Energy Levels, Classified Lines, and Zeeman Effect of Neutral Thorium

A list of about 9500 classified lines of Th I in the range 2345–29 662 Å is given. Lines in the range 2345–9239 A were observed and measured at NBS. Zeeman effect data for 2281 lines are listed. Lists of 254 even and 322 odd levels including their g values are presented. Among them there are 72 new levels, which were not contained in earlier publications.


Introduction
A re view of early researc h on the spec tra of ne utral and sin gly ionized thorium has been give n by Zalubas [1960] .1 I n the meantim e light sources a nd s pectro· graphs have been improved a nd more s tanda rd wave· le ngths of th orium have beco me available. Th eir developme nt is described by Giacche tti, Stanley, a nd Zalub as [1970] . Therefore, at NBS I undertook a new observation and description of Th I and Th II in orde r to expand the analyses of th ese spec tra. A co mplete line list of about 35 000 Th I a nd Th II lines will be publishe d separately. That li st will include all a vail· able thorium s tandards a nd will be useful for users of thorium stand ards. Zalub as and Corliss have published a list of classified lines in Th II [1974] .
In the present pa per I will revie w only the work on findin g the energy le vels and g factors of Th I.

Wavelengths and Classified Lines
Ne w observations were taken on the NBS 10.7 me ter Eagle spectrograph in the range 2000-12 000 A.
Electrodeless lamps and sliding sparks were used as light sources for the separation of Th I and Th II. Professor S. P . Davis suppli ed some s pec trograms for range 2600-4500 A produ ced on the Czerny·Turner spectrograph at the University of Californi a. I m eas· ured wavelengths in the 2800-12 000 A observation range. Stronger lines were meas ured on 3 to 6 plates, weake r lines (below inte nsity 5) on 2 to 3 plates. Small numb ers of lines of inte nsity 1 are include d which were meas ured only once. The wa vele ngth s in th e I Dates enclosed in brackets indicate literat ure references at the e nd of th is paper. range 2000-2800 A we re meas ured only on ce. Internal thorium standard s from Giac he tti [1966] , and Giac· c he tti, Stanley, and Zalubas [1970] were used.
All classified lin es fit the e nergy le vel sche me to within ± 0.05 cm -1 . More than 60 pe rcent of th e m do not exceed ± 0.02 e m -I.
The consis te ncy of wa vele ngths in thi s lis t can be de mon s trated by the following tes t. The a verage abo solute value of the diffe rence be twee n observed and calcula ted wa velengths expressed in c m -I was 0.008 fo r 50 lines around each of the wavele ngths 3000, 4000, 5000,6000, and 8000 A.
At the La boratoire Aim e Cotton , Orsay, France wavele ngths of Th I and Th II lin es in th e range 9239-29 662 A we re recorded by mean s of Fourier tran s· form s pectroscopy. Th e list of 3100 lines, includin g classifi ed lin es of Th I and Th II, was publis he d by Giacchetti , Blaise, Corliss, and Zalubas [1974].
The average de viation between the observed and calculated wavenumbers from this list is less than 0.002 cm -1 .
These two lists combined together have around 35 000 lines and excel earlier observations in accurac y and extent. In order to have all classified lines of Th I at hand, the 1900 Th I lines measured by Fourier tran s· form spectroscop'y are taken from Giacchetti et al. [1974] and incorporated in table 4. Table 4 contains about 9500 classified lin es in the wavelength range 2345-29 662 A. F or the wavele ngth range 2345-9239 A vis ually estimated relative intensi· ti es on an arbitrary scale from 1 to 5000 are giv e n for electrodeless lamp and sliding spark . For th e wa ve· length ra nge 9239-29 662 A only the electrod eless lamp inte nsity on a scale 0 to 9 is give n. The lette r symbols are use d to describe the c haracter of lines; b-blend (two wavelengths measured), d-double (one wavelength measured), hhazy, lshaded to longer wavele ngth, rreversed, sshaded to shorter wave· length, a nd wwide. In column 4 the wavenumber is give n in vacuum in units of cm -I. In the fifth and sixth columns the classification of a line is given by rounded·off values of the energy levels responsible for the transition, with their l·values as subscripts and a superscript degree symbol to indicate odd parity.

Zeeman EHect and g-Factors
B. E. Moore [1909] observed the splittings of thorium lines in the magnetic field in the range 3002-4721 A.
He observed mostly Th II lines and at that time of course did not derive land g values.
Lier [1939] observed the Zeeman effect of 40 lines of Th I in the range 3427-6182 A. Fifteen of these lines had resolved ZE patterns. These data were used by Schuurmans [1946], who derived g factors for all levels he found. Charles [1958] furnished ZE data for 46 lines in the range 3041-6457 A.
In the final list I used g factors derived from my observations. I have measured ca. 2281 Zeeman patterns in a magnetic field of 2.4 teslas, produced by the electromagnet at Argonne National Laboratory.
From completely resolved patterns I derived 1 values and g values, and from partly resolved patterns I obtained I1g, or p, or n, or a combination of these. I1g is the difference between two g values for that line. The distance from the no·field position to the most distant p component (parallel polarization) is denoted by p, and n is the distance from no·field position to the strongest n component (normal polarization). All-these measure· ments are given in Lorentz units. The type of Zeeman pattern is described by the following numbers: 1 indicates that the n pattern shades out, 2 indicates that the n pattern shades in, 3 indicates that the n pattern is symmetrical, II =l2, and the p pattern shades in, and 7 indicates that the pattern is either triplet or unresolved. The values of g derived from my measure· ments or measured I1g, p, and n are given in table 3. Values of g factors for the energy levels are derived from various numbers of patterns, from 60 for one level to only one for some levels. Therefore, the g·values are given with three, two, or one decimal digits. The estimated error is not greater than 2 in the last digit.

Energy Levels
In previous papers on Th I [Zalubas , 1959[Zalubas , , 1968 I gave an account of the earlier analyses of Th I. Here I will give just the total number of levels contributed by those early investigators that have proved to be real.
Schuurmans [1946] found 29 energy levels and properly identified the five lowest levels. Stuken· broeker and McNally [1953] contributed 7 levels. Charles [1958] found 2 levels. Steers [1967], using his far infrared observations, found 13 levels. Giac· chetti and Blaise [1970] found 10 levels. The remaining levels were found by the present author. Now the total number of known even levels is 254, and 322 odd levels .
The re are 72 levels given here for the first time . Th I has two very distinct systems of levels, whic h I have connected. One system starts with the ground state 6d 2 7s 23 F2 • The other starts 7795 c m -I above ground state with the 5J6d7s 2 3H~ level.
For energy level searches I treated the two systems separately. In such searches, I used a set of programs called COMBO, written by J. Tech for the analysis of spectra. The numbers of combinations and the avail· able ZE data were the critical factors in establishing new energy levels. I have not included levels which can be found by intercombinations between high odd levels, and high even levels , despite the fact that some chains are long. Some of them may be confirmed later by theoretical calculations and extension of the analysis. Even levels are given in table 1 and odd levels in table 2.
The uncertainty of the level values for levels below 15000 is 0.001 cm -I, and the un certainties of most of the high levels do not exceed 0.005 cm -I . Levels given with two decimal places have uncertainties of 0.01 to 0.02 cm -I .
I have identified the lowest lev els by using g values, and by comparison of the levels with calc ulations for the 6d 2 7s 2 and 6d 3 7s configurations by Trees [1960).
For the low odd levels I made preliminary ide ntifica· tions using Martin's [1963] Ce: I identifications, and also Klinkenberg's [1950] Th III identifications. Later I used Sugar's [1968] preliminary calculation s for the 5J6d7s 2 and 5J6d 2 7s configurations. Giacchetti and Blaise [1970] identified the 5F, 5G, 5H, 51 terms of the 5J5d7 s7p configuration. In addition , Brewer [1971] identified 14465 as 5G~ of 6d27s7p, and 18431 as 3F3 of 5J7s 2 7p. No attempt has been made to identify the tenus or configurations of the high energy levels.