Figure 1 1 . Influence of asymmetry of square wave Solution 3.75 X 1 0-4M CdCI? per 0 . 2 N KCI. 3
3
'
AF.100 mv. Frequency 500 cycles per second. E varies from right to left from 0.5 to 1.0 volt. Upper curve corresponds to square wave of Figure 8,A. tower curve to Figure 8,B
3
has to be corrected for the voltage drop across the transistors. t-sing transistors, with a loner resistance in a conducting condition (such as GE Type ZKl88A) will reduce this correction. The current losses b y the off resistance of the transistors may be too large in absolute measurements and in determinations where the currents are very small (belon 1 l a . ) , as in trace analysis. By uqiiig silicon transistors with a much higher off resistance, this difficulty may bc overcome. RESULTS
!
-1'
.
n
1
-
Varying the frequency \\ill vary the polarogranis obtained. As shoun in Figure 9, the oscillations caused by the dropping mercury electrode become more pronounced a t higher frequencies. The peak height is smaller a t higher and a t very lon frequencies. A frequency of about 20 cycles per second is the best compromise. Although a slightly higher frequency may be chosen. the influence
of the dropping electrode may 1,cconie too large, requiring a better ciainpirig of the recording system (a 2-second pen ,speed IO-mv. Brown recorder). The influence of the frequenct. on the peak height is given in Figure 10. For cadmium the peak height is faid>- constant between 30 and 70 cycles per second. For zinc this holds true between 15 and 80 cycles pcr second. K h e n t'he symmetry of the square wave is changed! the polarogram of Figure 11 are obtained. If the square wave is very asymmetric and the frequency is high (500 cycles per second) the duration of t'he shorter part of the whole cycle is so sniall that the reaction cannot take place and the i-E purl-e becomes similar to that in classic polarography. Consequently the tn.0 types of curve of Figure 11 are not obtained a t low frequencies. ACKNOWLEDGMENT
The author thanks Bot) Tuning for performing the measurcmeiits necessary. LITERATURE CITED
(1) Ishibashi, lI., Fujinaga, T., Bull. Chena. SOC.J a p a n 2 5 , 68 (193.)). ( 2 ) Krugers, J., Chem. W e e k h l a d 53, 672 (1957). RECEIVEDfor review LIay 28, 1038. A4cceptedOctober 14, 1958.
Porous Cup Technique in the Determination of Magnesium in Cast Iron A Statistical Study A. C. OTTOLlNl General Motors Corp., Defroif, Mich.
,A rapid and accurate solution method has been developed for the spectrochemical determination of magnesium in cast iron in the concentration A thorough range 0.01 to 0.16%. statistical study was carried out to evaluate instrumental errors, chemical preparation errors, and day-to-day reproducibility. The standard deviation of the method is 1.8370 ai the 0.075% concentration level; the accuracy is of the same order. Porous cup electrodes are used to introduce the sample into the excitation gap. Synthetically prepared standards facilitate calibration. The method has replaced chemical methods for magnesium in this laboratory.
R
a program was initiated for the routine control of magnesium in nodular cast iron. The requirements were that the method be rapid and capable of chemical accuracy. The best available chemical methods had a standard deviation of the order of 6.0%. They required about 2 days of elapsed time for analysis and were complex. Bryan, Sahstoll. and T'eldhuis ( 3 ) reported a spectrochemical solution method in 1949. Their reported average errors of &lo% or lcss indicated that i t might meet requirements. Prior to this program, three techniques for introducing solutions into the analytical gap had been studied in this ECENTLP
laboratory. Evaporation of the solution on the surface of a preheated electrode ( 7 ) had been evaluated for the analysis of steels. ;1 rotating disk electrode (6) had been found successful for the analysis of both new and used lubricating oils. Porous cup electrodes (5) had been used for the anal!-sis of steels, netv oils. lead-base alloys, and for the determination of hisniuth in cast iron. They n-ere selected for this investigation because of their inherent uniformity and case and simplicity in handling. Using this techniqur and a series of synthetic standards, a rapid, precise, and accurate method was developed for the determination of magneqium in cast VOL. 31,
NO. 3, MARCH 1 9 5 9
447
Emulsion Calibration. Calibrate t h e emulsion in accordance with t h e tentative practices for photographic photometry in spectrochemical analysis
~
Table I.
Standard Solutions
Std. Mg Solution, MI.
St andard
1 0 0
5
10 15 20
2 0 10
0 0 0 0
iron. A thorough statistical study has established a 1.83y0standard deviation. METHOD
Apparatus. B. R- L. large Littrom spectrograph having a reciprocal linear dispersion of 5.6 A. per nim. a t 3000 A. SSL Spec-Power source unit. ,TACO microphotometer. T y p e 200. Reagents a n d Materials. STANDARD ~IAGNESIC SOLUTION JI I (1 ml. = 0.1600 mg. of magnesium). Transfer 160.0 mg. of high-purity magnesium to a 250-nil. beaker, add 20 nil. of hydrochloric acid, and dissolve completely on a steam plate. Transfer t h e solution t o a 1-liter volumetric flask and dilute t o volume mith hydrochloric acid (1 t o 1). STAXDARD ~IAGNEQ.IUM SOLUTION2 (1 ml. = 0.0200 mg. of magnesium). Pipet 25 ml. of standard magnesium sample 1 into a 200-nil. volumetric flask and dilute to volume with hydrochloric acid (1 to 1). STASDARD I R O X SOLUTION (1 rill. = 20.0 me. of iron) Transfer a 10.0gram sample of high-purity iron to each of two 600-ml. beakers, add 100 ml. of hydrochloric acid (1 to 1) to each beaker, and heat until dissolution is complete. 4 d d concentrated nitric acid dropwise, with care, until iron is oxidized, and then 2 nil. in excess. Boil gently to remove oxides of nitrogen, cool, and combine by transferring to a 1-liter volumetric flask. Dilute to volume rTith distilled water. SAMPLE ELECTRODE. Cse a highpurity, preformed graphite porous cup as the upper and sample electrode. The cup has an outer diameter of 0.242 inch and an inner diameter of 0.156 inch and the bottom is 0.025 inch thick. These electrodes are commercially arailable from the h-ational Carbon Co. as L-3924 electrodes. LOWER ELECTRODE. 9 high-purity graphite rod 0.242 inch in diameter, machined to a 120" included-angle conical tip. Preparation of Standards. Add aliquote of t h e respective standard magnesium solutions t o six 250-ml. beakers (Table I). To each beaker, add 100-ml. aliquots of the standard iron solution and 25 ml. of hydrochloric acid (I to I). Reduce the volume in each beaker to 30 ml. by evaporation, cool to room temperature, and transfer to a 50-ml. volumetric flask. Dilute to volume with distilled water. Thus, each standard solution contains 2.00 grams of iron per 50 ml. of solution. The final hydrochloric acid 448
e
ANALYTICAL CHEMISTRY
0 0.200 0.800 1,600 2,400
3.200
Residual 0 0 0 0 0
01 0'4 08
12 16
concentration of these solutions is approximately 1 to 4. Preparation of Sample. Transfer a 2.00-gram saniple t o a 50-id. volumetric flask, add 15 nil. of hydrochloric acid (1 t o l), and place on steam plate until reaction is completed. Add concentiated nitric acid dropwise, with care, until oxidation of iron is complete. Cool t o room temperature and dilute to volume with hydrochloric acid (1 t o 2 ) . If insoluble matter is present, allow it t o settle and use a portion of the supernatant solution. Alternatively, dryfilter a portion of the solution before transferring i t to the cup electrode. Insoluble matter will retard the flow of solution through the electrode. Electrode System. Mount t h e lower electrode in t h e electrode holder of a normal Petrey stand, positioned so t h a t t h e lower electrode is in line optically with t h e entrance slit and t h e image of t h e source is focused on the collimating lens. Insert the porous cup electrode in an auxiliary electrode holder and position the bottom of cup oyer the point of the lower electrode. Maintain an analytical gap of 2 mm. between the electrodes. Add the sample solution to the porous cup b y means of a medicine dropper. Take care that no air bubbles are introduced with the sample. Excitation Conditions. Produce the spectrum according to t h e following conditions : Capacitance, pf. 0 0025 Inductance, Hh Residual Resistance, added, ohms None Current, r.f., amperes 4 5 Discharge breaks per half cycle 10 Auxiliarj- gap electrodes Magnesium Exposure Conditions. Record t h e spectrum according t o t h e following conditions : Spectral region, -k. Slit width, p Lens at slit, focal length, cm.
Preburn period, seconds Exposure period, seconds Emulsion type
2492 to 3510
10
21 5 20 20 SA-1
Replicate Exposures. Make duplicate exposures of each test sample t o obtain t h e concentration of magnesium. Photographic Processing. Process t h e emulsion in accordance with t h e tentative practices for photographic processing in spectrochemical analysis (1).
(9).
Photometry. Adjust t h e microphotometer slit width t o 10 microns. Measure transmittance measurements on the microphotometer using t h e magnesium 2852.13-A. arc line, excitation potential of 4.33 electron volts, and the iron 2851 30-A. arc line, excitation potential of 5.36 electron volts. S o background corrections are required. Analytical Curves. Spark each synthetic standard a t least five times t o establish the analytical curve. Obtain transmittance measurements of t h e analvtical line pair and convert these t o intensity ratios b y means of t h e emulsion calibration curve. Plot t h e intensity ratios as a function of concentration on log-log paper t o give a typical straight-line curve. Calculations. Evpose sufficient standards on each emulsion t o verify t h e position of t h e analytical curve. Convert t h e transmittance measurements of the analytical line pair to intensity ratios by means of the emulsion calibration curve. Refer t h e intensity ratio of the unknown t o t h e analytical curve t o obtain thc concentration of magnesium. RESULTS
To test the precision of this solution technique thoroughly, two methods of test w r e devised. The first was designed to establish the significance of the deviation due to chemical preparation. The experimental design made it possible to establish the significance of lineal segregation of magnesium in a typical cast iron bar.
A normal production heat test bar approximately 1 inch in diameter was used for the study. Starting a t top of the bar, a 10-gram sample of chips rvas machined and labeled Stratum I (S-I). This machining process lyas repeated orer the length of the bar until nine samples of 10 grams each were collected, labeled S-I to S-IX. Each sample as then thoroughly mixed and divided into two samples of &grams each, labeled C,, Cz CIS. A solution of 2 grams of sample was prepared from each C sample and sparked in duplicate, giving 36 spectral values labeled R1, Rz. RS . . R36. An analysis of variance was performed on the data obtained in this manner (4). Table I1 lists the statistical values obtained in carrying out the analysis of variance. The total degrees of freedom, 35, demonstrates the rigorous method of test employed. Significance tests were determined for the data (Table 111). Table I11 shows conclusively that at the 95% confidence level no significant error is introduced during the sample preparation. On the other hand, for
this specific test bar there is a significant deviation due to lineal segregation in the strata. However, the 2-gram sample employed would reduce the effect of transverse segregation present on the sparking surface of the bar. The second method of test was designed to establish instrumental repeatability, chemical repeatability during sample preparation, day-to-day reproducibility, and total over-all deviation. A proposed chemical cast iron standard was used. One 2-gram sample of this standard was taken into solution daily for 10 consecutive days. These solutions nere labeled A to J. On the first day of test, five spectrographic determinations were obtained for the magnesium content of sample A. On the second day sample A was again analyzed five times, along with five determinations of sample B On the third day, sample C was substituted for sample I$ and analyzed along with sample A. This testing was continued daily until the tenth day, on xhich sample J n a s analyzed along with sample A. Table IV shows a tabulation of these daily results, each a n average of five determinations. The instrumental repeatability can be determined b y the data shown in Table V, where each value represents the coefficient of variation for each group of five determinations. These data were analyzed statistically (Table VI). The instrumental deviation was calculated b y obtaining the root mean square of all the per cent standard deviations listed in Table V. The day-to-day, same solution, coefficient of variation was obtained by calculating the % u of the 10 values for sample A, each value representing five determinations. This 1.32% value thus included the deviation due to the instrument and day-to-day curve shifts. The day-today, different solutions, value contains the additional deviation of sample preparation. The over-all deviation value of 1.83% includes all variables and was obtained b y calculating the coefficient of variation on all of the 95 test data iaken over all the days and all the solutions. Any measure of accuracy is difficult with this method because of the lack of reliable methods for the determination of magnesium a t this level. However, the only sources of error to be encountered in addition to the deviations discussed are errors made in the preparation of the synthetic solution standards, and incomplete dissolution of the sample with a resulting lom- answer. Both
Table 11.
Analysis of Variance
MS
SS5
!kz)
!‘o-’b,
Deviation D.F. Strata or segregation (S) 502.8 8 62 85 Chemiral Preparation ( C j 9.5 9 1 06 Instrumental ( R ) 18 1 .00 Total 530.3 35 a SS. Sum of squares. D.F. Degrees of freedom. LIS, 3Iean square. ~
Table Ill.
Significance Tests
l\IS(S) -_ _
- MS(R) 62.85 (8, 18)
F (strata)
F (chemical preparation)
~~~~~~~~
Table IV.
B
A
C
Day 1 2 3 4
5 6 7
8
9 10
I
3
1 li 0 76
8
9 10
0.93 1.18
5
6
7
1.06 (9, 18) .it 95% confidence level: F (8, 18) = 3.71 F (9, 18) = 2 46
Day-to-Day Reproducibility
Sample D E F G Per Cent Magnesium
H
I
J
__~__
._
0 0776 0 0770 0 0758 0 0756 0 0754 0 0745 0 0736 0 0724 0 G758 0 0756 0 0760 0.0761 0.0756 0 0764 0 0760 0 0760 0,0758 0 0756
0 66 2 41 0 75 1 64 0 20 0 72
4
MS(R)
0 0764
Table V.
2
m(cj --
= -- -
1.59
0.i2
Instrument Repeatability Study
1 21
errors can be avoided with careful chemistry. TIME REQUIREMENTS
Time studies indicate that 12 samples can be analyzed in duplicate a t the rate of 12 minutes per sample. A single sample would require 40 minutes. This speed compares favorably with that of chemical techniques, and is sufficiently fast for the control of routine samples, especially where high accuracy is required. ACKNOWLEDGMENT
The author acknowledges the cooperation of his associates, J J. Schultz, A. H. Jones, M. D. Cooper, and D. L. F r y in t h e preparation of this paper. LITERATURE CITED
(1) A4m.Soc. Testing Materials, Philadelphia, Pa., “Methods for Chemical Analysis of Metals,” Designation E-115, p. 559, 1956.
2.51
0 73
0.93
Table VI.
Data on Precision
Type 70 u Instrumental 1.26 Day-to-day, same solution 1.32 Dag-to-day, different solu-l.ti1 tion 1.83 All variables
D.F. 76 9
-
Y
91
(2) Zbid., Designation E-116, p. 570. (3) Bryan, F. R., Nahstoll, G. A., Yeldhuis, H. D., ASTM Bull. 162, 69 (1949). (4) Dixon, IT. J., Massey, F. J., “Introduction to Statistical Analysis,” M c Graw-Hill, Sew York, 1951. (5) Feldman, C., -4x.4~.CHEY.2 1 , 1041 119411). (6) ,G?mbrill, C. M., Gassman, A. G., 0 heil, W. R., Zbid., 23, 1365 (1951). (7) Sloviter, H. A , , Sitkin, A., J. Opt. SOC. Am. 34,400 (1944). RECEIVEDfor review May 24, 1957. Accepted Kovember 17, 1958. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 1957. \ - - - - ,
VOL. 31, NO. 3, MARCH 1959
0
449
