W611 t1ESOLiJ‘e"!ON SPECTROSCOPY Am; H‘V’PERWW STRUCTURE ||I|II|IIIIIIIIII|IIIIIIIIIIIIIIIIIIZIIIII 3 1293 01770 9092 This is to certify that the thesis entitled HIGH RESOLUTION SPIGEBOBOOPY AID martin 'srmcm presented by Robert. Carney Honrydo has been accepted towards fulfillment of the requirements for Hagtor's degree in Pnsics 6.4.9M Major professor Date W5 . 0-169 \ - I If , ‘ - .. Y ' . ‘ _ . l 4.2.;Wv-NL10 ‘,,‘_:,‘ ‘ .. _ 2. ‘. 5 . .‘ ‘r ‘I ‘45 h . PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 1/98 cJCIRC/DateDuo.p65—p.14 HIGH ESOLUI‘ION SPECTROSCOPY AND HYPEEFIE STRUCTURE by Robert Carney McBryde l THESIS Submitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of “SM 01' scram Department of Phys ice 1951 Grateful acknowledgement is made to Dr. 0. Duane Reuse for the suggestion of this problem and his invaluable assistance during the emerimental and interpretation phases. To Dr. G. Kikuchi for helpful-discussions of line breadths and hyperfine structure appreciation is also expressed. 12.Lex— CNML4§MAK'&__ l {353{3?%53§% II. III. IV. 7. VII. TABLE OF COWS Introduction Line Breadth a. General 3. Natural width 0. Doppler width D. Collision broadening 3. Pressure broadening 3'. Start broadening Hollow Cathode Discharge Source 1. Introductory B. General design 0. Theory of operation D. Electrode design 3. Gas system and power supply General Theory of High Resolution Spectroscopy A . Introduc tory B. IabrybPerot interferometer G. Lummer-Gehrcke plate D. Summary General Eheory of Hyperfine Structure A. General B. Hyperfine structure in.Cu. Experimental. A. oPtioal systen 3. Results of investigation Conclusi on P380 no. 11 1% 17 17 18 27 31 36 37 37 MI 51+ I . Introduction . In the big: resolution investigation of the atomic spectra characteristic of elements having a nuclear spin. spectral lines which otherwise appear to be single are found to be complex groups of closely spaced lines. the hyperfine structure of the transition. Since know- ledge of such structure intervals and the information derivable from them - nuclear spin and magnetic moment, nucleus-electron interaction constants - leads to a better understanding of atomic structure. the measurement of hyperfine atructure is of importance. Hyperfine structure intervals are often small fractions of a wave number (cm'l) and hence light sources which produce lines having breadths suller than these intervals are required: actual resolution of structure being accom— lished using l'abry-Peret and Luner-Gehrcke plate interferometers or reflection and transmission type echelons. A knowledge of the physical conditions influencing spectral line breadths and control of their effects and the techniques of using interferometers as spectrometers are necessary. A summary will be given in this report of such factors and the type of broadening produced. design and construction details of a hollow cathode discharge source. and an indication of the theory underlying interfere-eter spectrometers - the l'abry-Perot etalon and Dunner-Gehroke plate - for use with such a source. he theory of hyperfine structure as arising from nucleus- electron interaction will be given and an attempt will be made to investigate the hyperfine structure of the "D - ‘P mltiplet in On. In addition. qualitative measurements of the influence of source operating conditions - gas pressure, cathode tesperature and discharge tube current - on the line breadth as observed in fringe width will be presented . A. II. Line Width. In order to reduce spectral line widths as produced by a source. the varirsis physical conditions influencing such line widths must be analysed and these factors utilised in the design of a suitable source. There are. in general. five factors influencing line width; these are. 1. Radiation daqing latural width 2. Doppler effect Doppler width 3. Collision damping Collision width ’4. Pressure effects Pressure width 5. Stark effect Stark width 2. Of these five effects the following statements my be made: Only the natural and doppler widths are independent of the density of the emitting or absorbing atoms. the remainder depending directly won the density. Quantitative discussions are possible from a theoretical viewpoint for only the natural and doppler widths: only a rough qualitative explanation being possible for the others. Collision and pressure width mechanisms are not fully understood; may differing views are evident in literature. latural widthl Classically. the natural width of a spectral line is due to radiation damping; a vibrating; therefore radiating. electric charge -h- continually loses energy and, as a result. the amplitude of vibration decreases while the frequency of vibration. 1/. . remains constant. Assuming an electron which may vibrate about an equilibriun position subject to a linear restoring force. a fourier analysis of the electric moment of the charge as a function of time t reveals that the radia- tion is not monochromatic but has a definite frequency distribution t/u‘n‘ (1) It») - __ z the quantity r is (14-?)1’rLE) given by ‘l. 2. (2) r= 41Tav. a...“ where MA- and a are BMC‘ the mass and charge of the particle in question. he form of this distribution is indicated in rig. l. 1'02.) I‘v) l‘ig. l v:I-—-————-———— The half width"| of this frequency distribution is given by the constant r .' since setting 1(1)): In /7_ , half the maxim intensity at 22.22, , and solving for 2/ gives u = 24,4.- r/z. Notice that X ‘ Dy half?idth we shall mean the width of the frequency distribution curve at half maximum intensity. is a constant and has the value I X= 41V: ..,. 3AMC7’ 01' (s) x = m . to“ A . a value which is independent of wavelength. Actually. observations of the natural width indicate that it is not a constant but depends upon wavelength. his nondependence of the natural line width upon wavelength is not realised when the problem is considered quantum mechanically. Here the energy levels. a. and E; , betwun which the transition occurs are not infinitely sharp but have finite widths AE. , and cat which are given by the Heisenberg uncertainty principle. AE. at no Ipw . at: being thought of as the average time spent by the atom in that particular energy state. Analysis of this con- dition shows that the total line width is the sum of the term widths of the two levels: Y= 3’. r X1. and is given by the relation (II) x = ‘33:. (:23... + 222:“) where 24. and I?“ are the frequency and “oscillator strength" corresponding to the transition 5:. «a. E" . the sumnntion being over all energ states lower than E. . Although the actual values for the oscillator strengths are unknown in the general case, the line width will depend upon the wavelength (frequency) of the transition and is not a constant as predicted classically. c . Deppler widthl’a In the classical Dcppler effect. the frequency of radiation 2/ -6- emitted or absorbed by' an atom in motion is (5) V: V. (\ _ m/c) where v. is the frtquency of radiation from the source and w the component of velocity in the direction of propagation of the light. Clearly then. in a gas with a random or thermal motion of the atoms or molecules this change of frequency with the nature of motion of a particle will produce a broadening of the spectral lines. In the case of a discharge in a gas the particle velocities, and since there is a preferred direction of motion an actual shift in the spectral line will be observed. Suppose light from a uniform, continuous source passes through a gas. the molecular or atomic velocities being only those due to thermal agitation. If all of the gas molecules were at rest the frequency of the light absorbed would be 2/. (assuming a gas of one type molecule only). a very nearly monochromatic frequency. If the molecules are in thernl equilibrium the frequency of the light absorbed is v as given by (5) and the fraction of the total intensity absorbed will be proportional to the number of molecules having velocities in the interval 0,. to vu+ «N. . Assuming a Harwell-Boltsman distribution of velocities. the number of molecules having this range of velocities is _. M 1 (6) 4N“ " N Ifirrzr “I I“ ”321-4" ) M,“ where N is the number of molecules having mass M at an absolute temperature T , 2 being the gas constant. he frequency distribution function of the absorbed light is 81'!!! by 2“- or v.‘ 7.9.1“ 2/: equation (5) being used to eliminate W and do. . The form of this distribution is indicated in Pig. 2. I hi.) T. (‘9) fig. 2. I -| I I I I I | 4 ‘V. 22—) The half width of this frequency distribution is Duh -_ 2. \Dflz‘lfiltor in terms of wavelength (8) ALA = 7. ioq 2. 3%.; k. A, being the wavelength corresponding to a frequency Va . Iron the expression for the doppler width. it is evident that inn/L depends upon the wavelength; huh being more pronounced for larger A . Unlike the natural line width. the doppler breadth may be experimentally controlled by merely varying the tesperature T of the absorbing or emit- ting atoms. Such a practive is very necessary for lighter atoms due to the dependence upon the inverse root of mass. D. Collision width Considerable controversy over the meclnnism of collision broad- ening exists. There is the impact theory of Lorents which visualises the -8... radiation V. omitted or absorbed by an atom as being terminated during the time of collision with another atom. this finite wave train giving a frequency distribution of intensity. Statistical theory predicts a collision broadening due to a perturbation of the energy levels of a radiating atom by the presence of the remaining atoms, the perturbation energy being a function of the positions of the perturbers. A brief discussion of each of these methods of attack is given. The impact or Lorents theory of collision broadening makes the assumptions: 1) Energy is emitted or absorbed only during the time between elastic collisions. the radiation process being resumed after collision with the same frequency and possibly a change in phase; 2) The time of collision is very short compared with the time between collisions. The radiating atom is pictured as colliding elastically with other atoms. the amplitude of radiation emitted during the period between collisions being of the form Aeiw’i- . Using the form of the amplitude as (9) Am = 1° . "W A. c c < t < "D the fourier integral of this function yields for the amplitude frequency distribution In...) the result {(wa‘w)© (10) £043) '-= L2. C. - \ ET? ’— i(w.-w) e Since in a gas the radiators are incoherent. the frequency distribution - 9 - of intensity is proportional to Z I Icusil and is given by (n) 1‘ Lu...) : C. one" LN". l”)? for ., ginglg (w- - “93" particle. Since Maxwell-Boltzman statistics obtain for the gas particles. the number of molecules which have a time T between collisions is 0.37:" . to being the collision frequency. On this assumption of a Maxwell-Boltsmn distribution of speeds for the gas molecules. ‘5. is given by (12) to = -——‘——;- _ o- - collision diameterof ‘1';- FG "W molecule m - density of molecule! 4'} - mean speed of molecules. The total intensity for a frequency a: is than -t a. -e. It...) = c S c k“ oin1(‘2'{' It at: which is of the dorm -00 “a" -w)‘ (13) The} , 1'an _ where In...) is the I + t: (ma-9‘")! axial intensity. From this intensity frequency distribution. the half width is seen to be (ll-I) AW. = 13-3,: where 1:0 is given by (12) and since «'3' = B‘s-T. becms “WM (15) 1W7. = 40“!w gm . Hence the line breadth on the Lorents theory of impact broadening depends directly upon the density of molecules and also upon the temperature and mass of the absorbing or radiating molecules . -10.. iAlthough this theory gives a quantitative expression for azgt its dependence upon a collision diameter c- is misleading. If obser- ved line breadths are used with (15) to calculate if the result does not agree with the molecular diameters as determined from viscosity measurements: in.almost all cases a factor of from 2 to 5 exists between 6' as determined spectroscopically'and that from viscosity effects. his points the need for some other mechanism of collision broadening. the statistical theory. In the statistical theory of collision broadening; the instantaneous distribution of molecules in space is analysed. the energy levels of the radiating or absorbing molecule being perturbed by interaction with neighboring molecules. The frequency of a spectral line arising from a transition between.any two levels will then be dif- fused over an interval corresponding to the range of such.intermolecular perturbations as are involved. In the range 2/ to V+ 41’ the intensity of a line is therefore proportional to the statistical probability'that the energy difference of the two levels involved lies between th and h(v+d'v) . In low pressures this interaction will be predasinantly one with nearest neighbors and weighting the frequency shift (“V- V-) in this case according to the probabilities of the appropriate inter- molecular separations. the intensity frequency distribution I (v) is found to vary inversely with (12- v.i'. Although this distribution has an infinity at resonance. :2: v. . the general form away from resonance is similar to the symmetric distributions already encountered. -11- Holsteinn in a paper “Pressure Broadening of Spectral Lines“ shows that the impact theory of collision broadening is valid near resonance conditions, that is near the line center. and also that the statistical theory is valid for frequencies v» v, . in the wings of the spectral line. In addition. the intensity frequency distribu- tion as predicted by both the impact and statistical theory treatments are continuous: that is. the distribution function for the center of the spectral line passes smoothly into that for the wings with no dis- continuity in the function or its first derivative. Collision broadening of spectral lines. for low pressures. results in a synetrical intensity frequency distribution similar to those for natural and Doppler breadths as in l'ig. l and 2. 2|. Pressure broadening. Increasing the density of the molecules surrounding a given emitting or absorbing molecule has the effect of producing an asymetri- cal broadening and an actual shift in the center of the spectral line. the shift being of the order of several angstroms in the case of extreme pressures. In all cases the asymetry in line shape is to the red. to lower frequencies. A qualitative explanation of the asymmetrical broadening is found on extension of the impact theory of collision broadening to take into account the altered radiation emitted or absorbed during the actual collision. One new uses for the amplitude function Mt) . analogous to (9). o t e 't (16) Mo = A.e' «in. A1 9. ‘txt <1: where 1‘. is the period betwun collisions and t-t. the duration of the collision process. in intensity frequency distribution function. To...) . should now be possible by proceeding as before in obtaining the fourier integral of (16) and. using this. determining n...) . However. the expression obtained for It») is extremely couple: and. unless the frequencies between and daring collisions are tnown,does not indicate the asymmetry in line shape. Only in the extreme case of near instantaneous collisions (relative to the time t. ) may Th») be obtained. In this case (13) is obtained since a low pressure is assumed in lung “1.7? (1‘ - ‘1) . Recently Margenaus has investigated the effects of pressure broadening on a statistical basis for any interaction law which varies inversely as the n'th power of the particle separation. The distance betvmen two energy levels involved in a transition is a function of the unperturbed energ so and the potential due to the surrounding perturb- ing particles. v where v = ; Vi . Hence the statistical treat- ment demands that the probability of an instantaneous configuration of particles giving V -- ; v.- mst be determined. this probability giving the spectral line intensity. From this analysis. Margenau finds that the probability WM that a total potential V is produced at the emitting or absorbing particle is given by .. 31M (11) WW = fin \er- QMLV.) 1mm du 0 where Vc represents the potential energy of two particles at the mean distance of separation and where U: is assumed to have the form -13- V.- = c a?“ u 6;) z u in) depending upon particle spin and all other variables other than M and in the analysis is assumed to possess a vanishing mean. he isportant result of this analysis is that the intensity at the center of a pressure broadened line. WU») . is -Ml3 Wtol °‘ dL where a is the number density of the particles. Comparing collision and pressure broadening it is evident that in the former case enly nearest neighbors are considered: the interaction of these two closest molecules being treated either as a collision process or from a statistical point of view as to the frequency shift. When higher particle densities are encountered. nearest neidibor interactions are no longer sufficint but the entire assemblage of molecules must be considered, as perturbing a single radiating molecule. the effect being one of a depression of the molecular energy levels. Due to the fact that inner levels are more tightly bound. the depression will be larger for higher. loosely bound states and the resultant spectral lines will be shifted to lower energies. that is to the longer wave lengths . the resultant intensity frequency distribution for pressure broadened spectral lines appears somewhat as indicated in fig. 3. which may be compared with the symmetrical line shapes of Zl'ig. l and 2. /P. 31;. 3 -1)..- Although fairly big pressures are required to produce an easily observable line center shift when using low dispersion instruments. when extremely high resolution is employed even moderate pressures give detectable line shifts. a serious defect indeed if quantitative investi- gations are to be made. 1'. Start breadth. Analogous to the Zeeman effect in spectral lines. the splitting of a spectral line into numerous components by the action of an external ugnetic field upon the angular momentum of an atom or molecule. there is a corresponding effect produced by an electric field. the Stark effect. In the Zeenn effect the total angular momentum of the particle is quanti- sed in the field direction and may take only the values may gr where the magaetic quantum number m1. takes the values MI: 3' T-I ... —‘S' 3 each orientation of the angular momentum vector representing a different particle energy. Although the total angular momentum vector is quantised in the electric field direction in the Stark effect. the magnetic quantum number takes the values my: o,..,tz... i3- . theplus'andminus sign indicating that the particle energy is the sue for either orientation: depends only upon the magnitude of m; . An exact solution for the effect of an electric field upon the energy levels of a one electron atom is found by solving the Schrodinger equation for the hydrogen atom. the interaction of electric field and atom them being of the form -15- or: AF+sF‘+cv3..-. where o'r isthechange in term value for a field strength F , A , a and c. being constants depending upon the quantum numbers. For low field strengths F . the term value shifts are very nearly given by 0T= A F and hence is the first order stark effect: as larger fields are used the second and third order terms become important. these being denoted as the second and third order stark effects. In a gaseous discharge many ions are formed which. upon col- lision or close approach. subject emitting or absorbing molecules to very intense electric fields with the result that the observed spectral lines. transitions between these altered term values. will be affected. Since the electric fields produced are not of constant value but vary during the collision or near-collision process reaching a maximum and diminishing once again. the net result on the observed spectral lines should be a symetric broadening. fhe electric fields through which the molecules move in a discharge should also produce a broadening of the lines and. in the case of molecules possessing permanent electric dipoles. qmdropoles or higher mltipoles. collisions or near-collisions between the molecules should have the effect of broadening the lines. If an average field F is assumed to exist during a collision of two particles. calculation6 has shown that such a field is given by ., (18) °~- ¢ M 5 ions F = at)“ M‘ dipoles «I: as 1, "* quadropoles -16.. where the a; are constants. M is the density of the particle. e. the electronic charge. [a dipole moment and ‘5 the quadropole moment. It will be noted that the I'average field'' produced by the ions or electric nultipoles depend strongly upon the density of particles in the discharge region. from than average fields can be calculated an intensity distribution function. no»). the form of which is similar to that of mg. (1) and having a half-width my.“ given by (19) 3.1g Amuue My, ion av.“ = 4.91 tum/um. dipole 6.37. Apfluim‘h quadropole . the quantity A.“ gives the total spread of the stark levels from the change in term value 01' = AF+ BF2 + C"; w-neglecting the second and third order terms. i'hese relations for the half width give only qualitative agreement when compared to experimentally observed stark broadened lines. Summining, then. the factors influencing line widths. nature of the influence - a broadening or shift. type of broadening - a symmetri- cal or asymmetrical. and the method of reducing or eliminating the effect are given in Table l. :- TABLE I. Width Type of effect Broadening Remedy ) Natural Broadening Symmetrical None Doppler . Broadening Symmetrical Use low . temperature Collision Broadening Symmetrical Use reduced pressures Pressure Broadening Asymetrical Use reduced pressures Stark - Shift-Broadening lither Reduce electric - ' field used .. 17 .. III. Hollow Cathode Discharge Source A. Introductory Of the several light sources devised which are useful in high resolution investigations. hyperfine structure and rotational structure of molecular band spectra. the hollow cathode discharge tube is simplest in construction and is ideally suited for such work. Vith this source it is possible to reduce spectral line breadth until. in some cases. the natural width is predominant. In addition to a small line width the hollow cathode discharge tube gives a source of extremely high intensity: sharp spectral lines and high intensity being the chief advantages of such a source. lssentially. the hollow cathode source is a modified Geissler discharge tube- a source in which an electrical discharge is maintained in a gas at reduced pressure. ruchen" modified the Geissler tube by introducing a hollow cathode and found that with correct operating conditions the negative glow receded to the cathode interior and there resulted in a very high intensity glow. The disadvantage of this source was the large Doppler broadening of the spectral lines which masked possible structure. Schuler8 redesigned the source and removed the cathode to the exterior whence very drastic cooling could be employed. With this possibility of using a low temperature coolant the source became one combining sharp spectral lines and high intensity. the correct combination for the attainment of low resolving limits. -18- B. General design. Iollowing design suggestions by Arroe and new". a hollow cathode source was designed and constructed which incorporated several of their design features. Figure it shows in sectional view the hollow cathode discharge tube proper. Since drastic cooling is employed. liquid air for best operation. the main body of the tube is of monel metal. an alloy which retains high tensile strength at low temperatures. Within this body is a glass tube which is closed by a thin glass obser- vation window at the upper end. Within the glass tube at the lower end is placed an aluminum anode. this then being electrically insulated from the cathode. A hollow cathode insert made of copper is placed in the remevable base; the vacuum seal at the base is made with a compressed fuse wire ring. On the gas inlet tube and at the tap of the monel body are placed water Jackets; the former to nintain the inlet tube at room temperature in order to protect the wax Joints and to prevent objection- able frosting of the system. the latter to prevent undue strains on the glass tube and also to keep the apeison wax at room temperature where its sealing properties are good. C. iheory of operation. Although the hollow cathode source may be used to investigate gaseous sampleslo. molecular band spectra then being obtained. it is commonly used with pure metallic or metallic oxide samples. As the vapour pressure of such materials are very low. the metal sample is sputtered to form a vapour cloud within the cathode which say then be f___: E f ‘— anure 4 Ho'low Cathoo‘c Discharge Tube -20.. excited and its characteristic energies observed. In such a source an inert gas. argon and heliun being ccssnonly used. is passed through the discharge region at reduced pressure. Apply- ing a potential between the anode and hollow cathode initiates a typical gaseous discharge. As the operating conditions of gas pressure and volt- age are altered the discharge positive glow disappears while the negative glow becones prominent and recedes to the cathode interior: the negative glow then assuming a very high intensity. Under these conditions the positive ions of the gas formed in the discharge are urged toward the cathode and upon striking the interior walls of the cathode sputter any material there; that is. a local evaporation of the surface occurs and a vapour cloud is formed. Due to the mechanism of the discharge large numbers of carrier gas atoms are excited to metastable energy states which. since the atoms are unaffected by the electric field. accumlate about the cathode. These metastable atone still possess energies of 19.? volts for helium and 11.5 volts for argon. Collisions between these atone and those of the vapour cloud result in a complete transfer of potential energy of excitation from the metastable atoa to the vapour atom. mch potential energy being transformed to kinetic energy of notion and. most important. potential energy of excitation. The excited vapour atoms then return to the ground state radiating their characteristic frequencies. !he observed sample spectra should consist of the are or spark excited lines depending upon whether He or A carrier gas is used. Due to the fact that the carrier gas ions are returned to the ground state with- out radiating energy. no evidence of the carrier gas spectrum should be -21- observed. At least the intensity of such lines will be very snll relative to the sample spectra if scouring at all. Also. the vapour cloud is at the tenperature of the coolant as is the carrier gas and hence Doppler broadening of the spectral lines will be entirely deter- mined by the coolant temperature. I). llectrode design. llectrode design in the hollow cathode source is the key to successful operation and to the proper elimination or reduction of con- trollable line breadth. 3y proper design the operating conditions - gas pressure and electrode potential - my easily be controlled: rare or minute saaples may be used due to recoverability of sample and a maximum utilisation of gas ions is effected. l. A nximum utilisation of ions ferned in the-discharge is secured since the electric field direction is such as to urge the ions to the cathode interior. In figure 8 is a diagram of the electrodes. the solid lines indicating approximately the equipotential traces. dotted lines the electric field. l'ig. 5. Iquipotential and electric field line traces of electrodes. -22.. Obviously then ions found in the anode interior will be directed downwards and 'focussed' by the electric field so that nearly all enter the cathode. A naxinmm sputtering will then occur fron the cathode interior wall and hence placing a sample of the mterial under investigation in this region insures that a vapour cloud of sputtered atoa will be formed which may undergo excitation. This focussing action occurs only when the discharge is initiated. During actual operation space charge effects about the cathode nay drastically change these conditions. 2. With such economy in utilising the ions for sputtering. extremely snall samples of material say be used since the ions nay be concentrated within a very snell area. depending upon the cathode diameter. An adequate vapour pressure of the material nay then be obtained in order t that sufficient intensity is available for observations. fhis feature of snall sample sise recommends the hollow cathode source for use with rare elements or with the now available enriched isotopic samples. 3. Referring again to Figure 5 it is noted that the carrier gas enters the hollow cathode near the upper end. the gas then accuaulates at the cathode opening and forms a protective umbrella over the cathode. This gas umbrella serves to prevent the vapour cloud of atoms from leaving the cathode thus assuring future sample recovery. Placing the anode near the cathode forces the carrier gas to flow throng: the central cylindrical region of the anode and escaped vapour atoms are deposited here. has in case of rare or enriched samples almost complete recovery is possible. -23- 3. Gas system and power supply. In conjunction with the discharge tube proper there is associated a gas system which allows the carrier gas to flow through the discharge region at a constant reproducible pressure. In addition provision for reducing the cathode temperature is lads. l'igure 7 and overlay show the complete hollow cathode discharge source and gas flow. be essential components of the gas system are a high pressure reservoir of gas. leak type reduction valve. discharge tube and hi@ vacuum sink. Lecture bottles of A and He at 1600 psi furnish the carrier gas. this pressure being reduced to 20 psi by a Hoke reduction \alve. In the discharge tube proper the gas pressure is controlled by means of tapered leak valves the design of which is indicated in Figure 6. These valves consist of a male taper which my be made to seat snugly or loosely into a corresponding female taper. the flow of gas then being completely adjustable. An.O—Ring furnishes the thread and seal. the two valves control the flow rate of gas in and out of the discharge tube and hence control the gas pressure. Gas pressure within the discharge region is indicated on a fig nanometer situated between.the discharge tube and out- let valve. Evacuation of the system is accomplished with a Cenco Megavac rotary oil pump running continuously. n filmy“ lino. "T -5. - - I; l'igure 6. WIMT: Lea-k I81”. Q) Carrier Gas Flow Diagram. -7.5" -25- to supply the necessary power for operation of the discharge tube a 300 volt 1.2 kw filtered full wave rectifier is used. the circuit diagram being given in figure 8. Since the cathode is electrically con- nected to the metal portions of the apparatus. the anode is operated above ground potential; no shocks then result to operator. In order to facil- itate starting the discharge 2700 ohms resistance is placed in series with the discharge tube. Since the tube resistance is very high at first almost the entire applied potential is across the electrodes but as the discharge coaaences the tube resistance decreases to about 800 ohs and only about one fourth of the applied potential is across the electrodes during oper- ation. this ballast resistance has the effect of stabilizing the current flow. shall fluctuations in tube resistance do not affect the total resistance appreciably and more nearly constant current flow results. : C “11005: Ltwd I.“ ‘q'flx‘ '« I F \V\\\ H H ”+1; //0 l’ A( A\V\\\\1’\’V\\\V figure 8. Power Supply. - 27 - yll IV. General theory of High Resolution Spectroscop 1. Introductory Hid: resolution spectroscopy as the term implies is concerned with attaining high resolving powers (low resolving limits) in order that hyperfine structure in spectral lines may be accurately analysed. The principle involved is that of interference properties of coherent light waves. Interference type optical instruments are chosen which utilise high orders of interference and produce interference fringes of very narrow breadth; these instruments meeting these requirements being the l. l'abry-Perot interferometer, 2. Lumer-Gehrcke plate. 3. Reflection and 1+. Irsnsmission echelons. Of these only the first two receive common application and are the instruents utilised in the detection of hyperfine structure in copper reported here. Since with an instrument of moderate dispersion hyperfine structure is not completely resolved. the group appearing as a single broadened line, an interference type instrument as listed above is used in order to separate the components. With each of these interferometers a single wavelength of radiation in will produce a set of characteristic interference fringes. If two wavelengths X and A art); are present there will be two superimposed fringe patterns. one corresponding to each wavelength. which are of slightly different sise. In practice radiation from a suitable source - one giving sharp spectral lines so that possible -28- structure is not masked - is first passed through one of these interfer- ometers and then throng: a suitable dispersing agent, constant deviation spectrometer. monochronter. etc.. the resultant observed spectrum will consist of interference fringes norsal to the direction of dispersion for each of the source spectral lines. Investigating the interference fringes of each wavelength for fringe components gives indication of the presence of hyperfine structure and analysis of these fringe separations allows the wavelength differences between the components to be determined. A brief consideration of the theory and use of the ll'abry-Perot and Imer-Gehroh plate will be given along with methods of reduction of fringe patterns to obtain component separations. ZB . Iabry—Perot interferometer. rho J'abry-Perot interferometer is essentially a plane parallel air plate having highly reflecting surfaces; the interference pattern being formed by the recombination of partial waves from a plane wave undergoing successive nultiple reflections within these surfaces. Irom the figure the phase difference between two consecutive partial rays is (1) “'1’ = 1" [1i mu? - Er: lane m 1-] where 1» is the order number of interference. Since A, = X. M, and M3 - bm‘p/pun‘f . the order nunber becomes W /: e—t-‘V -29.. re .. 21: “,4, . Integral values of ~5> will correspond to sarimum I. of intensity in the pattern and as 1’ = The) each of the integral values of ~3> correspond to an angle of incidence LPL' : hence all rays incident at an angle c9; will produce an intensity mximum in the field. the interference pattern then being composed of concentric rings. Since 10 changes by unity in going from one fringe to the next the wave number change is Just av: .L near the interference 'zt pattern center. If at every incidence upon the reflecting surfaces a portion a- of the amplitude (assuming for simplicity A=\ ) is transmitted and f reflected then the amplitude of the k'th partial wave will be (a- ) c‘ (I ' . or in terms of the intensity coefficients 5 and a. . Urn—0 n. the transmission and reflection power of the surface. s . Hence at a given point in the focal plane of the projecting lens. the amplitude {out i (091' — 1111’) of vibration is the sum of the individual waves: 3 e , 5 me a 1 ; (w‘r- cm 1-) s r» e , - - - where of? is the light frequency. T the time and 21!? the phase difference between successive waves. Because of the ssall angles of incidence this summation is essentially an infinite one. hence the resultant amplitude A is given by C 't is»? (la-l) -itn—ghe-O Act.» 2' 3 e h [L e t in)? | = 5 8 fig}; 2“? . no observable intensity Ins-R and hence is (2) 1e): ‘1 /“ "”1 l + 2:555)" cue-1T T which for 15> = integral is Tim = 3-: and "f = integral + "'5. . -30- I - 5“ m “ ~—_ ‘ (AA-(Q1 . Since 1» is a function of the interference ring diameters. along these diameters 111») fluctuates between these values of In“... and IN.“ . is n. becomes larger. a higter reflecting power. this difference in maxim and minimum intensity becomes greater and the width of the individual fringes decreases. Using a lens of focal length 4' to produce the interference ‘L fringes of diameter D; . the amp; term becomes one; = (' “ 0%.) and hence the order number 1’ is (3) w? = 11$ ( n — Ea) . If then two wavelengths X. and A1 are present each will produce its characteristic inter- ference pattern. the diameters b, and 131" being slightly different. It is this property of an interferometer which allows it to be used as a resolving instrument for by reducing the instrumental fringe width. increasing the surface reflecting power. hyperfine structure in the spectral lines may be observed as slightly displaced fringe cosponents. The resolving power of such an instrument may be obtained by assuming that two overlapping fringes of equal intensity will be Just resolved when the ratio of minimum to laximum intensity in the combined intensity distribution is 0.8106 ( 8/«2 ); this being the Rayleigh criterion. i'hen the resolving power becomes 2.9. a i = we 3;."- 3‘1 U-n-o) ference order and surface reflection coefficient. This resolving power a function of the inter- may be increased simply by increasing the plate separation. - increasing ~10 . - and increasing the reflection power of the surfaces. -31- Deduction of wavelength separations of hyperfine structure components from observed fringe patterns depends upon the fact that the order number «p is a function of of and between consecutive fringed there is a constant difference as} . Hence calling the order of interference at the pattern center P . P = 1:. + e where 19, is the first fringe order number and f: is a fraction of an order 1 'L 0 number: 6 being given by G = 1 — '1 . Hence for a two 5| component structure. An and A, . the central order numbers are 9.. and 9,, . Therefore Pu 11+ 6.. = 211‘ = 7-" “Va , Pp ?.+‘.= “vs where v is in cm‘l. The wave number difference is then ‘K'Va. = 2‘1 (1'.-’Fe.) * :;-*(Eb-€-,) - _ ..L ‘Dblt _ In; ) (u) A” ' it (fl 1°“) + 2* R 2'6: 0;: By taking several photographs of the fringe patterns at different values of t the order number difference 19.,- f._ say be determined and hence the wave number scpmtion calculated. Reduction of fringe patterns to give component separations are more involved for a larger number of components and hence for methods of reduction and attainment of accurate values reference is made to S. Tolansky. High Resolution Spectroscopy and also to K. V. Heissner. Journal of Optical Society of America. fl. hos (191a) and g, 185 (191m). 6. hummer-Gehrcke plate. The hummer-Gehrcke plate interferometer depends for its action upon the fact that the reflecting power of a surface is very large for angles of incidence near the critical angle. In order to secure such reflecting powers without the accompanying absorption as in the case of -52- metallic films a plane wave is made incident upon a plane parallel plate of glass or quarts in such a manner that the resultant internal reflections are at nearly the critical angle. In such a case. as illustrated in the diagram,a small portion of the wave is transmitted and the s-Jority reflected at each incidence. If the successive emergent waves are hrought to a common focus by using a lens as indicated a characteristic inter- ference pattern will be produced which consists of parallel fringes. the spacing of which decreases as the distance from fringe center increases. l'or a hummer plate of thickness t . length 0. and refractive index p. the necessary condition for constructive interference of the coherent partial waves is that the path difference be (5) put ma = «X where M is an integer. them the definition of [a , In Am a = M L . the relation for costructive interference becomes (6) 2*: Jj'- are? - Mk . Denoting the angle between consecutive fringes by A)". . at is found by squaring and z and differentiating (6) and is n; e _. Mt a». uni" MAIL 01‘ since a» ~. \ (7) Lu‘. : — My. : " XEL— “Mi: 2+."sz feed» it. Hence the angular separation of the fringes increases with wavelmgth and approaching grazing emergence and also varies inversely as the plate thickness. the plate length having no influence. The dispersion of the instrument is found by differentiating (6). /u being a variable. and is given by the relation (8) 9'5}; = __ [4c/u33'5f “ MIX—J /2tz.ule~1£. (9) 33—; = ZA/‘kafi — 29—ML) . A “in ‘LL Iron this the dispersion of the Lummsr plate is evidently independent of the plate dimensions. it depends only upon the optical properties of the plate. If the AC of (7) is equated to the oi of (8) then the resulting a}. is ... x‘ (10) AX = 1 z a or is the difference M )N - 4i/A 3A).. in wavelength that a component met have in order that it coincides with the main fringe of the next order. this is the spectral range of the inter- ferometer. An exact expression for the resolving power of a hummer plate is possible only at critical angle incidence in which case the Rayleigh criterion as to the smallest angle resolvable. Si - the angle between the central Inmum and first minimum - my be supplied. Since the wave front is of width Dani . SC is given by Si 5 X Iron - O flmi (9) is obtained a 3'1 corresponding to a 3k and hence equating gives -3h- (11) /(_1__ + I_.____.\; an“ ‘ 0"“ LM*| " La. LM' a“ BM“- dose The actual wave number separation of the components is then (15) $12: e 01’ where av is the wave number difference between n'th and (n + 1) 'st order of interference. l'rom lO. converting to wave nmbers. A)’ is M :: or close enough ‘ JJ‘“ ‘ Hence measurement (16) av __ #— . 2t (AKA-[LA eff) of the fringe component positions and knowledge of the Lummer plate optical preperties allows calculation of component spectral line separations. i'olansky13 gives methods of handling such calculations when a greater number of lines are present. "-36- D. Summary Resolving powers of upwards of 105 or 106 say be obtained with the J'abry—Perot type interferometer corresponding to resolving limits of the order of 0.005 cm‘1 in the visible spectrum. The resultant interference fringes are easily and reliably reduced to give the structure separations. With quarts Lumer plates used in the ultraviolet resolving limits far below those of a Il‘abry-Perot are attainable although applica- tion to coarse structure is difficult due to problem of determining interference orders of the various fringe components. l‘or the high resolving powers necessary in hyperfine structure investigations either Iabryu-Perot or hummer plate interferometers may be employed. the decision as to which depending upon the ultimate resolving limit required and the spectral range. - 37 - V. General Theory of Hyperfine Structure A. General. Detection of spectral lines which were definitely not single but composed of a number of closely spaced lines pointed the need for introduction of a nuclear spi'h angular momentum vector. such fine structure having been proven not due to an isotope effect in all cases. Associated with such a nuclear spin is a magnetic moment - about 1/2000 ‘th that of an electron - and the resulting interaction between the magnetic moment and valence electrons would be expected to prodme a splitting of the multiplet structure of the order observed. In addition to such evidence of a nuclear spin from atomic spectra) intensity alternation in the rotation- al structure of molecular band spectra in homonuclear molecules demanded that a nuclear spin :5— be introduced. 3. having the magnitude m 531“ and I itself being integral or half integral only. In order to determine the effect of a nuclear spin - actually the associated magnetic moment - upon the energy levels of an atom it is necessary to investigate the interaction of an arbitrary electron with such a nucleus. This interaction of electron and nucleus consists of two parts: 1) an interaction of electron orbit and nuclear spin and 2) 'an electron spin- nuclear spin interaction. 1. Spin, orbit interaction. Due to the orbital motion of the electron a ragnetic field is produced at the position of the nucleus about which the nuclear magnetic moment processes. This magnetic field is ..m A va .et- ” ¢ . "E; being the electric It (1) field there produced by the orbital electron at a distance 7? , (2) E =- 9;, 7: . Hence A; (3) TIM = 2. IL W . The electronic orbital C “3 angular momentum is quantised and hence may take only those values (h) m '1: *7? = t at which with (3) gives (5) Eu '4 fi : ‘1; (L‘) e M and 2. MC. 111' A. being the mass and charge of the valence electron. Since the nucleus has a magnetic moment [III . the ratio of magnetic to mechanical munents is KL 1. h I‘D factorand M the proton mass. Hence fir is 23:9... HI c where 31. is the nuclear 3 (6) )7, = at 95 if . The potential energy of this QWMc nuclear magnetic moment in the magnetic field produced by the electrons orbital motion, Um. . is (7) V” : Uu- fix 01‘ (8) = ‘31: 'TEMJ1 $117k c.) {13) if . Recalling that the nuclear and Bohr magnetons are defined by (9) [in 5 EL‘ e ILL 3 £5 9 then 3 41TML RIF/w“, Verde = 2%: [“3 E! t'f A" (10) v“; a a i .f -39.. 2. Spin-spin interaction Recalling that with the nuclear and electronic spin angular momenta are associated a nuclear and electronic magnetic moment. the spin-spin interaction is essentially the interaction of two mgnetic dipoles ,3, and )1” where j}, is the electronic magnetic moment. m- dipole interaction is of the rural“ (11) th‘uk a. __ [3/‘5 'AEAAT-n. _ if. ' .31] I‘- A3 The magnetic moment! ,3, and ff, are given. in terms of the nuclear and Bohr magistons. by .3 (12) III: = 31%" I ,3. = -?- /'*s s and hence (11) becomes V = 1 la" ' 3i A '3. If — 350-. M a: .5 [ ILL" S] (13) VMM- _ u [si- 1 £4: _ 3;] . “a. Hence the total interaction between a nucleus of spin ‘1? and a single valence electron is (11;) vmg = a'C-i’: 4, u [afi‘i-ji .. ‘f-a‘] where a. is 0- a '3: M . This rather cmbersome IL: expression is equivalent to one of the form vat-.1 = “3' I ’T . Recalling that the quantum mechanical average of a vector, 1: for instance. in the direction of anhther vector. say '2'!" . is Just the component of 1: along .1? if T. processes about '3‘ . from the diagram the magnitude of the component o H H .2. o 3 ullr is F i rt Ht Tfl and since the unit vector along if 1. %l s then < t >aw = Li: ? I(T+\) where 1" has been replaced by In“) . Using this result then. (10) becomes (15) V“; = a t 0E .- = 1 T.“ I ....) HIM) gamiufl‘) +L(L“)-‘(’"SJT'E.IBethe15 shows that (13) reduces to the form (15) V4 b = “ §_{I(T+\)-L(L+t)-s(s+n{Nth-Q‘I-HLM) up (anKzL-O Inn) _‘(g+‘) } .. L(L+\) {Thin} +s(s+t) - L(L“))] f of and hence the total interaction. Vat-u. s 1' (17) meJl = a]— 1"3' where 0.,- = 2: [Hun-tun“) —s(s+.)] .. «L 211:“) (1:3) (41L ...) I (37) . {}i{r(y+\) — L(L.+i) - s(s+\)} { J'(J’+I) 1- L(L+\) - suit 0; — L(L+t){1‘(1+\)t$($+\) —L(L+i)§] B. Hyperfine structure in On. The most important cases of this interaction - those in which the simplest expression for V,“ is obtained - are l. Estates. L=o and :r=s andhence (18) a: = O and no hyperfine structure results. ..hl- 2. s . V" e for bath T3 Lv 'I‘L and. T:- L “ '/1_ g (19) q]. = Lil-.2) & 'J'(l'+\) An application of these theoretical results will be made to the hyper- fine structure of the '0 - 1'P triplet in copper and an attempt to verify such structure using a hollow cathode diacharge source with both rabry-Perot and Lumer-Gehrcke plate interferometers as resolving in- struments will be described. Capper in both of its isotopic forms. cc“ and ct“ . 16 have a nuclear spin 1: - 3/2. and as a result the multiplet levels ‘9,“ . 19h , ‘D,,t and ‘9,“ are split into hyperfine structure levels by the interaction of nuclear spin and valence electron. the resulting energy levels may be calculated from (17) where a.r is defined by (19). Using these relations Table 2 is obtained in which the term change is given a: a function of the interaction constant a . In ll'igure 9 is depicted the energy level splitting and the resultant hyperfine structure transitions in the triplet. i‘his diagram is drawn to give the number of components. row a, and the resulting simplifica- tion gained by assuming that the ‘P level splitting constants are zero. row b. Existence of an isotope results in a greater complexity of the observed hyperfine structure. There are hyperfine structure components produced for each of these isotopes Cu"3 and Cu“ . the separations being slightly different and the entire group of each being displaced. Whereas with a single isotope the hyperfine structure should appear as in figure 9 the presence of an additional isotope and the accompanying components mks resolution of individual line difficult. Multiplet level ‘L P 3" OIL 5/1. ‘11. Table 2. En as ‘ -hg- Term value shift as function of interaction constant I and corresponding V values d'lid'fl'd'dd'll 4r sMS <5 1 -13.“ |S ' 1 u ‘3‘” 2 '12 a tit" -‘lo... -43 _ a... .88.. .8 3:25. a use: its ... ......» on t n." .e .23..» :3? : .: Zn : O - uh .. VI. Experimental A. Optical system. Radiation emanating from the hollow cathode is collimated and the direction of propagation changed from vertical to horizontal by means of a first surface Aluminum mirror. Such collimated light is then incident on a Iabry-Perot interferometer. aperture of 1 cm. and ‘plate separation of 10.1599 mm. and the emergent radiation is focussed on the entrance slit of a Hilger Constant Deviation Spectrometer. By tilting the etalop, rotating it about a horizontal axis parallel to the plane of the slit. the maximum of intensity is placed at the position of the second or third fringe and observations were possible. visual and photographic. A “.56 mm glass Lummer-Gehrcke plate was also used as a resolving instrumont, the optical train being identical although tilting of this instrument may introduce ghost inges which will appear to be structure. With the plate in a horizontal plane and using the same focussing lens only the first two orders of interference on either side of the center were available for observation. In the figure is a photograph of the optical system with the Lummer plate in position. The resulting interference fringes were photographed with a plate camera, the camera lens being placed adjacent to the spectrometer eyepiece: focusing was facilitated by viewing the fringes on the camera ground glass plate with an eyepiece. Interference fringe else on the photographic plates was controlled by the camera magnification and was set so as to allow reasonable exposure times and yet have the fringes far enough separated so that measurement of fringe separations was possible. Kodak Spectroscopic Plates III-1’3. 8 1/14 x 1: llh were used -45- - 1+6 .. and were developed in 13-19 for 3' 1/2 minutes and fixed in 0.76. B. Results of investigation. The results of this problem will be discussed in two parts. First the preliminary investigation of the fringe breadth, hence directly the spectral line breadth, as influenced by the hollow cathode source and secondly the hyperfine structure of the fiD-IP triplet in Cu. 1. Line breadth The interference patterns obtained for the 1D-'P triplet were repeated at three different temperatures of the refrigerant: cold water, solid 002 and acetone mixture and liquid nitrogen. 15°C. - 78°C and -l95°0 respectively. Figures 10 and 11 show these fringes and a comparison of identical fringes at different temperatures is possible. Raking this comparison between fringe breadths temperature reduction is seen to increase fringe sharpness. decrease the fringe width. Since the instrumental contribution to fringe width is constant over each group of fringe patterns and the gas pressure and tube current are constant. the conclusion is that lower temperatures give a reduction in the spectral line width. this being characteristic of the doppler effect broadening of lines. .At liquid nitrogen temperature the expected natural doppler and collision breadths are 1. natural ~ .0002 car’- 2. doppler "a 0.016 cm"1 8. collision ~ 0.002 cm”1 It is to be noted among the interference fringe patterns that the tube current for the Lummer plate analysis at liquid nitrogen - h? - Figure 10 Fabry-Perot Interference Fringes 0 r = 16.0 c .33 hr 2.: hr 1.2“ hr MOOme Lf-OOma ”00m 14 . 0mm- 2 .“mm—Hg ‘“ 2 .‘mm-Hg Hg RVUUA =782A o T = -78 C 16 hr 2.0 hr 1.0 hr 1+00ma l+00ma ‘300ma 2 . “mm- 3 .Omm-Hg 3 .Omm-Hg Hg " " ‘. avous ‘7”8A o T =2 -19'; C .0 hr 1 .0 hr 800m 800m 3 . tfirmly-Hg 3 .Omm-Hg Glofia svcon sagas - us - Figure 11 Lmer—Gehrcke Interference Fringes O Ta1doc .16 hr .33 hr 300m 300ma 2 . grum-Hg 2 .‘mm-Hg “103A canon “782A 0 T = --78 C .08 hr 1.0 hr .33 hr 300ma MOOma u00ms 2."!mn-H 2.3mm—Hg 2.3mm- Hg «10:; =7coa ”7825 *3 ll -19eoc 2.3 hr 7. .0 hr 130m 1"=Oma 2 .‘mm—Hg P ."o'mm-Hg <732A .16 hr 130 ma 2 .3mm-Hf'. 7107A I5700A ..hg- temperature has been reduced to 130 ms. this reduction in current giving a lower intensity but serving to increase the fringe sharpness. With the Fabryh-Perot etalon such currents gave intensities lower than were practicable to use and hence higher currents had to be used. This effect was not photographed but was noted visually with the spectrometer. Since all other conditions were identical, the conclusion reached is that lower tube currents give rise to sharper spectral lines; this effect is Just that to be expected on the basis of stark effect broadening of spectral lines. Confirmation of pressure and collision broadening effects were not obtained due to the limited range of pressures within which the discharge tube Operated correctly. 1 - 5 mm Hg. Within this range of pressures such broadening effects would not be apparent visually and probably not photographically. 2. Hyperfine structure of Cu. from the interference patterns obtained when the hollow cathode source was operated at ~195°0 the number of hyperfine structure components and their separations were calculated. The l'abry-Perot fringes allowed such a determination to be made for the 5700 and 5782 3 lines. the 510d being unresolved. With the Lesser plate patterns the hyperfine structure was deduced for all of the triplet lines. Ellie negatives were analysed by measuring the fringe positions with a traveling microscope (least count of 0.005 In), the settings of pointer on fringe center being made with the unaided eye. Two separate measurements were made and the fringe intervals agreed to within 0.02 mm for the aadority of cases. Using these fringe separations the component -QO- intervals were calculated using Behren's method of reduction for the Lumer-Gehrcke plate patterns and a similar method17 for the off-center l‘abry—Perot fringes. The results of these calculations are summarised in the following tables. a) Using l'abry-Perot etalon. Transition X l ‘Dsn - "’m. $105.55 to)" - 1P“ 5700 e25 ’0.“ - 'Ph 5782 .13 b) Using Lumerh-Gehrcke plate Transition A A 10"... ” ‘P3’l 510% 0:5 ‘th _ 1P,” 57m e25 ‘0"t - 2P7:— 5782.13 Components Not resolved 3 8 Components Component separations on":1 .181 .091 .171 .099 Conponent separations cm‘1 .091 .083 .IMS .119 .072 .097 Examination of Figure 12 in which the hyperfine structure groups are compared with those predicted and observed by Bitschl. leitschrift fur Physik 12. l (1932). and Schuler and Schmidt, (Zeitschrift fur Physik 2.92. 113 (1936), reveals that the resolving limit of the l‘abry—Perot etalon was not low enough. no components being resolved in the 310% and only three each in the 57001 and 5782A groups. This lack of resolving power is due to the fact that the Al film was thinner than required; this was used in the interests of sufficient fringe intensity. The resolving ee 0 h a. ~-e!e\.0\ tau 9. 3. he ’ g I 4 # § N\ bkb fink ... Q NQNM. w. 5N. 8km. .skvhkdk. ,3 k. 2 1 i , f. 1 :2, . ...... new .. \\b\\ swek‘sw .. EN 0 so \bx - . the. M‘hfifi t §§\§\\ xe¢ usthQ . «(fixexxsfiflo xawé \ks auie~,°\ l a , l , 3. ..l x, 3.. l .2 3 , nests“. ‘3 Rm Rib» \txktnfio S 3.“. x... 33.1%. {oxen - a k a... \a\.\aw\§u\hb .. \Kba‘. hSQxxtq __.‘——_._-———_ Fa—_~__‘_F_——4 - e2 - limit of a lbbrybPerot etalon depends directly'upon the surface fths reflection power; with a 1 cm etalon a resolving limit of 0.062 cm"1 is possible with 70 per cent reflecting surfaces and increasing this to 95 per cent reduces the resolving limit to 0.008 cm'l, theoretically. Hence with the etalon used. 1.01899 cm separation. the hyperfine structure of these three Cu.lines should be capable of resolution with a reasonably high reflecting film. However. such films reduce the fringe intensity to the point where, with this experimental arrangement at least. the necessary exposure time is impractical. The conclusion is then that unless larger etalon separations are used with these lower reflecting power films, this hyperfine structure is not resolvable. With.the glass Lummer plate the resolution.was sufficient to give three hyperfine components in the 5105A and 5700A groups and five in the 5782A one. The component intervals for the 5105s group are very nearly'those observed by Schuler and Schmidt. The separations of the 5782A set are also closely those reported. Since here the third through seventh calculated components have different intensities and the observed lines represent the resultant intensity distribution maxima. the agree- ment with theory is fair. In the 8700A group, as calculated, the inten- sities range from 3 to 100 and the resulting distribution would be expected to have marina between the first and second clusters of lines, about the third.cluster and again for the fourth. The last group have very low intensities and.probab1y are not detectable. On this basis the marina intervals would correspond roughly to those observed. Why these lines of 0.060 to 0.080 cm."1 intervals are not separated here as they are in the 5782A structure is a.pussle. Lower resolving limits with this -53- Lummer-Gehrcke plate are possible by using angles of incidence nearer the critical angle and by using only the perpendicular component of the light since the‘reflecting power of a surface is higher for this than for the parallel component at a given angle of incidence. This possibil- ity has not been explored as yet. Although hyperfine structure intervals are quoted to 0.001 on’1 the last figure is uncertain due to the closeness of the fringe components and to the method of'measuring fringe separations. Therefore these structure intervals must be viewed with a degree of caution. - 5b.- VII . Conclusion. A hollow cathode discharge source has been designed and constructed with which the spectral line breadths and the hyperfine structure in copper have been investigated. Temperature dependence of this line breadth and also a dependence upon discharge tube current have been noted: these effects being those expected on the basis of Dappler and Star]: effect broadening. Using both l‘abry-Perot and hammer- Gehrcke plate interferometers as resolving instruments the hyperfine structure of the 2D ”9 multiplet in copper have been observed, the resultant structure intervals as obtained from the Lunar plate fringes being in general agreement with previous observations. Ilo determination of the appropriate interaction constants is possible due to incomplete resolution. Due to the possibility of producing spectral lines of narrow breadth this type source could find application in l) Investigationbf spectral line breadths themselves and quantitative measurements of line breadth dependence upon gas pressure. discharge tube current and temperature: 2) analysis of hyperfine and isot0pe structure in spectral lines using the techniques employed here; and 3) production of molecular band spectra. particularly ionized diatomic molecules. for purposes of rotational‘and vibrational analys is. l. 2. z. u. S. 6. 7. s. 9. lo. 11. 12. 1h. 15. 16. 17. REFERENCES Margenau, H. and Watson, J. Rov. Mod. Phys. E. 22 (1936) White, H. Introduction to Atomic Spectra (McGraw—Hill, 1933) p. N19 Hargenau, H. Phys. Rev. 333. 156 (1981) Holstein. T. Phys. Rev. 12. 7143+ (1950) Hargenau. H. Phys. Rev. 5;. 156 (19:51) Born. a. Optik (Springer. 1933) Paschen. 1. Ann der Physik £52, 901 (1916) Schuler, a. Zeits. fur Physik fig. 323 (1926) Arroe. u. and Mack, J. J. 0. s. 1. fig. 387 (1950) McNally. J., Harrison. 0. and Rowe, l. J. 0. s. s. g. 93 (19117) Tolansky. S. High Resolution Spectroscopy (Pitman. 19117) Behrens, D. J. Journ. Sci. Instru. pg. 238 (19141) Tclansky. 3. High Resolution Spectroscopy (Pitssu. 19h?) p. 211 Weatherburn. C. B. Advanced Vector Analysis (Bell. 19ml») p. 161 Bethe, H. Handbuch der Physik vol. 311/1 p. 557 neck, J. r. Rev. Mod. Phys. _2_g. 611 (19%) Tolansky. S. High Resolution SpectroscOpy (Pitman, 191$?) p. 133 \s es '9 s ‘e "1111111131111 1111111111“