1. Introduction

The first description of the Hf1 and Hf 11 spectra was reported by W. F. Meggers in 1928[1]. This paper contains a history of the element hafnium, wavelength measurements of some 1500 lines in the arc and spark spectra of hafnium between 2155 A and 9250 A, and a list of 12 persistent lines in each spectrum.

He and B. F. Scribner discovered the first regularities in Hf I in 1930[2]. This was achieved with the aid of the temperature classification of the lines obtained by A. S. King[3] from a study of the electric furnace spectrum of hafnium, and by analogy with the known regularities in the similar spectra Ti 1[4] and Zr 1[5]. At this time the complexities in Hf1 were evident; the search for significant wave-number intervals of low terms had to be made with differences greater than 2000 cm-1. Thus the two leading intervals in the ground term a 3F, 2356.6 cm1 and 2211.0 cm-1, were detected. It was recognized that extension and confirmation of these first regularities would require further improvements in the basic descriptions of hafnium spectra and, also, observation of the Zeeman effect. Both requirements were partially met in 1932 when hafnium spectra were reobserved with purified oxide donated by G. von Hevesey, and the first Zeeman-effect spectrograms were made with a small rod of hafnium metal condensed on a tungsten filament, presented for this purpose by G. Holst of Eindhoven. The arc and spark spectra were remeasured, and extended from 1990 Å to 10637 Å. This list included nearly 1400 lines of Hf 1. Magnetic splitting was also observed for 70 Hf lines. The earlier regularities were fully confirmed. At this point, Meggers and Scribner

1 Figures in brackets indicate the literature references at the end of this paper.

turned their attention to Hf 11 and published their analysis of this spectrum in 1934[6] with the statement that "Further data and analysis of the HfI spectrum will be published after they have been supplemented by additional observations of Zeeman effects, since the data at present available have proven insufficient for the complete identification of HfI spectral terms."

Attempts to improve the description of the Hf 1 spectrum presented serious difficulties because of the presence of many metallic impurity lines as well as strong background HfO spectra. Finally, in 1958, C. H. Corliss and Meggers [7] succeeded in overcoming these complications by using "electrodeless metal-halide lamps excited at relatively low pressure and temperature, by microwaves. This light source was fully described in 1953 by C. H. Corliss, W. R. Bozman, and F. O. Westfall [8], "who showed that a bright emission spectrum of any metal could be obtained from outgassed quartz tubes inclosing a minute amount of any volatile compound of that metal when excited by

microwaves."

"Lamps of hafnium iodide, hafnium bromide, and hafnium chloride were compared" to detect atomic hafnium lines common to the different lamps. This "Improved Description of Hafnium Specta" extends from 1284 Å to 12043 Å and includes some 6200 lines in all as compared with the total of 2400 lines observed earlier.

When the present writer (CEM) requested data on Hf for inclusion in Volume III of "Atomic Energy Levels," [9] about 1957, Meggers generously provided all of his results to date, but he did not yet have the complete 1958 line list available. In August 1949, greatly improved Zeeman spectrograms were obtained, by courtesy of G. R. Harrison, with the Bitter Magnet at the Massachusetts Institute of Technology. These spectrograms were supplemented from 5000 Å to 9000 Å by later

of Standards, with a hafnium-iodide lamp as source. The beautiful standard patterns in Hfi and Hf II, reproduced in the 1958 paper on page 270, are from these Bureau spectrograms.

With the help from the Zeeman data, Meggers had classified approximately 2000 lines from his unpublished line list having the short-wave limit 1917 Å. His unpublished analysis at that time included 7 low even terms, 90 miscellaneous even levels, 16 odd terms, 58 miscellaneous odd levels and 132 g-values. These are recorded in reference [9]. All subsequent work on the analysis is based almost entirely on the description published in 1958. He continued his study of Hf1 with the aid of electronic computers until his death in 1966. At this time the material was turned over to C. E. Moore, with the request that she complete the unfinished work on which he had toiled for so many years. This assignment was accepted very reluctantly, with the feeling that it was an obligation she owed to her colleague, but, also, with the full realization that it would be a difficult task.

A square array had been prepared by electronic computer by W. R. Bozman, and a printout of the line list made by J. Sugar contained the classified lines as entered by Meggers, assisted by Isabel D. Murray. The line list totaled some 4700 lines of which about 2800 or 60 percent were classified.

From the regularities already known, it was obvious that Meggers intended to carry the analysis much further on the basis of LS-coupling, by analogy with Tii and Zri and with the help from Zeeman data. Consequently, the work has been carried forward with the terms arranged by configurations exactly according to his plan. A serious problem was to determine how completely the Zeeman observations had been utilized in obtaining final g-values. Many resolved patterns had been used to derive the gvalues recorded in the 1966 list of levels. The writer was, however, unable to locate a complete summary of g-values similar to the meticulously prepared list Meggers left for Yb 11, for example. With the idea that more g-values could be obtained from the original observations, she examined the observing books of Zeeman data, recorded all observed patterns in a duplicate ledger of the line list, and derived as complete a set of g-values as possible. There are now 198 levels for which g-values are known as compared with about 137 listed in 1966. The details of the reduction are described below.

2. The Analysis

Throughout the work on extending the analysis, a great effort has been made to differentiate between the assignments by Meggers and those added later by the writer. When a complete square array by terms had been prepared, and the combining properties of the levels with regard to intensity, limit terms and g-values were studied, it was evident that some of the earlier tentative designations needed revision.

adopted are given in tables 1 and 2, for even and odd terms, respectively. The tables are arranged similarly, with the terms grouped by configuration. The successive columns give the configuration, term designation, J-value, level, interval, observed gvalue, Landé g-value, number of combinations, and notes described at the end of table 1. Every user is requested to give serious attention to these notes. They have been devised to show as explicitly as possible how far the analysis had been carried by Meggers in 1966 in contrast to additions and revisions made by the writer in order to present a less fragmentary interpretation. The same notes apply to tables 2, 4, and 5.

All levels having note "a" were found by Meggers so far as can be ascertained, and those noted as "b" by the writer. Note "c" is particularly important; it signifies that the configuration and term assignments were by Meggers. Notes "e, f, g" indicate changes made in Meggers' lists. To summarize, notes "a, c" represent Meggers' work, "b, d" indicate interpretation by the writer, and other notes are intermediate in type.

-1

In general, many intervals are irregular, which is not surprising for this complex spectrum in which both sets of levels, odd and even, overlap in position. The LS-coupling does not hold for the more complex configurations. The present terms have been assembled on the basis of their combining properties, intensities of the combinations, and gvalues. Many designations are subject to question and may well need revision. For example, the level 25678.61 cm1 is very low for the 'F term in the configuration 5d3 (b 2F)6s, but the combinations are good. Similarly, the level 42061.60 cm-1 ascribed to 3D in the configuration 5d 6s2 (a 2D)7s has a g-value of 0.832? based on one observation. This g-value, if correct, fits 'G, and the line in question may be masked. Further search for this 3D ̧ level has been fruitless. There is, also, no check on the level 28586.39 cm designated as y Pi, which is based on only two combinations. The levels y G and y G, also need further confirmation. Colons following the level values in tables 1 and 2 indicate dubious entries. The number of combinations listed in column 8 also provide a check on the reliability of the level.

-1

A number of gaps still exist in completing the various groups of terms. This is regrettable, but at present further searching by hand is not rewarding.

An array of observed and predicted terms of Hf 1 is presented in table 3. This table is condensed in form, but it is designed for use by comparison with tables 1 and 2. Predicted terms are entered only for those configurations having some observed terms. Many predicted terms could be added, as shown, for example, in the array given on page XXII of reference [9]. The line list is not exhausted (see table 8), but most of the leading lines have been classified (see sec. 6).

Because of the complications that arise when an attempt is made to carry the LS-coupling scheme

writer has repeated the term lists (tables 1 and 2), by level, in the general format that Meggers was using in the course of his work on the analysis. All levels are tabulated in numerical order in tables 4 and 5. Table 4 contains the even levels, 164 in all. Table 5 gives the odd levels, which total 159.

In each table the respective columns list the Jvalue, the level value (cm-1), the notes described at the end of table 1, and the number of combinations. A single note, "a" or "b" in column 4 indicates the miscellaneous levels, i.e., those not yet assigned a term designation. The number of levels totals 323 of which all but 26 were found by Meggers.

3. The Ionization Limit

4

No good series are known as yet in Hf 1. By using the term 5d26s2a 3F, and the center of gravity of the terms ascribed to 5d2 (a 3F4)6s 7s as series terms, J. Sugar as calculated an ionization limit of 54700+600 cm-1 by methods described in reference[10]. With the conversion factor 0.000123981, the ionization potential is 6.78±0.07 eV.

An experimental value of the ionization potential of Hf 1, 6.65±0.1 eV, obtained by an electron impact method, has been reported by E. G. Rauh and R. J. Ackerman[11]. This value agrees well with the spectroscopic value derived above.

4. Observed g-Values

A test of the correctness of the interpretation and of the accuracy of the Zeeman data is to be found if the g-sum rule of Pauli is applied. As Meggers stated in the paper on Hf 11[6], this rule "is expected to be valid for all spectra no matter what the nature of the vector coupling may be." It should be applied to complete groups.

A comparison of observed and Landé g-values for LS-coupling is given for the leading even terms in table 6, and similarly, for the odd terms in table 7. The Landé values are taken from the detailed "Tables of Theoretical Zeeman Effects"[12] published by C. C. Kiess and W. F. Meggers in 1928.

The terms are arranged by configuration and the g-sums are given for each J-value in the separate groups. The agreement is excellent for the low groups in each table. The remaining configurations show larger deviations, as is to be expected, since the groups are not so complete, some designations are open to question, and there is a departure from LScoupling. These factors are inherent in such a complex spectrum.

In general, these results are a real tribute to Meggers, whose splendid observations and measurements are revealed. In these tables only a few dubious g-values are entered and these are marked with a colon.

The history of the Zeeman observations has been described above briefly, and need not be repeated here. In averaging the final g-values from the individual Zeeman observations triple weight was given to the best fully-resolved patterns, double weight to many other resolved patterns and less weight to those that were not resolved. With the final averages thus obtained for as many levels as possible, the remaining blended patterns were carefully examined. For classified lines having an observed g-value for one level, this value was used to derive the g-value for the other level, on the assumption that the strongest component had been observed in the unresolved pattern.

The final observed Zeeman data are given in appendix A where the separate columns contain: (1) Wavelength, (2) Intensity (Tube), (3) Intensity (Spark), (4) Classification, (5) Type of Pattern, (6) Ag, (7) 1st g, (8) 2nd g, (9) the strong p-component, and (10) the strong n-component. The Zeeman type is quoted mostly from reference [7], but some have been taken directly from the observing books. The observed g-values used as described above for blended or unresolved patterns are entered in parentheses. The "1st g" refers to the first entry in column 4, i.e., the lower level of the transition. The "2nd g" refers to the higher level involved. The respective J-values are given as subscripts in column 4. The Zeeman effect has been observed for 531 lines.

6. The Classified Lines

Appendix B contains the final list of observed and classified lines. Meggers' line list of 1966 contained more than 4700 lines in which 3217 classifications are entered. There are 88 lines which are blends having 2 classifications. The number of classified lines is, therefore, 3129 or roughly 67 percent of the total.

The successive columns contain: (1) Wavelength, (2) Intensity (Arc or Tube), (3) Intensity (Spark), (4) Wave number (cm-1), (5) Wave number o-c (cm-1), (6) Classification. The levels involved in the transition are entered in whole numbers, with the respective J-values as subscripts. The "odd" level of the transition is in italics.

The observed data are mostly from reference [7]. In column 2 "Z" has been added for the lines having an observed Zeeman effect. For these lines, the Zeeman data are given in appendix A.

The leading unclassified lines are listed in table 8. Lines of intensity 30 or greater are included, and the total number is 31. They are taken from appendix B. The wavelengths and intensities are entered in this table with "Z" added in column 2 for those lines having an observed Zeeman effect (app. A).

Throughout the years since W. F. Meggers envisaged a Monograph on Hf 1, a number of colleagues have generously supported the effort. W. R. Bozman and J. Sugar assisted with the computer preparation of the square array and line list which have been used in bringing the work to its present conclusion. The late Ruth Peterson furnished the latest revised computer list of the energy levels. Much of the careful editing of the final list of classified lines was ably handled by the late Isabel D. Murray. The generous contributions of all who have participated in the work and helped to make this publication a tribute to W. F. Meggers are gratefully acknowledged. The writer is especially indebted to her husband, B. W. Sitterly, for his cordial cooperation throughout the years.

Special thanks are due to K. G. Kessler and D. R. Lide for their support and encouragement during the preparation of this Monograph, and to J. L. Tech for his valuable suggestions.

Washington, D.C. December 9, 1975

[1] Meggers, W. F., Bur. Std. J. Res. 1, 151-187, RP8 (1928).

[2] Meggers, W. F., Bur. Std. J. Res. 4, 169-175 RP139 (1930).

[3] King, A. S., Astroph. J. 70, 105–113 (1929).

[4] Russell, H. N., Astroph. J. 66, 347–438 (1927).

[5] Kiess, C. C., and Kiess, H. K., J. Res. Natl. Bur. Std. 6, 621-672, RP296 (1931).

[6] Meggers, W. F., and Scribner, B. F., J. Res. Natl. Bur. Std. 13, 625-657, RP732 (1934).

[7] Corliss, C. H., and Meggers, W. F., J. Res. Natl. Bur. Std. 61, No. 4, 269-324, RP2904 (1958).

[8] Corliss, C. H., Bozman, W. R., and Westfall, F. O., J. Opt. Soc. Am. 43, No. 5, 398–400 (1953).

[9] Moore, C. E., "Atomic Energy Levels" Circ. Natl. Bur. Std. 467, III, 143–146 (1958).

[10] Sugar, J., J. Chem. Phys. 59, 788-791 (1973).

[11] Rauh, E. G., and Ackermann, R. J., J. Chem. Phys. 60, 1396-1400 (1974).

[12] Kiess, C. C., and Meggers, W. F., Bur. Std. J. Res. 1, 641-684, RP23 (1928).