mas-s-·sp:ectrometer but cori'ven i ence is to be preferred to such precis.ion .in the present work. ,certain features of l.Aston• s apparatus .were not entirely ·suited to the itudy of ionization and were modified :as will be discus-sed below. It was particularly produced by electron introduced desired to irivestigate .ions bombardment ·so ·a bot fil:ament .was as a ·source of electrons. filament was supported pyrex mounting Orig,inally the on tungsten books .imb:edded in a but radiation from the filament heated the pyrex to the decomposition point so fused q~artz of a different design was substituted. "The new design shielded the filament much les·s completely ·so that when thoriated tung·st:en .was used it was ·soon d~acti'vated by LA ·still better style bombardment ,with neg ati've ions. of filament mounting. was made from lavit:e which rigidly • and coriveniently shield. ~ supports filament:, grid, It is very and filament corivenient to join the lavite to the pyrex through a ground joint but the thermal expansion of lavite is greater than of pyrex ·so the lavi te should surround the pyrex any apprec~able temperature change. if the joint is to undergo ·The .impro'vement now needed most in the fi1ament is to elimiqate the potential drop through the fi1ament ·so that the bombard- ing electrons will be of more nearly uniform energy and 4 this will be attempted in the new de·sign. ·The filament used is circular and surrounds ,a concentric cylindrical ~rid of wire gauze ·so th at the electrons are ;acceler ated radially inward in the plane filament shield is of thin hole slightly larger of the fil:ament. ·sheet metal and than the grid. distance between contains a It is supported parallel to the fi 1 amen t and ·slightly filament than the The fur th er from the filament and grid. ·The shield may be operated :at the potential of the grid but it is con·venient to be able to ·viary it if de·sired. The filament current is ·supplied by a ·storage battery; the accelerating ·voltage for the electrons is obtained from a small supplies motor generator set and both these must be insulated tial applied to accelerate is necessary to have from ground for the poten·the positive ions since it the body of the •apgaratus ,at zero potential. The accelerating potential for the positive ions is obtained from •a small dynamotor using _a storage battery for the primary input. between the ·This potential is _applied filament shield and the collimator. Experience using a filter to cut down commutator energies imparted within a fairly has shown front slit of the that ~ven without ripple etc. the to the ions by the potential all lie narrow region prdvided the conditions in the discharge pre·vent ·any 'visible glow. by suitable choice of the deflecting the entire collimated potential almost beam will traverse thus yielding ·very good intensity. · ·Therefore The the analyser condition of no 'visible discharge places limits on the electron cur- 5 rent, the bo mbarding ·volta~e, and the 9-;as pre-ssure, the·se :pu an ti ti e·s a re so interrelate d one will change the permi ssi bl e and that ·v ari atio n o·f ·val ue·s of the others. Due to the co mparatively low maximum energy of the ions it is possible to use larger deflections than were feasible . in t.Aston'-s work so an electric deflection of one quarter radian and a magnetic deflection of one radian were chosen. 'The ratio of four to one between the deflections was presetved po·s-si bil i ty time, of using because of the rather remote pho togr aphi c in .which 0ase this ratio recording •at ·some gives linear mas·s dis- persion. Oareful attention was ·al'sO g.iven the mas-s-spectrome ter as worked and R. H. . Fowler in ·an ·attempt to the theory o·f out j o.in tly by lAston to utilize ·second order focussing ·and this resulted in making the trailing edge o·f the magnet cone-ave with ,a radius of 84mm but :as will be ·shown l iater the focus·sing was ·very ·seriously in :er- ror. 'The upper panel of plate II ·shows a photograph of the old •appaI'latus ·.as set up w.i th the nec:e·s-sary auxili;ary equipment. The details of the mass~s~ectrometer ·are rather obscured by the mass of the magnetic circuit but the collimator is contained in the horizontal tube extending to the left shown is from the electric ·field. tAl·so the triple walled cylindriqal .iron shield for protecting the electrometer from the ·stray field of the magnet. cally Just ,in front of the shield i·s the magneti- controll:,d switch for the ·various electrometer 6 Because the rnas·s-spe.c·trometer j : s di'vided potentiials. compartments by the .into three collimator slits it is necessary to us:e a multiple pump.ing ·system. i ·s being constantly rsin.ce girn ,admitted to the ioni zrnt.ion ch amber gr~ater pumping .cap;ac.ity is req,,uired there and an extra large second stag:e pump is .combined with :a regultar ·s.i.ze first ·stage. ;a single ~wo regul:ar second ·stage pumps ·setved by first ·stage pump ·are us:ed as ·scavengers, one connected to the collimator ·and one to the magnet box. The gas to be studied is contained in 1a two liter reser'vo.ir :and !admitted .Qapillary tube ·s:een to the ,appar. atus rising ·vertically through the :at the extreme left of the picture. This tube is continued within the ionization chamber rand its end lie·s on the ,axis of the grid in the pl:ane of the filiament ·so that the con cen- tration of gas molecules is highest ,at the po,int o·f ,impact w.ith the electrons. The de·sign data ·for this app:al'!atus .it ·should be oapable indicated that of very good resolution ·so it was ·a great disappointment to find that -~ven under the most favorable circumstances the peaks were ~so broad ,as to extend o'ver ·se·veral uni ts pro'vision had b:een made ·slit along of mole.cular weight. tAlso for .an :adjustment of the fo.cal the median ray ·so that it .could be located at the po,int of sharpe·st focus ,and e·ven this :adjustment failed to have :any :effect. the paths of the ions ,careful consideration o·f seemed to indicate that the dif-< ficul ty l:ay in the magnetic field ' SO it ;was in'vesti g a- 7 tea more carefully. ·The uniformity of the field be- tween the pole·s was seen to be rather poor due pre·sumably to the fact that the pole :fiace·s .were not ·so1id but .were made of two pieces ·soldered together because ·stock of ·sufficient ·s.ize was not was not at :all sufficient av;ailable. to account Howe·ver this for the observed error in focus ·so ·a graphical in'vestigat.ion o·f the form of the pole faces was undertaken. of perfect collimation On the assumption a rough approxi~ation could be obtained for the energy of 1.a particle .in terms of ·the deflection it ·suffered in the electric f.i eld :and this ·value of the energy permitted the calculation of the radius of curvature of the path in the magnetic field. Neglecting edge effects in the magnetic field this circle of curvature .at the would be tangent to the ·straight path leading edge gents at the o·f the field and a ·system of tan- trailing edge should show the ·sub-sequent behavior of the rays. The lower panel of plate TI .in- dicates the result. The electric and magnetic fields and the focal slit are shown, ·al·so the emergent t:.a n- gents of four rays obtained ·as abo've, :and the focal error is indeed surprising. all, There ,is no point focus :at and the best focus of the caustic i ·s almo·st 1<4 cm temporary mounting for the from the .intended fo CtrS. lA focal slit and collector was made to ·fit the new loca- was found to b:e much .impro'ved but ·still poor on account of the caustic. 'The new location tion :and the focus of the collector :also added considerable capac.i ty to the tube circuit and made readings ·very ·slow. to the _g raphical in·ves tig at ion Pre·vious ·slits had been tried :at PLATE 11 8 the edge·s of the magnetic field and produc·ed benefit when used :at the trai1ing edge. ing edge lAt the lead- they would produce ·sharper peaks the s,acri·fic:e from the of much intensity, little but only at as would be expected great divergence of the rays at the intended focus. The theory of the mas-s-spectrometer ·sho.ws that the molecular weight of an ion is proportion al to the ·sec- ond power of the magnetic field focal ·slit, which brings d t to the ·so the identification of the ions depends on a knowledge o·f the magnetic field. 'Sit1ee it would be ·very conv e nient .indication of the field to have a co ntinuou-s it was originally hoped that measurement of the cureent in the magnet would give some rough idea of the ·strength of the field, prd~ided each cycle started from ·some standard magnetic state. However it was ·soon found that even if there had been exact proportionality between current and field ·strength, much greater precision in the measurement -was required than could be at- tained with an ordinary ammeter, ,and furthermore hysteresis was not negligible. was improvis:ed ployed Next an external :air gap and ·a c:a1ibrated b.ismuth ·spiral was em- whose resistance was measured with a 1Carey-:Fo·s- ter bridge. 'This give usable results corrections were required for temperature. measured is a change in resistance but laborious ·The effect due to the magnetic field but this effect has a temper.ature co:eff.icient :and there is al·so the usQal change ,in resistance due to the 9 temperature alone ·so the necessary corrections were of ,~uadratic form. tAn es'sential datum .is the re·sistance of the spiral in zero field at the temperature of the rsince in the ·air g·ap the sp.iral subsequent readin g s. was closely surrounded by massive iron, .its temperature there might be quite different air outside the gap, the field and ,it .was impracticable to reduce to zero at this quantity. from that found in the frequent intervals 'Thus it was found to determine necessary to build a ·very flat platinum re s istance thermo met e r which could be perm an e n tly inst al led in the air g ap be-side the bismuth spiral and its resistance was ured on the ·same bridge. coriveniently meas- IAll this greatly compliqated the obs:ervations and the calculations ·so it w;as firnally given up when it was found that the calibration of the spiral was not entirely reliable. 'The next step was the us:e o·f a 'snap coil connected to a ballistic ~al'v:anometer. .indirect compari·son Thi·s was .calibrated by with the large .Zeeman effect ·sole- noid ·and g;ave rather consistent results but was unsat- isfactory because it required to be manually cocked •and tripped for each reading, ·and :also jr arred the p1iotrorrs sufficiently to interfere with their operation. tA brief ·attempt was next made the magnet circuit portional with a wattmeter in for .it would give to the square of the indiqations pro- cureent 1and so approx- imately to the ·sq:uare of the field .but ,again ·suffic.ient precision in r~ading could not be attained. vanel of plate III, 'The lower in the curve running upward to the right, ·shows how the precision falls off •at low ·fields. 10 ~11 four curves are plotted against a rheo·stat settings, and the break .in each .i's due to the change in the material of the w.indin g . linear scale of In the group of three cur'ves, the lowest represents watt's; ballistic g alvano meter deflections; ate curve represents and the intermedi- the 'sqaare of the ·vanometer deflections. the highest, ballistic gal- ·The fourth curve is the recip- rocal square root of the watts, and so is neces·sarily proportion al to re-si'stance, consequently .it ·should continue straight to the end if the readin ~s weEe accurate enough. The satisfactory and final solution of the problem is discus·sed in connection with the new apparatus. Because of the difficulty of quantitative .inter- pretation of photographic records of mass ·spectra (ex- cept in the case of ,isotopes of the ·same element) electrometric regi'stration was planned fro m the -start. first attempt was made with "The the Hoffmann ·~acuum binant electrometer and much time was expended to operaJe. Only amber or quartz .insulation could be permitted on the lead from tromet er and the le ad had to be on getting it the collector to the elecrun .in ·vacuum all the way to prevent its pickin g up atmo·spher,ic .ions. trouble was experienced Much w.i th the tilting ·sw.i tch due to its inferior design, and a ·stable mounting ·for the ,in- strument was arranged only by ·suspending crete block from the ceiling. mission o·f ·vi brat ion electrometer a heavy con- tAlso _to pr~vent trans- from the mass-spectrometer to the the ·vacuum shield around the lead had to 11 be divided into two parts and the connection between the parts run exposed to the air although it was closely shielded so that it could pick up iorrs ·very small ·volume. f actor ily t.All this was from only a aceompl ished ·s atis- but the really serious difficulty came from the stray field of the magnet. The mechanical control in the Hoffmann and is celled by an the needle is quite small electrical control, is made of a partially can- consequently, diamagnetic since platinum-iridium alloy, it i·s subject to quite large deflections in even a weak magnetic field. It was first hoped to o·vercome this by surrounding the whole electrometer with a large triple walled cylindrical iron shield, reduced the effect but while this it was still too great. step was tbe making of a ·The next much n:ore int ri c,ate shield to ~o inside the case of the Hof f mann and completely s ur- round the binants which contain the system. The sides and top were all protected by three shells of iron separated by two shells of brass, shells of iron and the bottom This assembly re- placed the heavy copper shield originally surrounding the binants and looked very impressive, but failed to work. and one of brass. by two It seems quite reasonable to attribute tility to the in the weak poor permeability fields to which its fu- of the inner shields they were exposed due to the successful workin~ of the outer ·shield. 'Some such substance as permalloy having high permeability .in weak fields, might pr6ve more suitable for the inner members of such a system of shields. tem of shields The failure of this sys- caused the abandonment of the Hoffmann 12 for at this time pl io tron, it appeared that the low grid current FP,54, offered more hope of ·succes·s, which is discussed in connection with the new apparatus. In order to obtain ·sentation the cleare·st graphical of the observed data, been tried. repr:e- ·variou·s methods have ·The plotting of electrometer deflections against magnet current was impracticable because ·sufficient prec.ision was not attainable in the current reading-s. !Although it presented ·some be·st arrangement ·seemed to be con·venient parameter di"sadvant_age·s the to plot in terms of some which could be ·eas.ily read, and the settings of the rheostat used to control the magnet current were admirably ,adapted to this. is wound with a hundred turns of wire The rheo·s tat and ·each turn i ·s readily ·subdi'vided ,into a thousand parts:, crements are easily read. rheostat ·settings ·sulted lA linear would be ·very in s~vere crowding ·so small in- distribution of con·venient but it re- of the peaks at one end of the scale and made interpolation ·very difficult because the distribution of molecular weight ·an in'verse square law. ,was according to !After a couple of unsuccessful approximations ,were tried a table of three hundred setting·s of the rheostat was ,worked out which, when spaced uniformly along the ·abscis-sa, would gi've a linear dis- tribution of the square of the field ;and, hence, of the molecular weight. 'This table i"s reproduced ,in the up- per panel of plate III. By ·varying the ·voltage of the magnet battery this table could be made to co'ver 1a ;wide range of desired dispersions without disturbing their PLATE 0 l'f ~ 292165 J 5 72 1559, 2 2'46 J 8 J9 16t' 0 i! 18 i! 46 JJ?. J 6 4 J9 4 16 511 5 54 95 6 41 68~8 7'89 8244 870() I W 1ftl 5786 6404 1029 1654 8288 8922 9564 4 0206 0857 1509 2168 2827 J496 4/65 4842 5520 6208 6896 7592 8289 8996 9104 5 04i!1 IIJ8 1865 259J JJJ1 4069 4818 5567 6327 7088 7859 86JO 9414 6 0198 Oft9J 1789 2597 J)f05 4225 ~~ 0086 0552 1022 1492 211$7 ~1 3891 4318 4169 5 60 5 57 63514 6156 T. 59 7. 67 BJ '(i i~ig 9~25 204!96 0 71 '.,,i4.9Jo 25 ~¥ J~O/ + 5 Bi! 5 J4 ~92 7 15 8 81 8 SJ ~ I ,.I t~ ~~ •, + 5 ·580 501 6 14 7. 61 84' ·g 9 69 7 o JO I 05 2 69 7l88t J 4 2 546 5 86 7. 06 8l4' 917. 8 0'?8 /080 2048 J0/1 4002 4988 5990 699J 8015 9031 9 0071 1117 2111 J2J1 4J/6 5.396 6496 1591 8718 98.39 10 098J 21i!8 Ji!94 4461 5650 6840 805J 9261 II 0506 1747 JOI/ 4i!76 5569 6862 818i! 9502 12 0851 2i!OO J518 4951 6367 7778 92i!O /J066J i!jJ9 J 15 5126 14 66i· 8 BJ 9 JO I 15 14 424 6 7, /J 94 15 II i! 4 4 46 6J9 8 7 9 97 16 1845 J693 5591 1489 m.Ji'" J,fOO 5408 1459 9539 1652 JJJ1 4292 5262 6254 7247 8271 9295 19 OJ51 1408 i!499 J590 4719 5848 1015 818i! 9391 20 0600 1852 J/05 440J 5701 7049 8397 9798 21 1199 i!655 41/i! si,2s 7 45 8 'i!S i!i!OJ 6 I 55 J 05 5 i!8 70 8 55 2J O 58 i! 46 18 2J 4 '57 '.50 9 I 2:7~ 16 60 2J,/ ~427 ' 19 5 1 46 BJ 9 28 ~9 58!86 29 J, ,;>2 JOO?Z 6t2 4 18 9 2 Jlcilj72 J2 20 T.8i! JJ 88Ji7i! I ~1 !J 014 OJ J64() 5 J1 l 5 8 25 J8 Ti 55 J951sr 40 5012 41 4:89 42 5 69 4,1.727!: 44 9 94 46 J 07 ~ii!7 51 J 37 SJ J 27 55 6 44 t + I r Ill 13 linearity. tion This permitted ·very con·venient interpol:a- for the -. identification of peaks from their loca- tiorrs, but .it was necessary that the same scale be presetved throughout the range. rangement for plotting The latest and best ar- is independent of the locations of the peaks along the abscissa for it makes use of a ·second cutve plotted concurrently with the first which gives a continuous indication of the square of the magnetic field. Thus only the ·value this cur·ve is required ·for the peak since the relation weight is constant. plotted of the ordinate of identification between field and molecular "The points for both cutves are as ohser·vedJ for this is much less laborious than tabulating readings and plotting ·afterward. a point of ·any which d~viates from the cutve tAlso is immediately apparent and can be checked promptly, whiled t might be dverlooked in the tabulation. The precision with which the points are read .i·s best real.ized iin the plotting by pricking them in with a firre needle, and the important regiorrs of the cutve are emphasised later by drawing a ·small circle about each point with :a ri'vet pen. 14 ~ representatfve selection of the results obtained i ·s reproduced below; some of the cur·ve·s ·showing the .results obtained under ·showing the ·strang:e ·:favorable conditions :_ and ·somet.imes inexplicable eff:ects accomp:anying the results ·The pLates reproduced results under ·and othe~s ,when conditions are :_ adverse. have been chosen to .illustrate a ·variety of conditions rather than to ·show a complete ·study o·f any on:e ·su,hs~_anc:e. Plate I'V ·shows :in the upper panel :an early run on nitrogen bombarded with 90 ·volt electrons. ·The range o·f molecular weight co'vered is from roughly :300 down to almo·st .zero·, decre as.ing ·from 1 eft to r.i ght. ·These ob- ser'vatiorrs ,were made w.i th the Ho·ffmann electrometer :and ,are ·somewhat more pliotron .circuit. Lirregul ar th an tho·se made with the ·The Largest peak i ·s 1a bout where Hi+ would be expected but the uncertainty of :identifiiQation is ·such that it may represent either :~• or T~+ from the ·filament. When more certain identifi~ation became pos- sible the larg·e peak no long·er :app·e_ared ' SO the question was not ·settled. "The ·sharp peak ,at the right -. is due ' to N:;, and the other to N'+ or to N'. ;• which is ,indistinguishable ·from ·.it by this method. can be att,ached to the the ·various peaks. No real ·s.ignifi.c,ance irreguLaritie·s •.in the ·shapes of ·The upper curve on this 'sheet .i's ·a plot of the magnetic field :as determin·ed by a balli·st.ic gal'v an.ometer. ·The lower panel of plate I'V ·shows i. a L_ater run on PLATE I I tt-+ft I I -- I ' iI I ~ - I i L I I ' ;-- - - j •t• I I .-r· I I I I I ... I I I . •,.; I I Irf o .... I - I I .. .. l •, ~ ( ' . I I I l .... . I : · t ..... I 1 ,---f jl I I L I ',.I • I~ -+ I I :i . - I I I ! I I ' j I'- + f---------- I ~ l I I l .. t j ~~ "'' ... I',._ •, I '\ " 'l I'; !"'1•··•!.............:7•"'''1" -~·r- ._ I ·················• ........ ·.'.,.· . ......... ·-· ......... •····· ......... , .... IV 15 nitrogen bombarded with 65 ·volt electrons. tAgain ·an unidentified peak due to ·some heavy .ion ·appears but it is definitely lighter ·and les·s intens·e ,in proportion to ~;. "The run was not continued the~• peak. far enough to include The upper cutve indicates the current in the magnet winding, the ·si.ze of step the breaks being due to c~anges .in taken w:ith the rheostat in an effort to preser·ve ;approximately line,ar mas-s di·spers.i on. ·The upper panel of pLate 'V shows ·still :another run on nitrogen, this time bombarded w:i th 40 ·volt .electrons but the scale of molecular .we.igbts ·.include a much smaller reg,ion. has been changed to 'The left hand peak is the familiar~; and the right is~• but the largest was for some time unexplained, to be 23, al though .its weight ·appeiared agreeing with Na•. Finally it was traced to some borax that had been used ·as a flux dering the grid. in silver sol- In the hope that this might ·setve as a ·source for other ions some magnesium borate was pre- pared and placed ·at the same point but no .ions could ·be found. 'The lower panel of plate 'V i's reproduced to ·show the effect produced on the peaks when the ·focussing error was disco·vered ·and partially corre.cted, be compared with the upper panel. and should !Although part of the region between the peaks has been deleted the ·scale is unchanged. ·somewhat The electron current has been .increased but the .re·sul ting ioni~ation i ·s ·still not in proportion to the height of the pe,aks, the greater .in- tensity being largely due to the the focal sli't. imprdved location of "The peaks are all taken in nitrogen, PLATE ,--- t---- ............ :· . -· . _ _____........ •• - :lo. ------1- - '---------+----'--.,______, \ '' • I • .. ..~: :_ - - + -,,'-~ -+-- I ... - - ) 1 V 16 '+ 'The an d in each pair the left hand one is due to N2 • first pair :and the were taken ,at 45 ·vol ts:, third ·at :35 ·vol ts the ·s:econd at 113, but in this pair the right hand peak was marred by the extinction of the discharge current. The left hand peak of the last group ,is the strongest of all and was taken with 60 ·volt bombardment but its high intensity is due to using more than twice the electron current pr~viously employed. In each case the part of a peak lying off the top of the page is reduced ten times by ·shunting the gal'v:anometer. "The last peak on the ·sheet is the ·same ·as the next pre'vious has been conderrsed four times ·vertically but so it can all be shown to the ·same scale. "The upper panel of plate 'VI shows one peak of ,an entirely similar run, but plotted to a horizontal ·scale fifteen times as great more det ai 1. lAsi de from ·verifying the pronounced ;asym- metry of the peak little ·added information was gained. The lower panel the degree in order to examine the peak in of plate 'VI of ·separation obtained in molecular weight is included for unit difference in the reg.ion 12, 1;3, •+ ' +· + 14. "These '+ peaks are due to N ·, iCN , for the really ·satis~actory first time the to ·show and ;C , ,and this sheet shows method of identifying peaks made possible by the magnetic balance :although here the ·simple. problem of identification ·The deflections of the balance •are plotted as a nearly straight line the peaks, ,was quite slightly below the ~as:e line of and the ordinate of this line corresponding to the crest of a peak is proportional to the molecular weight of the peak. PLATE I T ! I I I I I I j t I I I I I I .. -~1 I I i \ ; .•..,. I I I I I I T T LL I --+-- - ~ ; ♦ .. .... _/ I I '+ I .•: I ! t . I t T -- i : I I ·.·.• ·' : I I I I I I I ~ : - I •••••♦ : - r~ - i I I . ! I--'-- '-- . 1 - T I f--- 1 . l I '• ..,. . ·. •. I t~ i"·\ 1---------< ....... + I . - I - .. I---'- '-'-~ -·-f--- ••• 1 I W--- 4 . . ...·······•••• t Vt ·The next three panels, including all of pLate 'VII and the upper one on plate 'VIII, belong to a single run in carbon monoxide at 132 ·volts, under conditions which permitted an left hand peak arc discharge. In the first is bifurcated and the panel the higher branch is due to 1Co:; but the other branch is one unit heavier and has not been accounted for. The ·very weak peak is :al- most certainly due to ;C2 0 •. .The higher branch of the right hand peak agrees very well with ico·• panion is unexplained. cession O' + , but .its com- ·The second panel ·shows in ·sucwhile the last shows faint but reliable indications of ,c•·• and o•·+. The lower panel of plate 'VIII shows part of a run on ammonia at i:31 ·vol ts without arc di-sch arge. If run in the usual direction the retentivity of the iron does not permit the field to fall low enough to focus the i• and~; peaks ·so this was made ~in the r~verse direction. . + . '' ♦ 1 + 'I ♦ ·+ From the left the peaks shown are; H, H2 , N, NH, NH~. In :all the work on ammoni·a the intensities of NH; and NH; are consistently comparable, NH• and NH 4 ' + being much .weaker. with There is no meaning to the break in the cutve at the extreme right. Plate TX shows a similar run on ammonia taken in the normal direction and shows ·very nicely the equality between N~; and NH;. The large peak .i"s due to~; and its companion may be due to NiH• but thi"s is uncertain. PLATE VII . .. .. . - ..... .... '------,-------,---·-... i- ,.... •.. ---+----+--,,...+---- ---,.-----,~ J,.._,,' I\,.._ i ..... : •.._ .... PLATE 1---------t--t-,-----i------ ~ - - t t I ~ j I , -t t f----i--+-+-+---+-+-+----t-----1~ L T .. .. . --:-- ... .:,. :--.........,.. . ·, ---+-+-- VIII PLATE ~.,--j---,-1.---1-- -l- • + -4- t _,___ - •, ;. j_ ;. ;. t .... IX 18 Pl ate :x shows ,a run under arc con di t.ions on a mixture of three parts hydrogen to one of nitrogen. "The bombarding ·voltage was 80 ·vol t ·s. The peaks due to the pure components :are quite strong but tho·s:e correspond- ing to compound·s :are weak al though ·still certain. peaks shown are H•, •♦ ... I-I;, H°;, I ♦ • ... panel, with N , NH , NH: 2 , NH 3 , 1 C\ The and ,en• ir1 the upper I .. and N2 below. tA ·faint companion to the left of~; is unexplained. Plate :xr shows a run in all ways clo·sely similar to p1'ate -X except th at the gas mixture i·s contaminated with some atmospheric ·air. entirely different How~ver the results look for the peaks are .in m~ny instances twinned, greatly complicating the interpretation of the re·sul ts. • tAl though the po·si tive ion current .in this run is somewhat greater it i ·s difficult to ·see how that"can be responsible for the effects found. It is hoped the new apparatus can shed some light on this by the elimination of focussing error. PLATE I I I 1 I j j b--+ _J ~ I -1- t I r-+- r----1- - >------- - + - - ! - t - + H ~- i - -+-- -1 r I - - I I. - ----l-- - + I l I : ! : + 1: I I ~- - - : j I 1 ·.".= ~ . I i • ....-1 ··~ : - - i -- I .-. L.--L. .._,.;•..,. -·- -+-- - I + r--- ----+---,--+- ---+ : ! . l--- r -t--- -r- -~/ J-_ =t~ X PLATE + T + _J + -,...-.:.__l . . .. .. . . :·.--.;__.-.. I + t . XI 19 t.Although the old •apparatu-s failed completely only in the error in focussing due to the wrong shape of the pole ·face·s, it s·e:emed impracticable to correct that ·error and use the re·st of whole the existing •apparatu·s. apparatus was ·a:s-sembled ·The .with soft solder ·and in ·spite of oareful use of liquid ·air trap·s the ·solder had become ,impregnated w.i th mercury from the pumping ·system so that the mech~nical ·strength low and ·vacuum leaks were us:e of Glyp t al ·varnish. of the joints was ·very prevented only t.Also the by the free homo gene.i ty of the field between the pole face·s was not ·s.atisf.actory, ·various •additional :and impro'vements were of ·suffic.ient 5.rj- portance to .warrant complete rebuilding. 'The only important change in the method of producing the ions is the ·substitution of an ·equipotential type emitter for the old tungsten filament in order to hive the bombarding electrons of more n~arly uniform :energy. •Considerable difficulty has also been experi- enced from the bursting of the outer shell of the large ground joint carrying the fil!ament •ass:embly. ·This ;JHrs due to the fact that the glassblowing e qui pmen:t l't as inadequate for working the necessary amount of pyrex for a ·shell of the prop:er thickness. been o'vercome by making 'This difficulty bas the ioni.2iat.ion d11_amb:er :and ·the ·shell of the ground joint of ·ste·el, using ·sil'ver "solder for assembling the parts. The grinding of the ground joint is ·somewhat more difficult with ·this construction 20 but this is more than offset by its ·advantages. lA ·slight modification in the measurement rent in the high potential circuit has been of cur- introduced to eliminate the effect of leakage currents. lA perfo- rated graphite di·sk is ·supported .in ·front of the collimator and ,is ·suitably insuJ:ated ·so that the current recefved on it can be measured :and l~_akage currents will not pass through the microammeter. In tAston's appavatus the plate·s of the electric deflecting field ·are flat ·and parLallel and inclined ·at the same angle to both the .incident ·and emergent median r:ay. This ·arrangement that the backward has the ·very prolongations useful property of the emergent rays w:ill ·all prass through a ·single po.int, :and this po.int i's quite con·veni ent field. .in the consideration However when the electric deflection creased to one quarter nadian of flat of the mag.n:etic panallel plates i·s ,in- the neces&ary ·separation becomes excessive, requiring higher ·voltage to produce the ·s.ame field •and exaggerating the edge effects of the field. effects are of unknown magnitude 'Since these :edge and can not he calcu- lated readily it was considered advisable to substitute for the flat plates equiangular sectors of coaxial cylinders which would permit much closer s~acing. These plates are made of brass, gold plated to pr~vent corro·sion:, and are held ·a :fiixed distance ·ap,art by ·spacers o·f glass fixed to each ·side by screws. 'The lower plate is .in electrical contact with the body of the ap~aratus PLATE 1 I ... ........ ··· ·····• • - +- + -1---,-__,_______--1 XII 21 and is :_accurately located by dowel·s ·so i t may mo·ved ;and replaced readily. ·be re_: tA photograph of the elec- tric field ·assembly is reproduced in the upper panel o·f plate XII. Neglecting edge effects the field between the two plates is :an in'vers:e first power field and the ·solutio.n o·f the motion of a charged part.icle .in ·such a ·field is of a ·very in·vol'ved nature but h:as re.cently been worked out for the c a s:e required here. ('Se:e ;Doctor' ·s :Disser- tation, G. 1C. Munro, )California Institute, 1i9:33) tAll the rays are assumed to be pa~allel and to enter the field :at the ·same point, on the mid~ay between the plates. leading edge It is required to find the emergent paths and ·velocities:, for the ·veloc.i ty deter- mines the radius of cutvature in the magnetic field and the emergent path determines the magnetic field. the point of .incidence to ·The emergent path is described by the angle ,it make·s with the medi·an ray and the point at which it cuts the trailing edge of the tAll solutions are gi'ven in terms electric field. of ,a parameter, x='<f>o/ 2E lg ( p:2/ Pi ,) •cp0 is the ·volt age •applied to the electric field, :E is the energy of the particle in electron-·vol ts :and p: 2 and Pt, are the outer and ,inner radii of the electric field. If the center of cut~ature of the plates of the field be ·taken as the origin, the median ray describes a circular arc of riadi us rcx ·about this point and the r:adius to the point of emergence of any ray is p=ucx. of the characteristic u is one quantities of the emergent path and is gi'ven by u= 1+( 1-x) •·f ( x). The other character- 22 is tic quantity of the the angle emergent path ,is the tangent o·f between the :emergent ray and the medi1an ray, and is expressed by (1,-x) • F(x). l.A po's.i ti've ·s.ign for this quantity indicates that the ray undergoes a greater deflection in the electric field than does the median ray. ·The relation between ·the emergent ·velocity 'V and the .incident ·velocity ·v i ·s expre·ssed by ·v=·. vv[ 1-x( 1-x) · ·<ti( x) l ·The serie·s in·vol'ved in these expre·ss.ion·s are: · f(x.)=.032,085,023,984, 212 2 - . 0 0 1, 20 4 , 28 9 , 20:3x:. +. 0 00 , 0 4 6 , 88 0 , 155x· - • 0 0 0 , 0 0 l , 8 53 , 9 6 x 3 ·. +.000,000,0 h1 ,3,95x 4 -.000,000,002,9 7x 5 :+.OOO,OOO.,OOO.,l.11x 6 :F( x.) =. 255,-341, 921., 221, 1:33 -.010,957,952,9 ·2 X :+.000.,458,41'9,9 X 4 :+.000,000,790,8 x. -.000,000,032, ,5 x. e-; 3 -.000,019,040.,1 x• , '6 :+.000.,000,001,,3 x. ·cp ( X) = • o:3t. ,•581, 05 1, 247, 468 -.ooo,67,3, .377,81x. ·. +.ooo,01{a,721,99x.· . 4 :+. 000, 000, 01 ,9 , 2 x. - . 000,000, 000, 65x 2 ~ -.ooo,ooo.,627,•57x. 3 :+. 000, 000., 000,Q 2x 6 'The par;ameter x has the ·value unity for the median ray, is less than unity for rays of greater energy, ·and is greater than unity for rays of less energy. .considered ·ad'visable to extend the calculations the range x=0 .,5 to x=l<. 5 to aid in .ing cutve of the magnetic ·field It was o·ver defining the bound- ~ven though the usable part of the :f.ield is included in the rang·e x=0.7 to 1:c.;3 It .can :also be shown from these results that while the backward prolongations of the emergent rays do not pass exactly through :a ·single point, it .is ,a ·very 'Sat- isfactory ·approxim_at.ion to assume th:at they do for each 23 ~ay i.ntersects e·very other within a ·very ·small region. In connection with the .complete redesigning of the apparatus it seemed :advisable better results to in·vestigate whether might not be pos-sible .with 'some differ- ent ratio of magnetic to ·electric deflection. The same electric deflection that had heen previously used, cr;uarter radian, was again cho·sen and fi've one different ·values of magnetic deflection .were .combined with it in a graphical solution .in order to examine the resulting mass dispersion. ·The cases ·selected were for defle.c- tiorrs of one-half and one radian .in the same direction as the electric deflection, and one-half, one, and rc/2 radians in the opposite direction. used, one radi,an in The case previously the reverse direct.ion, ,was found to be best for no other case gave any better di·spersion and mast required ·shapes of the magnetic more difficult to construct. field much The de·s.ir:able restric- tions on the shape of the magnetic field are; the incident rays ·should be approximately normally .incident and the angle of incidence should have only a ·small ·varia- tion o'ver the us·able range of the pole face, bounding curve of the pole face should 1 and the preferably be one -that can be generated mechanically, or :_ at least one of small with and regul·ar curvature reasonable precision 'Considerable proper select.ion ·since ·so much There seem ·so that tit can on ordinary importance is machine tools. also attached of matenial for the be cut to the magnetic circuit depends on the homogeneity of the field. to be only about three suitable materials 24 for a large magnet; !.Armco iron, 'Swedish or Norway iron, and Rema iron. In the present instance neither Nor~ay nor lArmco wcrs av ail able in the re qui red size so the ·select ion of Rema was indicated. Fortunately it is :al·so probably the best of the three. 1An average ·analysi·s of Rema dead ·soft electric ·furnace charcoal iron, kindly ·supplied by the local agent, limits for .impurities: 10arbon, .7%; Phosphorus, .002-.005%; 'Silicon, .1,5-.25%. in :a ·very gives the following .02-.05%; Mangane·se, .2·sulphur, .015-.020%; and 'Since this iron was available o.nly wasteful form the local agent, had the necessary ,amount forged Ryer, Inc. :, ·from a ·six inch Gothic ingot to a four ,inch ·square section. "Through the cour- tesy of Mr. ,D ·s 1Clark ,a rr.etallographic examination w.as made of a ·specimen ·as received and aga:irn after thorough normalization and this r~vealed that no advantage would result from further annealling ·so the iron .was used a:s received. ~lthough quite ·soft it works extremely well and seems completely homog·eneous, iapparently ·an 1admirable material for magnetic work. 'Since the analysis of the electric field ~ad ·shown that no analytic expression existed ·for the character- istics of the beam ,after tri_a:versing the electric fi:eld, the only a:~aiLable method for determining the proper shi_ape for the magnetic field l 1 ay .in :as·s.ign.ing :appropriate ·v:alue·s to the g~ramet:er x ries of points on the des.ired :and so calculating 1a ·securve. Tt .w,_as found that the calculation could be greatly ·simplified ' by introducing the further restriction on the magnetic field that it be ·symmetrical with re·spect to the perpendicu- 25 lar hisector of the line joining the focus to the point in the electric ·field which ,is 1 as-sumed to be :a ·virtual ·source for the emergent ~ays. oate·s the .center of curvature o·f ·symmetry. Thi~ coriveniently lo- ·for any ray on thts axis "The necessary geometrical consi der:ations ·are shown ,in the figur.e in which BE i ·s the :axis of ·symmetry, ~D :and lA the ·v,irtual ·sourc:e .in the electric ·field. is ·a ray which makes an ;angle ra ,with the med~an ~ay .AB, ,and ·s.ince L'B.AG=½ radi1an, Ls :FIDC, 1 6rach ·equal to ½-,a. DF'C, ;and :UAG are EFJJAD and i's of length R, and :FlC..LRE B i·s taken :as the orig.in of a ·system of rectangul!ar coordini,ate·s with x po·si tive upward positive to the left. ·and y for con'venience y=:FlC='R ·sin ( ½-1a); t.AB sina ·sec(½-10t)+R sin(½-1a) tan(½-10t). x::;BC= BD+:DC= lAll .needed quan- ti t:1.e·s iare known .with the exception of :R which is g.i'ven by the relation the I'l_adius o·f ·R=R 0 v'[1/x-2( 1-x) •'<p(x)] .in which R0 curvature of the is medi1an ray .in the mag- 26 netic field. 'Such :a qalcul::ation W[ZS used to determine the x and y coordinates of ·fltftyffve points on the edge of the magnetic field, 1_ and ~the lower panel o·f pl:ate '. XII ·shows a plot of these po.ints . -' the origin1al points ·.The upper curve ·shows plottelf ~·t o "·fi've tiirre-s full ·si.ze in each dimension ,and the low.er curve iis ·t.ime·s fi've on the abscissas and times ·fifty on the 6fdi~ates to .check the smoothness of the curve. the same points 'The :intermed t ate .curve ·shows ·after the :edge ·corre.ct.ion ·h:ad b:e:en ,ap- plied. 'The original points wer.e based that the m_agnetic field b:etween on ·the tawsumption the pole ·fi,ace-s ti's uni- form and everywhere els:e ;is zero·, but unfortunately ·the actual ·situation .is different. is to weaken the field 'The :effe.ct of ·the edge ·somewhat .near the :edg:e ,irrs.ide., and the field 1also fringes into the ·sp,_ace outside ·the pole·s. 'This required the computat.ion of ·two edge cor- rections, which ,was done by Mr. Munro. abo've) (iSee reference 'The results obtained ·show that to a reasonable :approximation the line ,integ~al of the along the p_ath of ta p .a rticle magnetic ·tield .when calcul:ated by me,ans of :a !Schwartz tr:ansform_ation i ·s the s;ame ·as d·f the uniform field w:ere taken to :extend beyond the edge a dis- tance equal to eight--sevenths of the tair gap. ·However .it must he noted that all rays do not have the ·same anl gle of .incidence ·so the .correction must ·be modified ·ac- cordingly. 'This .is now final ·form of the edge the parameter x sufficient to determine the but the points as determined by do not represent equal increments .in x 27 and would ·be ,incon·venient to ·set up in 1actually cutting the shape of the ·face. '. For this r~ason ·the origin o·f coord.inate-s to lie in the ;was shifted the pole face 1and an interpolation bottom edge o·f .~as made ·for ,a new ·series of points lying uniformly 0. 040J.i. RJti 1apart in the x direction. LAn :attempt wias made curve by means of in the to •approximate the an el1ip·s:e ;which could millilig machine. fitted to the points but when a be gener,at:ed g· eneral conic .was by the method of least squares it pro·ved to be a hyperbola having ·an ecc:entrici ty far remdved ·from a possible ellipse. X X 0.000 1.7889 040 7881 080 787,3 120 7865 160 7857 200 7849 240 7841 280 73:3 2 78 24 ·320 :360 7816 400 440 7808 7801 7794 7786 7779 480 520 ,560 600 640 680 720 760 800 ) 840 880 920 0.960 1,. 000 1 1• 7714 7771 7765 7758 7752 7745 7740 7734 7728 7722 7718 7709 7704 7700 7696 769:3 7691 The ,actual points are: · X 2.000 1.7703 7709 7714 7722 7729 7738 7688 7685 76&3 7681 7680 7679 7678 7677 7677 X y J.ooo Lc.8084 S:~<4 8145 8179 82:L3 8249 8288 8328 8/368 8408 8458 8510 8563 7746 7756 7765 7775 7788 7800 78 :L3 7828 7844 8615 8668 7858 8720 7678 7877 877,3 7680 7895 7915 8825 8878 39;30 393;3 7679 7682 7684 7687 7690 7694 7698 79:35 7956 7978 '8005 9035 9088 8029 80 55 1 ·The ,actual cutting .was done on the shaper. utilizing a long ·stroke indicator on the feed motion regular indicator for the depth of cut. and a :Whenever the 28 feed indicator pointed to a multiple of 0.040 the depth of cut .was quickly adjusted to the right value for that point, ·so that a ·serie·s of ·step·s was cut who·se en·velope was the desired cutve. ·The final finish was obtained by scraping by hand until the tool marks were just ·vi·sIt ·s:eems likely that :a precision of 0.001 or at i ble. mast 0.002 inch ·should be realized by this means. When it ,was finally decided tron in place of the Hoffmann already on hand so it to use the '. FP,54 plio- there happened to be one was quickly tnied out .in a one tube c.ircuit, al though it w:as hardly rea-sonable to ex- pect one tube to be -sufficient. would he necessary to use ·a two tube circuit to combine sufficient stability si ti'vi ty. ·This she.wed that it in order with the required sen- However when the second tube was obt,ained ' SO great was the disparity between it '. and the first that the usual circuit would not permit compensation for the difference, but a determination of the complete tube charactenistics ·showed that a slight modification would suffice. !As originally ·set up ·se·veral radio potent-iometers and the c.ircui t •.included ·vari:able re·sis~ance·s which were ·a continual ·source of trouble because of the poor contact between the rotating :arirt :ahd the dvinding. Increasing the contact pressure displacing worse. the winding, frequently resulted .in thereby making matters ·still "The trouble was fin:ally :eliminated by -substi- tuting ·soldered connections wherever po-ssible, ·vising a new type adj ustment-s. of potentiometer ·for the 1and denece·s~ary 29 The pre·sent form ,is shown in the "The upper port.ion of the of the circuit upper panel of plate ,XIII. drawing represents the part of the circuit which is included within the mass-spectrometer, part shows the five while the arrangement of the control box. long horizontal lines in the tute the main potentiometer, "The control box consti- the .current through which supplies all the potent.tals for the tube·s. tiometer consists lower of five long ·str:and·s resistance wire having a total "This poten- o·f ·a compo·si te resistance ,in ·series of about 8Ow. Thi·s wire consists copper core 6verlaid with a setving of asbestos fiber, on which in turn of a biack enamelled is wound the actual re·sistance w.ire. This ·succes-sfully avo.ids the nece·ssi ty and frag,ile wire ·si.ze·s for using ·srq_all or .incon·venient lengths tain the necessary resistance. to ob- ·This wire i ·s regularly used in certain types of Electriad resistors ;and was obt·ained from one of them. with Eby ·Cont;act .with the w.ire .is made Midget b.inding po·sts which ·are ·so made as to permit firm contact w.i th no danger of cutting the .wire. The circuit battery cons.is ts of 18 ·vol t·s of hour :Exide radio batteries 100 ampere •and supplie·s quite constant ·voltage, needing to be charged :about once in two .weeks. The potentiometer is seen to be broken n~ar the left end of the ·s:econd ·strand ;and this gap ,is bridged by the filament·s current of ·the two tubes, is the sum of the -Because of ·the ·so that the potentiometer filament currents ·for both. difference in filament ·some additional resistance cha~acteristicrs is included .in the filament of the dummy tube. ·series w:i th ·&, is ·a 2Ow ·vari:able resistance rent. for fine control of the potentiometer cur- R 2 i's a 2000w ·variable re·si"stance for fine ·ad- justment of the galvanometer zero by -slightly ·shifting the space charge grid potential of the second tube. and R4 are 25000-w .wire wound fixed resistances, R3 .which, with the plate impedances of the tubes, form the bridge across .which the galvanometer is connected. 100000,w wire wound resistor impedance .in parallel ,with the plate of tube 2 to bring it to o f t u b e 1. R6 · , 2 , x 10: I ,w, 1 . ·3 Rf; i ·s a equality with that 1 an d •'R7 · , 5 • '. A'Jx 10 HI•w, ·are h.1g h resistance shunts from the 'S. 'S. :white 1Co. ·The balancing of this ·venient than circuit i·s rather less con- for a circuit containing but its real advantage more ·variables, lies .in the fact that reba}anc- ing is less often required for there ,are fewer contacts subject to spontaneous ·variation. ·The most con·ven.ient method is the u-se of the tube character.istic·s, particu- 1 arly the curves of plate current ·vs plate ·vol tag:e and plate current ·vs filament ·voltage. The proper poten- tiometer ·settings filament terminals for the positive are determined from the filament ·voltage charactevistic and the prep er ·value of the pl:ate shunt comparison of the plate impedances. i ·s o ht ained by Then the control grid and plate potentials are ·set at their recommended ·value·s, for the ·voltage drop due :allowance being made through ·R3 and R4. If the ·space charge gr.id potentials are set ·at the.ir recommended ·values the pl ate currents may be decidedly unequal ·so one or both must be shifted to obtain this equality. ·Then by the use of a galV!an- ometer the final balance is obtained end lAyrton shunt by further ·variation of the space charge g rid·s, the final steps being made on R2 • 'The ·very best electrical ·shielding of all parts of the circuit is of the greatest importance in order to secure st 2ble operation. -The stora ~e hattery i·s placed in ·a copper sheathed box surrounded by a double walled celotex box for temperature control, although there is some doubt now if temperature ·v:ar i at ions •ar:e as ·serious as they were once thought. tAll control de'vice·s ,are contained in a box sheathed inside land out with copper. -The pl io t rons and the.ir high res.is t ances are mounted in high ·vacuum inside of the mass-spectrometer, fully protected tagainst moisture, and thus and picking up of at- mo·spheric ions '. and :also electrical disturb:ance·s. al·so ohv.i :ates the neces-s.i ty ·This for bringing out ;a highly insulated electrometer lead capacity. lAll leads are also -well shielded, but flex- ible me~al gas tubing be worthles·s with its resulting which ~as first tried against high frequency tubuiar copper braid i's pro·vided prdved to disturoances ·so ,a ~ade by the Roehling company ·substituted with good results. nection :added was lA special ground con- outside the building :and a heavy ·shielded lead connects it with the circuit at the com- mon negati've filament terminal, while the ·shield 1_around the ground le_ad is used ·as all other shields. the ground connection for It was also found tadvtantageou-s to j1acket the cylindrical portion o·f the tu.bes with nickel gauze which is connected to the control grid l~ad be- hind the ·shunt as shown. Many different type·s of high res.istance ·shunts were tried such a:s ,in di a ink 1 i ne on brfrs tol board, the ·same impregnated under ·vacuum with paraffin or cere·sin, ·alcohol-xylol mixtures in ·sealed glass cells with platinum electrode·s, rods made of granulated · carbon with ·an organic binder, and pencil line on :amber. had to be abandoned for ·v:arious reasons; ·variable, others were too "Thes:e i all some were too hard to ·adjust, ,introduced too much capacity, exhibited polari~ation etc. a ·satisfactory solution .was ·found Finally in the us:e of molded compo·si tion re·s.istors ·secured from the 'S 'S :White :Dental Manuf:acturing Co., L~52 West 42nd ·st., New York. No at- 1 tempt has been made to check the constancy or the ·value of the re·sistance·, but th~ fact that they permit ·stable 117 • opera t 10n :a t ·a ·sens1·t·· 1 v1·t y o f 2.x 10ampere·s per "Saa1 e di'vision is taken as suf fi ci en t .indication of the.ir suitability for this work. iCertain special features have been incorporated .in the new apparatus to fill needs that have ;ar.isen in the operat.ion of the old ·apparatus. ·These :are de·scr:ibed below to indicate their probable usefulness. lA shutter ·system has be:en introduced into the beam just :ah~ad of the magnetic :fiield riable width ·so that a ·slit of ·va- m_ay be p~aced at ·any de·s.ired po·si tion in the beam, .with both the width and the po·s,ition .controll _able from outside will permit the analysis terms of the energy will be ·very through :a dou;ble cone joint. of the ·v~arious ion peaks ,in of the component particles intere·sting to learn ·Thirs and .it ,if certain .ions :are produced at ·such a location in the discharge that the.ir energies are significantly different from the maximum available. tAl·so the shutters will permit :a test of the correctness of the form of the pole faces by irivestigating .whether all zones o·f the magnet same point. If certain regions :are in error also be . po·ssible to eliminate cepting focus at the their contribution their ef f:ec t it w.ill by ,inter- with the shutters, making good results possible from the remainder. Previously the first time, when new peaks were being ·sought ·for as when :a new ·suhstanc·e had bee.n •. intro- duced or instrument adjustments changed, it was quite a Laboriou·s job to find the approximate locations, for 1a great many points had to be ta ken clo·s e together in order not to dverlook a peak. In the new apparatus the focal slit is double and either of two s.i ,ze-s of can be s·elec ted by an extern al adjustment. broad ·slit is used for determining ·s1i ts ·Thus the approximate looa- tions of peaks, when the narrow ·slit is ·suhsti tuted for their analysts. ·There is ·also the po·ss.ib.ility of com- plete closing of the slit when it is desired the zero of the pliotron circuit. to check lA photograph of the slit is shown in the upper panel of plate ,XI'V. ·To permit obtaining the best focus :a radi:al motion of the entire collector ·system is prdvided for by using a sylphon connection between the the body of the instrument . collector system and t.Also :a tangential motion is permitted by hinges opposite the sylphon in the rods supporting the collector system, gent screw varies the inclination. and a micrometer tanThus with the mag- neti6 field fixed the same peak may he repeatedly plot- PLATE ,, ,. -= l XIII ted under varied conditions by using the tangent screw. l./ln en ti rely research unexpected in this laboratory de'velopmen t from another was a s.imple ·and elegant means of obtaining a continuous indication of the field strength in the air gap. plication of the Gouy ·susceptibilities. in suspending a The method i"s an :in·verse :ap- method of measuiing In brief diamagnetic the Gouy method consists cylinder of diamagnetic ·substance hav- ing its axis ·vertical so that one end is in a region of uniform field and the other is in zero field. exerted on the specimen by the field 'Force is then m~asured and this force is proportional to the susceptibility of the substance ·Thus and to the square of the field ·strength. by using a substance of known ·susceptibility the force is proportional to the square of the field. order to obtain good sensitivity a substance ·susceptibility is used, In of high and pseudo-crystalline graphite has pro'ved ·very sati·sfactory. ·This is prepared by al- lowing flake graphite to ·settle through a suitable me- di um so that the fl ake·s are all oriented par all el. ·sLa.b of thi·s material is suspended ance which tA from ·a torsion bal- by means of a suitable optioal ·system .indi- cates on a scale the ·square of the field. molecular weight of a peak square of the field 'Since the i"s also proportional to the this can be calibrated directly .in terms of molecular weight. ·To avoid mechanical hyste- resis the made of fused quartz·, torsion fiber is the whole de'v.ice is mount3d ins.ide the ·vacuum. •and lA sim- ilar device mounted in an external :air gap has b:een entirely satisfactory on the old ·apparatus. tAn isometric ·sketch of the magnetic balance is ·shown ,in the lower panel of pl.ate :XIII. In order to obtain sufficiently clo·se control of the current through the electromagnet, it was necessary to design and build a large rheostat type. It was des.ired of :a ·very ·special to be able to co·ver the .whole :r;ange from minimum to maximum field with continuous ·variation of current because any system irivolve trouble with view of the completed p,anel of plate :XIV. of steps would magnetic hysteresis. A partial rheostat .is shown in the lower ·The main feature of the construc- tion .is the brass drum twenty six inches long and eight and a fourth inches in diameter, 6ver the whole length of which is cut a specially shaped double thread of one quarter inch pitch. wire·s are Two number ten ~au~e resistance wound side by side in this thread 1 ated from the drum by a layer of ·varnished In order to obtain the desired teristic it was necessary and insucambric. resistance cha~ac- to make the first eighteen turns of each wire of ~dvance, ·and the remainder of Nichrome, the joints being ·silver soldered. ·vance end·s The free Ad- are connected to collector r,ings as ·shown, with which connection is made through ~aminated bronze brushes, and the other ends are anchored ·separately to an insulating block so tests for short circuits ,and for ground·s may be made. ·stalled Mercury cups were originally ·.in- on the collector rings irrstead of brushes ~ut almost caused the loss of the rheostat fo.r e·ven the :agitation of the mercury by the rings cause .it to emulsify with the oil, was sufficient to thereby threatening PLATE XIV the amalga~ation of the whole instrument. are ·short circuited The w.ire·s at :any de·sired point by a wedge of copper mounted on the carriage which is .capable of mo- tion along the rails parallel with the •axis of the cylinder. This wedge is ·sufficient to drive the carriage so that it follows the wire but it rwhowed :a slight tendency to lo·se contact with one w.ire so ,a laminated contact was :added factory~ and the combination is entirely ·satis- ·The min.imum resistance obtainable between the terminals is less than one tenth ohm and the maximum .is about twenty two ohms. tA dial graduated in 100 di'vi- sions is mounted 011 the shaft ;arid a ·slow motion gear of ratio 100: '1 permits settings to .0001 revolution of the In order to keep the surface of the wire clean drum. the drum and carriage which ·also are immersed in transformer oil, practically eliminates be.cause the thermal capacity temperature changes of the rheo·st:at ,is enor- mous compared with the energy dissipated in it. It is indeed constant interest this problem. 'Smythe, a pleasure to acknowledge displ:ayed by here the Professor Millikan in 'Special thanks a:ee ·also extended to ,Dr. who willi11gly assisted -in the difficulties encountered o·vercoming many of !and who kindly crit.ici,zed the pl:ans for the new ·apparatus. Much credit ,is 1also due the ·staff of the .instrument ·shop for the.ir ·skillful assistance in .building apparatus, has ·exerted his ,ingenuity on many occasiorrs. and to Mr. :Clancy who to good purpose in my beqalf