V O L U M E 2 4 , NO. 5, M A Y 1 9 5 2
807
Above 2.00% another graph must be drawn, using either a smaller sample or smaller aliquots from the dissolved siliron solution already in use.
Table 111. Data for Silicon Curve (0.2500-gram sample a n d 20-ml. aliquots)
NBS
T,
Standard Cast Irone
405 mp
I
Si Value,
ACKNOWLEDGMENT
%
The author wishes to thank his associates in the laboratory, who assisted in checking results during the pears the method was being tested. LITERATURE CITED
(1) American Foundrymen’s Association, Chicago, “Handbook of Cuoola Ooeration.” 1946.
* 0.1250 gram of each for a determination.
(2) Bolts; D.
F ,with Aiellon. If.G.. IND.EKG.CREM.,ANAL. ED..
19,876 (1947).
containing over about 0.50% combined carbon, required extra time and extra potassium nitrate for solution. Irons common in the United States do not usually contain over 0.90% phosphorus, and up to this value the final acidity suitably euppresses the phosphomolybdate colored complex. Manganese up to the values shown in Table I1 does not appear to have appreciable effect on the silicon values. Ferric iron, which has appreciable transmittance in the region of 405 mp, is cancelled out by the use of the solution, which has not been treated with molybdate, to set the transmittance of the instrument to 100%. The relatively dark background of the silicomolybdate complex may be somewhat objectionable, but to attempt to get rid of it is time-consuming and is not necessary for the average cast iron foundry. For eutectivity determinations silicon has only about one third the effect of carbon, as indicated by the formula, E, = % T.C. 0.3 (% Si % P), which is approximate and used extensively in one form or another in the industry (1). For total carbon the accepted accuracy in the foundry industry is 1 0 . 0 3 or 0.04%, depending upon the rarbon content; this would give a considerable leeway for the same effect from silicon, if the fonnula were applied. Actually, the accuracy for silicon is urually regarded as about the same as for carbon. I t is not considered advisable to try to run by this method silirons that are much above 3.00%, as the curve begins to flatten out above 2.00%, so that the smallest differences in transmittance represent too great a difference in silicon.
+
+
(3) Clausen, D. F.,and Roussopoulas. H. D.. Aitachem ~ V P W S 6,. 41 (1946). J . SOC.Chem. Ind., (4) Gentry, C. H. R., and Sherrington, I,. 0.. 65,90 (1946). ( 5 ) Haywood, F. W., and Wood, A. A . R., “Mrtallurgical Analysis
by Means of the Spekker Phot,oelectric Absorpt,iometer,” London, Adam Hilger, 1944. (6) Hill, 5’. T.,ANAL.CHEM.,21, 589 (1949). ( 7 ) Hilson, H. D., I N D . END.CHEM.,ANAL. ED.,16, 560 (1944). Little, J., J . SOC.Chem. Ind., 64, 118 (1945). 1 Lord and Demorest, “Metallurgical Analysis.” New Tork, JIoCraw-Hill Book Co., 1924. ’ Lundell, Hoffman, and Bright, “Chemical Analysis of It,on and Steel.” New York. John Wiles & Sons. 1931. Malov, S. I., Yakovlev, P. Ya., and Eliseev. A . A , . Zauodskaya 1
Lab., 5, 665 (1936).
Marshall, H., Chem. Sews, 83, i6 (1901). Pinsl, H., Arch. Eisenht’iltenw. 8, 97 (1934) Ibid., 9,223 (1935). Ibid., 10,139 (1936).
Ridsdale, K. D.,Ridsdale & C’o., Middlesbrough. England Analoid Descriotive Booklet 336 (Januarv 1948). Rozental, D., an; Campbell, H.. I m EN;. CHEM.,ANAI E D , 17,222 (1945).
Schwarte, M.C., and Morris, 1,. W., I h i d . , 15,20 (1943). Thayer, L..I.,Ibid., 2, 276 (1930). Walters, H. E., Chem. .Veu.s, 84,239 (1901). Weihrich, R.,and Schwarts. W. Arch. Eisenhlittenw., 14, ,501 (19311. RECEIVED for review March 28, 1951. Sccepted January 2 5 , 1952. Presented a t Pittsburgh conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa.,March 7, 1951.
CaIcuIator for SpectrochemicaI AnaIysis Application of Seidel Function of Density in Its Development GEORGE OPLINGER
Solvay Process Division, ‘4llied Chemical 61: Dye Corp., Syraciise: N. Y .
T
HE use of calculators in modern spectrochemical analysis has been demonstrated as a practical necessity in regard to both the saving of time and the reduction of error. Of the many types described (2-6, 10-12, 14, 16-20, 83,2 4 ) , no one calculator was found to provide all the features desired to correct for the variables encountered in spectrochemical analysis. This paper describes a calculator that has proved highly satisfactory in reRolving variables introduced by the characteristics of the photographic emulsion, while still permitting adjustments to correct for the recognized shifts of the analytical curve The successful application of any calculator depends fundamentally on reliable calibration of the photographic emulsion, the essential prerequisite for all photographic intensity measurements. A study of the various means of accomplishing a precise calibration, as summarized by Churchill (6) and paralleled in the author’s work, resulted in the choice of the two-line method using pairs of iron lines selected from those classified by Dieke and Crosswhite (7-9) In applying Crosswhite’s ( 7 ) and Dieke’s
(8) data oil the intensity relationships of various lines in the iron spectrum, the Seidel (13) function of density, log ( I o / I - I ) , hereafter termed the Seidel density, was employed in expressing spectral line blackening. This function was first suggested by Baker (1). 10represents the incident light being used to measure the emulsion and I is the light transmitted by the spectral line being measured. This function of density, explored by Srhmidt (%’I), makes possible a much more precise method of establishing the characteristic curve of the photographic emulsion. His experiments showed that the plot on rectangular coordinates of the Seidel density for the stronger line against the Seidel density for the Feaker line was linear for all emulsions tested in the density range 0.1 to 2.0, and beloa 3200 A. It was desirable to use this method in extending the wave-length region to 4600 A As no data were given for any emulsion above 3200 A. and the emulsions tested were not listed by Schmidt, an experimental study of Eastnian Spectrum Analysis S o . 1 and Type 103-0 was undertaken. For the No 1 emulsion, the wavelength re-
808
A N A L Y T I C A L CHEMISTRY
gions 3100 to 3300 A., 3350 to 3450 A., 3900 to 4100 A., and 4200 to 4300 A. were explored; for the 103-0 emulsion, the 3100 to 3300 A. and 4500 to 4600 A. regions were of interest. The calculators, one in the plant laboratory and one in the research laboratory, have been in use for over three years. During that time no evidence of any significant departure from symmetry has been apparent for the emulsions below 4100 A. An estimated saving of 75% of the time spent in calculating analytical results has been realized, as well as a reduction in calculating errors. In addition, control of the response of each photographic plate or film is provided.
Table I.
Emulsion
$A No. 1
Data Concerning Iron Line Pairs, Slope, and Intercept of Seidel Preliminary Curves Length 3178.01 3175.44 3286.76 3306.36 3413,14 3399,34 4021.87 3951.16 4219.36 4247.43
103-0 plates
3178.01 3175.44
103-Ofilms
3178.01 3175.44
103-0 plates
4592.65
4547.85 4611 28
4625.05
Log I 2 . 08ga 1.968' 2,255s 2.164b 2,199b 2.036s 2 . 046a 1.902a 2.144a 2.01Sn 2.089a 1.968a 2 08ga 1.968" 1. 524a 1.368O 1.375O 1.2020
ALogl
Elope
Intercept
Coefficient of Correlation
0.121
1.001
0.325
0.999-J-
0.091
1.005
0.250
0.999f
0,163
1,001
0.456
0.999-t
0.144
1.004
0.541
0.999f
0.125
1.055
0.540
0.997
0,121
1.001
0.275
0.995
0.121
0.994
0.220
0.994
0.156
1.026
0.286
0.996
0,173
1.026
0.336
0,998
Crosswhite (7). b.Dieke (8).
EQIJIPMENT AYD EMULSION PROCESSING
Spectrograph. Jarrell-Bsh 21-foot grating, Wadsworth monnting. Bausch and Lomb, medium Density Comparator. Jarrell- A ~ ~ ~ ~ r e c o r d i n g . Excitation. ARL alternating current arc unit. 2.5 kv., 3.0 amperes, pure iron electrodes, 0.25 inch diameter. Emulsions. Eastman Spectrum Analysis No. 1. Eastmnn Spectroscopic Type 103-0. Photographic Processing. Developer D.19 (Eastman Kodak). Acetic acid short stoD. Kodak rapid fixer. Four-minute development in rocking h a y at 20" C.EXPERIMENTAL
Iron arc spectra were photographed on film and plates in a wide range of intensity levels employing neutral screens of various transmittances. Exposure times were maintained at values commensurate with those experienced in normal analytical practice. The spectral lines chosen from Crosswhite's ( 7 )and Dieke's (8) data are given in Table I, the choice being based on practical considerations of log I (intensity) ratios and density of lines in the spectra. Statistical study showed that the line pairs were reproducible from exposure to exposure. For example, the stand3178 ard deviation of the line pair Fe log I ratio 0.121, was 3175 0.006 as determined from a series of 28 exposures on a single plate. A second plate showed a similar standard deviation, but on comparison of two plates, the standard deviation rose to 0.015. Spectral line blackness was measured as a galvanometer deflection (on a scale 0 to 100) for each line of the selected pairs of iron lines The density comparator in this case is adjusted to read zero for full light intensity (clear plate) and 100 for maximum density. Each of the galvanometer readings was converted to the Seidel density. This value may be calculated, but in this instance a curve relating galvsriometer deflection to Seidel density was found to be more convenient for the conversion
A calculator for spectrochemical analysis was desired to simplify the tedious photographic emulsion calibration procedure and to provide a means of obtaining analytical data directly. A calculator was developed that makes use of a fixed photographic emulsion calibration scale, with provision for operation at any contrast within the limits of the emulsion for the determination of analytical data. It has been demonstrated experimentally in the case of two photographic emulsions that their response in terms of the Seidel function of density, d, = log
($- 1). can be expressed as a straight-line
curve with a slope of unity between certain wavelength limits. The intercepts of the curves for a particular emulsion are determined by the wavelength response; they are also determined by the emulsion. These findings provide the basis for the type of calculator described and suggest possibilities in future attempts to relate the response of an emulsion at one wave length to the response at a different wave length.
In plotting on rectangular coordinates the Seidel density of the stronger line against the Seidel density of the weaker line for each line pair in the several spectral regions for both emulsions over the density range 0.1 to 2.0, straight-line Seidel preliminary curves were obtained. The method of least squares was applied in each case to check the position and slope of each curve. For all line pairs the curves were in excellent agreement. The coefficient of correlation was better than 0.999 for SA No. 1 emulsions and 0.995 for 103-0 emulsions below 4100 A. By applying the least squares method, the slope for each curve was found to be approximately unity for the SA No. 1 and 103-0 emulsions below 4100 A Above 4100 A. the slopes increased to l.0e55for the SA S o . 1 emulsion at 4200 A. and to 1.026 for the 103-0 emulsion a t 4600 A. Several of the Seidel preliminary curves are shown in Figure 1. In cases where the slope approaches unity, the value of one was assumed. Experience has indicated that a departure of &0.005 from unit slope can be tolerated. The data are summarized in Table I. Characteristic curves were then derived from each Seidel preliminary curve, The procedure for this involves taking an arbitrary minimum Seidel density value on the abscissa, - 1.900, for example, from curve C in Figure 1, assigning it to the w a k e r line and reading the corresponding Seidel density value, - 1.680, for the stronger line on the ordinate. These values are converted to galvanometer deflections by calculation or by referenre to the conversion curve. The values obtained, 1.2 and 2 0. respec-
Table 11. Typical Data for Plotting Characteristic Curve (Deri\ed f r o m Seidel preliminary curve C of Figure 1)
-J
I,og
(:
-
1)
- 1.900 - 1.680
-1.460 -1.240 - 1.020 -0,800
-0,580 -0.360 -0.110 f0.080 co.300 +0.520
1-0.740
+ 0 960
+i.iso
~a~vano~iieter Deflection 1 2 2.0 3 3 5.1 8.6
13.: 20. ' 30.3 42.0 54.6 66.6 76.8 84.7 90.1 93.8
Log I 0 0,121 0.242 0,363 0.481 0.605 0.726 0.847 0.968 1.089 1.210 1.331 1.452 1 . 573 1.691
809
V O L U M E 2 4 , NO. 5, M A Y 1 9 5 2
tively, are plotted on rectangular coordinates against an arbitrary log Z value of zero for the weaker line and zero plus the difference in log I of the two lines, 0 121, as in Figure 2. The second Seidel density value, -1.680, is then assigned as the weaker line on the abscissa and the corresponding value for the stronger line, -1.460, is read on the ordinate. After conversion to galvanometer deflection, 3.3, it is plotted against 0 plus 0.121 plus 0.121 or 0.242. The procedure for obtaining successive Seidel density values is continued for the entire span of the Seidel preliminary curve to the point of maximum expected density. After conversion to galvanometer deflections, they are plotted against successive increments of the log I difference for the two lines, Values for drawing the characteristic curve shown in Figure 2 are given in Table 11. Churchill (6) describes this procedure in reference to the use of a preliminary curve plot in the two-line calibration method. The characteristic curves derived from the Seidel preliminary curves whose slopes were unity were symmetrical, as would be Q expected. In these cases the lower part or “toe” of the charac-i teristic curve has the same shape as the upper part, or ‘[shoulder.” The inflection point is a t the center of rotation. A rotation of the LOG ( f p - I ) F O R WEAKER LINE upper part of the curve through 180” with the center of rotation Figure 1. Seidel Preliminary Curves fixed results in the upper part coinciding with the lower part of the curve. These results are in agreement with Silberstein (22) 4219.36 A. SA-1 emulsion, Fe 4247.43 in regard to the symmetry of unsensitized emulsions. Charac3413.14 teristic curves derived from Seidel preliminary curves whose B . SA-1 emulsion, Fe 3399.34 slopes departed from unity were not symmetrical. The toe of the 3178.01 C. 103-0 emulsion, Fe curve was extended over a greater range than the shoulder. I t is 3115.44 bevond the scope of this paper _ . to postulate on the departure from symmetry of the characteristic curves above 4100 A. Table 111. Data for Plotting Characteristic Curves for Iron Line Pair The values obtained above may be obtained 3178.01 a-ithout recourse to the Seidel preliminary curve, 3175.44 providing their respective slopes are unity. By 103-0 Film, SA 1 Plates, 103-0 Plates definition, successive Seidel density values, repreIntercept 0.2?5 Intercept 0.220 Intercept 0.325 L ~ ~ ( -$ 1) falv. L , ~ ~ (-+1) Galv. senting integral increments of the log Z difference ALog I = 0.121, L , ~ ~ ( $ ! - 1) Log r efl. defl. of the respective line pair, increase by the value 0 -1.900 1.2 -1.900 1.2 -1.900 1.2 of the intercept. These successive values, after 2.3 -1.680 2.0 0.121 - 1.575 2.6 -1.625 0.242 -1.250 5.3 -1.350 4.3 -1.460 3.3 conversion to galvanometer deflections, are then -0.925 10.5 -1.075 7.7 -1.240 5.4 0.363 plotted as above. A tabulation of values thua 0.484 -0,600 20.0 -0,800 13.5 -1,020 8.6 22.9 -0.800 13.5 -0,275 34.7 -0,525 0.605 obtained for all the line pairs and emulsiona 36.0 -0.580 20.7 +0.050 52.8 -0.260 0.726 70.3 +0.025 0.847 +0.375 51.4 -0.360 30.3 where the Seidel preliminary curve slope is unity 0.968 +0.700 83.4 +0.300 66.6 -0,140 42.0 1.089 +1.025 91.4 +0.575 79.0 i-0.080 54.6 is given in Tables I11 and IV. In the tabula1.210 -1,350 95.7 +0.850 87.7 f0.300 66.6 tion, the initial log I = 0 for each case. For 1.331 +1.675 97.9 +1.125 93.0 +0.520 76.8 1,452 +1.400 96.2 $0.740 84.7 the Seidel preliminary curves departing from 1.573 +1.675 97.9 +0.960 90.1 unity, these values also may be obtained by 1.694 +1.180 93.8 1.815 i-1.400 96.2 calculation. Using the equation for a straight 1.936 +1.620 97.8 line, y = mx b, successive values of r and y are determined and related after conversion to galvanometer deTable IV. Data for Plotting Characteristic Curves for Additional Iron Line flections to the log I increments of the Pairs line pair as above. If instead of taking a m i n i m u m density for the weaker line in deriving values for the svmmetrical characteris~
+
tic curve, a value of log 0 0.091 0.182 0.273 0.364 0.455 0.546 0.637 0.728 0.819 0.910 1.001 1.092 1.183 1.274
-
-1.900 -1,650 -1.400 -1,150 -0,900 -0,650 -0,400 -0.150 +0.100 f0.350 +0.600 +0.850 +1.100 +1.350 f1.600
1.2 2.2 3.8 6.6 11.2 18.2 28.4 41.4 55.7 69.1 80.0 87.7 92.6 95.7 97.5
0 0.160 0.320 0.480 0.640 0.800 0.960 1.120 1.280 0.080
0.240 0.400 0.560 0.720 0.880 1.040 1.200
-1.900 -1.444 -0.988 -0.532 -0.076 f0.380 f0.836 41.292 +1.748 -1.672
-1.216 -0.760 -0.304 +0.152 +0.608 f1.064 +1.520
1.2 3.4 9.2 22.5 45.7 70.5 87.3 95.2 98.3 2.1 5.7 14.7 33.2 58.7 80.2 92.0 97.1
0 0.144 0.288 0.432 0.576 0.720 0.864 1.008 0.072 0.216 0.360 0.504 0.648 0.792 0,936
-1.900 -1,360 -0.820 -0.280 +0.260 +0.800
4-1.340 +1.880 -1.630 -1.090 -0.550 -0.010 $0.530 +1.070 +1.610
1.2 4.2 13.1 34.3 64.5 86.4 95.6 98.8 2.3 7.4 21.8 49.4 77.3 92.1 97.6
(p -
I)
p:
0
(equivalent to a galvanometer deflection of 50) was assigned and successive values were then taken in both directions on the Seidel preliminary curves, the characteristic curves for various gamma values would cross symmetrically a t a common point as shown in Figure 3. This illustrates a phase of the work of Levy (16) and provides the basis for the construction of a calculator.
ANALYTICAL CHEMISTRY
810
Provision for incorporating the cases where the characteristic curve is not symmetrical in the calculator is described below. An interesting phase of the Seidel function of density is that when Seidel density values are used in place of galvanometer deflections when plotting the characteristic curve, a straight-line curve results, providing of course that the Seidel preliminary curve slope is unity. This fact has intriguing possibilities. Had the galvanometer scale been engraved with Seidel density values instead of linear values, this phase would have been explored further. It is hoped that t,his can be done in the future.
shown by the several dashed lines in Figure 4. In transferring this to a basis suitable for constructing the calculator, the radial lines passing through the various scale divisiws were extended a short distance above the low gamma scale, so that a somewhat extended scale could be drawn parallel to the original scales. This extended scale becomes the fixed calibration scale of the calculator through which all galvanometer deflections are referred.
DEVELOPMENT OF CALCULATOR
Application to Symmetrical Characteristic Curves. The idea in developing the calculator was t o provide a means of locating the response of the photographic emulsion in respect to an established scale, and to make any corrections for recognized shifts of the analytical curve prior to reading off analytical results. The second condition is easily satisfied by the familiar slide rule application to analytical scales. In regard to the first condition, a procedure for correlating mechanically the various characteristic curves is involved.
80 'Ot
z70 -
51 c W 0
-
-160
II. W
n
K
E50
Y 0
z 9 0
-
Figure 3.
Characteristic Curves
Drawn through same inflection point
-
J
3
I
0.6
Figure 2.
I
I
LO
LOG INTENSITY
I
I
I
4
14
Typical Characteristic Curve
Derived from Seidel preliminary curve C of Figure 1
The following developnieiit will be confined to the cases where the characteristic curves are symmetrical. The lineal scales representing the lowest and highest gamma curves, prepared as described by Churchill ( 6 ) , are placed on a drawing board so that the lowest gamma scale is approximately 5 inches above and parallel to the highest gamma scale, and a line joining their midpoints i8 a t right angles to either scale, as in Figure 4. Each scale division (1.5 through 98.5) of the two scales can then be joined, with the result that the extended lines intersect a t a common focus below the high gamma scale. Thus in the trapezoidal figure, ABCD, formed between the two scales, the characteristic curve scale representing any gamma value within the limits of the two scales can be located, the lines radiating from the common focus passing through like scale values of each as
A full size photoengraving was then made, reproducing the fixed scale and common focus point. The photoengraved plate waa 20.5 inches wide and 15.5 inches high. Later, another photoengraving was made for a second calculator reproducing only the fixed scale. The common focus point in tdis case was located by measurement when the calculator was fabricated. At the common focus point, a Lucite arm with a hair line inscribed along its length is pivoted. Rotation of the arm carries the upper end across the fixed scale of the calculator. The hair line permits the location of any given scale value in the area below the fixed scale. A double slide rule is mounted in the area below the fixed scale parallel to and ea able of vertical movement in respect to the fixed scale. The doube slide rule is constructed from two Dietsgen No. 1767P National IO-inch wooden slide rules. The body of one is mounted a t right angles on two vertical strips which slide in ways underneath the fixed scale. The body of the second is glued to the slide of the first (or lower) slide rule to permit horizontal motion in respect to the lower slide. A setscrew holds the slide rule in position after the gamma or contrast adjustment has been made. A vertical scale beneath the slide rule is used in evaluating relative contrast during the adjustment. The double slide rule provides the horizontal motion necessary in making the initial contrast adjustment, the subsequent analytical scale correction, and the ultimate determination of anal ical results Removable slides are provided for the top side rule which carry the various analytical scales for use. A slide bearing a linear log Z scale is also provided. The completed calculator is shown in Figure 5.
v
Application to Nonsymmetrical Characteristic Curves. Since in the 4250 A. region for Spectrum Analysis No. 1emulsion and in the 4500 to 4600 A. region for 103-0 emulsion a departure from unity of the Seidel density curve occurred, the calculator aa originally designed was not effeetive. The method adopted to
V O L U M E 24, NO. 5, M A Y 1 9 5 2
811
correct for this departure involves rotating the calibration scale through a small angle proportional to the slope of the Seidel density curve. The point of rotation was determined empirically. This rotation distorts the calibration scale to the degree necessary to provide a representative calibration. Experience has shown that log I ratios obtained by this distortion vary no more than 5 % from their true value. This factor is not serious, since an* lytical curves are determined on the same distorted scale. A cornpasison of a series of log I ratios obtained from the distorted calibrat:-the t N e calibration Table V. Adjustment for Us contrast adjustmeni line pairs employed oslibration are used. spectra are included plate 01film for this --e1-
.
--_I
Figure 5.
After measuring the blackness at the several iron line pairs OD the density comparator, set the arm of the calculator on the cali-
Table V.
Comparison of Log I Ratios from Emulsion Curve and Distorted Calculator Scale Cr. P.P.M. 4.0
8.0
18.0
42.0 120.0 400.0
Completed Calculator
hrittion scale a t the blackness of the weaker of the two lines of a pair. Then move the log I scale of the slide rule so that the zero mark is under the hair line. Now move the arm to read the hlackne88 of the heavier line on the calibration scale. Note the log I value on the log I soale where the hair line crosses. If the v&lue obtained is leas than the given value of the log I ratio of the two lines, the slide rule must be moved vertically upward; if greater, the slide rule must he moved vertically downward. Repeat the adjustment by trial and error until the spacing on the lag I scale corresponds to the given ratio. Note the position of the slide rule in respect to the vertical linear scale. Repeat this adjustr ment far several pairs of lines from other spectra, noting the sliderule vertical position. Average these values and set the vertical position of the slide rule a t this averaged value. The setscrew then holds the slide rule in thia position, which ia the contrast response of the particular emulsion within the particular wave length region. Log I ratios of lines in the analytical spectra.may now he determined, using the log I scale. When using analytical scales in the calculator, correction for any Bhift of the anslytical curve can now be made. The spectra of one or two standard samples are included on each analytical plate or film for this purpose. Having read the hlaokness of the spectral lines of interest, the appropriate anslytical scale is inserted in the slide rule. Move the arm of the calculator so that the hair line reads the blackness of the spectral line of the test element in the standard on the calibration scale. Now slide the analytical scale so that the mncentrationaf the test element in the atandurd is under the hair line. Move the arm to intersect the cdihration scale B at the blackness of the internal standssd line. The point where the hair line intersects the anslyt,ical scale is the corrected index of the analytical curve. Averaged data from duplicate or triplicate spectra. are normally used for this adjustment. The analytical scale is then moved in respect t o the Elide rule so that the corrected index valueis alimed with the ~ e r oindex of the slide " rule. The scale is now ready for use in determining analytical results. ACKNOWLEDGMENT
Figure 4.
Layout for Calculator
The author wish= to express his appreciation to Marianne Divers and Lucille R. Smith for assistance in the
812
A N A L Y T I C A L CHEMISTRY
experimental work, Fred Bruton for the fabrication of the calculator, and Howard A. Bewick for helpful criticism in the preparation of the manuscript.
(11) Henderson-Hamilton, J. C., and Laurie, A . J., J . Sac. Chem.
I n d . (London), 64, 309-12 (1945). (12) Hughes, H. K., and hfurphy, R. R., J . Optical SOC.Am., 39, 501 (1949). (13) Kaiser, H., Spectrochirn. Acta, 2, 1-17 (1941). (14) King, Carl, J . Optical Soc. Am., 32, 112 (1942). (15) Levy, Saul, Ibid., 34,447 (1944). (16) Muller, R. H., IND.ENG.CHEM.,ANAL.ED.,13, GG7 (1941). (17) op1inger,~ . , I b i d . , 19,444 (1947). Ibid., 10,664 (1938). (18) Owens, J. S., (19) Owens, J. S., “Proceedings of Fifth Summer Conference on Spectroscopy and Its Applications,” h’ew Tork, p. 17, John Wiley & Sons, 1938. (20) Sampson, -4. M., Ibid , p. 8. (21) Schmidt, R.7 Ret. truu. ch@tn.,679 737 (1948). (22) Silberstein, Ludwik, J . Optical SOC.Am., 32, 474 (1942). . ~ 34, 689 (1944). (23) Sinclair, D. ~ 4Ibid., (24) Vanselow, A. P., and Liebig, G . F., Jr., Ibid., 34, 219 (1944).
LITERATURE CITED (1) Baker, E. A., Proc. Rw.SOC.Edinburgh, 45, 166 (1925). (2) Breckpot, R., Spectrochim. Acta, 1, 137 (1939). (3) Brommelle, N. s., and Clayton, H. R., J . Soc. Chem. I d . (London),63,83-9 (1944). (4) Carlason, C. G., Jernkontorets Ann., 132, 467-84 (1948). (5) Chong, E. Y., Metallw’rtschaft, 22, 562 (Dec. 20, 1944). (6) Churchill, J. R., IND. EXG.CHEM.,ANAL.ED., 16, 662-70 (1944). (7) Crosswhite,H. H., Spectrochim. Acta, 4, 122 (1950). (8) Dieke, G. H., War Production Board, Office of Research and Development, Rept. E-54 (Sept. 24, 1943). (9) Dieke, G. H., and Crosswhite, H. H., J . Optical SOC.Am., 33, 425 (1943). (10) Hale, C., Eighth Pittsburgh Conference on Applied Spectroscopy, 1947.
RECEIVED for review August 20, 1931.
Accepted February 6, 1962.
Physical Properties of Natural and Synthetic Rubber Materials at low Temperatures J . Z. LICHTMAN
AND
C. K. CIIATTEN
Material Laboratory, New Y o r k Nazal Shipyard, Brooklyn 1 ,
Although a ,number of elastomers resistant to low temperatures have recently been developed, relatively little work has been done in standardizing the instruments and procedures used in evaluating their low temperature properties. Accordingly, a survey of all such equipment and methods was undertaken in order to standardize appropriate ones and to determine the degree of correlation in data obtained in the tests. A comparison of data derived in evaluating typical stocks exposed at low temperatures and tests using a torsion wire ap-
I
N T H E past few years both government and industrial rubber laboratories have been actively engaged in developing elastomer materials that will function properly under low temperature service conditions. Extensive research programs carried out by these laboratories have resulted in the formulation of a considerable number of low temperature-resistant stocks such as those used in hose, rubberized fabrics, and gaskets. At the same time many different types of apparatus and procedures have been devised for evaluating the low temperature properties of the stocks. Each of these teste was, naturally, favored by the organization responsible for its development. This condition resulted in the adoption of a large variety of testing instruments, many of which were designed for use in evaluating the same basic properties, although by slightly different means. I n view of the importance of standardization of evaluation procedures to be incorporated in procurement specifications, the Bureau of Ships, Department of the Navy, authorized the Material Laboratory, New York Naval Shipyard, to investigate all known devices and procedures used in evaluating the low temperature properties of rubber materials and, if considered desirable, to modify or revise appropriate ones The over-all objective of the program was to select and standardize equipment and procedures considered to be the most suitable for incorporation in military procurement specifications. An analysis of the manner of employing the major number of rubber items in shipboard low
N. Y
paratus and a hardness indentation device confirms the existence of a mathematical relationship between flexural modulus and hardness indentation. The iovestigations indicate the feasibility of using either instrument to evaluate the stiffness or the hardness properties of elastomers over a range of exposure conditions. The small T-50 type specimens used with the torsion wire apparatus are well adapted for use in evaluating changes in stiffness of elastomers due to exposing the materials to solvents.
temperature service 4iows the following physical properties to be the most significant: (1) flexibility, or the magnitude of stress required to produce an observed degree of deformation; (2) compression set, including the rate and amount of dimensional recovery of a material after being held under constant deformation; and (3) brittleness or structural failure of a material under rapid deformation. The significance of these properties may vary from one application to another, each specification thus requiring the evaluation of the property or properties that are pertinent to the service performance of the material. REVIEW OF TEST PROCEDURES
A review was made of the methods used in evaluating the three classes of properties. The second property, compression set, is usually evaluated after the specimens have been aged a t elevated temperatures under constant deflection (8). A basically similar method has been adopted and standardized by the Navy for carrying out low temperature evaluations, a modification being made in that the dimensional recovery of the specimen is evaluated a t the test temperature and a t two time intervals-Le., 10 seconds and 30 minutes after the specimen is released from the clamping plates, In this manner both the rate and amount of recovery of the specimen are evaluated and the data so obtained can be used in differentiating between first- and second-order transition effects. The evaluation of the third property, brittleness, has like-
