INDUSTRIAL AND ENGINEERING CHEMISTRY
734 c
TABLE IV. STANDARD DEVIATIONS Cr 35 32.21 27.72
No of Sam le8 Ma'ximum concentration Minimum QI concentration Spectrograpfic/chemical ratio Maximum Minimum Average Standard deviation % deviation
B/,
Ni 35 24.53 14.26
1.118 0.947 1.032 0.034 3.3
1.131 0.907 1 002
0,042 4.2
T~~~ V. pRECIsIoN AND R~~~~~~~~~~~~~ T~~~ON Per Cent Transmirrsion Cr3169 Ni3087 Fe3259 14.6 19.6 49.2
ii:: 13.9
15.4 16.4 15.0 14.0
i;::
f;:; 44;:" 8 . 0 18.2 22.1 22.6
19.8
g;:
u
.
1.760 1.790 1.793 1.778
g;;:
Cr
u Ni
--
1.835
i:;;: 1.790
52.7 53.6 51.3 50.0
21.0
SAMPLE X Y
0.223 0.237
%
cr
A
sINaLn % Ni
0.624
28.35
17.68
0.616 0.612 0.635 0.616 0.802
29.05 28.59 28.75 28.99
17.70 17.78 18.16 17.70 17.45
::::;
ii:;: 28.75
~:~~~ ;:g;
deviation - 0 . 7 7 2 deviation - 1.33
i;:;
:;:;$
Conclusions The precision of the method is sufficient for the control of melting operations of stainless steel. The primary advantage
Vol. 15, No. 12
claimed for the method is speed. Analyses have been made in this laboratory with total time of 9 minutes elapsed between taking the sample at the furnace and rcporting the results. A test of the reproducibility of results and the precision attainable WM made by making a stepped-sector exposure, a serics of exposures for standard samples, and ten exposures for a single sample on the same plate. The complete series of curves and constants were determined from this plate. The data for the ten determinations on the single sample are shown in Table V. The per cent deviation for a series of seven determinations by chemical methods on the same sample used for the determinations listed in Table V was 1.71 per cent for chromium and 0.89 per cent for nickel. Obviously the spectrographic method compares favorably with the chemical method when thenecessary careis takeninmakingaseriesof determinations.
Literature Cited S.,"A Practical Grating Spectrograph for Industrial Use",Proc. Sixth M. I. T. Summer Conference on Spectroscopy,
(1) Baird, W.
1939. (2) Harrison, G. R., "M. I. T. Wavelength Tables", New York,
John Wiley & Sons, 1939. (3) King, Carl, J . Opticd SOC.Am., 32, 112 (1912). (4) Pierce, W. C., and Nachtrieb, N. H., IND. ENQ.C H ~ M ANAL. ., ED., 13, 774 (1941). PBESENTBD before the Alabama Section of the AMZRICAN CHEMICAL socrrrT.
A Fused Salt Technique in Spectrochemical Analysis N. H. NACHTRIEB', D. H. JOHNSON, ANDK. S. DRESS Pittsburgh Plate Glass Co., Barberton, Ohio
F
REQUENTLY the chemical spectroscopist is called upon to
analyze powdered substances for minor constituents. Broadly speaking, analyses of material in the solid state present more difficulties than determinations of constituents in solution. Substances which are soluble in water, acids, or other common solvents entail few problems; in such cases it is usually feasible to evaporate a drop of a solution of the material, containing an added internal standard when necessary, on the surfaces of smoothfaced graphite electrodes and to excite the residue in a high-voltage alternating current arc ( 2 ) . Substances which do not dissolve in the common solvents are generally treated according to the time-honored practice of vaporizing and exciting them from the cored anode of a direct current arc. As is well known, analyses carried out by means of the direct current arc often fall short of the accuracy which is possible in spectrochemical analysis. Not all of the blame can be laid to the intrinsic properties of the direct current arc, however. A common source of error, both in the preparation of standard powder samples and in the analysis of unknown eamples, is lack of homogeneity. I t is an exceedingly arduous task to disperse small amounts of added material uniformly throughout a matrix in the preparation of standard powders. Confrontcd with this problem in the analysis of titanium dioxide, the authors have developed a method in which samples are fused with potassium bisulfate, and graphite electrodes are 1 Present address. Univeriity of Chicago Metallurgioal Laboratory, Chicago, Ill.
. .
i i
0010
0 040
IO
1
I
PERCENT \ ,NADIUM FIQURE 1. WORKING Cmw FOR VANADIUM IN TITANIUM DIOXIDE
C 10
ANALYTICAL EDITION
December 15, 1943 ..h
TABLE I. TYPICALRESULTS OF TITANIUM DIOXIDEANALYSIS Antimony Added Found
%
Chromium Added Found
%
Vanadium Addsd Found
Iron Added Found
%
%
%
0.100
0.112 0.073 0.101 0.111 0.096 0.099
0.200
Av.
1 .oo 0.67 1.10 0.95 0.92 0.93
0.192 0.171 0.120 0.150 0.170 0.161
0.050
0.056 0.045 0.051 0.045 0.059 0.051
0.100
0.106 0.097 0.039 0.091 0.092 0.095
0.050
AI..
0.57 0.48 0.49 0.55 0.61 0.54
0.056 0.046 0.054 0.050 0.055 0.052
0.024
0.050
0.050 0.043 0.057 0.040 0.042
0.025
0.049 0.022 0.033 0.019 0.037 0.032
%
1 .oo
0.50
0.25
0.22 0.28 0.26 0.29 0.35 0.28
0.025
0.025 0.020 0.025 0.030 0.026
% 0.100
% 0.090 0.114 0.109 0.111 0.108 0.106
73s
50.0 nrams of ootassium bisulfate and 2.50 nrams of titanium dioxide. In similar fashion, succeeding standard mclts and electrodes were prepared, each containing one half the impurity content of the formcr. Spectra were recorded by a Iarqe quartz Bausch & Lomb Littrow spectrograph set for the region 2500 t o 3400 A. The electrodes, held in water-cooled nickel electrode holders, were accurately adjusted to a scparation of 1.0 mm. and arced at 4.5 amperes (2200volt alternating current arc). The exposure time was 1 minute (no pre-arc), with a spectrograph slit of 30 microns. Eastman Spectrum Analysis No. 1 plates, calibrnted bx means of an iron arc and a rotating stepped sector, were developed for 2 minutes a t 18" C. in D-72.
Treatment of Data
The line pair3 chosen for the construction of working curves were: Sb 2598/Ti 2590.3 b., V 3183.4/Ti 3179.3 b., Fe 3020.6/Ti 3002.7 b., and Cr 3021.6/Ti 0.014 0.12, 0.15 0.013 0.025 0.025 0.0125 0.016 0.012 0.15 0.024 0.013 3002.7 A. It was found desirable, in the interests of 0.013 0.13 0,009 0.034 0,008 0.05 0.022 precision, to apply background corrections to all lines 0.012 0.14 0,008 0.029 0.009 except Sb 2598 b. and Ti 2590.3 b., which lie in a 0,011 Av. 0.12 0.027 0.012 region of low background. The background correcAv. error tion, discussed by Pierce and Nachtrieb (9), consists (5 detns.) 7.8 10.1 4.8 10.5 in converting line and background transmissions (or c densities) into relative intensities, and subtracting the background intensities from the line plus background intensities for both analysis and internal standcoated by dipping them in the molten solution. In addition to ard lines. The logarithms of the background-corrected line providing a simple means of preparing homogeneous standards, intensity ratios (analysis line f internal standard line) are then the procedure eliminates the undesirable sputtering characterfound to be a linear function of the logarithms of the conistics of the titanium dioxide arc. centrations down to rather low concentration limits. The fact that the working curves depart from linearity a t lo? concentraPreparation of Standard Samples tions indicates that the titanium dioxide was not pure. The residual concentrations of impurities in the titanium dioxide could be Exactly 100.0 grams of ppectroscopically pure potassium bisulfate and 5.00 grams of pure titanium dioxidc were fused todetermined by a method of successive approximations (8) or by gether in n 115-mm. porcclain dish over a MBker burner. To the construction of intemity-concentration graphs (1, 3). melt were added 0.0600 gram of antimony trioxide, 0.0352 gram shows the effect of background correction on the Figure of ferrous ammonium sulfate. 0.0240 nrnm of vanadous chloride, vanadium working curve; the increased sensitivity (slope) ocand 0.0141 gram of potassium diclirc&ate. The resulting clear casioned by the correction is typical. Moreover, it is noteworthy melt contained 1.00 per cent antimony, 0.201 per cent vanadium, 0.10 per cent iron, and 0.10 per cent chromium (all based on titanium dioxide). After tlie melt had b e e n t h o r o u g h l y stirred, smooth-faced graphite electrodes were dipped therein. The ends of the electrodes were > barcly touched to the surface of t the melt. A smooth thin coat Y of salt was nssurcd by preheatz W ing the electrodes before immerc sion of their tips. and by shak2 ing off the excess salt immediately after their removal from the m-clt. A thin vitreous coating was found to lead to better reproducibility, since electrode separations could be more accurat,cly controlled. When the remaining melt had aolidificd, it was crushed nnd w e i g h e d . O n e h a l f of t h e crushed salt was re-fused with Av.
0.046
-
---f
FIQURE2. WORKINGCURVES FOR IRON, CHROMIUM, AND ANTIMONY IN
TITANIUM DIOXIDE
INDUSTRIAL AND ENGINEERING CHEMISTRY
736
Vol. 15, No. 12
value of unity. Figure 2 showv the working curves obtained for antimony, iron, and chromium. Figure 3 shows the final working curves, obtained from the curves of Figure 2 by correction for the residual impurities in the titanium dioxide. Table I lists the results of four analyses in quintuplicate for each of the elements, together with the average deviation from the true values. The time required for the eighty determinations was about 5 hours. Table I1 gives a comparison of results for iron in sixteen samples of titanium dioxide by chemical analysis and by the spectrochemical procedure desrribed in this paper.
Summary A technique has been described for the analysis of powders which may be dissolved in molten potassium acid PERCENT sulfate. The method has been applied FIGURE 3. WORKING CURVESCORRECTED FOR RESIDUAL IMPURITIES -_ to the determination of antimony, vanadium, chromium, and iron in titanium dioxide, analysis conditions and working curves for TABLE 11. COMPARISON OF SPECTROCHEMICAL AND CHEMICAL which are presented. The advantages derived from the appliANALYSES OF IRON IN TITANIUM DIOXIDE cation of background corrections are illustrated by working Sample Iron Found curves for a vanadium line. Chemicala Spectrochemical No. The fused salt technique of preparing samples and standards % % may be extended to many other substances which are difficult 1 0.003 0.003 0.006 2 0,006 to dissolve in the common solvents or to arc as such. The seleo0.016 0.016 3 4 0.027 0.033 tion of a flux in a particular case will be determined largely by 0.029 0.029 5 solubility considerations and by ita suitability as a spectroscopic 0.003 6 0.002 7 0.006 0.007 buffer. 0.005 8 0,005
a
aio
0 010
0.001
9 10 11 12 13 14
0.002
15 16
0.009 0.006
0.003 0,005 0.004 0.009
0.004
0.004 0.003 0.004 0.004 0,011 0,006 0.012 0.006
Colorimetric thiooyanate method.
Acknowledgment
The authors wish to express their appreciation to Leavitt Gard and to John Darby for carrying out the chemical analyses required in the testing of the method.
Literature Cited
that the slope of the background-corrected vanadium working curve is 1.063, in good agreement with the theoretically predicted
(1) Cholak, J.. and Story, R. V., J . Optical SOC. Am.. 32, 502-5 (1942). (2) Duffendack, 0. S., and Wolfe, R. A., IND.ENQ.C H ~ MANAL. ., ED.,10,161-4 (1938). (3) Pierce, W. C., and Nachtrieb, N. H., Ibdd., 13, 774-81 (1941).
Identification of Rust on Iron and Steel SIR: I n a recent article [IND. ENG.CHEM.,ANAL.ED.,15,464 (1943)l I described a sensitive test for iron rust which involved the use of gelatin-coated paper. Processed, single-weight, glossy photographic paper was recommended for the test. It has been called to my attention that two nonsensitized gelatin papers, Eastman Kodak Imbibition Paper and Defender Backing S, are available and my reasons for not using such papers were questioned. The two nonsensitieed papers mentioned are available in doubleweight stock only. I tried double-weight papers in the original tests and found that they did not adhere well to the surface being
tested unless constant pressure was applied. The single-weight paper described in the article adhered perfectly without pressure, even when applied to curved surfaces of short radius. I still prefer the desensitized single-weight paper for the reason given. I n all other respects the double-weight nonsensitized papers are satisfactory. I regret that I neglected to mention this point in the article. RALPH0. CLARK Gulf Research & Development Co. P. 0. Box 2038 Pittsburgh 30, Penna.
