V O L U M E 24, NO. 7, J U L Y 1 9 5 2 Table I.
Titration of Zinc with Potassium Ferrocyanide
(Drop time 1.8 seconds. Room temp. E = -1.30 volts. 0.0495 M KIFe(CiY)s. 0.073 .If ZnSOd) Zn++ 6 0.1 KCI. s KdFe(CN)o, Soh., HzS04, NO. hI1. MI. nr1. 111. 1 2.00 0.0 30 0 1.82 2 2.00 0 0 30 0 1.78 3 2.00 0.0 30.0 1.825 4 2.00 0.0 30.0 1.81 5 2.00 0.0 30.0 1.81 6 30.0 2.00 0.0 1.825 7 30.0 2.00 1.92 1.0 8 30.0 2.00 2.0 1.95 9 30.0 1,925 2.00 3.0 10 30.0 2.00 4.0 1.94 11 2.00 5.0 30.0 1.93 12 2.00 1.0 30.0 1.92 2.00 1.0 30.0 1.93 , 13 14 2.00 1. 0 30.0 1.94 15 1.0 30.0 1.95 2.00 16 2.00 1.0 30.0 1.94 17 2.00 2.0 30.0 1.95 18 2.00 2.0 30.0 1.93 19 2.00 2.0 30.0 1.93 20 2.00 2.0 30.0 1.95 21 1.94 22 9.70 23 9.72 24 9.72 25 9.70 26 4.85 27 4.85 28 4.84 29 4.84 1.939 m1./2 ml. of Zn A\7
Experiments 7 to 11 were performed to study the effect of free acid on the end point of the titration. Graphs of these and all other titrations showed, before the end point had been reached, the same initial curvature f o l l o ~ e dby a straight line. This curvature became more extensive as the free acid content was increased. Thus those titrations in which larger amounts of acid were employed gave a smaller number of points on the straight-line portion of the curve. Several preliminary titrations RThich were performed a t a temperature of 25" f 0.1" C. gave similar graphs.
1199
As the ratio of the volume of the zinc solution to the potassium chloride was constant, it is probable that the solubility of the precipitate in the free acid was responsible for this initial curvature. The end points in these experiments do not show much variation. As the end points of titrations in which smaller amounts of free acid were present were easy t o ascertain, all the remaining experiments were performed with 1 to 2 ml. of free acid per 30 ml. of supporting electrolyte. The mean of experiments 12 to 29 is 1.94 ml. of potassium ferrocyanide solution per 2 ml. of zinc solution, with a standard deviation of 0.015 ml. The ratio of the moles of titrant to zinc is 1 to 1.52 or 2 to 3.04. This indicates that the reaction was probably taking place according to the following equation: 3Zn++
+ 2Fe(CS)---- + 2 R f
+ IinZna[Fe(CN)6]2
The range of acidity over which this equation appears t o hold and ab the same time to give easily interpreted graphs is approximately 0.18 to 0.35 S in sulfuric acid. Finally, the errors are compatible with this type of procedure. SUMMARY
Small amounts of zinc have been titrated with potassium ferrocyanide in the presence of free sulfuric acid with a fair degree of accuracy. The reaction appeared to follow the accepted equation of these reactants, in that KtZna[Fe(C?;)6]2 was a product of the titration. LITERATURE CITED
Neuberger, A , , Z . anal. Cheni., 116, 1 (1939). (2) Nimer, E. L., Hamm, R. E., and Lee, G. L., ANAL.CHEM.,22,790
(1)
(1950). (3)
Riccoboni, L., and Goldschmied, P., Proc. X l t h Intern. Congr.
(4)
Spalenka, hI., Collection Czechosloa. Chem. Communs., 11, 146
Pure and Applied Chem. ( L o n d o n ) , 1, 199 (1947). (1939).
RECEIVED for review January 18, 1952. Accepted April
1 1 , 1952.
Analysis of Mixtures of Chloride and Bromide by Ion-Exchange Chromatography WILLIAM RIEMAN I11 AND SIEGFRIED LINDENBAUM School of Chemistry, Rutgers University, Neu, Brunswick, N. J .
HE determination of chloride and bromide in a mixture of the two has long been a difficult analytical problem. -4tteberry and Boyd ( 1 ) have demonstrated that the halides can be separated by ion-exchange chromatography, but their data are expressed only as a graph which leaves some doubt concerning the completeness of the separation. This note reports an investigation, started before the publication of the paper by Atteberry and Boyd, on the quantitative analysis of chloride-bromide mixtures. APPARATUS AND REAGENTS
Amberlite XE-67, a strong-base anion-exchange resin, was used in a 16.8 em. X 3.83 sq. cm. column. It required no preliminary treatment except the removal of the fine particles by decantation and treatment with 0.60 M sodium nitrate in a column until the eluate contained no chloride. The eluant was 0.60 M sodium nitrate containing 0.40 ml. of Cutscum per liter. This is a nonionic wetting agent purchased from Fisher Scientific Co., N. Y . Cutscum was added a t first to ensure thorough wetting of the siphon pipet and hence a constant volume of delivery. It was found later that the addition of Cutscum permitted an increase in flow rate without the undesirable effects (flattening and tailing of the elution graphs) that usually accompany such an increase. Silver nitrate, approximately 0.1 M , was standardized by the titration of pure sodium chloride or potassium bromide dissolved in the eluant solution, with potassium chromate as indicator. Thus the conditions of standardization closely resembled those of the analysis.
D
-1 siphon pipet, used to deliver equa volumes of eluate, is illustrated in Figure 1 The drainage tube, D,must be large enough (4-to 6-mm. internal diameter) so that the entire sample is delivered before a drop forms a t T . The overflow tube, 8, must have a small diameter (2- to 3-mm. internal diameter) to prevent the solution from draining instead of siphoning. The overflow tube should be close to the neck, N , of the receiver, R, to minimize the effect of small changes in the position of the siphon pipet. The end of the delivery tube, E, should be ground open on one side for about 2 cm. to give complete drainage. PROCEDURE
F i g u r e 1. Siphon Pipet
Weighed quantities of sodium chloride and potassium bromide (not over 2.5 millimoles of either) were mixed and dissolved in 2 ml. of water in a small beaker. The solution was poured on the top of the column, and the beaker was rinsed with a small volume of eluant. Then the eluant was passed through the column at a rate of 1.0 cm. per minute. Fractions of 7.85 ml. of the eluate were collected with the siphon pipet. Each fraction (sometimes a mixture of two con-
ANALYTICAL CHEMISTRY
1200 secutive fractions) was titrated with standard silver nitrate after the addition of 1 drop of 1 M potassium chromate. The capacity and interstitial volume, Ti, of the column were determined as follows: hydrochloric acid, 0.60 M , was passed through the column until the eluate had the same composition as the eluant. After draining the liquid in the tube t o the level of the resin, 0.60 .$I sodiuni nitrate was passed through the column until all the hydrogen and chloride ions were removed. The total quantity of hydrogen ion in the eluate expressed in millimoles (determined by titration with sodium hydroxide) was divided by 0.60 to find the interstitial volume. It was 35.5 ml. The total quantity of chloride ion (determined by titration with silver nitrate) minus the quantity of hydrogen ion represents the chloride originally present as RCl-i.e., the total capacity of the column. It was 78.0 me. Since the column contained 27.1 grams of oven-dried resin, the capacity \vas 2.88 me. per gram of dry, chloride-form resin. RESULT 9
Khen 1.447 niniol. of chloride and 1.744 mmol. of bromide xere taken, no halide was found in the first five fractions of 7.85 ml. each. Successive fractions contained the following amounts of halide: 2, 16, 68, 176, 325, 389, 290, 124, 44, 10, 4, 0, and 0 micromoles. The total is 1.448 nimol. and represents the chloride content of the mixture. Thereafter, fractions of two pipetfuls (15.70 mi.) were titrated. The quantities of halide found in successive fractions were 22, 98, 246, 368, 386, 303, 189, 90, 36, 10, 4, 0, and 0. This total is 1.i52 mmol. and represents the bromide content of the sample. -4plot of ill,the molarity of the halide in a fraction, against Li, the total volume of eluate, reveals a fairly close conformity to Equation 1
bromide. Three analyses by this method of a sample of reagentgrade potassium bromide yielded 0.24, 0.23, and 0.26% chloride. The label indicated that the reagent contained not over 0.3% c,hloride. A correction was applied for the chloride content of this reagent when it was used in halide mixtures. The major source of inaccuracy in the foregoing determinations is undoubtedly the cumulative error in the titmtion of ten or more fractions for the determination of each halide. Seven mixtures of chloride, bromide, and iodide ivere analyzed by a slightly different tcvhnique in which all the eluate fractions containing any one halide were coinbiried before the tit,ration, which was done potentiometrically. Thc errors for chloride ranged from +0.002 to +0.012 mmol., for bromide from 0,001 to -0.008 mniol. The Imwncc of iodide in these mixtures does not interfere with the determination of chloride arid bromide, but 1800 ml. of 0.6 .If sodium nitrate must be passed through the column in order to remove the iodide bcfore the column can he used again. The iodide is contained in so large a fraction of the eluate that its tlct~erminationby this method is not recommended. The research is being cont’inuedin order to develop a good ion-exchange procedure for the determination of each halide in a mixture of all three. SUMMART AND CONCLUSIONS
The method described for the determination of chloride and bromide in halide niixtures by ion-exchange chroniatography requires less than 3 hours. The error in a single det,ermination of any one halide seldom exceeds 0.2 mole % of the total halide in the sample. ACKh-OWLEDGMENT
The authors are indebted t,o Robert Kunin of Rohni 8: Haas Co., Philadelphia, for the sample of rimberlite XE-67. with the following values for the parameters:
LITERATURE CITED
Cci = 2.2i, pci = 74, C B ~= 5.91, p~~ = 65
(1)
Atteberry, R. W., and Boyd, G. E., J . Am. Chem. Soc., 72, 4805
Other similar elutions yielded errors of +0.017, +0.004, and +0.008 mmol. for chloride and +0.010, -0.039, and $0.010 for
(2)
Beukenkamp, J., and Rieman, IT., SNAL. CHEM.,22, 582 (1950).
(1950). RECEIVED f o r re\-ie\v .January 11. 1852. Accepted April 17, 1952.
Derivative Polarographic Titration of Glucose RALPH N. ADAMS, CH.4RLES N. REILLEY,
AND N.
HOW-ELL FURMAN
Princeton L-niversity, Princeton, IV. J .
THE direct potentiometric titration of alkaline ferricyanide with glucose was investigated by Brit,ton and Phillips in 1940 ( 3 ) . .&]though this method seems to afford a rapid and precise determination of reducing sugars, it appears t o have found little application. This is probably due in part to the cumbersome experimental requirements of potentiometric t i h t i o n s at high temperatures. The method of derivative polarographic titration recently developed in this laboratory is ideally suited t o high temperature titrations (8). This method eliminates t.he reference half-cell and the connecting salt bridge, which are sources of experimental difficulty in high temperature titrations, provides continuous indication, and eliminates plotting of end points. Complete details of the theory and experimental technique have been published (8). The applicability of the derivative polarographic method to the ferricyanide-glucose titration was predicted from a stud? of typical polarogram of the ferricyanide-glucose system during the course of the titration. Such polarograms were realized experiment,ally using stationary platinum wire electrodes under conditions similar to the various stages in the derivative titration. r
Figure 1, A , shows t’hepolarogram of “pure” ferricyanide at the beginning of the titration. The slope of the polarographic xvave,
d i / d E , as it crosses the zero current axis is very small and thus the experimentally measured derivative voltage, dE/di, is high. The lower dotted portion of this curve indieat- a point early iii the titration, where a small amount of glucose has been added. Ferrocyanide formation is indicated by the anodic portion of t h e wave. A sharp drop in derivative voltage is produced. In Figure 1, B, about, midway through the titration, the polarogram has the form shown, \>-herethe slope of the wave as it crosses t h e zero current line is very large and the measured d E / d i is close to, zero. Figure 1, C, indicates thP polarogram at the equivalence point. The dotted line represents either the reduction of un-knoivn electroactive species of the glucose system or the reductioii of hydrogen ions. In either case the derivative voltage is large and remains so after the equivalence point due to the irreversible nature of the glucose system. The expected course of the dei,iva-. tive voltage during the titration is shown in Figure 1, D. Thew predictions were verified esperimentally. In Figure 2 are plotted the data for a typical titration follon.ed under potentiomet,ric control and by the derivative method. I n all cases studied the derivat,ive end point was coincident with the: ordinary potentiometric equivalence point. APPARATUS AND EXPERIMENTAL
The apparatus used for derivative polarographic titration has been fully described (8). A 150-ml. beaker served as titration
