UNITED STATES DEPARTMENT OF COMMERCE • Luther H. Hodges, Secretary NATIONAL BUREAU OF STANDARDS • A. V. Astin, Director Tables of Spectral-Line Intensities Part II Arranged by Wavelengths The intensity, character, wavelength, and spectrum of 39,000 lines between 2000 A and 9000 A observed in copper arcs containing 0.1 atomic percent of each of 70 elements. William F. Meggers, Charles H. Corliss, and Bourdon F. Scribner U' ^ National Bureau of Standards Monograph 32 — Part II Issued October 2, 1961 For sale by the Superintendent of Documents, U.S. Government i'rinting Otfice, Washington 25, D.C. - Price $3 "Vational Bureau of Standards IU0V8 1961 QC/OQ ^^^^^^ Co f^y Z Contents Page 1. Introduction III 2. Experiments V 2.1. Dilution in silver VI 2.2. Dilution in copper VI 2.3. Arc electrodes VIII 2.4. Spectrograph VIII 2.5. Photographic plates viii 2.6. Energy calibration of copper lines VIII 3. Results IX 4. References XII Table 1. The strong lines in order of wavelength i 1 Table 2. All observed lines in order of wavelength , 11 II Tables of Spectral-Line Intensities William F. Meggers, Charles H. Corliss, and Bourdon F. Scribner The relative intensities, or radiant powers, of 39,000 spectral lines with wavelengths be- tween 2000 and 9000 Angstroms have been determined on a uniform energy scale for seventy chemical elements. This was done by mixing 0.1 atomic percent of each element in powdered copper, pressing the powder-mixture to form solid electrodes which were burned in a 10 ampere, 220 volt direct-current arc, and photographing the spectra with a stigmatic concave grating while a step sector was rotating in front of the slit. The sectored spectrograms facilitated the estimation of intensities of all element lines relative to copper lines which were then calibrated on an energy scale provided by standardized lamps, and all estimated line intensities were finally adjusted to fit this calibration. Comparisons with other intensity measurements in indi- vidual spectra indicate that the National Bureau of Standards spectral-line intensities may have average errors of 20 percent, but first of all they provide uniform quantitative values for the seventy chemical elements commonly determined by spectrochemists. These data are presented by element in part I, and all 39,000 observed lines are given in order of wavelength in part II. 1. Introduction Spectrochemistry was born a century ago when of spectra not only in the visible but also in the in- Kirchhoff and Bunsen [1] ^ definitely demonstrated visible ultraviolet and infrared regions. But even if that chemical elements were uniquely identified by the light source is reproducible and standardized it is spectral radiations, or lines as seen in a spectroscope not easy to evaluate the spectral efficiencies of spectro- provided with a slit. This led immediately to the iden- graphs and photographic emulsions so the usual pro- tification of many chemical elements in the sun and cedure has been to make subjective visual estimates of to the discovery of several new elements, but no quan- relative intensities of spectral lines on an arbitrary titative chemical analyses were made until much later. scale based on the relative blackness and/or width of In 1874, Lockyer [2] stated that "while the qualita- spectral-line images appearing on a developed photo- tive spectrum analysis depends upon the positions of graphic plate. Consequently, in thousands of individual the lines, the quantitative analysis depends not upon papers and in numerous comprehensive compilations of their position but upon their length, brightness, thick- spectral data we find only qualitative data on inten- ness, and number as compared with the number visible sities which may have some meaning for adjacent lines in the spectrum of a pure vapor". Thus, position (or in a given spectrum but none at all when comparing wavelength) and brightness (or intensity) are recog- widely spaced lines, or lines of different spectra of the nized as being the two most important properties of same element or of different chemical elements. spectral lines; wavelengths identify chemical elements In the beginning, most intensity data were reported and intensities indicate the concentrations of identified on an arbitrary scale of 10 steps, weak lines being elements in mixtures or chemical compounds. assigned an intensity of 1, and the strongest line in- During the past century there has been spectacular tensity 10. Even as late as 1945 extensive new spectral improvement in the accuracy of spectral wavelength tables prepared by Gatterer and Junkes [3] displayed determinations; the early ones were limited to 3 or 4 estimated intensities on this limited 1 to 10 scale. Since figures, the later use of diffraction gratings and wave- 1910 some spectroscopists have arbitrarily expanded length standards permitted the specification of 5 or 6 this arbitrarily compressed scale. For example, in the figures. Since 1900 the application of interferometers very extensive spectral tables published by Exner and and better gratings has refined many wavelengths to 7 Haschek [4] the estimated intensities range from 1 figures, and recently some 8-figure values of wavelength to 1000. In wavelength tables compiled by Twyman standards have been provided. Unfortunately during and Smith [5] the maximum intensity is 20, in the this past century very little progress has been made in compilation of Kayser and Ritschl [6] estimated inten- assigning uniform quantitative intensity values to sities rise to 4000, and in the well-known M.I.T. Wave- spectral radiations. The great bulk of spectral observa- length Tables [7] they soar to 9000. The most recent tions have been made photographically because photo- compilation of Tables of Spectrum Lines by Zaidel, graphic emulsions provide detailed, permanent records Prokof'ev, and Raiskii [8] quotes data from the M.I.T. Tables and more modern sources but adds nothing ^ Figures in brackets indicate the literature references on page XII. new on spectral line intensities. Ill In or about the year 1925, microdensitometers were to physical intensities of 39,000 spectral lines repre- developed for the purpose of quantitative measurement senting 70 elements, all on the same energy scale. of relative intensities among related lines in multiplets These experiments and results are based on the follow- to test the sum rules derived from the quantum theory ing propositions, regarded as fundamental for the of spectral structure, but no general applications v^^ere quantitative description of residual spectra of diluted made. Since then thousands of spectrochemists have elements excited in ordinary d-c arcs. applied microdensitometers to quantitative chemical 1. The limiting detectability of any line is defined analyses by calibrating intensity ratios of analysis- and as the atomic concentration that ensures positive de- internal-standard lines, but such measurements have tection of the line. This limit is determined mainly by contributed nothing to the basic data on spectral line unavoidable background on a fully exposed spectro- intensities, tdkewise, with few exceptions, the mod- gram. The spectrum of an arc burning in air consists ern substitution of electronic photodetectors for photo- of discrete lines due to atoms, and of more or less graphic emulsions has added nothing to our knowledge extensive band systems from transient compounds of true line intensities over long ranges of different (usually monoxides), all superposed on a continuous spectra of many chemical elements. background arising from thermal radiation of incan- How may one hope to obtain, with a reasonable descent oxides, from transitions in the continuum, and amount of labor, quantitative intensity data on the possibly from scattered light. This background sets a same scale for thousands of spectral lines representing limit to the exposure for faint lines that may be given practically all of the metallic elements? A hint was by any actual spectrograph. If this were not true, the given in 1874 by Lockyer [2] who observed that "the exposure could be increased indefinitely to compensate lines of any constituent of a mechanical mixture dis- for unlimited reduction in concentration, and detect- appeared from the spectrum as its percentage was ability would always be infinite. Faint lines are not reduced." Acting on this suggestion. Hartley [9], in recorded by underexposure, and they cannot be recog- 1884, began to study the spark spectra of metals in nized on a very dense background produced by over- solutions with concentrations of 1 percent, 0.1 percent, exposure. In order to guarantee positive recognition 0.01 percent, and 0.001 percent, and proposed a method and unambiguous chemical identification a spectral of quantitative spectrochemical analysis based on the line should be sufficiently well defined to permit ac- lines that could be detected at each dilution. Similar curate wavelength measurement. Experience shows studies were later made by Pollok and Leonard [10], that the minimum photographic density that meets by de Gramont [11], and by Lowe [12], all showing this requirement is of the order of 0.05 above that of that with progressive dilution of an element its spec- the background. tral lines weakened and vanished until only the most 2. The limiting detectability of any element in an sensitive line remained to reveal its presence. In all arc depends on the matrix in which the element finds these works the principle of quantitative spectrochem- itself. There is no doubt that in the conventional arc istry appeared to rest on the number of lines detectable relative volatilities of the chemical elements as well as rather than on their individual intensities. Casual ob- relative ionization potentials affect the relative servation must have shown lines of equal strength in strengths of their mixed spectra. In general, the ele- spectra of solutions differing 1000 fold in concentration ments with high-vapor pressure and/or low-ionization but no one mentioned it.