841
N O V E M B E R 1947 hydi ochloridc solution, using a separate portion of the sample, and the resulting solution is titratcd to pH 7 with a standard acid or base. h corresponding blank correction is applied to the titer of potassium hydroxide solutioh. None of thesc interfering compounds \vas encountered in the synthetic samples. Other ketones and aldchydes will interfere with the determination of acetone; hon evt’r, thtlir presence in light hvdrocarbons is rare. LITERATURE CITED
(1) Barthauer, G . L., Jones, F. V., and Metler, A. V., I N D .ENG. CHEM.,ANAL.ED., 18, 354 (1946). (2) Bennett, A . H., and Donovan, F. K., Analyst, 47, 146 (1922). (3) Berthoud, A., and Eichelberger, W.,Helv. Chim. Acta, 17, 23
(1934). (4) Busey, R. H., Barthauer, G. L., and Metler, A. V., 1x11. ENG. CHEY.,ANAL.ED.,18, 407 (1946). (5) Cassar, H. A, I n d . Eng. Chem., 19, 1061 (1927). (6) Duke, F. R., IND. ENG.CHEY.,ANAL.ED., 16, 110 (1044). (7) Fenwick, F., Zbid.,4, 144 (1932).
(8) Funk H. Schormuller J., and Hensinger, W., 2. anorg. allgem. Chem., 205, 361 (1932). (9) Hahn, F. L., 2. anal. Chem., 69,417 (1926). (10) Haughton, c. o., IND. ENG.CHEM.,-4NAL. ED., 9, 167 (1937). (11) Hickman, K. C. D., and Linstead, R. P., J . Chem. Soc., 121,
2502 (1922). (12) Iddles, H. A., and Jackson, C. E., IND. ENG.CHEM.,ANAL.ED., 6 , 454 (1934). (13) Kebler, L. F., J . Am. Chem. Soc., 19, 316 (1897). (14) Kepder, G.. 2.anuew. Chem., 18, 464 (1005). (15) M&asco, M , I n d . Eng. Chem., 18,701 (1926). (16) hlessinger, J., B e r . , 21, 3366 (1888). (17) Morton, A. A , , and Mark, J. G., I N D ENG. . CHEM.,A N ~ J ,ED.. . 6, 151 (1934). (18) Ogg, R. A., Jr., J . Am. Chem. Soc., 57, 2728 (1935). (19) Rakshit, J., Analyst, 41, 245 (1916). (20) Scarborough, J. B., “Numerical Mathematical Analysis,” p. 60, Baltimore, Md., Johns Hopkins Press, 1930. (21) Willard, H. H., and Furman, N. H., “Elementary Quantitative Analysis,” 3rd ed., p. 91, New York, D. Van Nostrand Co , 1940. RECEIVED August 19, 1946
Rapid Method for Analysis of Chlorides R. B. DEAN’ AND R. L. HAWLEYl Department of Chemistry, University of Hawaii, Honolulu, Hawaii A new electrometric end point for the titration of chlorides with silver nitrate is detected by the potential between a silver wire and a copper wire in a solution of copper sulfate containing the unknown chloride. A simple potentiometer and a microammeter are the only electrical equipment necessary. The errors are of the order of 0.2%. Of 44 ions tested for possible interference, 30 introduce no error. Of the 14 ions which do interfere, 2 (silver and mercurous) are incompatible with chloride ions, and 7 of the remaining 12 can be easily remoied.
T
HE elect,rometric determination of an end point has many advantages over classical colorimetric methods, since it does not depend upon a subjective impression and can be used under adverse lighting conditions. The determination of chlorides is one field where electrometric detection of the end point is particularly advantageous. Although many methods for the elcctrometric titration of chlorides have been described (1, 2, 9, 10, 12, 14, 1.5) most of them require a salt bridge to make connection between the reference electrode and the solution containing the indicator electrode ( 7 ) . A salt bridge introduces an error because chloride ions will diffuse out of and int,o the bridge. hlost salt bridges are also inconvenient, to use and not well adapted to multiple determinations. Perhaps the best way of avoiding the use of a salt bridge is t.o use the glass electrode of a pH meter in a solution of the sample that has been made about 1 11 nitric acid. A silver indicator electrode is attached in place of the calomel electrode and the apparent pII or voltage, E, is read on the pH meter as the chloride sample is tit’rated with silvcr nitrate. There is a rapid changc in E as the end point, is passed and the E corresponding to the largest value of dE/dVol. is the end point,. Once this value has been established for any particular system, the meter can be set a t this value and titrations carried out until the indicating needle or eye shows that the potential or apparent pH corresponding to the end point has been reached. This method, of course, requires an expensive vacuum tube voltmeter. Tungsten electrodes have been used as reference electrodes ( I d ) , but they are not stable and the potent’ial corresponding to 1 Present address, Department of Chemistry, University of Oregon, Eugene, Ore. 1 Present address, 6811 9th St., N. W.. Washington. D.C.
the end point drifts. This makes the tungsten electrode unsuitable for routine determinations. The method described in this paper makes use of a copper xire reference electrode in a strong solution of copper sulfate. This electrode system is completely reversible and reproducible. It has a low resistance and the potentials can be measured with a portable needle galvanometer and a simple potentiometer circuit. APPARATUS AND PROCEDURE
The copper electrode was a piece of 16-gage copper wire. Thc silver electrode was 22-gage and was sealed into a glass capillary for protection. It is possible to make direct seals b e b e e n 22gage silver wire and 3-mm. soft glass tubing, although some of the seals may crack on cooling. I t is convenient to attach the silver wire to a length of copper or bronze wire with silver solder before sealing on the glass tube. The copper sulfate reagent was made by adding 1.5% of 6 iV sulfuric acid to a saturated solution of copper sulfate. A small amount of chloride in this solution will do no harm if it is titrated as a blank before the unknown solution is added. Copper nitrate might be used to advantage, although it is more expensive. The electrical circuit, shown in Figure 1, consists of a simple potentiometer made from radio-resistors, a dry cell, and a microammeter used as a galvanometer. Two methods of calibration are possible. If the variablk rheostat (a 500-ohm tone-control potentiometer) is fitted with a calibrated dial the dial readings may be plotted against the buret>readings to produce a conventional potentiometric titration curve. The end point occurs at the dial reading which is in the center of the steepest portion of the curve. With saturated copper sulfate the end point is close to 200 millivolts. If the variable rheostat does not have a Calibrated dial the galvanometer deflection can be noted after successive increments of silver nitrate, and the rheostat then adjusted SO that the deflection of the galvanometer after addition of 1 microliter (0.001 ml.) of the silver nitrate solution is a maxi-
V O L U M E 19, NO. 1 1 ’
842 mum. The rheostat is, of course, adjusted to bring the galvanometer back to zero after each increment of silver nitrate. The end point in subsequent titrations is determined by the Muller method (6, 11). The circuit illustrated draws only 0.6 milliampere and the endpoint setting has remained constant for several weeks when checked against a standard cell. A commercial potentiometer calibrated with a standard cell could be used but would be more expensive. The method was tested with the apparatus contained in the portable analysis kit described elsewhere ( 5 ) .
.
Silver nitrate was added from a micrometer syringe buret (3, 4, IS) containing 1.6 ml. of acidified silver nitrate. Ten milliliters of copper sulfate reagent were placed in a short wide test tube which was suspeqded under the two electrodes and the buret tip, so that they all dipped beloTy the surface of the solution. A tube delivering air for stirring was also introduced into the test tube, as indicated in Figure 1. The galvanometer ordinarily indicated some chloride in the solution and silver nitrate was added from the buret until the galvanometer indicated zero. This buret reading was recorded. I n a series of determinations this first reading should be constant. The chloride sample was then added from a Krogh syringe pipet ( S , 8 ) . Approximately 0.2 ml. of a 0.563 N sodium chloride solution was used for standardization. This solution contains 20 grams of chloride per liter and corresponds to sea water in total halides. Silver nitrate was added until the galvanometer again indicated zero. The difference in the two buret readings is proportional to the added chloride. If the unknown sample is delivered from the same syringe pipet and titrated in the same way, th’e chloride content can be obtained by simple proportion from the buret readings. It may be convenient to adjust the strength of the silver nitrate solution, so that the buret reads directly in moles or grams of chloride per liter. The apparatus was calibrated by running a series of known solutions of sodium chloride and the silver nitrate used was found to be linearly proportional t o the concentration of chloride up to 0.563 N for samples containing 0.2 ml. of solution.
Micrometer-syringe buret
Hlltt-
Y
I
I
Figure 1. Electrical Circuit
Table I.
Ions Examined for Possible Interference with Analysis for Chloride
(Copper-copper sulfate reference electrode.) Interference As silver As chloride Arsenate Ammonium Silver Bromide Arsenite Bismuth Mercurous Cyanide Carbonate Cerous Nitrite Iodide Citrate Chromous Permanganate Thiocyanate Fluoride Cobaltous Chromate Sulfide Hydroxyl Cupric Dichromate Thiosulfate Nitrate Ferric Ferrocyanidea Ferricyanide Oxalate Ferrous Picrate Hydrogen Ortho hosphate Manganous Pyropfosphate Mercuric Sulfate Plumbous Peroxydisulfate Potassium Sulfite Sodium Tartrate Zinc a Drops of ferrocyanide solution become covered with semipermeable cupric ferrocyanide membranes and probably sequester some chloride.
K O Interference
INTERFERING SUBSTARCES
A number of ions have been tested for possible interference. I n all case ions that are reported as not interfering were present in amounts equal to or greater than the chloride ion (0.563 N ) . Only qualitative tests were run on the extent of interference, since this will depend in many cases on the total amount of chloride present in the sample. Some ions reacted almost quantitatively as chloride; a few reacted as a negative quantity of of chloride (or as silver) but not always quantitatively. Some of these may have oxidized the chloride t o chlorine since the solution was about 0.01 N in sulfuric acid. Table I lists the ions examined and indicates the nature of the interference. Ions a t the top of the list interfered more than ions further down. There is a slight error produced by the diluting effect of the added sample and the silver nitrate solution from the buret. Dilution changes the potential of the copper-copper sulfate electrode, but the effect is aithin the limits of error of the method if the volume of the added solutions is kept small. Less concentrated copper sulfate can be used with only slight increase in this error. If copper sulfate formed an ideal solution the potential change a t the copper electrode produced by 20% dilution would be (RTIBF)In 1.20 which a t 25’ is about 2.5 mv. Saturated copper sulfate is very far from an ideal solution and the actual potential change is about 1 mv. when 2 ml. of water are added. This produces an error of less than 0.2%. Many of the ions which interfere, including pyanide, sulfide, thiosulfate, nitrite, permanganate, chromate, and dichromate, can be removed by warming the unknown solution with sulfuric and oxalic acids. The current flow between the electrodes before the end point is reached deposits silver chloride of the order of 10-8 equivalent per minute on the silver electrode and copper on the copper electrode. This is the chloride contained in 1 ml. of 10-6 molar chloride and produces a negligible error. The plating, however, constantly renews the electrode surfaces and probably enhances the stability of the system. The titration solution should be
removed as soon as the end point is reached. The electrodes need only light rinsing to remove silver chloride clumps, since there is neither excess silver nor chloride in the solution a t the end point.. ACKNOWLEDGMENT
The authors wish to acknowledge with thanks the assistance of
J. Y . Nitta in preparing the illustration. LITERATURE CITED
Best, R. J., J . Agr. Sci., 19,533 (1929). Cunningham, B., Kirk, P. L., and Brooks, 9. C., J . BWZ. Chem., 139, 11 (1941). Dean, R. B., ANAL.CHEM.,in press. Dean, R. B., and Fetcher, S. F., Science, 96,237 (1942). Dean, R. B.,and Hawley, R. L., Pacific Sci., 1, 108 (1947). Kolthoff, I. M.,and Laitinen, H. H., “pH and Electro Titrations,” 2nd ed., pp. ~.111-13, New York, John Wiley & Sons, 1941.
Koltoff, I. M., and Sandell, E. B., ”Textbook of Quantitative Inorganic Analysis,” rev. ed., pp. 505-9, New York, Macmillan Co., 1943. Krogh, A,, IND.ENG.CHEM.,ANAL.ED.,7, 130 (1935). Lehman, J., Acta Paediat., 26,258 (1939). Linderstr6m-Lang, K., Palmer, A. H., and Holter, H., Z . physiol. Chem., 231, 226 (1935).
Mtlller, E., “Die electrometrische Massanalyse,” 5th ed., Dresden, Theodore Steinkopf, 1932. Reid, A., Sen Gupta, N.C., and Gogoi, N. N., J. Indian Chem. Soc., 21, 154 (1944). Trevor, J. W., Biochem. J., 19,1111 (1925). (14).W e s t .. L. E.. and Robinson. R. J.. J. Marine Research (Sears Foundation), 4, 1 (1941). (15) Yeck, R. P., and Kissin, G. H., IND. ESG. CHEM.,ANAL.ED., 17, 692 (1945). RECEIVED January 18, 1947. Presented i n part before the North Western Regional Meeting, AMERICAXCHEMICALSOCIETY,M a y 2 and 3, 1917. Research paper 6, Cooperative Fisheries Research Staff, Territorial Board of Agriculture and Forestry, and Cniversity of Hawaii.