Precision and Accuracy of Colorimetric Procedures as Analytical Control Methods Determination of Silica ALLEN L. OLSEN, EDWIN A. GEE, VERDA McLENDON, AND DELWIN D. BLUE Bureau of Mines, Eastern Experiment Station, College Park, Md. Working Standard. Dilute 2 ml. of standard to 1OOOml. (1ml. = 0.02 mg. of silica) and store in a rubber bottle. Reducing Solution. Dissolve 2 rams of 1 2 4-aminonaphtholsulfonic acid, 120 grams of sojium metabkhfite, and 24 grams of sodium sulfite in water and dilute to 1 liter. Store solution in a dark bottle in a cool place. Ammonium molybdate, 5% solution in water. Hydrochloric acid, 0.5 N . Indicator, 2,4dinitrophenol or 2,6-dinitrophenol, Saturated solution in water. Ammonia Gas. Bubble clean air through ammonium hydroxide. PROCEDURE. Discharge an aliquot of the acidified solution containing 0.02 to 0.10 mg. of silica, into a 25-ml. calibrated blood-sugar test tube (1.3) by means of a pipet. Add 2 or 3 dro s of indicator, and bubble ammonia gas into the solution to tge yellow end oint. Add 5 ml. of 0.5 N hydrochloric acid and enough d i s t i e d water to bring the meniscus to the 25-ml. mark, and mix the contents of the tube thoroughly. Add 2 ml. of ammonium molybdate by means of a pipet, mix thoroughly, and permit the solution to stand 5 minutes. Add 3 ml. of reducing solution by means of a pipet, mix again, and determine the color absorption with the Klett-Summerson photoelectric colorimeter equipped tvith a filter having a transmission range of 640 to 700 mp. It is absolutely essential to run a blank on the reagents, the correction being applied to the colorimeter reading. A standardization curve is prepared from a series of oints of known silica content. Since the curve is linear througgout the concentration range used in this study (0.01 to 0.28 mg. of silica), it will be found convenient to calculate a conversion factor. A quantity, in triplicate, of the working standard, 2 or 3 ml., is discharged into blood-sugar test tubes and the color is produced in the usual manner. After the blank correction is made, the scale division is evaluated in terms of milligrams of silica.
A colorimetric procedure for the rapid determination of small amounts of silica, involving the reduction of the silicomolybdate complex to the intense molybdenum blue, has been developed to meet the special requirements of analyzing leach liquors. The quantities and concentrations of reagents specified are considered to be best for color stability, sensitivity, and speed of color development, Factors influencing color intensities have been investigated, and techniques for a precision and accuracy of a control character are described. Precision and accuracy studies on data from ordinary routine analyses of leach liquors have been made. The average precision measured b y the average deviation of single results from while the over-all accuracy is of the mean i s of the order of *I%, the order of * 1%.
A
PREVIOUS article (f 2) outlined an experimental procedure for the colorimetric determination of aluminum and described the requisite techniques for precision and accuracy of a control character. Further studies on methods of analytical control have resulted in the development of a procedure for the colorimetric determination of silica. This is a modification of previously published methods, and the investigation evaluates the factors that influence precision and accuracy as employed in routine analyses. Statistical reasoning based on the standard deviation is applied to the acquired data (1). The usual procedure for the colorimetric determination of soluble silica depends on either the formation of the yellow silicomolybdate color produced when ammonium molybdate reacts in an acid media with the soluble silica (4, 10, 13) or the reduction of the silicomolybdate complex to the intense molybdenum blue color (3, 7). Edwards (6),noting the difficulties inherent in the determination of silica in calcined alumina, suggests the use of the silicomolybdate as a means of rapid and accurate estimations of small amounts of silica. The procedure employing the formation of the molybdenum blue color was selected for the rapid analysis of small amounts of silica in leach liquors in pilot-plant operations, a leaching step in the lime-soda process for the extraction of alumina from siliceous bauxites and clays. It is believed that an extended sensitivity of t h e reaction will be reflected in more accurate results. The interferences cited by Knudson, Juday, and Meloche (10) were not present in the sample. Kahler (7) develops the silicomolybdate complex by the use of hydrochloric acid, considering a pH range of 2.4 to 2.7 as optimum prior to reduction, and reduces the yellow complex with sodium sulfite. However, this method is unsatisfactory for the analysis of leach liquors, since the pH (6.8) of the final solution closely approaches optimum conditions for precipitation of aluminum hydroxide. To minimize color progression and prevent precipitation of aluminum hydroxide, an organic reducing agent (8, 9) has been investigated and found to be applicable.
The conversion factor used in this investigation, 0.000311 mg. of silica per scale division, should not be selected as an arbitrary constant since the value of the factor would not necessarily hold for other instruments. A factor must be regarded as valid only under the conditions of calibration. The product of a given scale reading and 0.002, an arbitrarily selected proportionality factor (14), results in an approximate density reading where density = log l/transmission. FACTORS INFLUENCING RESULTS
I n a study on the reproducibilities of blanks, the foregoing procedure was applied to several milliliters of slightly acidified, distilled water. I t has been found that the presence of impurities on walls of blood-sugar test tubes, supposedly chemically clean, has a pronounced effect upon the colorimeter reading. It is highly advisable to label new test tubes and use these specifically for colorimetric silica. Chromic-sulfuric acid mixture effectively cleans the new tubes; the use of alkaline cleaning mixture not only is unnecessary but is inadvisable, since etched surfaces induce irregular results. I t is extremely important to make blank corrections in routine qnalyses, as impurities in test tubes and/or reagents can markedly affect the final reading. The instability of the blue color has been cited (2, 10) as an objection to the use of reduced silicomolybdate in colorimetric silica. As shown in Table I, the color is stable for several hours. I n this particular range of the colorimeter scale, the 2-unit increase in reading represents one scale division.
The recommended quantities and concentrations of reagents in the following procedure are considered to be best for color stability, sensitivity, and speed of color development. ANALYTICAL PROCEDURE
REAQENTS.Standard Silica Solution. Dissolve 47.32 grams of sodium metasilicate (Na&4iO*.9H10)in 1 liter of water (1ml. = 10 mg. of silica), filter, and standardize gravimetrically (6). Store in rubber bottle (14). 462
ANALYTICAL EDITION
July, 1944
Table 1. Stability of Molybdenum Blue (Colorimeter readings on standard silica, uncorrected for blank) No. 1 No.2 No.3 Elapsed Time, Minutes
Table
II. Effect of pH on Colorimeter Reading
Color (End of Colorimeter Titration), Reading Sample 2,6-DinitropH Final (Corrected No. phenol 25 ml. volume for Blank) Standard Soluble Silica: 0.0590 mg. of Si08 1 Very faint yellow 1.02 1.98 188 2 Very light yellow 1.01 1.99 188 3 Light yellow 1.02 1.99 188
Error,
%
Yellow
- -
+
dition of aluminum and sodium ions indicated the absence of soluble silica in these reagents. The data in the second part of Table I1 indicate that the overstepping of end point of the standard in the presence of these ions results in decreased accuracy of a negative order. The intensity of yellow in tubes l a and 2a was obtained by stopping the titration a t the appearance of a white precipitate, aluminum hydroxide, with no evidence of yellow color, and flushing out the residual gas in the tube with a small amount of water. The formation of this precipitate probably results from localized neutralization; however, since the precipitate disappears on shaking the tube, its presence does not appreciably impair the value of the procedure. The samples in the remaining tubes were titrated to various shades of yellow before the residual ammonia gas was rinsed out. However, the A120r:Si0* ratio in this synthetic mixture is much higher than that usually encountered in routine analysis.
-0.5 -0.5 -0.5
1. 134 2 02 189 0 _. ... Deep yellow 1.06 2.08 189 0 Deep yellow 1.21 2.39 185 -2.1 Standard Soluble Silica: 0.0590 mg. of Si02 and Ions' Very faint yellow 1.02 1.91 189 0 la 1.07 Faint yellow 1.98 188 -0.5 2a 30 1.10 Li ht yellow 2.04 185 -2.1 Y3low 1.11 2.11 183 -3.2 4a ~~. Yellow i.is 2.20 180 -4.7 Sa 6 s Yellow 1.27 2.42 177 -6.3 a . 1 ml. of standard 0.01 mg. of Si02 10 mg. of AlzOa 10 mg. of NaaO (Si0z:AhOa 0.1%). 4 5 6
+
Since the pH in this analytical procedure is critical, it becomes necessary to adjust all samples to approximately the same hydrogen-ion concentration. I n the presence of a suitable indicator the previously acidified aliquot is neutralized with gaseous ammonia, a technique designed to eliminate the introduction of measurable amounts of silica in the use of dilute ammonium hydroxide. Either 2,4-dinitrophenol (pH 2.6 to 4.4) or 2,6-dinitrophenol (pH 2.0 to 4.0, 11) may be used as the indicator, since the yellow color changes to colorless a t about pH 2, thereby resulting in no complications from added color. The effect of varying degrees of neutralization on standard soluble silica with and without the addition of aluminum and sodium ions was studied in terms of the resultant colorimeter reading. I n Table 11, six test tubes, corresponding to samples 1 to 6, each contained 3 ml. of working standard and each sample was treated with successively greater amounts of ammonia gas. I t appears that the end point is not easily overstepped in the case of the standard. Only sample 6 shows appreciable changes. In carrying out a titration, two-color changes usually are observed, first a light yellow and next a deep yellow. A satisfactory procedure is established by stopping the titration a t the appearance of a faint yellow since, on rinsing the gas tube, the residual ammonia gas changes the color to a deep yellow. The pH of these solutions was measured by means of a glass electrode and Beckman pH meter after dilution had brought the meniscus to the 25-ml. mark and after all ingredients had been added. The numerical value for pH, although not initially reflected in the colorimetric reading, shows a gradual change as the series is descended. Only No. 6 , representing a definite excess of ammonia, shows significant changes. I n ordinary laboratory practice, it is doubtful if titration would be carried to the extremes of Nos. 5 and 6 . The second set of test tubes in Table 11, samples l a to 6a, represents a study on the effect of varying degrees of neutralization on standard soluble silica in the presence of aluminum and sodium ions. Aluminum chloride hexahydrate and sodium chloride were added to the working standard in amounts calculated to contain 10 mg. per ml. each of aluminum trioxide and sodium oxide. Blank determinations with and without the ad-
463
PRECISION AND ACCURACY
The precision of the colorimetric method may be studied by applying the procedure to the analysis of leach liquors. Three typical leach liquors previously acidified and diluted (10 ml. in 250 ml.) were selected, and a 1-ml. portion of each sample was taken for analysis. The silica content was measured over a period of several days. Table I11 shows what might be expected in the way of precision for 10 determinations on each sample. The method of evaluating the factor of precision is that derived from consideration of results set up by A.S.T.M. (1). The accuracy of the method is determined by making a comparative study of standard soluble silica with and without the addition of aluminum and sodium ions (Table IV). DISCUSSION
In computing the * values for the results in Table 111, a value for a was chosen such that, in 99 chances out of 100, one
Table 111. Precision of Method under Routine Conditions (Analysis of leach solutions, colorimetric silica) -Sample 1 Sam le 2, Sam l e 3 Test No. Total Si02 d d t X 10' Totar SiOa Totaf SiOi Mo. Mg. Mg. 1
2
3 4 5 6 7 8 9 10
-- a- ?$* -
Av.
0.0899 0,0899 0.0893 0.0915 0.0899 0.0899 0.0899 0,0899 0,0893 0.0915
Zd2
0 0901 544 x 10-3
uio
0.0007
F .
I
5u
-0.0002 -0.0002 -0.0008 +0.0014 -0.0002 -0.0002 -0,0002 0.0002 0.0008 f0.0014
--
0.0901 * 0 0008
(P.
-
4 4
64 196 4 4 4 4 64 196
--
0.1240 0.1256 0.1240 0.1256 0.1240 0.1256 0.1240 0.1240 0.1225 0.1240
0 0009 0.1243* 0.0010
Sample No. 1
2 3 4 5 6 7 8 9 10
11
12
-
0,1562 2308 X 10 - 8 0.0015 0.1562* 0.0016
0.99, 10 observations) ~
Table
--
0.1243 885 X 10 -9
0.1581 0.1550 0.1550 0.1550 0.1581 0.1581 0.1581 0.1550 0.1550 0.1550
~~
IV. Accuracy of Method under Routine Conditions (Colorimetric silica) Analysis, Total Silica Si0r:AlrOa Calculated Found
%
Ma.
0.1 0.1 0.1 0.5 1.0
0.0393 0.0590 0.0786 0,0590 n ,0590 0.0590 0.0590 0.0590 0,0590 0.0590 0.0590 0.0590
1.o
1.0
1.o
2.0 3.0 4.0 5.0
MO. 0.0392 0.0587 0.0778 0.0589 0.0590 0.0590 0.0596 0.0500 0.0593 0.0593 0.0593 0.0593
Error
% +0.3 -0.6 -1.1 -0.2 0 0 +1.2 0 +0.8 +0.8 +0.8 +0.8
INDUSTRIAL AND ENGINEERING CHEMISTRY
464
might expect the ranges bounded by the computed limits to include, of the universe sampled, the objective average, p‘. From a consideration of the data in Table IV, it is apparent that the over-all bccuracy is of the order of 2%. .- Since the SiO2:AhOS ratio markedly affects the accuracy, it is believed that the ratios investigated in Table IV cover the ranges usually encountered in routine work. Repetition of 0.0590 mg. of silica per ml. in this study is explained by the fact that this concentration provides a reading in the most satisfactory portion of the scale for colorimetric measurement (200). For this study, 3 ml. of working standard were admixed with varying proportions of aluminum and sodium ions. In the range of 0.1% (SiOs:AlnOa), the results invariably are low, probably indicating that the end point had been overstepped. I n the range of low silica to high alumina ratios, the color change in the titration procedure is not too abrupt. I n this regard it has been found that 2,6-dinitrop,henol, owing to depth of color, furnishes a more distinct end point than 2,4-dinitrophenol. However, in the range of 1.0 to 5.0%, no particular difficulty is experienced in observing color changes of the indicator.
from J. E. Conley, Chief, Chemical Engineering Unit, Bureau of Mines. They are indebted to R. MacMillan, former junior chemist, Bureau of Mines, who suggested the use of gaseous tmmonia in PH adjustments. LITERATURE CITED
Am. SOC.Testing Materials, “Manual on Presentation of Data”, 3rd printing, p. 41,Philadelphia, 1940. Bertrand, G., Bull. aoc. chim. biol., 6, 157 (1924). Boyle, A. J., and Hughey, V. V., IND.ENQ.CBEM.,ANAL.ED., 15, 618 (1943).
Dienert, F., and Wandebulcke, F., Compt. rend., 176, 1478 (1923).
Edwards, J. E., IND. ENQ.CHEM.,ANAL.ED., 13,70 (1941). Hillebrand, W. F., and Lundell, G. E. F., “Applied Inorganic Analysis”, p. 721, New York, John Wiley & Co., 1929. Kahler, H. L.,IND. ENG.CHEM.,ANAL.ED., 13, 536 (1941). King, E.J., and Dolan, M., Can. M e d . Assoc. J . , 31, 21 (1934). King, E. J., and Stantial, H., Biochem.J . , 27, 990 (1933). Knudson, H. W.,Juday, C., and Meloche, V. W., IND.ENQ. CHEM.,ANAL.ED., 12,270 (1940). Mellan, I., “Organic Reagents in Inorganic Analysis”, p. 225, Philadelphia, Blakiston Co.,1941. Olsen, A. L.,Gee, E. A., and McLendon, V., IND.ENG.CHEM., ANAL.ED., 16, 169 (1944). Schwartz,M. C., Ibid.,14,893 (1942). Summerson, W. H., J. Biol. Chem., 130, 149 (1939).
ACKNOWLEDGMENT
The authors gratefully assistance in the direction of this work and the preparation of this paper
Vol. 16, No. 7
PoBLxsHmD
D.
c,
by permiasion of the Director, Bureau of Minea, Washington,
Determination of Fatty A c i d Monoesters of /-Ascorbic and d-lsoascorbic Acids in Fats and
Oils
JACK TURER AND R. M. SPECK Eastern Regional Research Laboratory, Philadelphia, Pa.
The P,6-dichlorophenolindophenol reagent in acetone can b e successfully used for the determination of the fatty acid monoesters of I-ascorbic and d-isoascorbic acids in fat and oil substrates.
M
ONOESTERS of Z-ascorbic and d-isoascorbic acids have been prepared (4) by direct esterification of the ascorbic acids with fatty acids. According to the experimental evidence, the asterification probably takes place on the primary alcohol group of the ascorbic acids, as illustrated by the following formulas: 0
//
0
//
?r &OH 7 11 I
&-OH I I
HC HO-AH
0
1
I
H-&-OH
~ H I O O C ( C H ~ ) I ~ C H ~h L O O C (CH&rCHs l-Ascorbyl palmitate d-Isoascorbyl palmitate
* I n previous work in this laboratory ($ it was found that these ascorbyl esters were effective antioxidants for fats and oils. (For convenience of expression, the fatty acid monoesters of Gascorbic acid and d-isoascorbic acid are referred to in this paper as the ascorbyl esters. It has been suggested that these esters be named as derivatives of the ascorbic acids-for example, palmitoyl Z-as-
corbic and palmitoyl d-isoascorbic acids.) I n connection with an investigation of the antioxidant action of the ascorbyl esters, the resulfs of which will be published later, it was essential to have an analytical method for determining ascorbyl esters in fat substrates. Ascorbyl esters are somewhat soluble in fats and oils but insoluble in water, whereas the ascorbic acids are insoluble in fats and oils but very soluble in water. Since the enediol group is unchanged when ascorbic acid is converted to the monoester (d), the methods for determining ascorbic acid which are dependent upon the reactivity of the enediol group s h d d be applicable for the determination of the ascorbyl ester. One method that has been used for determining ascorbic acid involves titration with iodine solution (3). It is obvious, however, that this method would be unsuitable with fat and oil substrates, since these contain unsaturated groups, which absorb iodine, The most widely used method today is that originally suggested by Tillmans (6),in which the ascorbic acid in aqueous solution is titrated with 2,6dichlorophenolindophenol. In order to use this method for rtscorbyl esters in fat substrates, considerable modification was necessary, since these esters are insoluble in water. Various solvents (absolute ethyl alcohol, ethyl ether, dioxane, and acetone) were tried and acetone was found to be preferable. STANDARDIZATION
Standardization of the indophenol solution is carried out as follows: A known amount of the sodium salt of 2,6-djchlorophenolindophenol is dissolved in dry acetone of AMERICAN CHEMICAL SOCIETYgrade. A solution containing 0.25 gram of the indophenol salt in 1 liter of acetone may be used for deter-
