Coulometric Titration of Arsenic( Ill) Solutions with Cerium( Ill) Spectrophotometric End-Point Detection N. HOWELL FURMAN and A. JAMES FENTON,
JR.
Frick Chemical Laboratory, Princeton University, Princeton,
N. 1.
Their background medium was l N sulfuric acid. The work described here is similar, but a 1iniit:ttion mas encountered on their method when it was applied to coulometric procedures. I n coulometric generations of ceric ion from a saturated cerous solution, 1 N in sulfuric acid, a definite reagent blank was found which was not reproducible. This lack of reproducibility is ~ of ceric ion with the oxidizable very likely due to s l o reaction species present in the cerous sulfate or acid and solvent. The blank was eliminated by generating ceric ion to an arbitrary absorbance, allowing the impurities to react, adding an excess of arsenious acid, and titrating the excess in the usual manner.
The titration of arsenious acid with ceric sulfate has long been an excellent method for standardizing ceric solutions and for determining macro quantities of arsenic in acid solutions. No previous attempt has been made to test the applicability of this classical titration to coulometric ceric oxidinietry. The otherw-ise slow7 reaction is catalyzed by a trace of osmium tetraoxide. Spectrophotometric end-point detection with an unniodified Beckman DU instrument is shown to be accurate, providing the limitations of the method are understood. The end-point procedure depends upon the strong ceric absorbance maximum at 320 mp. The method thus consists of detecting the absorbance due to excess ceric ion and extrapolating to zero absorbance. There is no absorption by the components of the solution until the equivalence point is passed. The results compare favorably with those obtained in the titration of reduced arsenic solutions with electrically generated iodine, bromine, and chlorine in which amperometric, potentiometric, or photonietric end-point procedures were used.
S”
VERAL investigators have reported the determinations of reduced arsenic solutions by electrically generated reagents. Swift and others have used iodine (IS), chlorine (e), and bromine (11) t o determine arsenite solutions coulometrically. Amperometric end-point procedures were employed throughout by these investigators. Everett and Reilley (6) have described a method for determining arsenite with electrically generated iodine, utilizing spectrophotometric measurements of excess iodine a t 342 mp t o determine the end point. An automatic coulometric titrator with photometric indication in the visible region has been shown applicable to the arseniteiodine reaction (16). On a macro scale potentiometric indication has been described as accurate for the titration of arsenite with externally generated iodine, bromine, and chlorine (12). A semimicro method with electrically generated permanganate is reported to give reproducible results when o-phenanthroline ferrous sulfate is employed as indicator (14). Electrically generated cerium(1V) in sulfate media has been used t o determine ferrous ion coulometrically. The end points can be determined potentiometrically (9),or with the sensitive end-point procedure of Cooke, Reilley, and Furman ( 2 ) . Hydroquinone, p-aminophenol (7), uranium(1V) (8), and ferrocyanide ( 4 ) have also been successfully determined coulometrically with cerium(1V). The end points in these determinations were found using the “sensitive end-point’’ procedure. The coulometric ceric-arsenious acid reaction has not hitherto been reported, nor has a spectrophotometric end-point procedure been attempted previously for coulometric ceric titrations. This investigation was initiated to test the feasibility of applying spectrophotometric end-point detection to coulometric ceric titrations. The method was tested by titrating microgram quantities of arsenious acid, which in itself is a new application for coulometric ceric oxidimetry. Bricker and Sweetser (1) describe a spectrophotometric method for determining microgram quantities of arsenious acid by direct titration i%-ithceric sulfate solutions ns dilute as 4 X 4
515
Figure 1. Titration cell A.
Light path
C.
Lead amalgam isolated elec-
D.
Capillary trode inlet tube for gas stirring
B. Generating electrode, 2 X 2 cm.
The current-time product for the blank plus excess arsenious acid is then subtracted from the first determination. Several subsequent samples may be analyzed in the same generating medium. APPARATUS
All work described here was done with a Beckman DU spectrophotometer. The titration cell is shown in Figure 1; it is identical with that described by Bricker and Sweetser (1). There was sufficient space above the light path t o accommodate the necessary electrodes while making it possible t o minimize the solution volume. A felt cover with a slit kept stray light from the cell chamber; nitrogen gas passing through a capillary stirred the solution. No galvanometer unsteadiness was experienced if bubbles did not enter the light path. The electrical current was supplied by a bank of 45-volt radio B batteries delivering a total of 225 volts. Constancy of current was obtained a t the higher current levels by using additional batteries in a series-parallel arrangement. The current was meas-
ANALYTICAL CHEMISTRY
516
The osmium tetroxide (G. F. Smith Chemical Co.) was a 0.01M solution, 0.1N in sulfuric acid.
Table I. Titration of Arsenic(II1) with Cerium(1V)
Av.
934.0
+1.4
+O.l5
Generator Current, Ala. 4.718 4.708 4.683 4,820 4.800 4.71G
Av.
372.7 374.4 371.G 372.9
0.3 1.4 1.4 -0.1
-0.08 +0.37 -0.37 -0.03
4.799 4,790 4.788 4.792
-0.27
1,051 1.050 1.050 1.053 1,052 1 051 71.050 1.051
Arsenic, y Taken Found
Error Y
___ % +0.44 -0.05 +0.03
4-0.27 fO.O.5
373.0
+o.oo
-0.27 +0.34 +0.69 -0.07 f0.27
AI,. 1 4 5 . 6 110.8
Av.
112.1 112.2 111.6 110.4 111.2 111.2 111.4
+0.1
+ O . 15
1.3 1.4 0.8 0.4 0.4 0.4 +O.G
+l.Z $1.3 +0.7 -0.4 f0.4 f0.4 +O.G
0.4710 0.4710 0.4710 0.4721 0.4721 0.4721 0.4715
+1.6 +0.4
+1.3 +O.G
0.4710 0.4710 0.4710 0.4700 0.4700 0.4700 0.4700 0.4700 0,4704
f2.1 +2.7 +1.2 i l . 9 +0.2 -0.5 $1.4
1.055 1.054 1.054 1.054 1.054 1.053 1.054
+0.4
-0.01
+o
9
+0.01 +0.4
Av.
74.39
+0.6
1.2 1.G 0.7 1.1 0.1 0.3
Av.
59.40 59,77 58.91 59.31 58.32 57.93 58.96
58.18
+0.8
PROCEDURE
Approximately 50 ml. of saturated cerous sulfate generating solution were placed in the titration cell. The nitrogen was turned on to mix and deaerate the contents of the cell. The wave length was set a t 320, 300, or 375 mp, depending on the current level, and the absorbance dial was set a t 0.200. With the dark current balance a t zero, the photocell mas exposed and the slit width adjusted t o zero galvanometer deflection. The sensitivity ' knob was maintained a t a constant position 3 revolutions from its counterclockwise limit. The current was applied and the reagent blank measured. After generating ceric ion until the slope of the absorbance-time plot appeared constant, an aliquot of arsenious acid sample and 1 to 2 drops of osmium tetroxide solution were added and the contents of the cell mixed thoroughly. The absorbance mas noted and sample A titrated. Subtracting the current-time product for the blank from the product for sample A should result in the theoretical product for the amount of arsenious acid present. However, it was not found possible to determine a reliable blank correction for any given amount of generating medium; therefore, it was found necessary to generate ceric ion to an arbitrary absorbance, add an excess amount of arsenious acid, and generate ceric ion to an end point determined by a plot of absorbance us. time. Then, upon adding a succession of samples and titrating coulometrically, it was found that good results mere obtained by integrating the currenttime values between any two successive end points after the first. The initial absorbance setting must be set arbitrarily a t some positive value--e.g., 0.200-because the absorbance fa113 due to dilution after each new addition of arsenious acid. Six or seven 2-ml. samples could be analyzed in the same generating solution before the slope of absorbance us. time became too flat t o allow an accurate determination of the end point. Figure 2 shows part of a typical run. It is possible to take the initial absorbance after each addition of sample as the horizontal portion of the graph because nothing present in the solution absorbs light a t the wave length employed until the end point is passed. At 360 and 375 mp the drop in absorbance through dilution was negligible until after three 3-ml. samples had been added. RESULTS
ured by taking the voltage drop across the resistance of an Otto Woulf decade box with a Leeds & hTorthrup No. 7665 potentiometer. An S-0 clock, Standard Electric Time Co., having a start-stop error of 0.01 second, was used to measure the generation time. The circuitry was similar to that described by Cooke, Reilley, and Furman ( 3 ) . A dummy resistance of 1000 ohms was added to the circuit shown there to eliminate current fluctuation on interruption. The generating anode was a piece of bright platinum-iridium foil (10% iridium) measuring 2 X 2 cm. The isolated cathode was a lead amalgam-lead sulfate-1N sulfuric acid half cell. The side arm plug was constructed of paper saturated with I S sulfuric acid (7).
The results are presented in Table I. KO results are reported for quantities of arsenic less than 58 y, although smaller samples could be done by this method a t lower generating current levels. The limit of sensitivity is governed by the relatively smaller absorbance increase in lomer generating current levels leading t o
REAGENTS
The generating medium was a saturated solution of cerous sulfate in 1N sulfuric acid prepared from the reagent grade octahydrate (G. F. Smith Chemical Co., Columbus, Ohio). The concentrated sulfuric acid was distilled in a n effort to minimize the blank. The arsenious acid was prepared from National Bureau of Standards primary standard arsenious oxide in the usual manner to obtain 0.liV stock solutions. Because arsenious oxide is a primary standard, the normalities were calculated by weight and checked by titration with standard 0.099i9iV potassium permanganate, using osmium tetroxide as catalyst. The titrimetric standardizations agreed with the calculated values to an average deviation of 3 parts in 10,000 for six titrations. Aliquots of a stock solution were measured out with calibrated glassware and diluted to volume with 1N sulfuric acid to obtain solutions of the correct normality. These dilute stock solutions were made up daily. Portions of diluted stock were transferred by cilibrated pipet directly into the titration cell.
TIME (SEC.) Fii:ure 2.
Titration graphs for 58.18 y of arsenious acid
BL is attempted titration of blank: A. B , and C correspond to titration of known quantities of arsenio(II1).
V O L U M E 28, NO. 4, A P R I L 1 9 5 6 inaccuracy in finding the true end point. Larger samples are possible but the simple battery arrangement used here limited usable constant current level. DISCUSSION
Accurate results are attainable with this method if care is taken to oxidize the trace impurities present. I n previous work on ceric oxidimetry the effect of trace impurities has not been evident, largely because the sensitive end-point procedure of Cooke and others (2) has been applied (4,7 , 8). Any reagent blank is eliminated in this method by the initial generatifon of ceric ion to the preset end-point potential as determined by a semimicro potentiometric titration. Although several papers have appeared on the use of photometric techniques to determine the end points in coulometric titrations, this paper is the first t o deal with the use of the high ceric absorbance in the near ultraviolet region. The determinations reported here, some unpublished data on brominations done in this laboratory ( l o ) ,and the recent work of Everett and Reilley (5) represent the first in which advantage has been taken of strong absorbances in the near ultraviolet to give sensitive end points. With few exceptions, coulometric methods show an absolute error of the general order of 1 y. Under exceptionally favorable circumstances significant results may be obtained in the submicrogram range, as in the titration of permanganate with ferrous
517 ion ( I S ) . Contributing factors which tend to limit the loxvest range for any particular method appear to be (1) unfavorable reaction kinetics, (2) trace inipurities in reagents, and (3) possible consumption of the oxidant to form the initial oxide layer on the inert generating electrode or reduction of this oxide layer by the reductant. LITERATURE CITED
(1) Bricker, C. E., Sweetser, P. B., A N ~ L CIIEU. . 24, 409 (1952). (2) Cooke, W. D., Reilley, C. S . , Furman, X. I I . , Ibid., 23, 1662 (1951). (3) I b i d . , 24, 205 (1952). (4) Dilts, R. V., Furman, K. H., Ibid., 27, 1275 (1955). (5) Everett, G. W., Ileilley, C. N., Ibid., 26, I750 (1954). (6) Farrington, P.S., Swift, E. H., Ibid., 22, 889 (1950). (7) Furman, N. H., Adanis, It. N., Ibid., 25, 1504 (1953). (8) Furman, X. H., Bricker, C. E., Dilts, It. V., Ibid., 482 (1953). (9) Furman, N. H., Cooke, W. D., Iteillcy, C. S., Ibid., 23, 945 (1951).
(10) Handleman, J. W., Princeton University, Princeton, N. J., un-
published data. (11) Myers, R. J., Swift, E. H., J . Am. Chem. SOC.70, 1047 (1948). (12) Pitts, J. N., Jr., DeFord, D. D., Martin, T. W., Schmall, E. A., ANAL. CHEM.26, 628 (1951). (13) Ramsey, W. J., Farrington, P. S., Swift, E. H., Ibid., 22, 332 (1950). (14) Tutundaic, P. S., AIladenovic, S., Anal. Chem. Acta 12, 390 (1955). (15) TT’ise, E. N., Gilles, P.TV., Reynolds, C. A., Jr., AN.4L. CHEM. 25, 1344 (1953). RECEIVED
for review November 19, 1955.
Accepted February 1, 1956.
Table Suitable for Mounting a Microchemical Balance AL STEYERMARK,
E.
D. INGALLS, and J. W. WILKENFELDT
Hoffmenn-Le Roche / n e , Nutley, N . J.
A table suitable for mounting a niicrochemical balance in an area in which there are vibration disturbance& has been constructed, the essential parts being a reinforced concrete block on deflected springs. The springs reduce transmission of the vibrations present, while the block acts as an inertia mass to reduce the movement that is caused by normal manipulation of the balance. Oscillograph recordings showed that the system has an isolation efficiency of about 9370. To prevent the operator from coming in contact with the resiliently mounted system, except through normal manipulation of the balance, protective woodwork was incorporated. Ten microchemical balances (including four different makes), mounted on tables of this type, have precisions which range from 1.1 to 2.6 y , in spite of the fact that they are located on the floor above a manufacturing area. ,
T
HE importance of a suitable table for mounting a micro-
cheniical balance has been stressed by a number of authors (1-3, 6, 8-12, 14, 18, 21, 24, 24), because it is recognized that vibrations and shocks from esternal sources cause damage to the knife edges. Pregl(18, 21), Emich (6),and Alber and Harand (1, 2) stressed the generous use of lead as a nieans of absorbing vibrations. Various types of spring systems (large and small, damped and undamped) (7, 10, 12, 16, 17, 22); t x o heavy plates separated by dry sand (26); and rubber ( I S ) , including balls (IS, 14), have been used. Tables constructed of cork, brick, n-ood, lead, and stone slabs have proved very satisfactory in the laboratories
of this company ($3, 24) as well as in others ( 6 , 19, 25). However, during the process of expanding this company’s laboratories, a new location was selected in n-hich there is considerable vibration caused by the presence of a number of centrifuges and pumps. A table of the type used in the old laboratories, as well as various modifications, did not absorb vibrations to the desired extent. Consulting engineers, specializing in vibration problems (Korfund Co., Inc., 48-15 32nd Place, Long Island City 1, N. Y . ) , advised that the use of springs, which would reduce transmission of the vibrations, would offer the best me:ins of isolation. Balances thus mounted are, liowxw, in danger of being damaged owing to movement of the system caused by the force exerted on it by the operator during normal operation of the balance. To minimize this effect, each balance mas placed on a separate reinforced concrete inertia block mounted on deflected springs, the block acting as an inertia mass to reduce this movement. It is desirable to have the mass as large and the center of gravity of each block as low as practical, Tr-ith the springs spaced as far apart as possible for increased stability. Each location presents its own vibration problem which should be solved independently, but the same principles are applicable arid the selection of the proper combination of springs and inertia mass should make it possible to construct a suitable table. The manner of solving the problem in, these laboratories mill serve as an example. Vibration isolation depends upon the ratio of the frequency of the disturbing vibration, F D , and the natural frequency of the isolation mountings, Fs, the efficiency being expressed by the formula ( I 6 ) ,