509
V O L U M E 24, NO, 3, M A R C H 1 9 5 2 in the adsorption will overcome the difficulties encountered in analyzing residua. However, no investigation of the use of clay was undertaken in the present work. CONCLUSION
The strong affinity of silica gel for aromatic compounds has been utilized to obtain an accurate, relatively rapid method of analysis for total aromatics in heavy petroleum distillates. Though the method is similar to that described some time ago by Lipkin et al., certain modifications make it suitable for routine determinations in laboratories and relatively rapid when several determinations are made simultaneously by the same operator.
ACKNOWLEDGMENT
Permission of the management of the Humble Oil and Refining Co. to release the material presented herein is gratefully acknowledged. An expression of thanks is also due V. H. Rushing, whose assist,ance throughout the work was invaluable. LITERATURE CITED
(1)
Lipkin, M. R., Hoffeoker, W. A., Martin, C. C., and Ledley, R. E.,
ANAL. CHEM.,20, 130 (1948). (2) Mills, I. W., Proc. Am. Petroleum Inst., 29M (III), 50 (1949). (3) Wanless. G. G... Ebv. . L. T.. and Rehner. John, Jr., . ~ N A L .CHEM.,
23, 563 (1951). RECEIVED for review April 12, 1951.
Accepted January 14, 1952.
Rapid Potentiometric Determination of Chloride at low Concentrations tl
t n Solutions of High lonic Strength W. J. BLAEDEL, W. B. LEWIS', AND J. W. THOMAS2 University of Wisconsin, Madison, Wis. st stand rd methods of chloride determination possess severe shortcomings when applied to the rapid and routine determination of low concentrations of chloride in solutions of high ionic strength. A potentiometric method is described which allows detection of a limiting chloride concentration of 2 X lo-' M (0.7 microgram per 10 ml.) in solutions containing about 1 M sulfuric acid and/or sodium sulfate. The relative error is about 270 at chloride concentrations from 3 X 10-4 to 0.05 M . The method
T
HERE exist many methods for the analysis of chloride, but only a few are applicable to the rapid, routine determination of low chloride concentrations in solutions of high ionic strength. Such determinations are involved in the analysis of chemicals for traces of chloride, of air for chlorine-containing gases after absorption and hydrolysis in absorbing solutions, and of organic compounds after Eschka or Parr bomb fusion. Several analytical methods were investigated with a view toward &ding one that was capable of detecting a limiting concentration of chloride of loa M or less in solutions containing acids and/or salts a t moderately high concentrations-Le., equivalent to about 1 M sulfuric acid or sodium sulfate. As the objective was to apply the procedure to routine determinations on large numbers of samples, speed and simplicity were important requirements. For concentrations of chloride well above the limiting concentration, a relative error of 1 to 3% waR considered tolerable for most of the kinds of analyses mentioned above. A great deal of preliminary experimental work indicated that most accepted methods of chloride analysis did not satisfy these requisites. The high ionic strength rendered turbidimetric and nephelometric methods very unreliable. Direct titration with standard silver nitrate, using adsorption indicators like dichlorofluorescein or diphenylamine blue, was unsuccessful because of the high ionic strength and- low chloride concentration. The Mohr method, using chromate as an indicator, was not sensitive enough. Indirect argentimetric methods, such as those of Volhard, McPresent address, Los Alamos Soientific Laboratory, Los Alamos, N. M. 'Present address, Bureau of Dairy Industry, U. S. Department of Agriculture, Beltsville, Md.
involves measuring the voltage difference between two silver-silver chloride electrodes in a concentration cell, one arm of which contains the unknown solution and the other a standard solution. The chloride content of the unknown is found by the aid of a nomogram. Application to the determination of phosgene or cyanogen chloride in air after absorption in alcoholic sodium hydroxide is described. A pair of workers may prepare and analyze over 200 samples per day.
Lean and Van Slyke, and Liebig, were satisfactory from the standpoint of sensitivity and accuracy; but the manipulation and care required made them poor for routine use. The precision of such indirect methods, Khen used routinely, falls much farther below the nonroutine precision by which the method is gaged than does the precision of direct methods. The mercurimetric method, involving direct titration of the chloride with standard mercuric nitrate, was not sensitive enough and required large and variable corrections for low chloride concentrations. Potentiometric titration methods, both direct and differential, possessed sufficient sensitivity and precision, but required too much time. The common potentiometric methods consist of measuring the voltage of a concentration cell, one half of which is an appropriate electrode immersed in the unknorn solution, and the other half is a standard half-cell. Most of these involve troublesome and time-consuming corrections for liquid junction potentials and ionic strength effects. Furman and Low (8), however, have devised a simple differential method which eliminates error due to these two effects. The method was conceived primarily for measuring low chloride concentrations in solutions of high ionic strengths, and may be easily adapted to routine analysis. THEORY O F DIFFERENTIAL POTENTIOMETRIC METHOD
Briefly, the method consists of measuring the voltage of a silver-silver chloride electrode immersed in the unknown solution against an identical electrode immersed in a standard 0.00100 M sodium chloride solution. Both solutions must have practically
ANALYTICAL CHEMISTRY
510 identical composition except for chloride concentration, which contributes only negligibly to ionic strength, eo activity and junction effects cancel. From the voltage of the cell, the unknown chloride concentration is computed. The theory is fully discussed in the original articles ( 2 ) , but is outlined below in altered form to suit the procedure used in this article.
If the molar chloride concentration in the unknown solution is originally x,it becomes x y a t the electrode surface after reaching equilibrium with the silver chloride electrode, y being the contribution due to solubility of the electrode. When x is low (less than 10-4 M ) , y is appreciable and may even exceed x. The molar concentration, a, of chloride in the standard solution is made so large (0.00100 $1)that the solubility of silver chloride contributes negligibly to the chloride concentration. As both solutions are almost e ual in ionic strength and composition, except for the small chqoride concentration, activity effects and junction potentials need not be considered. In the following experiments, the supporting electrolyte contains 0 20 M sodium sulfate and 0.40 M sulfuric acid. The cell and its voltage, E, at 25” C. may be written
+
Solving Equation 4 for x, a
x = 10E/0.05912 -
sx
10E/0.05912 a
(5)
For routine analyses of the kinds mentioned in the &st paragraph of this article, it is easy to keep composition and temperature constant in all determinations; hence S is constant. By measuring the voltage for solutions of known low chloride content, S may be found for the conditions used. Since S and a are known, the unknown chloride Concentration, x, is calculable from the observed voltage, E, according to Equation 5. The calculations involved in the use of Equation 5 are timeconsuming. However, the use of Figure 1 reduces the time of calculation to a few seconds per sample. Figure 1 consists of two parts. At low chloride concentrations (less than 25 X 10- M ) , where the last term in Equation 5 is appreciable, x depends upon the two variables, S and E, and may be calculated using the nomogram of three vertical scales At higher concentrations of chloride (greater than 25 X 10” M), the last term of Equation 5 is negligible, and z depends upon E only, being found by using the horizontal colinear scales in three segmgnts in the upper part of Figure 1.
E = 0.05912 logloL
X+Y
PROCEDURE
Figure 2 is a diagram of the concenttation cell in which the determinations are carried out.
LOWER SCALES,
CHLORIDE MOLARITYx 10:o -
55-
:
E,(mL4110 Figure 1. Nomogram and Scales for Computing Molar Chloride Concentrations
The molar solubility, y, of silver chloride in the unknown solution is a variable quantity, dependent upon x and the solubility product, S.
+
?Ax Y) = s Eliminating y between Equations 2 and 3,
(3)
The electrodes, d and d‘, are easy to prepare and keep well. The cell is mounted on a board and immersed in ct thermostat a t 25” zt 0.5” C. to level c-c’ in Figure 2. To make an analysis, the left arm of the cell is filled well above level c-c’ with standard soluti:n and the right arm with unknown solution, with stopcocks a, a , and b all closed. The bridge is made by momentarily opening stopcock bot0 the position shown in Figure 2, and then turning through 180 . Stopcocks a and a‘ are momentarily opened to equalize the solution levels in the two arms of the cell, and then closed. Siphoning between arms is not appreciable if c and c’ are on the same level. The cell voltage is read after about a minute to the nearest few tenths of a millivolt. The unknown solution is then pulled out of the right cell arm by momentarily turning stopcock b to a positiot 90” clockwise from that shown, and then back to the position 180 from that shown in Figure 2. -4little more standard solution is added to the left cell arm to bring the level back above c-c’, and the cell is ready for the next sample. The voltage read from the potentiometer must be corrected for any asymmetry potential between the electrodes. This corrected voltage is then wed to determine the chloride molarity in the unknoxn solution using Figure 1, knowing the value of S for the system under consideration. The chloride concentration found in this way must be corrected by subtracting the blank, which is determined in the same way as an unknown. Notes to Procedure. 1. The electrodes used in this work were prepared by the method of Brown (1). These were 1 sq. cm. platinum sheet electrodes, first coated electrolytically with silver and then chloridized. Electrodes prepared in this way are plum-colored, and show differences less than 0.5 mv. They are stable for several months when not in use, and last for hundreds of analyses. They may be used in daylight, but not direct sunlight. When not in use, a pair should be short-circuited in acidulated potassium chloride. Several pairs should be kept on hand. They should never be allowed to dry. In recent work, good results were obtained using for electrodes five-turn helices (6 mm. in diameter X 2 cm. long) made from silver wire (1 mm. in diameter), soldered to copper wire leads, and sealed into the glass tubing with Apiezon cement. These electrodes were chloridized as described above, were much simpler to prepare than the platinum ones, and showed even smaller asymmetry potentials. The stability over a week of use seemed ae good as for the platinum electrodes, but the long-term stability was not studied. These electrodes are worth further investigation. Admittedly, this method requires some time in preparation prior to analysis, but this is not considered a great disadvantage. In routine analysis, where results are desired as soon as possible after collecting samples, it is the time required during analysis which is a t a premium. 2. Temperature control need not be critical. Measurements
V O L U M E 2 4 , NO. 3, M A R C H 1 9 5 2
51 1
LLI
-
I
both level
i d
SCALE
Ob-----+/“
.
W Figure 2.
Concentration Cell
L
throughout the range of chloride concentratioons from 10-5 to 0.05 M showed variation less than 0.3 mv. per C. around 25” C . A relatively crude thermostat suffices. 3. The standard solution should be prepared to have the same composition as the unknown solution, within 2 to 4%, except for chloride concentration. In routine analyses, where samples are always prepared in a certain way, this is easy to do simply by measuring the quantities of materials involved. The composition need not be critically controlled. Voltages were read between a standard solution (composition: 0.00100 M sodium chloride, 0.30 M sodium mlfate, and 0.60 M sulfuric acid) and pairs of “unknown” solutions containing chloride conto 10-6 M. One member of each pair had centrations from the same concentration of sodium sulfate and sulfuric acid as the standard, and the other member had 10% less. The differences between members of all pairs of unknown solutions were only 0.2 to 0.4 mv. To use Figure 1, the concentration of sodium chloride in the standard solution should be 0.00100 M ,within 1%. I t is best to standardize this solution after preparation, because of chloride impurities in other reagents used to prepare the standard solution. 4. By storing the unknown and standard solutions in the thermostat for a short time prior to introducing them into the cell, no appreciable time is required for temperature e uilibrium. 5. A student potentiometer is satisfactory. UBe a storage battery instead of a dry cell reduces the frequency of checking the potentiometer against the standard cell. If there is a galvanometer light, it should have its own power source. A switch for reversing electrode polarity saves much time when solutions having chloride concentrations around 10-3 M are being analyzed. 6. When consecutive analyses are made, the old solution may be rinsed from the unknown arm with portions of the new unknown to be analyzed. The standard solution need not be withdrawn completely with each sample, but small portions of standard should be added to keep the solution fresh and the level up. When the apparatus is not used for a time (as overnight), both arms should be rinsed and filled with the same acidulated potasBium chloride solution, and the electrodes short-circuited. 7. The asymmetry potential is easily determined by filling both cell arms with standard solution and measuring the voltage.
07
This voltage is then subtracted from the voltage reading on the unknown, and the resulting difference is applied to Figure 1 to calculate the chloride concentration of the unknown. The asymmetry potential does not depend upon chloride concentration. The dependence of this potential upon large changes in composition was not studied, but small-i.e., lO%-changes in composition have no appreciable effect. When the asymmetry potential becomes greater than 1 to 2, mv., or when voltage readings drift or fluctuate during an analysis, one or both electrodes have deteriorated and must be replaced. As the asymmetry potential may also change slightly during the first few hours of use after being short-circuited, it 1‘ well to check it each half-hour or so until constancy is reached. 8. Experience with many electrodes has shown that S is the hame for all electrodes, even those with high asymmetry potentials and different colors. However, S does increase with composition of the solution and temperature. S is best determined experimentally for a particular solution by measuring the vcltages of two solutions of the folloming chloride contents in the desired supporting electrolyte against the st:tnd:ird chlo~idesolution in the same supporting electrolyte: (1) blank, (2) blank plus a low, known concentration of chloride (snv 2.0 X 10-j X ) . These two voltages give two simultaneouu tJquations of the type of Equation 5, which may be solved for iS and the value of the blank. Or, more rapidly, the value of S may be found by trial and error from these two voltages using thp nomogram of Figure 1. As checks, it is desirable to run a third, or even fourth, solution a t somewhat different chloride concentrations. This procedure establishes S with a precision about i o . 5 x 1o-~oi112. The value of S need not be accurately known, because all chloride concentrations are obtained as differences by subtracting the blank from the total chloride found in the unknown. A particular error in S causes error in the same direction in both quantities involved in the difference, and the error in the soughtfor quantity is compensated. Because the term involving S rapidly becomes less significant as the chloride concentration increases, the error in S also becomes less significant. Mathematical analysis has shown that an absolute error of 1.0 x 10-10 .If*in S produces a relative error of only 1%in the chloride concentration when this concentration is around 10-4 M , and a relative error of only about 10% when the chloride concentration is around 10-6 AI. These errors are less than those involved in the reproducibility of the blank. 9. In almost all cases, the reproducibility of the blank (ca. 10-5 M ) determines the lowest limit of chloride detectable by the method. For solutions that contain about 1M concentrations of supporting electrolyte?, Considerable chloride is usually introduced even jvith “chloride-free” reagents, especially if sodium carbonate or sodium hydroxide is used in the procedure. The blank can be lowered only by troublesome purification of the reagents. Unlike S, which need be determined only once for a particular system, the blank should be checked frequently in a series of analyses. Table I. Determination of Chloride Concentrationa of Solutions by Differential Potentiometric AIethod E.X.F. of Solution us. Stand- Total NaCl NaCl Added ard Solutionb, in Solutlon JM X 105 % hZv. Coriected hf X 105 (Calcd.as Errol in KaC1 Added, for Asym(Calcd. Total Minus XaCl Solution ,1f X 105 metry Potential from Fig. 1) Blank) Added ___ -_1 (blank) 0.0 89.3 2.4 2 1.0 82.1 3.2 0.8 -20 3 2.0 77.9 4.5 2.1 5 4.0 69.7 6..5 ? 49 .. ;, 10.0 54.0 12.1 6 40.0 22.2 42.0 39.6 1 7 400 -36.0 407 405 8 1000 -58 8 980 980 9 4000 --s5.2 4100 4100 +‘2 a Besides chloride, solutions contained 20% by volume of methanol, 0.20 M p i y S O 4 , and 0.40 .If &So