APRIL 15, 1940 ANALYTICAL EDITION 195

perature and a stream of air passed in, either bv pressure 011 the air was admitted to keep the deposited solid from melting. flask side tube or by suction at the upper end of the distilling The beta-nap]lthol TTas deposited in pure colldition head, The cool air condenses the solid and carries it to the dis- tilling head, where it is filtered off by the cotton plug. easily dislodged from the tube. Compared with ordinary

sublimation or atmospheric or vacuum distillation, with the splattering and solidification in the condenser tube and the neck and sides of the flask, the method is more rapid and con- venient, achieves purity, and avoids complicated apparatus and troublesome operation.

Beta-naphthol, which has a vapor pressure of approxi- mately 3 mm. a t its melt,ing point (122' C.), in this device sublimed a t a rate of almost 1 gram per minute. The tempera- ture was maintained above the melting point, but enough cool

Quantitative Spectrochemical Analysis by Measurement of Relative Intensities

E. K . JaYCOX AND A. E. RUEHLE Bell Telephone Laboratories, Inc., New York, 3. Y.

-4 method is described which combines flexibility of application with the improved precision resulting from modern methods of photometry. Applications have been made to samples in which the main com- ponent is lead, aluminum, iron, copper, nickel, and the alkaline earth oxides, re- spectively, with an average precision of 50 to 100 parts per thousand of element determined.

HE rapid extension of the application of the spectro- T graph to chemical analysis in recent years may be traced to improved methods of reproducing excitation con- ditions and improved methods of photomet'ry. In earlier work the condensed spark betv-een elect'rodes made from the sample itself was the principal met,hod of excitation used. The main draxbacks of this method are relatively lon- spec- tral sensitivity and uncertainty in exciting representative portions of samples and standards. Both of these difficulties were circumvented by Sitchie (6) x h o used arc excitat'ion of dried solutions. S o t only are most metals more sensitive in the arc, but i t is much more convenient to make up reliable synthetic standards in solution form.

Sitchie's met'hod is widely used today, hut its precision is limited by t'he inherent lack of reproducibilit,y of the arc, which is subject to "wandering" and current fluctuations. Sitchie and Standen ( 7 ) applied Gerlach's (4) internal stand- ard method in an attempt to eliminate the effect of such variations. This particular application involves two as- sumptions: (1) that any variability in the arc nil1 affect the excitation of the test element and the reference element in exactly the same manner; ( 2 ) that, the difference in den- sities of two lines on the plate is proportional to the logarit,hm of the ratio of the intensities of these lines in the source. Experience indicates that the first assumption is reasonahly justifiable in many cases. I n some cases, however, a change in conditions will materially change the relative intensities of a pair of lines. For this reason the internal standard method must not be applied indiscriminately, but the fundamental assumption must be justified in each case.

The second assumption is also a potential source of t,rouble, since it holds only for the straight-line portion of the charac- teristic curre of the plate. For this reason one must either

use a given line pair only a t or near the concentration of test, element giving equal intensity for bobh lines (method of homologous pair?) or else one must correct for any deviations from the straight-line port,ion of the characteri-t' 5 ic curve through a calibration of each plate. The equal density method is n-idely used abroad (1, 4 , I O ) , while the practice of plate calibration has gained in favor with American

LOGl R E L A T I V E E X P O S U R E (E= It)

FIGURE 1. TYPICAL C4LIBRATIOX AND J T O R K I S G CURVES FOR TIN IS LEAD

196 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 12, NO. 4

On the processed and dried plate measure the densities of TABLE I. PRECISION DATA FOR TIN IN LEAD suitable lines of B , m, n, 0, etc., and of each step of the same line

of B in the step spectrogram. Plot the density of each step against step numbers (log relative exposure to the base equal to the auerture ratio). From this curve and the measured

(Sn 2863.3/Pb 2657.1) Aliquot Tin Added Tin Found Deviation Deviation

m

1 2 3 4 5 6 7 8 9 10

0.047 0.047 0.047 0.047 0.047 0.0096 0.0096 0.0096 0.0096 0.0096

0.051 0.046 0.049 0.048 0.046 0.0102 0.0093 0.0093 0.0101 0.0094

densities 'of the chosen line pairs in the standards construct-a working curve of log concentration us. difference in log relative +0.004 8 , 5

-0.001 2 . 1 +0.003 6 . 3 exposure (Figure 1). Using the same curves and the measured +0.002 4 . 2 densities of corresponding line pairs in the samples, read off the

concentrations of m, n, 0, etc. In general the calibration curves -0 .001 2 . 1 +0.0006 5 . 9 - 0.0003 3 . 0 will vary from plate to plate but the working curves will be the - 0.0003 3 .0 same or a t worst slightly displaced but parallel. +0.0005 4 . 9 -0.0002 2 . 0

Av. t 4 . 2

70

Base, Im- Material purity

BaO-SrO Sr/Bab

Pb

P b P b

Pb P b P b Pb A1

A1

A1

A1

A1

A1

Fe Fe Fe Fe Fe Fe Fe Fe Fe c u c u Xi Ni

A g

Bi c u

Fe Sb Sn Zn c u

Fe

hl n

P b

Sn

Zn

AI R.l g Ca c u c o Cr Mn Ni Si P b T1 Ba Mg

TABLE 11. REPRESENTATIVE APPLICATIONS

Range

0.001-0.03

0.001-0.3 0.001-0.3

0.001-0.3 0.001-0.3 0 .001-0 .3 0.001-0.3 0.001-1

0.001-1

0.001-1

0.001-1

0.001-1

0.02-1

0.003-1 .O 0.003-1.0 0 .1-1 .0 d.001-1.0 0.01-1 .o 0.01-1.0 0.001-1.0 0.01-1 .o 0.003-0.1 0.001-1.0 0.05-1.0 0.003-1.0 0.003-1.0 0.11-9.0

Source

Solution d.c. or a.c. arc, radial sector

Solution d.c. arc Solution d.c. or a.c. arc, radial

Solution d.c. arc Solution d.c. arc Solution d.c. or a.c, arc Dry nitrate d.c. arc Dry nitrate + carbon dust +

HNOa d.c. aro Dry nitiate + carbon dust +

" 0 8 . d.c. arc

sector

Dry nitrate + oarbon

Dry nitrate + carbon

Dry nitrate + carbon

Dry nitrate + carbon

Chloride soln. a.0. arc Chloride soln. a.0. aro Chloride soln. a.0. aro

"01, d.c. arc

HKOa, d.c. arc

"03, d.c. arc

"08, d.c. arc

Chloride soln. 8.0. aro Chloride soln. 8.c. aro Chloride soh . a.c. arc Chloride s o h . 8.0. arc Chloride soln. 8.0. arc Chloride soh . a.c. aro i'iitrate soln. d.0. arc Nitrate aoln. d.0. arc Nitrate soh . d.c. arc Nitrate s o h d.c. aro Soln. on Cu d.c. arc

Chloride soln. 8.0. aro Chloride soln. 8.c. aro Chloride soh . a.c. arc Chloride s o h . 8.0. arc Chloride soln. 8.0. arc Chloride soh . a.c. aro i'iitrate soln. d.0. arc Nitrate aoln. d.0. arc Nitrate soh . d.c. arc Nitrate s o h d.c. aro Soln. on Cu d.c. arc

dust + dust + dust + dust +

Impurity Line

Ag 3280.7 Bi 3067.7

Cu 3274.0 Fe 2719.0 Sb 2598.1 Sn 2863.3 Zn 4810.5

Cu 3274.0

Fe 2719.0

M n 2798.3

P b 2833.1

Sn 2840.0

Zn 3345.0 A1 3082.2 M g 2802.7 Ca 3158.9 Cu 3274.0 Co 3044.0 Cr 4254.3 hfn 4030.8 Xi 3493.0 Si 2881.6 P b 2833.1 TI 2767.9 Ba 6141.7 Mg 2779.9

Reference Line

P b 3220.5 P b 2657.1

P b 3220.5 Pb 2657.1 P b 2657.1 P b 2657.1 P b 5005.4

A1 3050.1

A1 3050.1

A1 3050.1

A1 3050.1

A1 3050.1

A1 3050.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Fe 3298.1 Cu 2768.9 Cu 2768.9 Ni 6176.8 Ni 2770.2 Ba 4554.0

Precision0 P. p . 1000

* 67 * 70

* 85 * 70 * 70 i 42 * 80

d100

*loo *loo *loo *loo *loo *loo * 100 *loo * 80 * 80

80 * 80 * so =k 80 * 65 * 65 * 80 * 67 * 75

a Precision listed represents average deviation from mean of six or more independent determinations. samples were of known composition, average devigtion from true value was no greater than this figure.

b Ratio of major components.

Where

analysts since the work of Duffendack (3) emphasized its importance. The authors have applied the modified method described below to a variety of cases with uniform success.. It combines flexibility of application, convenience, and speed with a degree of precision satisfactory for most work.

Apparatus Suitable apparatus for application of this method consists of

a large spectrograph and its usual accessories, a step sector (Bausch & Lomb, ratio 1.5, is suitable), a rotating electrode assembly ( d ) , a densitometer (modified Moll in this case), and Eastman process plates.

Experimental Method Dissolve the sample in an acid or mixture of acids suitable

to hold all known components in solution. Prepare a solution of pure base metal, B, in the same concentration as the sample. Add known amounts of impurities m, IZ, 0, etc., t o a portion of the base solution. Dilute portions of this highest standard with more base solution to give a graded series of standard solutions. Add aliquot portions (0.1 cc.) of each standard and sample to y p h i t e electrodes which have previously been shaped, puri-

ed (8, 11) (if necessary), cups coated with collodion, and pre- heated to 100" C. Continue heating on an aluminum block until dry. Arc at suitable current [direct or alternating cur- rent (9)] while rotating the lower electrode at 600 r. p. m. ( 2 ) . Focus the image of the source on the collimator or beyond (light approximately parallel). After arcing each standard and sample and recording their spectra on a suitable plate, take an additional spectrum of the highest standard through a step sector, using a slightly longer exposure.

It will be seen from the data in Tables I and I1 that the high precision claimed by some investigators for similar methods has not been attained in the present work. While a higher degree of precision may be expected by intensive study of any given application, the authors' experience in- dicates tha t for run-of-the-mill samples a precision of * 50 parts per thousand is about the best to be expected with arc excitation. This is satisfactory for many purposes and is approximately twice as good as the precision attainable by the comparison standard method with the same preparation and excitation of the samples studied (5).

Literature Cited (1) Brownsdon, H. W., and Someren, E. H. S. van, J . Inst . Metals,

(2) Clarke, B. L., and Ruehle, A. E., Be22 System Tech. J., 17,

(3) Duffendack, 0. S., Wolfe, R. A., gnd Smith, R. W., IKD. EXG.

(4) Gerlach, W., 2. anorg. allgem. Chem., 142, 389 (1925). (5) Jaycox, E. K., and Ruehle, A. E., 1939 Conference on Spectros-

(6) Nitchie, C. C., IND. EKQ. CHEN., Anal. E d , 1, 1 (1929). (7) Nitchie, C. C., and Standen, G. W., Ibid., 4, 182 (1932). (8) Owens, J. S., Metals & Alloys, 9, 15 (1938). (9) Ruehle, A. E., and Jaycox, E. K., IND. ENG. CHEM., Anal. Ed.,

46, 97 (1931).

381 (1938).

CHEM., Anal. Ed., 5, 226 (1933).

copy, New York, John Wiley & Sons (in press).

in press. (10) Smith, D. M., Trans. Faraday Soc., 26 101 (1930) (11) Staud, A. H., and Ruehle, A. E., IND. ENG CHEM., Anal. Ed.,

10, 59 (1938).