Effect of Constituent Materials upon Spectrographic Measurement of Seven Impurity Elements Ill. C. BACHELDER' Los Alamos Scientijic Laboratory, University of California, Los Alamos, N. M .
Inorganic powder samples weighing 350 micrograms were burned to completion in a direct current arc and the effects of constituent elements upon the spectrographic measurement of the seven impurity elements, tin, vanadium, beryllium, cadmium, cobalt, antimony, and manganese were studied.
T
HE work of Goldschmidt and Peters ( 2 ) brought into sharp focus the fact of differential volatilization, that elements present in a carbon arc do not emit their spectra simultaneously. They completely volatilized in a carbon arc a cupeled lead bead containing other metals. During the 170 seconds required for complete volatilization, lead appeared in the spectra within the fist 20 seconds and disappeared in 50 seconds. Gold did not appear until after 20 seconds, rhodium until after 75 seconds of
In this work, attention was directed to the following points as a basis for a method of semiquantitative analysis of inorganic powders. Weight of Sample. It was desired to use a minimum weight which would completely volatilize in a reasonable period. At the same time the weight must be large enough to give an accurate intercomparison of samples having different matrices and containing varying numbers of impurity elements. Both 50and 100-microgram samples were tried but failed to give reproducible results; 350 micrograms were finally used. The region between 100 and 350 micrograms has not been examined. Presence of Impurity Elements in Sample, excited in the direct current arc. Of the 37 elements used as impurities, seven elements of common occurrence in the laboratory mere studied. The choice of seven simply brought the initial investigation within a reasonable scope. Use of Base Materials (sodium chloride, lead oxide, iron powder, magnesium oxide, and carbon). Because of the limitations of the work these base materials were arbitrarily chosen to cover a wide range of volatility and are not intended to be completely representative. Sodium chloride volatilizes far more readily than many of the impurity elements added to it. Carbon, on the other hand, represents the opposite condition, and lead oxide, iron powder, and magnesium oxide fall between the two extremes of sodium chloride and carbon. Preparation of Standard Curves for each of the seven elements in each of the five matrices. This was followed by an examination of these 35 curves to determine the effect of the impurities and matrices upon the spectrographic measurement of the seven impurity elements studied.
arcing. If one volatilizes a sample in a carbon arc, the initial time of emission of a particular element will vary with the volatility of other elements present. Slavin ( 4 ) has pointed out the failure of intensity methods when a constant exposure time is used to record photographically the emission spectrum of the same element in materials of different composition. He proposed, as a basis for quantitative analysis, the total energy of the emission until all the specimen has been burned in a carbon arc, eliminating the time factor. Consideration of the foregoing indicates that the use of the oarbon arc as a source in quantitative work calls for complete volatilization of the sample. It is desirable, then, to use a sample weight small enough so that it can be burned to completion in a reasonable time. That a very small sample size need not limit quantitative measurements has been shown by Fitz and Murray (1). Concerned with the rapid quantitative measurement of minute samples of inorganic powders, they developed a method by which they determined tin, silicon, aluminum, iron, magnesium, copper, manganese, nickel, calcium, and titanium in various silicate materials with an accuracy of 10%. In general, the sample weight was 1.0 mg. but the method encompasses weights ranging from 0.1 to 5.0 mg. The sample was mixed with a large excess of pure powder mixture which serves as buffer and internal standard. Weighed proportions of the mixture were pressed into a pellet and burned completely in a direct current arc. The line intensity ratios between selected lines of sample constituent and internal standard mere determined photometrically and converted to weight by reference to calibration curves which were prepared from synthetic standards containing oxides of the ten metals in varying proportions. These factors point to a broader approach to analytical problems, which is further emphasized by the semiquantitative method of Harvey (3). He uses for analysis a IO-mg. sample which is completely volatilized in a direct current carbon arc. Ratios are densitometrically established between selected lines and background and are multiplied by sensitivity factors that have been previously determined for particular spectrum lines in a particular matrix. Thus, prepared standards are eliminated and only a plate calibration curve is needed. The limit of error is 30 to 50%.
PREPARATION OF STANDARDS
Two grams of each standard mere prepared by weighing out the requisite amounts of base material and impurity elements
Table I.
Composition of Standards
(Eleven series of standards: three having carbon matrix, three in sodium chloride matrix, three in magnesium oxide matrix, one i n lead oxide matrix. a n d one in powdered iron matrix) Range of Impurity Series Base Material Elements, P.P.M. Impurity Elements Present -4 Carbon 10,000-100,000 Sb. Au, Ph,,Sn, Cd, Rln, Co B Sodium chloride (7 impurity elements) C Magnesium oxide 5000-50,000 Zn Be Sb 4 u P h Sn Cd, D Carbon E Sodium chloride r(fn.'co: 1 7 , ' ~ s . ' A; (12 F Magnesium oxide impurity elements) B , C a C d Fe Si Be Zn. 50-10.000 G Carbon Li 'K t!r k n ' A S ' Ag H Sodium chloride I Magnesium oxide Sn' 1; SA A; Pd T1' A1 Xi: Cu, Cd, Pd, Ge V Se, ' RRb, h , ' HSr, g:
1 Present address, Institute for t h e S t u d y of Metals. University of Chioago, Chicago 37, Ill.
1366
J
Lead oxide
100-50,000
K
Iron powder
100-50,000
R;, T e , C i , BB, W, Ir (37 impurity elements) Sb 4u Sn Cd n I n Co V hi. Ag, b e , ' N E , 'Zn '(12 impurity elements) S h . A u Sn Cd M n Co V As, kg,'Zn,'Be. 'Na 'Cli impurity elements)
V O L U M E 21, NO. 1 1 , N O V E M B E R 1949 as the metal, oxide, or chloride. These were mixed and handground in an agate mortar for 15 minutes, then placed on a mixing mill for 1 hour. and finally hand-ground for an additional 15 minutes. 4 mm. I.D. X
2 mm. deep
4 mm. I.D. X 3mm. deep
1367 num spatula, about 6 cm. in length, was used to handle the powders. The material to be weighed was placed on a watch glass 3 cm. in diameter, in which an off-center hole (2 to 3 mm. in diameter) had been blown toward the underside of the glass. This provided a slight lip which fitted easily into the mouth of the electrode crater, and the material was transferred to the crater by brushing it carefully through the hole with a fine camel's-hair brush; 350 micrograms of standard and 250 micrograms of internal standard (2070 molybdenum and 10% bismuth in a powdered carbon base) were weighed separately, transferred to the crater, and distributed evenly over the crater floor by tapping gently. To avoid loss of charge with initial arcing, a drop of collodion (ether and collodion, 1 to 1 by volume) was added. Charges were burned for 2 minutes a t 11.2 amperes and 250 volts input, with a 4-mm. electrode separation, kept constant b manual operation. The upper electrode was changed for eacl sample. ?astman SA-1 plates were used and developed in D-19 a t 18 * 1 ' C. for 3 minutes. Lines Used B e 3 1 3 0 . 4 1 6 / M o 3132.594 .Mn 2576.104/Bi 2897.975 Sb 2877.915/Bi 2897.975 Sn 3 1 7 5 . 0 1 9 / M o 3170.347 V 3 1 8 3 . 9 8 2 / M o 3170.347 Co 3 0 4 4 . 0 0 5 / M o 3170.347 Cd 3 2 6 1 . 0 5 7 / M o 3170.347
Figure 1. Dimensions of Graphite Electrode PREPARATION AND U S E OF CURVES
Eleven series of standards in five base materials were made up (Table I). Three series were of carbon base containing, respectively, 7, 12, and 37 impurity elements. Three each were of aodium chloride and magnesium oxide base prepared in the same manner as the carbon series. The lead oxide base contained twelve impurities; the iron powder, twelve. Consideration of Table I will make clear the fact that for impurity element concentration ranges of 20,000 to 50,000 p.p.m., the combined amounts of elements added as impurities are of the same order of magnitude as the base material. Intensity ratio values derived from series A, B, and C were not used in the final calculations because the three standard series so closely resembled series D, E, and F in number and range of impurities. For every set of standards, moving plates were run to determine maximum time necessary for the complete burning of the sample. The sample vias made up of the mixture of the base material and impurity elements as standard, and the graphite containing the internal standard elements. molybdenum and bismuth. The time was set a t 2 minutes. DETAILS OF ANALYTICAL PROCEDURE
The Wadsworth fully automatic stigmatic grating spectrograph, 21-foot grating 15,000lines per inch (2.5 cm.) (Jarrell-Ash Company) was used. The optical system consisted of a 350-mm. quartz lens focused beyond the slit, and a step sector. The slit height was 14 mni.; the slit width, 25 microns. Electrodes were prepared from special spectrographic graphite (5ational Carbon Company). Figure 1 illustrates the electrode setup. The lower electrode (positive) consisted of a 15-mm. length of '/?-inch carbon rod, in the lower end of which a 3-mm. crater with a 4-mm. inside diameter was drilled, for mounting on a carbon post. In the upper end a 2-mm. crater, witha 4-mm. inside diameter, was drilled. In this the charge was placed. The upper electrode, made from inch carbon rod, was cut to a sharp point and inserted in the upper carbon electrode holder. Both upper and lower electrodes were preburned for 35 seconds, All weighings were made on an assay halance. A thin plati-
Plate calibration curves were run and intensity ratios calculated from transmi
