609
V O L U M E 21, NO. 5, M A Y 1 9 4 9 Reasons for avoiding wave lengths below 410 millimicrons have been shown. T o avoid entirely the tungstate ion interference would require selecting a wave length of approximately 435 millimicrons and result in a serious sacrifice of sensitivity. In this sense the selection of 410 millimicrons was the result of a compromise.
steel. Two alternative procedures for routine analysis have been shown to have a precision far more satisfactory than routine methods u-hich were known in the past; their adaptability to routine application in laboratories faced with the problem of handling large numbers of these determinations is immediately suggested.
SUMMARY
LITERATURE CITED
The demand for a precise method for the determination of vanadium in the low concentration ranges characteristic of lowalloy steels has been fulfilled by this investigation. Two methods have been demonstrated from the standpoint of precision to be comparable with well established methods for the determination of the more common alloying elements in this type of
(1) Smith, G. F., and Geta, C. 4.,ISD. ENG.CHEM., ANAL.ED.,10, 191-5 (1938). (2) Wright, E. R., with Mellon, M . G., Ibid., 9,251-4 (1937). RECEIVED M a y 1, 1948. Presented before the Division of iinalytioal a n d Micro Chemistry a t t h e 112th Meeting of the .4wmxcAx C H E M I C ASoL CIETY. S e w York, N.Y.
Colorimetric Determination of Aluminum in Steel Use of 8-Hydroxyquinoline STEPHEN E. WIBERLEY AND LEWIS G. BASSETT, Rensselaer Polytechnic Institute, Troy, N. Y . The development of a simple, rapid, and accurate oclorimetric method for the determination of small quantities of aluminum in steels is described. The sample is dissolved in nitric and perchloric acids, fumed, cooled, diluted, and electrolyzed in a water-jacketed mercury cathode cell until iron-free. The aluminum salt of 8-hydroxyquinoline (8-quinolinol) is formed in a buffered acetic acid solution at a pH of 6.0 * 1.0. The salt i s extracted with chloroform and the color intensity of the resulting solution is measured with a suitable colorimeter or spectrophotometer. The application of the method to various types of samples is described. The effects of interfering elements and of such variables as wave length, pH, and amount of reagent are evaluated.
AT
,sent there is a recognized need for a simple, rapid, and accurate method for the determination of small amounts of aluminum in steels. The generally used gravimetric procedure (3)is tedious and of doubtful accuracy when used for samples containing less than 0.1% aluminum. The small amounts of aluminum under consideration, and the need for simplicity and rapidity of measurement, indicate the desirability of a colorimetric method. ‘4survey of the literature shows several possibilities. Aluminon (5, I d ) , alizarin red S (4),and hematoxylin ( I f ) form colored lakes with aluminum hydroxide. These reagents have thc disadvantage common to many lake-forming compounds, relative instability of the lakes over a period of time. Fluorometric methods using morin ( I O , 19) and pontachrome blue black R (18) have been developed. However, common anions such as phosphate, fluoride, and sulfate cause marked interference with morin, and pontachrome blue black R , although extremely sensitive, requires approximately 1 hour before full intensity of fluorescenre is attained. 8-Hydroxyquinoline (&quinolinol), has been used extensively as a precipitant for many metals, including aluminum. Colorimetric methods (2, 8) based on solution of the quinolate in acid with subsequent conversion to an azo dye, require the separation of the precipitate. Alexander ( I ) , Moeller (16), and Gentry and Sherrington (6) have determined aluminum directly by measuring the color intensity of the yellow solution obtained by the extraction of the aluminum quinolate with chloroform. This method appeared most promising to the authors and has been used with certain modifications, as the basis of the method described. The determination of aluminum in steel poses a particular problem, because iron interferes with 110th the formation and extraction of the quinolate. Electrolysis with a mercury cathode has been chosen as the means of removing most of the interfering ele-
ments including iron. Melaven ( 1 5 ) .
The cell used is a modification of khat of
REAGEIYTS AND ERUlPiMENS
Solution .4,2% 8-Hydroxyquinoline in 1 A’Acetic Acid. Dissolve 20 grams of 8-hydroxyquinoline (hlallinckrodt reagent grade) in 1 liter of 1 iV acetic acid (60 ml. of glacial acetic acid diluted to 1 liter). Solution B, Buffer Solution. Dissolve 200 grams of ammonium acetate and 70 ml. of concentrated ammonium hydroxide (approximately 15 M ) in a total volume of 1 liter. Solution C, Standard llluminum Solution. Weigh 13.9 grams of reagent grade A1(N03)3 9 H z 0 or 17.6 grams of A12(S04)3KzS04 2 4 H ~ 0 dissolve , in water, make up the volume to 1 liter, and standardize by ammonia precipitation and ignition to the oxide. (Solution should contain 1.000 mg. of aluminum Solution D. Pipet 20 ml. of the above solution and c%:tTk: 500 ml. in a volumetric flask. This solution now contains 0.040 mg. (40 micrograms) of aluminum per ml. Reagent grade chloroform. Beckman quartz spectrophotometer, Model DU. ,411 measurements were made in 1-em cells. Beckman pH meter, laboratory Model G. Mercury cathode cells (see Figure 1 ) . DEVELOPME3T OF PROCEDURE
Previous investigators of the extraction colorimetric method for aluminum have added a chloroform solution of the 8-hydroxyquinoline reagent to a buffered aqueous solution of an aluminum salt, and then agitated the mixture. The aluminum quinolate formed remains in the chloroform layer on settling. The yellow chloroform layer is then separated, and diluted to a definite volume with chloroform, and its color intensity is measured. There is some disagreement concerning the proper conditions for the extraction. .4lexander (1) found complete extraction a t pH 3.5,
610
ANALYTICAL CHEMISTRY
Moeller (16) only in the pH interval of 4.3 to 4.6, and Gentry and Sherrington (6) in a pH range of 4.5 to 11.5 except between 6.5 to 8 where incomplete extraction was reported. An alternative procedure to that given above is first to form the quinolate by adding an acetic acid solution of the reagent to the buffered aluminum salt solution, then to adjust the pH, and estract with chloroform. This should extend the permissible range of pH, as complete extraction should be obtained a t any pH a t which the quinolate is normally insoluble in the aqueous layer. The conditions for this latter procedure have been investigated.
pH. Solutions containing 4.00 ml. of the standard aluminum solution (Solution D) in 50 ml. of water were treated with 10 ml. of a buffer solution (30 grams of ammonium acetate and 30 ml. of glacial acetic acid in 650 ml. of water), 2 ml. of 2% 8-hydroxyquinoline (Solution A), and sufficient ammonium hydroxide or acetic acid to attain the desired pH. Three extractions were made on each sample with 10- to 15-ml. portions of chloroform. The chloroform layers were combined and made up to a total volume of 50 ml. The optical density was determined on the Beckman spectrophotometer a t 390 millimicrons versus chloroform. Samples containing the same amounts of the reagents except for the aluminum were run in a similar manner.
0.~00
0.600
z
3
I
O'-O
0.200
0.wo
I
I
7 .o
9.0
PH
Figure 2.
Effect of pH on Extraction of A l u m i n u m Quinolate
Table I.
Effect of Reagent on Optical Density
8-Hydroxyquinoline, 111.
1 (Blank) 1 2 (Blank) 2 3 (Blank) 3 4 (Blank) 4
5 (Blank)
5 6 (Blank) 6
Figure 1.
Water-Jacketed Slercury Cathode Cell
The results obtained are shown in Figure 2. Extractions are wen to be complete over a pH range of 4.9 to 9.4. These results , found in workcheck reasonably well with those of Goto ( i ) who ing with macroamounts of aluminum that precipitation was complete over the p H range 4.2 to 9.8. The amount of reagent extracted increases more rapidly a t the higher pH values, resulting in higher blanks. The blanks are reasonably constant in the p H range 5.0 to 7.0 which has therefore been chosen as the optimum range for the method.
Optical Density
Optical Den-ity (Corrected for Blank: 0:ioo 0:i02
o 1 iig 0:iis 0:iio 0:iia
Wave Length. Solutions containing 2 ml. of reagent Solution A, 2 ml. of buffer Solution B, and 0, 1, and 2 ml. of Solution D corresponding to 0, 40, and 80 micrograms of aluminum, respectively, n-ere estracted with chloroform, and diluted to 50 ml., and optical densities were measured over a wave length range of 360 to 430 millimicrons. The data obtained are shown in Figure 3. Although there is a relatively broad absorption band, the curve for 80 micrograms shows a slight maximum at 390 millimicrons. Accordingly, this wave length has been chosen for the method. Amount of Reagent. Solutions containing 4 ml. of Solution I) (160 micrograms of aluminum), 2 ml. of buffer Solution B, and 1, 2, 3, 4, 5 , and 6 ml. of 2y0 8-hydroxyquinoline (Solution respectively, were extracted a t a pH of 6.3 and the optical densities determined. A similar set but with no aluminum present was also run. The results are shown in Table I.
The optical densities, after correction for the blank ri~adings, Two are constant over the whole range of reagent ronc~l~tration. milliliters of reagent Solution A have heen selected as furnishing 110th a sufficient excess of reagent and a loa blank. Stability. Optical densities of the chloroform extracts obtained during the investigation of the effect of pH were measured over a period of 24 hours. S o variations were observed, regardless of the pH a t which thc t%xtractionsTwre made.
611
V O L U M E 21, NO. 5, M A Y 1 9 4 9 Preparation of Calibration Curve. Based on the above study of conditions the following procedure was used for the preparation of a calibration curve. Standard solution samples containing 0, 40, SO, 120, and 160 micrograms of aluminum were each treated with 2 ml. of reagent Solution A and 2 ml. of buffer Solution B, made up to a total volume of approximately 60 ml. with distilled water, and extracted in a separatory funnel with three 10- to 15-ml. portions of chloroform. The chloroform layers were drawn off through a 7-cm. No. 40 n’hatman filter into a 50-ml. volumetric flask, and diluted to the mark with chloroform. Optical densities were then measured on the Beckman spectrophotometer at 390 millimicrons. The results corrected for the blank (see Table I ) art’
