Accuracy In Spectrochemical Analysis

VoL. 13, No. 4, 1959

85

APPLIED SPECTROSCOPY Published by the Society for Applied Spectroscopy No. 4, 1959

Volume 13

Accuracy In Spectrochemical Analysis Arno Arrak Belmont Smelting & Refining Works, Inc., Brooklyn, New York and

A. J. Mitteldor~ Spex Industries, Inc., Scotch Plains, New Jersey* Abstract Accuracy is one of the broad areas of spectrochemtcal analys*s where improvement by an order of magmtude would result in vast

new fields of apphcat*on. This paper discusses the subject under the topics of sampling, the source, electrodes, the spectrograph, photometry and calculations New techmques for reducing variables will be d,scussed These include optics for reducing the effects of arc wanderlng, the Elpae for reproducible packing of powders into electrodes, devices :for mechamcally traversing a flat sample during sparking, the Stallwood jet and enclosed arc chamber. Introduction

Many potential applications of spectrographic analysis he dormant simply because present-day accuracy is not good enough. N o t only is this true in industry where improved accuracy would mean superior quality control as well as savings in money, but m biology and medicine better accuracy could result in a new diagnostic tool to tell the researcher why nature goes wrong.

Sampling The scapegoat of all analytical techniques is sampling. After a sample is consumed, no one can argue that it once truly represented the average composition of the material being analyzed. The reader is referred to an excellent treatise on the subject (1, p. 51) which discusses the reduction of a production batch to a laboratory size sample. The spectrographic problem really begins at this point, however, because the spectrograph "sees" only a few milligrams of the actual sample and any segregation whatsoever is greatly magnified.

determination and, further, that certain elements, which ordinarily spark off at the beginning of the exposure, have a constant intensity with time when new areas of the sample are continuously being sparked. He cited (4) A1 and Mg determinations in zinc-base alloys. Ordinarily, spectral line intensity of these elements drops sharply after the first few seconds, but, by using a moving specimen, the sparking-off effect is removed. Improved results were confirmed in the analysis of cast irons by Clark (5) who moved the sample linearly during the sparking so that the sparked area was equivalent to eight stationary burnings. As shown in Table I, the technique aids markedly in matching spectrographic with wet chemical determinations. Note that, although better than individual sparkings, the sweep method is not quite as precise as the multiple exposure method in which four individual burnings are superimposed. The principal reason for this is statlst~cal. Theoretically, any determination can be improved by the square root of n (the number of determinations) by taking the arithmetic mean of n values. TABLE I. SPECTROCHEMICAL ANALYSES OF CAST IRON Devtatson from Wet Chemtcal Values, % Element

Mo Cr Nl Mn $1

Szngle Sweep Determznatzon Method

5 6 4 5 13

3 3 2 4 4

Four Supertmopsed

Spectra 2 2 1 3 5

One method of reducing segregation effects m point-toplane techniques is that of taking multiple, superimposed exposures at different places along the sample. Another, and one that is readily automated, is to move the sample slowly and umformly during sparking. Hurwitz (2,3) showed how traversing segregates aids in obtaining a true

To take advantage of the sweep technique, a Petrey Stand with a slowly rotating table is manufactured by Spex Industries (Figure 1). The speed of traverse is determined by the choice of motor but about 90 ° per 30 seconds - - the normal sparking time - - is suggested.

'Presented at the 10th Annual Symposium on Spectroscopy, Society for Applied Spectroscopy, Chicago, IlL, June, 1959, and at the Denver Conference, Rocky Mountain Spectroscopy Society, August. 1959.

For powders the sampling problem has a number of aspects often overlooked: particle size differences, adsorption and absorption of water vapor and atmospheric gases, packing of the powder into the electrode, to name a few.

86

APPLIED SPECTROSCO1'Y

TAL~LE III F E R C E N T S T A N D A R D D E V I A T I O N O F A S E R I E S OF RELATIVE INTENSITIES OF ANALYSIS LINESa (7) PackeJ

Fe

AI

Ca

Sn

Na

Operator 1 Operator 2 Operator 3 Machine

23 5 21.7 21.6 19.0

17 7 14 8 9 5 7.4

9 7 16 4 7.6 6.6

17 5

1 ~8

16.9

17 3

7 7 7.5

11.1 8 3

aTwenty samplesexcited

burnmgs themselves are more reproducible when the Elpac used.

IS

Incidentally, Strashelm, the inventor of the Elpac, has stated that much better results are achieved wJth baked powders, the effects of gas and moisture being exaggerated when automatic packing is used. FIG. 1. THE PETREY TABLE USED WITH THE SpEx 9010 ARc-SPARK STAND Specimens m a y be placed on a t u r n t a b l e so t h a t a large area is s a m p l e d A jet ( n o t s h o w n ) , m a y be d i r e c t e d at t h e s p a r k to r e d u c e m a t r i x effects.

Recognizing these problems, Tingle and Matocha (6), m an excellent pmce of work, converted a method normally considered but semi-quantitative into one with quantitative accuracy. Their technique involves fusion of samples with lithmm borate serving to homogenize the sample and convert it to a common matrix simultaneously. The sample thus treated in ground and screened so that only a selectively chosen particle range is used in the analysis. Packing electrodes is a problem neatly handled with the Elpac, a new machine which automatically tamps craters full (7). A n excess of the material is placed m a disposable polyethylene funnel perched snugly around the neck of the electrode. A tamper is lowered into position and a fixed vibration packs the crater in a few seconds. Both uniformity and tight packing are achieved with the instrument as illustrated in Table II. In three out of four typical samples, the amount of material packed is reproduced within around 1%. Furthermore, the amount packed averages around 50% more than that obtained by hand packing. Note, too, that two of the electrodes chosen were ¼ " d. and center-post types which are, of course, extremely difficult to pack by hand.

TABLE I1. ~'ELPACKED" VS HAND-pACKED ELECTRODES:~ Quantity Packed, mg Coef of Variation, % Material Ellmc Hand Elpac Hand Potassium Carbonate plus g r a p M t e , 1 1 Iron Oxide Llthmm Carbonate plus graph*te, 1 1 Lithium Carbonate plus g r a p h i t e , 1 1

221 a 283 a

171 a 167 a

1.2 2.5

4 0 4.0

82 b

661)

1 3

8.1

10 e

1.0

24 e

10

~Average of 10 runs "¼" d electrode with undercut crater h¼,, d electrode, crater undercut with center post ' ~ " d electrode, crater ¼'" deep Table III shows that superior packing with the Elpac does, indeed, result in improved precision. The figures in this table represent the coefficient of variation of lntensltives of individual lines, dlustratmg the fact that d.c. arc

The Source Spark sources have been improved steaddy in recent years. The use of an air blast at the auxihary gap helps stabdize the spark. Controls of the auxiliary gap spacing and oscilloscopic observation of rate of &scharge of the spark are helpful in reducing variables Superior materials and shapes of the auxiliary gap electrodes also aid in maintainmg spark stability. While too new to evaluate, the Bardocz electromcally controlled spark (6) appears promisIng. Feldman and Ellenburg formulated a procedure (9) by which, they claimed, optamum r e p r o d u o b d i t y was obtained from their spark sourceS: 1 Auxiliary gap opened to 10 m m or wider. 2 Autotransformer m primary circuit set at zero. 3 Source turned on. 4 Autotransformer adjusted to give a predetermined voltage across its variable arm. 5 Auxiliary gap narrowed until one breakdown occurred on each half cycle. 6 Autotransformer adjusted to give the desired number of breaks per half cycle. Variables such as cleanliness, smoothness, and shape of the auxiliary electrodes, as well as resistivity of the amblent atmosphere which is temperature and humidity dependent, are avoided by using this procedure. In the opinion of the authors, it aids in reproducing the curve of current vs time from one discharge to the next. A t t e m p t s have been made to couple the advantages of the spark's stability with the arc's sensitivity and such hybrid sources may now be obtained commercially. In pure arcs - - still the sources of highest sensitivity - several new developments are notable. The constant current d.c. arc is one. Through electronic control, the d.c. arc may be maintained constant within a few percent. Since the drift of an arc as the sample is consumed and the gap widens may amount to 100%, a constant current d.c. arc should result in superior overall accuracy. A commerreal source is not yet available b u t three spectrographers have independently built such units. They are L. E. Owen (10) of Goodyear Atomic Corporation, E. Ziemendorf (unpublished) of the Carborundum Company in Niagara Falls and D. O. Landon (unpublished) of Spex Industries. Incidentally, a bonus obtained with the constant current d.c. arc is potentiometer control of the current instead of the cumbersome motor-driven heavy iron core reactor. The potentiometer permits accurate settings before a run is made and rapid changes m current without overshooting. tReprmted by courtesy of "Analytwal Chemistry "

VOL. 13, N o . 4, 1959

•

87

ERR GAS

I

BODY

Z CAP -SNAPSOFF 5 HOSE CONNECTION 4 SLEEVE-HELD IN ELECTRODE JAWS .5 LOCK NUT 6 STUD-ADJUSTABLE FOR DIFFERENT PRE FORMS 7 BUSHING- USED WITH 3/16 ~ ELECTRODES

Fio. 2. CROss-SEcTION OF THE SVEX STALLWOOD JET Reversing the stud [6] permits the use of either ¼" d. or 3/16" d electrodes. A push-rod ~s used to advance a deep electrode as it burns away O v e r t h e years, there h a v e been m a n y a t t e m p t s to t a m e the d.c. arc t h r o u g h t h e use of e x t e r n a l c o n t r i vances. M a g n e t s h a v e been used b o t h t o c o n s t r a i n the arc c o l u m n a n d to r o t a t e it at a fixed speed. Gas jets, double arcs, a n d c e n t e r - p o s t electrodes are still o t h e r i n n o v a t i o n s aimed at t r a n q u i l i z i n g the easily excitable arc. O f these gadgets, one t h a t " c a u g h t o n " is the Stallwood jet (11), n o w used in dozens of laboratories. W i t h it, an a n n u l a r c u r t a i n of gas is b l o w n u p w a r d s a r o u n d t h e arc c o l u m n to r e s t r a i n arc wander. Compressed air m a y be used b u t , to achieve the added benefit of increased sensit i v i t y a n d suppression of c y a n o g e n bands, m i x t u r e s of o x y g e n and a r g o n (30 t o 5 0 % o.f the f o r m e r ) are r e c o m -

FIG. 3. T H E SPEX ENCLOSED ARC CHAMBER Although pictured w~th the Stallwood jet reside, either accessory may be used alone. The chamber consists of two closely fitting Pyrex@ cyhnders which telescope to permit the operator to change electrodes through an access port. The port ~s sealed when the tuner cyhnder *s turned back to :ts operating position. mended. T h e Spex version of t h e jet is d i a g r a m m e d in F i g u r e 2. F i g u r e 3 shows i t placed in the lower electrode jaws and enclosed to exclude air. I m p r o v e m e n t in a c c u r a c y t h r o u g h the use of the Stallwood jet has been c o n f i r m e d b y a n u m b e r of spectrographers since the original article was w r i t t e n m 1954. J o e n s u u ( 1 2 ) , describing its use in the analysis c:f rare earths o n a r o u t i n e bas,s, cites the a d & t i o n a l a d v a n t a g e t h a t f u m e s are b l o w n o u t c o n t i n u o u s l y so there is less self-absorption. T h e results of the a u t h o r s ' e x p e r i m e n t s are s h o w n in F i g u r e 4 a n d T a b l e IV. T h e s p e c t r o g r a m s

@ ~.~

?

C~ FIG. 4. A , B, AND C ARE THREE SPECTROGRAMS TAKEN UNDER ALMOST iDENTICAL CONDITIONS OF THE SPEX G2

STANDARD CONTAINING 0 . 0 1 ~ OF 43 ELEMENTS The burmng for C was made m an open d.c. arc, in B a Stallwood Jet was used, C the combination Stallwood Jet and Enclosed Arc Chamber Note the enhancement of many hnes and freedom from background, making the elements many times more sensitive. Note, too, the reduction m the C:N.o band m B and its complete ehmmatlon m A

88

APPLIED

TABLE

IV.

COMPARISON

READINGS

Element

WITH

OF

PRECISION

AND WITHOUT

Wavelength

Average Coef

INTENSITY JET

Coef. of Variation, % With Wtthout Stallwood Jet Stallwood Jet

A Copper Tltamum Sliver Cadm*um Tm Zirconium

OF

STALLWOOD

3274 3235 3280 3261 3263 3273

11.5 2.8 2.7 4.6 2.7 3.5

21 0 13.7 13.4 10.0 16 7 20.0

of VanatLon

4 6

15.8

dramatically illustrate the improvement in hne-to-background ratio, especially in the region of the C2N2 bands. Other major advantages of the Stallwood jet are reduction of matrix effects and prevention of selective volatilization of low boiling elements. This is accomplished through the use of deep crater electrodes in which successive layers of sample are burned away while lower layers remain cool. The electrode is placed so that it protrudes by 1-2 m m above the jet housing. As it is consumed in the arc, a rod is used to push it upwards through the hollow stud. Inert gas atmospheres have been the subject of numerous papers and the general opinion seems to be that decided improvement in accuracy and sensitivity may often be achieved. The reader is referred to the article by Thiers and Vallee (13) for detailed information. A recent article by Hammaker and associates (14) illustrates the advantages of using an atmosphere of argon-oxygen in suppressing the cyanogen band intensities. From the viewpoint of accuracy, the result is to remove background almost completely from behind many sensitive spectral lines and so permit their use without having to make inaccurate background corrections. Fry and Schreiber demonstrate (15) how analytical accuracy can be improved when miscellaneous samples of steel are run by the use of an air blast directed at the spark gap. Variations in chemical composition, physical size and shape, and metallurgical history affect the analytical curves less when an air blast is used. Electrodes

An unpublished report by S. R. Wiley illustrates how prec.jsion can be materially improved step-wise by going f r o m ' ¼ " to 3/16" to ~ " diameter electrodes. With his Wadsworth Spectrograph, cylindrical optics were used so that the arc column was focused at the slit in the horizontal plane. Accordingly, wandering of the arc off-axis caused large changes in the intensity at the slit. By resmcting the wander with a narrow electrode, this effect was strikingly reduced. TABLE

V. PHYSICAL

Red Stock Nat. AGKSP Nat.AGKS Nat. SPK Rmgsdorff RW I ek Rmgsdorff RW I extra Rmgsdorff RW IV UCP U-1 UCP U-2 UCP UF4 UCP UF4S UCP Spectrotech

PROPERTIES

Water Absorption % 14.8 18.6 6 9 27 13.3 15.1

14.8

OF GRAPHITE

Hardness (diamond penetratzon)

RODS

Electrwal Reststwtty ohm in.

-6 1 -6 0 9.2 9.0 8.1 4.2 ---

2.7 x 2.6 x 4.8 x 3 1x 3.2 x 3.2 x 3.6x 4 7 x 4.7 x 4 9 x

10 -4 10 -4 10 -4 10 -~ 10 -4 10 -4 10 -4 10 -4 10 -4 10 -4

7.0

4.5 x 10-4

SPECTROSCOPY

Whde discussing electrodes, the subject of physical properties of graphite should be considered. Taking the electrode for granted, the spectrographer does not often question their uniformity from one to the next. Yet W. R. Kennedy (16), American Cast Iron Pipe Co., discovered some surprising differences (Table V). Using eight types of graphite rods of American and foreign manufacture, he sparked the same sample of cast iron nine times with each rod, machining it after every exposure to a hemispherical tip of 30 °. Results for Si and Mn determinations are given m Table VI. Kennedy tried to track down the cause of varying reproducibility from one brand or type of electrode to another. Porosity variations seemed to be a logical offender and so he tried to obtain information on this indirectly by measuring the water absorption after boiling. Although quite wide variations were found (.27-19%), he could not correlate them with spectrographic results. Kennedy also found differences in electrical resistivity and hardness but again could not correlate the results. But he could demonstrate that shifts in working curves and attained precision were directly attributable to the type of electrode employed. TABLE

VI.

EFFECT

OF ON

Rod Stock b RW RW RW UCP UCP UCP UCP Nat. Nat.

GRAPHITE

COUNTER

ELECTRODE

PRECISION a

Coe/flczent of Varlatzon, % e St Mn

I ek (Rmgsdorff) I extra (Rmgsdorff) IV (Rmgsdorff) U-1 U-2 UF4 UF4S AGKSP SPK

3.90 2 02 2.79 2.57 3.2 0 2.71

2.99 3.78 1.50

1.54 2.72 1.74 1.87 2.74 2.42 3.02 2.48 1.90

a M a t e n a l - - Steel Std. N B S 1162 b G r a p h l t e r o d - - m a c h i n e d to 30 ° h e m i s p h e r i c a l ttp CNo. o f d e t e r m i n a t i o n s - - 9 :for each r u n

The shape of an electrode is, as can be judged from the tremendous number of preforms now available, meaningful in attaining high accuracy. Selective volatilization is, for example, very directly related to accuracy. If one were running a sample for magnesium and tungsten in a conventional undercut electrode, magnesium would volatilize out during the first few seconds of arcing, tungsten during the last few. Using such electrodes, the determination of these elements is thus subject to inherent errors. In addition, refractory elements often wind up as beads which drop out of the cup before they are completely consumed. For the determination of many elements of varying volatilization rates, Landon (17) found that a very deep ( ½ " ) crater in a narrow thin-walled electrode produced similar burn-off curves for many elements of varying volatility. As already mentioned, Stallwood (11) used deep, narrow electrodes with his 3et to reduce selective volatilization and, consequently, matrix effects. Weather

One effect of weather variations is on a spark source, and this has already been mentioned. The main effect of temperature variation on the spectrograph is to defocus it to a degree which, though visually undetectable, may be enough to change the line widths and, consequently, their density. Here, temperature gradients are particularly troublesome because, by heating parts unequally, they affect different regions on the plate unequally. The slit and

VOL. 13, No. 4, 1959

89

optical components may also distort to give rise to changes in line density. Apart from the spectrograph proper, temperature variations change the sens*tivity of photographic emulsions and the parameters of electronic components in the source unit and the microphotometer. H u m l & t y affects the sensitivity of photographic emulsions, too, not only by disturbing the formation rate of the latent image but also by absorbing radiation at the shorter wavelengths.

G

La R'

....

Considered a permanent fixture, aligned, tested and sealed by the manufacturer, the optical system is largely ignored by spectrographers. Periodic checking is, nevertheless, recommended to avoid needless loss of precision. Improperly ahgned external and internal optics, poor focusing, spectrographic aberrations and slit eccentricities all may give rise to drifting data.

Q

The purpose of the external optics is to fill the spectrograph aperture with light and so obtain the maximum speed and resolving power. The images should be regularly inspected for vignetting of the source Image by some aperture in the spectrograph (commonly the collimator). This is particularly important if a long slit is used (eg. with a stepped sector or filtei') since vignetting results in uneven illumination along the length of a line, affecting _ arc and spark lines in radically different ways.

,ca

)(

R'

[,

)(

[~

Ri

8

K

FIG. J . S T E I N H E I L ' S " L I N S E N R A S T E R " T h e & a g r a m shows h o w h g h t f r o m a w,de source (a) ~s broken up into an a r r a y o f smaller ~mages which, m t u r n , are all sent t h r o u g h the sht (S) and focussed on the c o l h m a t o r ( K ) . T h e optical system *s valuable m offsetting arc wander.

rective measures were proposed. One is the Lmsenraster, announced by Steinheil (West Germany), a lens system which produces an array of small intermediate images of the source Instead of one as m the conventional three-lens system (Figure 5). As the arc wanders, each small image wanders but to a correspondingly smaller extent. Excellent control of arc wandering is said to be achieved in this fashion although at the expense of some loss of speed. A second system o£ correcting arc wandering is the use of a bi-prism as suggested by L. J. Linder (unpublished) of the Alcoa Research Laboratories in E. St. Louis (Figure 6). An aperture is placed at the intermediate image position so that the image here is focused on the collimator. The apex angle of the bi-prism is so chosen that the image of the source, which would ordinarily fill the aperture in

Other causes of uneven illumination along the length of the lines are chromatic aberration of the optics and dispersion of any prisms in the optical system. (Here, it is interesting to note that Bausch & Lomb has recently changed its hi-prism system of dividing the beam to the gratings of the Dual Grating Spectrograph to a mirror system which eliminates both problems.) In the commonly used three-lens system of illumination, the first lens, used for producing an intermediate image of the source, should be achromatically corrected to obtain uniform line illumination at all wavelengths (18). External optics are plagued with still another annoyance - - arc wandering - - and recently several novel cor-

I

PLAN

K

II'~' II

Optics

Oouble Imaoo -

LI3

Rf L b

I

1

0

O

I

I

8~ o_Az Z~ o

to uJ

Jg

Screen or

~ o

_

-

_

-

E L EVATIOI~P

Counter

PIG.

£lectrode

0I

6. L I N D E R ' S BIPRISM SYSTEM }-,'OR S U P P R E S S I N G ,ARC W A N D E R

Left: Inverted electrode nnages and aperture relationship. R i g h t one moves off-axis, the other returns.

Optical system A double image of the electrodes is focussed on an aperture and as

90

APPLIED SPECTROSCOPY

the screen, is split mto two adjacent images with only half of each image falling in the aperture. When, as a result of arc wandering, one image moves out of the aperture, its twin moves into it from the other side. Light from all parts of the source is thus always entering the spectrograph but the intensity of illumination is, of course, reduced by a factor of two. The Linsenraster has the disadvantage of chromatic aberration, the bi-prism of chromatic d~spersion. As has been mentioned under the heading of '~Weather", improper focusing and changes in sht width disrupt relative line intensities. They also adversely affect the sensitivaty limit of the spectrograph because the maximum lineto-background ratio is reached only when the slit is made narrow enough to allow the natural line breadth to be recorded. Again the necessity of maintaining constant and optimum positioning of the optics is emphasized.

Astigmatism and coma are two more optical ~mperfections which take their toll in accuracy. These are generally greater off-axis: lines at the end of the plate are affected more than at the center. Astigmatism changes the intensity of a line by discarding some of the hght vertically. Coma, on the other hand, widens a line asymmetrically. Accordingly, it is good practice for the spectrographer to maintain the same wavelength coverage on a plate for a given procedure to assure that all aberrations act on a particular line in the same way each time. Photometry Since quantitative spectrochemical analysis depends upon the evaluation of relative line intensities which, an turn, are obtained through photographic density measurements, its accuracy depends upon the methods used for calibrating emulsions and interpreting the data. From the standpoint of accuracy, a calibration procedure using two lines and two steps of a sector or filter ~s the best practical choice. The use of several steps requires a slit too long for uniform illumination. Other methods are hard to set up. With a modern microphotometer, Seidel cordinate paper should yield a reasonably linear graph from about 4-90% T, the lower limit being set by the scattered light in the instrument. The scattered light originates largely from field illumination which, in modern instruments, allows the spectrographer the decided advantage of watching the line as it is being measured. Some designs wall, however, scatter more hght than others. There are two affects of this scatter: 1) the observed gamma will be lower than the true gamma; 2) at very low transmittance values, the H&D characteristic curve will deviate from a straight line. For accurate work, no transmittance readings should be taken below that where the H&D curve is no longer straight. The linearity of a microphotometer is another pertinent variable. By linearity is meant the ability of a macrophotometer to reproduce the transmittance value of a spectral line regardless of its sensitivity setting. A simple test may be performed as follows: Set the background reading in a clear portion of the plate at 100% T and read the transmittances of several lines at levels of about 5%, 30% and 70% T. Next, reduce the overall light to the detector in some manner, by using a dimmer switch often provided on microphotometers or with a neutral filter (uniformly fogged film or smoked glass) taped to the condenser lens. Two or three such filters should be used having transmittances of about 25%, 50% and 75%. Take readings of the

spectral lines 1) without resettmg the background at 100% T; 2) after setting the background at 100%. In all Instances, the relative amount of hght transmitted by the lines should be the same. In practice, of course, variations are obtained when the above procedure as followed. A study of these will set practical limitations of the macrophotometer for the spectrographer. Periodic checks are suggested as electromc components age. Analagous to this electromc check is one which should be performed on the optics, especially w~th regard to focus. On many microphotometers the focus settmg is much more critical than can be detected by the eye. In fact the best way of focusing the instrument is by settmg it to the maximum effective density of a hne: Choose a hne of around 30% T in the middle of a plate. The width of the line should be at least three times the width of the scanning sht. Take several readmgs of the line at different focus settings and pencil a marking on the lens in the position of maximum density (best focus). Repeat for lines near the four edges of the plate. The optimum focus should be the same, provided the carriage moves in a plane perpendicular to the optical components.

Calculations Evaluations of errors due to calculations made at various steps leading to the final % concentration in spectroscopy are, necessarily, indeterminate because so many calculations are graphical and, therefore, subjective. Reading and preparing a graph, the choice of scale, and the thickness and closeness of fit of the line all depend on the skill of the operator. Here a recent trend is quite helpful. It is to convert scales wherever possible, to straight line graphs. Thus, the familiar S-shaped H&D curve appears on many calculating boards today as a straight line through the simple expedient of using a Seidel, or similar, transformation. Incidentally, such a transformatmn corrects only for the toe of the curve. The shoulder - - at low transmittance values - - is usually the result of scattered light in the microphotometer which enters the slit from portions of the spectrogram around the line itself. It is best for the spectrographer to steer clear of this region of high density in his work. Analytical curves are normally straight but background may introduce a curved portion at values where the line and background began to merge. To straighten out this region, background corrections are advisable and several ha~;e been proposed (I, p. 79). Analytical curves are dependent on the presence of a third element and this effect can be minimized through the use of calculations involving ratios of concentrations rather than the calculations themselves. This, too, is explained by ASTM (1, p. 74). Conclusions

After a few months indoctrination, a freshman spectrographer will often wonder why all of his analyses are not perfectly accurate. With all the precautions he is taught to take and with the wonderful, expensive equipment at his disposal, he cannot imagine anything less than perfection. Some of the reasons for loss of accuracy are contained in this report. Some, like "matrix effect" have been largely ignored because they are too involved to be described in generahties. Still others are sure to have escaped us altogether and we shall welcome their bemg brought to our attention. Through such exchanges of information, it is

VOL. 13, No. 4, 1959

91

9. C. Feldman and J. Y. Ellenberg, Anal. Chem. 30, 420 (1958) 10. L. E. Owen, Apphed Spectroscopy 12, 178 (1958) 11. B. J. Stallwood, ]. opt. Soc. Am. 44, 172 (1954) 12. O. Joensuu, Pittsburgh Conference, 1959 (unpublished) 13. R. E. Thiers and B. L. Vallee, Spectrochim. Acta 11, 179 (1957) 14. E. M. Hammaker, G. W. Pope, Y. G. Ishida and W. F. Wagner, Apphed Spectroscopy 12, 161 (1958) 15. D. L. Fry and T. P. Schrelber, Apphed Spectroscopy 11, 1 (1957) 16. W. L. Kennedy, Pzttsburgh Conference, 1959 (unpublished) 17. D. O. Landon, Am. Assoc. Spectrographers Meeting, Chic ago, 1956 (unpublished) 18. K. D. Mielenz, Spectrochim. Acta 10, 99 (1957)

hoped that present accuracy limitations can be continually extended to make the emission spectrograph an increasingly useful tool in industry and science. Literature Cited

1. "Methods for Emission Spectrochemwal Analysis", 2.

3. 4. 5. 6. 7. 8.

American Society for Testing Materials, Philadelphia, Pa., 1957 J. K. Hurwitz, Spectrochim. Acta 9, 3 (1957) Ibzd 12, 211 (1958) J. K. Hurwltz, Applied Spectroscopy 10, 124 (1956) O. G. Clark, Pittsburgh Conference, 1959 (unpublished) W. H. Tingle and C. K. Matocha, Anal. Chem. 30, 494 (1958) A. Strashelm and E. J. Tappere, Apphed Spectroscopy 13, 12 (1959) A. Bardocz, Applied Spectroscopy 10, 183 (1956)

S u b m i t t e d A u g u s t 10, 1959 Y

A Comparison of Precision For Solid, Liquid and Powder Sampling Techniques in The X-Ray Fluorescence Analysis of High Temperature Alloys* Stanley Friedlander and Alan Goldblatt Chicago Spectra Service Laboratory, Inc., Chicago 32, Illinois Abstract D a t a o b t a i n e d m t h e r o u t i n e use o f a m u l t i - c h a n n e l x - r a y s p e c t r o m e t e r are presented to demonstrate precision on stainless steel, nickel

and cobalt base alloys as sohds, liquids and bnquetted powders Coeffic*ents of vananon of 0.5% and less for the major const*tuents in high temperature alloys are readdy attainable. The combmat*on of hqmd and powder techmques is particularly sigmficant m that it prowdes an independent method for standar&zmg complex alloys by use of synthet*c standards. Introduction In 1957, ~t was concluded that x-ray fluorescence instrumentation met the requirements of this laboratory for use as a routine tool. The need for fast, reliable methods for analyzing h~gh temperature alloys was increasing rapidly. There was not only the problem of trying to apply chemxcal and optical emission spectrographic methods to all the new alloys, but the need for a tool to aid in arriving at a representative analysis, since in some cases the highly alloyed samples received exhibxt considerable segregation. The stringent requirement of speed was met with the installation of a multi-channel x-ray spectrometer designed to analyze seven elements simultaneously in one minute. This instrument also provided a sampling advantage, because xt analyzed a circular area nearly one inch in diameter. If the equipment could also provide routinely the precision necessary for analyzing high concentrations of elements, then only the problem of accurate standardization would remain. The need for large numbers of well analyzed solid specimens to establish a reliable calibration emphasizes the ~Presented at the "Pzttsburgh Confere~tce on Analytwal Chemistry and Apphed Spectroscopy", Pittsburgh, Pa, March 1959

empirical nature of x-ray spectrographic techniques. The enhancement or absorption effect of one alloying element can materially influence the analysis of another. Frequently it is necessary to use a set of standards in which only one element is varied at a time. To prepare such standards as sohd speciments is time consuming and requires extensive chemical analysis. It has been shown by many workrs (I) that ddution wall reduce or completely eliminate absorption and enhancement effects. Davis and Clark (2) reported excellent agreement with chemical analyses using oxides and graphite mixtures for analyzing various mckel alloys. Others (3, 4, 5) have reported on solution and powder techniques using synthetic standards. Thus, if adequate precision could be obtained with solution and powders, then these techniques hold much promise. They could serve both as a rapid method for directly analyzing metal samples regardless of form and as a prim.ary method of standardization of solid samples. Experimental The work described in this paper undertook to evaluate the precision attainable in the routine analysis of solid, liquid and briquetted powder samples. To evaluate precision, three representative sohd samples were selected:

1. National Bureau of Standards Sample #D846, a 300 series stainless steel, which will be identified as Sample S. 2. A mckel base alloy of the Monel type, identified as

Sample N. 3. A cobalt base alloy of the Stellite type, identified as

Sample C.