Eighth Conference on Applied Spectroscopy L. T. HALLETT, Associate Editor using a multisource unit, Thc range of elements determined with an accuracy of 5 to 7Tc was:
HE Eighth Conference on Applied Spectroscopy, sponsored T by the Spectroscopy Society of Pittsburgh, was held November 13, 14, and 15 at the Mellon Institute. These local conferences were started in 1940 when George Harrison presented one paper on industrial applications of spectroscopy and were expanded over the intervening years t,o one-day sessions of several papers dealing with emission spectroscopy, mostly applied to metallurgy. Last year a two-day session with twelve papers on emission and absorption vr-as held, with an attendance of 200. This year it consisted of a three-day session on emission and absorption with 24 papers and an attendance of close to 300. Far from being local, this conference at,tracted many from outside the Pittsburgh area. The continued growth of this group is due not only to its progressive membership but also to the untiring efforts and spirit of Mary E. Warga on the staff of the Physics Department of the Gniversity of Pittsburgh. Most, of the papers presented are of interest to readers of this journal. The majority of the authors do not, at'this time, plan publication, but they will be glad to give details of their papers to those who wish to R i t e t o them. In those cases where authors plan publication, the journal is mentioned in the abstract.
Emission Spectroscopy 1. A Spectrochemical Method for the Determination of A 1 2 0 3 in Steel. RAYMOND H. COLINAND D. A. GIIRDXER, CarnegieIllinois Steel Corp., Gary, Ind. A chemical separation of the acid-soluble and acid-insoluble components of the steel is made. The insoluble portion which contains that part of the aluminum present in the sample as A1203 is fused and taken up in a weak acid solution. Iron is added as an internal shndard. Spectrographic examination of the solution is made by means of the well-knoivn dropping electrode, whereby fresh solution is continuously supplied to the spark gap during exposure. The chemical preparation of the samples requires 2.5 t o 3 hours with 10 minutes allowed for the spectroscopic examination. The dropping electrode is slightly modified and uses a graduated reservoir to which a hollow carbon electrode is attached by means of a rubber tubing. The solution is allowed t o drop at a rate of 10 to 20 drops per minute onto a pointed solid carbon electrode. The range covered is 0.500 to 0.005c7,. Calibration is based on synthetic standards. Spectrochemical Analysis of Pig Tin. WILEY E. $TEAKWeirton Steel Co., Weirton, W. Va. This method employs the conventional internal standard principle, using a single tin line with the alloying element lines to obtain relative intensity ratios. Pin samples approximately 7 / 8 2 inch in diameter are used as self-electrodes. An electronic spark is used for excitation and spectra are photographed with the Baird Associates 3-meter grating spectrograph on Eastman S.A. So. 1 platrs. The elements and ranges determined are as follows: 2.
LEY,
0,010 t o 0.015 to 0.005 to 0.002 to 0.002 to 0.002 t o
0.1
to2
0 005 t o 0 02 0.6 to 1 0 07 t o 0 . 7 5
4. A Calculating Board for Determining Relative Densities without the Use of a Calibration Curve. CHARLES HALE,Copperweld Steel Co., Warren. Ohio. The straight-line portion of an emulsion ralibration curve can be construct,ed for any wave-lcngth region from the densitometer readings of two selected lines in that region, one of low and one of high density, provided that both readings lie within the predetermined straight-line portion of the curve. A calculat,ing board based on this princple is described for use in steel analysis. This board requires the densitometer readings of two iron lines and one element line for each analysis, and yields results corrected for variations in emulsion contrast from one spectrum to another and from one wave-length region to another.
5. Practical Aspects of Establishing Analytical Working Curves in Spectrographic Analysis. R O B ~ RW. T SMITH,General Motors Corp., Flint, hIich. The following t,opics were discussed: The practical considerations which govern choice of instrumentation and t,echnique for specific problems. Factors influencing choice of equipment, plates, lines, photonietric procedures, etc., in the light of speed cs. accuracy requirements. The problem of securing and checking adequate primary and secondary standard samples. The random and systematic errors entering into t h r actual analysis, together with means for their evaluation. Thc economic factors which govern the choice of analytical method. Thc author points out that in determining cost, all factors must be considered, both direct and indirect. A great deal depends on whether the sample requires a routine determination or whether special methods must be used for its analysis. Where there is a wide variety of samples and determinations, chemical analyses may prove cheaper. The Gentsral SIotors Corp. has found on the average that spectrographic analysis is onc sixth to one eighth the cost of chemical methods. The spectrographic method can give more information concerning trace elements than can chemical analysis. The author made a special point that analysis is a broad subject. Because this is so, spectroscopy \vi11 sometimes give only part of the answer required, and other methods must he used. I n no sense, therefore, should the spect,rographer fecl that he is complete in his approach, and he should seek t,he advice of help of others \\-oi,king in the analytical field with different approaches. 6 . The Basis of Quantitative Computation in Spectrographic Analysis. J. H. COULLIETTE, University of Chattanooga, Chattanooga, Tenn. Innumerable experiments have shown that thp intensity of the radiant energy emitted by a volume of incandescent vapor is a function of the concentration of sourcc' atoms in the vapor.
% Lead Antimony Areenic Bismuth Copper Aluminuin
0 08 t o 2 0 05 t o 0 . 2 0.01 0 1 t o 02 . 8
Tin Lead Nickel Iron Aluminum Silicon Manganese Antimony
0.50 0.26 0.06 0.06 0.05 0.008
I
=
f(.Y)
The form of the function is dcterniined by the physical conditions existing in the vapor. Then for a niikture of two kinds of atoms within the same volumc o f irieandescent vapor the relations may he written
Standard samples were prepartld by dilution of a master alloy with pure tin. The master alloy was made by adding calculated amounts of the desired impurity elements t o pure tin. Several methods of sampling molten tin were tried. Homogeneous and representative samples were obtained most conveniently by the use of U-shaped Pyrex tubes. A notable increase in analytical precision was found when the samples were kept cool during sparking by using massive brass block electrode holders.
Il
3. Spectrographic Analysis of Copper-Base Alloys. .IRNO TUTEUR, Duquesne Smelting Corp., Pittsburgh, Pa. A metbod is presented for the analysis of copper-base alloys
=
,f(.Yl) and I r =
f(-Vp)
whrrr I , is the intensity of a given wave length cmittcd by the atoms LVl, and Ipis the intensity of a second wave length emitted by the atoms Sa. By maintaining the conditions of excitation constant and varying the concentrations of the atoms, S i and iY2, within the vapor, it can be shown empirically that I1112 = f(A-1 / S p ) .
1042
If the values of the intensity ratio obhined from measurements
D E C E M B E R 1947
1043
on a series of samples of known conce.ntrations are plotted against the concentration ratios, a calibration curve is obtained from ivhich the concentration ratios of these two kinds of atoms for unknown samples may be determined. This constitutes the basis for the so-called “internal standard method” (4) of quantitative determinations, where a known concentration (SZ) of a “standard” material is added t o the sample. When the sample consists of one major constituent and one or more minor constituents, it has become customary to assume that the small deviation of this major constituent from 1 0 0 ~ is o negligible, and the basic equation is taken as
replicate analyses ( I ) . Emphasis is placed on computational procedure and on the significance of results rather than on the statistical theory. It is fashionable nowadays to apply statistical methods to almost all forms of analvsis. and the author made a Dlea that its application t o spectrochemical analysis could yield *valuable information which present-day spectrographers are not in the habit of seeking. The average spectrographer usually does not concern himself as to whether his errors are due to chance or are real. The paper cites examples from results obtained from normal analysis to illustrate the above points. The bibliography will permit the reader to gain an idea of the statistical methods employed.
For materials consisting of three major constituents, one constituent ( N 1 ) may be taken as the reference element, and the relations m i t t e n as
LITERATURE CITED
I J I z = f(.VL/lYz) and 13/Iz = f(-\’,jS2) From the curves representing these functions for a series of known samples, the values of the concentration ratios can be determined. It, is then necessary to calculate the concentrations of the individual elements from these concentration ratios. This is done by means of the solution of the equations:
+ +v* + s,= 100% .\-,/.Yz = a
1Yl
The simultaneous algebraic solutions of’ these equations gives S Z =
100/(1
+ a + h)
This method requires the simultaneous evaluation of all the elements in the sample. If all the elements are t,hus determined, the precision of the method depends only upon the observational errors, since no approximations are involved. This method of computation from the concentration ratios has been applied to the analysis of aluminum ore ( I ) , furnace slags of various types ( 5 ) ,stainless steel ( 6 ) ,ferrosilicon ( S ) , and barium tit anat e ceramic materials. LITERATURE CITED
(1) Churchill. J. R., and Russell, R. G., IND.ENG.CHEM..AXAL.ED., 16, 653 (1944). (2) Coulliet,te, J. H., Zhid., 15, 732 (1913). (3) Coulliette. J. H., J. Optical SOC.Am.,37,609 (1947). (4) Gerlsch, W., and Schweitzer, E., “Foundations and Methods of Chemical Analysis by the Emission Spectrum,” London, Adam Hilger, 1931. ( 5 ) Rozsa, J. T., M e t a l Progress, 51 (April 1947).
7. Stable Working Curves as an Approach to Standard Methods of Spectrographic Analysis. WILLIAMH. AIAGRUM, Simonds 6sw and Steel Co., Lockport, iT.Y. This paper is the result of an investigation undertaken at the Gimonds Saw and Steel Co. to eliminate the so-called norking curve shifts so famiiiar to spectrographers. This was accomplished by studying the response of various analytical line-pairs t o changing source conditions. From these data the “ideal” source for these line-pairs is determined. This ideal source does not give rise t o shifting working curves. There are certain definite limits of circuit parameter settings within xvhich the analytical results are not affected even if the exposure is made under one set of parameters and the working curves under another. This permissible parameter change immediately suggests a possibility of duplicating curves from one laboratory to another, which would eventually give rise to standard methods of spectrographic analysis. The author made a point that focus is an inlportant factor, but more important is the accurate calibration of each emulsion. He pointed out that, if the source is properly nininrained, there will be no change in working curves. 8. Some Applications of Analysis of Variance to Spectrochemical Analysis. JOSEPH GEFFNER, Weirton Steel Co., Weirton, W.Va: The nell-known statistical technique of analysis of variance ( d ) is applied t o the following problems in spectrochemical analysis: testing for homogeneity, evaluating the relative contributions of fundamental causes of deviations in analyses (S), and estimating the increase in precision obtained by running
(1) Mandel, J., IND.ENG.CHEM.,ANAL.ED.,18, 280 (1946). (2) Snedecor, G. W., “Statistical Methods.” Chapters 10 and 11,
Ames, Iowa State College Press, 1946. (3) Vincent, H. B.. and Sawyer, R. A , , J . Optical Soe. Am.,32, 686 (1942).
Absorption Spectroscopy
9. Prisms Versus Gratings, 1 to 6 Microns. II. B. Aluminum Alloys. €3. F. SCRJBXER, CAVANAGH, National Bureau of Standards, Washington, D. C. The spectrochemical analysis of aluminum alloys is complicated by shifts in analytical curves as a result of major differences in composition of the alloys. These shifts have necessitated the application of a wide variety of standard samples for accurate results. The literature (1-4) concerning this effect was reviewed, and the results of an investigation of the effect, observed with the point-to-plane spark excitation, were presented. Observations of the specirn excited by this spark in atmospheres of air, nitrqgen, helium, and carbon dioxide have shown that the effect is a function of thc amount of oxygen present and may be eliminated by the application of an inert atmosphere. Studies of the spark excitation undcr various discharge conditions with counter electrodes of zinc, cadmium, and graphite and with the four atmospheres hav0 shown the spark in a nitrogen atmosphere to be capable of highest accuracy. -4simple apparatus for sparking disk samples i n a controlled atniosphere was described. The apparatus consisted of an ordinary Petri spark stand surrounded by a bell jar through which nitrogen flowed t o maintain an atmosphere free of oxygen. I t is planned t o publish this paper in the Journal of Research of the Sational Bureazc of Standards. LITERATURE C I T E D
(1) Balz, G., %. Metallkunde, 50,106-11 (1938). (2) Churchill, H. V., and Churchill, J. R., J . Optical S O C . Am., 31, 611-19 (1941). (3) Seith, W., and Hessling, H., 2 . Elektrochem., 49, 2JO-15 (1943). (4) Walsh, A., “Collected Papers on Metallurgical Analysis ty the Spectrograph,” pp. 65-81, London, Briiish Non-Ferrous Metals Research Association, 1945. Direct Rending Instruments ( Q u a n t o m e t e r s ) 21.
A Direct-Reading Spectrometer for the Analysis of Steel.
E. R. VASCE,Timken Roller Bearing Co., Canton, Ohio.
The first direct-reading spectrometer to be used for the analysis of stccl has been installed and is in use 21 hours daily to control the product of seven electric and t’hree open hearth furnaces. There are approximately 150 tests on lvhich 1300 determinations arc’ made for each %-hour period. Elements reported are mangancsc, silicon, chromium, nickel, molybdenum, copper, vanadium, and tin. One operator is used on each 8-hour turn. Cast steel pins, i/32 inch, are used for analysis which are taken with the aid of a glass tube which has been sealed undcr vacuum. Samplrs are quenched in nater, cut in two, and sent in a pneumatic tube to the laboratory. The pins are ground t o a point on both ends in order to permit duplicate exposure; then placed in the electrode holders, and power is applied, after which the operation of the spectrometer is fully automatic. Kithin 40 seconds after the power is turned on, a report for eight elements is read from the calibrated clock dials and reported to the melting department. The Dow-Baird instrument is used. The author pointed out that this instrument is being used for quick control work. However, it has not been used a sufficient length of time to rely upon it entirely for final specification work. For this they still use the chemical methods, but found that the spectrographic method in practically all cases checked the chemical procedure very closely, and it is hoped that when more data are obtained the spectrographic method may be used for final specification work. 22. Applied Quantometry in the Aluminum Industry. F. R. POTTER,Aluminum Research Laboratories, Xew Kensington,
Pa. Quantonietry has become an integral part of production, controlliug as it does the entire output of one of the largest aluminum sheet niills in the world. I n seven months a single instrument (Applied Research Laboratories model) has turned out over 350,000 determinations, faster, more accurately, and more economically than the chemical or spectrographic methods previously used. Both metallurgists and production men now believe the Quantometer or its equivalent t o be a necessity for efficient operation. Thc Qumtometer has provided valuable research informat ion and has made possible exact determinations of magnesium and silicon in the range of 0.0001%. Direct-reading spectroscopic instruments will have an increasingly important influence on the orgrnization and production methods of metallurgical plants in the future. All large fabricating and castings plants in the aluminum industry will eventually be equipped with control equipment of this type. The major aim will be product improvement and production economy with analytical costs and instrument costs as secondary considerations.
1045 23. Details in the Analysis of Steels with the Quantometet. A I . F. HASLER, Applied Research Laboratories, Glendale, Calif. The problem of providing the optimum in accurate, direct-reading spectrochemical analyses for the steel industry has been given further study with the Quantometer. -4variety of spectrum lines for each element have been examined under several source conditions and choices made for each concentrational range. Reproducibility figures for some of the most important ranges of various elements were discussed. The author pointed out that methods do exist, arid are practical for, the analysis of steels. Proper working curves, of course, must be xvorked out which correct for the influence of one element on another. B typical example was cited for the analysis of chromium, nickel, and molybdenum alloys. Xaturally, spectrographers are watching with interest the application of the Quantometer to analytical problems mainly in the metallurgical field. The papers presented gave considerable data and background as to their usefulnes as well as to t.heir limitations. The elimination of photographic film with its variability is, of course, a great step forward. The correction for background is easily accomplished n-ith these insrruments. The fundamental weakness of these inst,ruments is their lack of versatility when changing from alloy to alloy. However, considerable progress has been made in this direction, and it would seem that the establishment of working curves with standard samples solves this problem in the main. The prepared calibration strips as used with the .Ipplied Research Laboratory model give extreme versatility with constantly changing alloys. There is no doubt that for product control the Quantometer will speed ‘up plant operations and give a closer control of quality with less loss of spoiled melt. However, it is our feeling that these iris:ruments mill not displace chemical methods nor photographic emission instruments. These will be used more and more for special research on methods and in the establishing of working conditions for the rather expensive Quantometer. I t would seem a t present that a very careful analysis of analytical practice in relation to plant control must be undertaken before a Quantometer can be justified. After studying all the factors (cost, the number of alloys to be run, etc.), if the time of analysis can be considerably shortened and plant practice can be speeded up, then the installation of a Quantometer can be justified. The precision and accuracy obtained with the apparatus appear equal to, and in some cases better than, those formerly obtained by using nhotographic film.
BOOK R E V I E W Colorimetric Methods for Rapid Analysis of Silicate Materials. Rune Hedin. 110 pages. Proceedings, S o . 8. Swedish Cement and Concrete Institute, Royal Institute of Technology, Stockholm, Sweden, 1947. Price, 10 kr. This treatise describes a procedure for the determination of SiOl, FezOS, AlzOs, TiO,, CaO, JIgO, NaTO, and K?O in a total analysis of silicate material$, with all except CaO determined colorimetrically. The colorimeter is of unique design. -1mercury discharge lamp is used with filters selected to screen out individual spectral lines. TNO photoelectric cells of the selenium barrier type are connected in a compensating circuit and balanced optically, by means of iris diaphragms, to secure zero deflection of a galvanometer. Comparison solutions of known concentration are not used in the determinations. The author describes methods of preparing three sample solutions from an original sample: (1) for &termination of XazO and K20; (2) for &On, FetOa, -41~03,and TiOz; (3) for CaO and hlgO. The reagents for colorimetric determinations in these solutions are selected for each constituent with reference to possible interference from other constituents present. Extensive eupeiimects are reported, in which the effects of such interferences are considered, and also the effects of volume of sample solution. quantity of reagent, variations in pH, and other factors bearing upon the accuracy of the determinations. A time study, reported in detail, indicates L. A . DAHL that the total analysis can be completed in 8 hours.
