Determination of Dissolved Oxygen in Aqueous Solutions G. A. PERLEY Leads & Northrup Co., Philadelphia, Penna.
HE Winkler method is recognized as reliable except where very small quantities of dissolved oxygen are involved. Three types of errors may arise when applying this method to the estimation of traces of dissolved oxygen in water: (1) Impurities in the water such as nitrites, sulfites, iron ions, and organic matter may cause inaccuracies; (2) the correction for the dissolved oxygen added in the reagents may introduce uncertainties; (3) the dependence of the titration end point of the iodine upon the nature of the starch and upon the temperature introduces other uncertainties. Theriault (8) has discussed the determination of dissolved oxygen by the Winkler method; an excellent bibliography on the determination of dissolved oxygen in water by all methods has been presented by Schwartz (6); and the problem as applied to boiler feed water has been considered by White, Leland, and Button (10). With such excellent reviews available, it seems unnecessary to discuss the earlier work. Our problem involves the analyses of power plant waters, particularly where the oxygen concentrations are less than 0.5 ml. per liter. When the author had occasionko calibrate industrial dissolved oxygen recorders, the serious limitations of the starch-iodide titration end point became apparent. The sampling method suggested by Swartz and Gurney (7) was adopted in an effort to correct for the errors from impurities and from the oxygen added in the reagents. When the work was started, the author did not have available the sampling procedure of White, Leland, and Button (IO),who collected the dissolved oxygen in the distillate from the sample in an effort to eliminate the nonvolatile substances which affect the accuracy. They used essentially the author’s titration method (3’).
Kolthoff and Furman (6) considered that the thiosulfatetetrathionate electrode is irreversible. The platinum-saturated calomel electrode is one of the most satisfactory for electrometric oxidation-reduction systems. It seemed that the electrometric method should be particularly applicable to the determination in question.
Method of Analysis SAMPLING.All oxygen must be removed from the sampling system before sampling is begun. When the water is a t elevated temperatures and pressures, it is necessary to avoid flashing of the sample. The flow to the sampling bottles should be controlled by adjusting a valve on the discharge side of an adequate cooling coil. The temperature of the water at the sampling time should be slightly below that of the room temperature. The temperature of the sample bottle should not be allowed to decrease l’ C. between the time of sampling and the time of analysis. Hence, the temperature a t the sampling point preferably should not exceed 30’ C. (86 ’F.). Three samples are collected in series in sampling bottles equipped with rubber stoppers and with inlet and outlet tubes. The sampling bottles should be slightly oversize and should have narrow mouths designed t o accommodate ground-glass stoppers. The end of the glass stoppers should be ground to a semiconical shape to prevent trapping of air bubbles when replacing stoppers. No trace of a gas bubble can be tolerated in the top of the sampling bottles. Two 250-ml. samples and one 500-ml. sample should be taken. The water sample should enter one of the 250-ml. bottles through a tube extending t o the bottom. The sample overflows through an outlet tube at the top into the 500-ml. bottle intake tube. This overflows into the second 250-ml. bottle. McLean sampling tubes may be used in place of the bottles. Glass-toglass butt connections should be provided for all connections. Rubber tubing is used merely to hold the connections in place. At least seven times the total volume of the three sampling bottles should be withdrawn before the sampling is stopped.
Iodometry by Electrometric Methods Willard and Fenwick (11) and Van Name and Fenwick (9) suggested a bimetallic electrode system for use in the electrometric titration of iodine. Foulk and Bowden ( I ) modified the previous bimetallic electrode titration and described a “dead-stop end point” method for iodometry. Hewson and Rees (2) applied the above method to the determination of iodine liberated by the Winkler reagents. A consideration of the electrometric determination of iodine in solution leads to the following: I2
Iz + 2 (e)
REAGENTS. The following stock reagents are desirable: 0.1 molar stock sodium thiosulfate solution, 24.82 grams of Na2SzOs.5H20 per liter (stored in a brown bottle protected by a soda-lime tube). A 0.01 molar sodium thiosulfate solution is made up from the above solution and standardized each day by means of a 0.01 molar potassium biiodate solution. 0.01 molar stock potassium biiodate solution, 0.3250 gram of KIOs.HI03 per liter. Manganous chloride solution, 412 grams of MnC12.4H20 dissolved in distilled water and made up to 1 liter. Alkaline-iodide reagent, 700 grams of potassium hydroxide and 150 grams of potassium iodide per liter. The reagent should be free from carbonates, as manganese carbonate does not react with dissolved oxygen. A paraffined glass stopper should be used in the bottle. Concentrated sulfuric acid of specific gravity 1.84, diluted with an equal volume of water.
which is a case of the general class Oxidant (e) C reductant
Kolthoff (4) showed that the potential of the iodine electrode a t 25’ C. is represented by the equation: 0 059 [I21 E = Ea log 2
PROCEDURE. The 500-ml. sample and one of the 250-ml. samples are treated as follows:
According to Jones and Kaplan (3),when using a saturated iodine electrode, we have at 25’ C. E = 0.5362 - 0.059 log [I-]
Remove the rubber stopper with inlet and outlet tubes. By means of separate ipets rapidly add 2 ml. of the alkaline reagent, and then 2 ml. of the manganous chloride reagent. Both reagents sink t o the bottom of the bottle without excessive air contamination when the tip of the pipet is held below the surface of the sample. Rapidly insert the glass stopper. The slightly air contaminated upper portion of the sample is elimi-
Kolthoff (4) pointed out that the potential of the iodine electrode depends upon the iodide concentration. 240
MAY 15, 1939
nated by the overflow when the glass stopper is replaced. Shake thoroughly for 2 minutes and let stand until the precipitate settles three quarters of the height of the sampling bottle. Remove stopper, add 2 ml. of sulfuric acid reagent, replace stopper, and sfiake thoroughly until the precipitate dissolves. Iodide solutions are slowly oxidized by air. Accordingly, the glass stopper should not be removed from the sample bottles until all is prepared for the subsequent titration. The titration should be easily completed in less than 5 minutes; hence the error due t o the liberation of iodine by air oxidation is very small. Measure out exactly 250 ml. of the sample from the 250-ml. bottle. Add exactly 250 ml. of untreated water from the other 250-ml. bottle. Then add 0.5 ml. of 0.01 N potassium biiodate solution from a pipet. The addition of the standard biiodate solution serves as a check on the titration, particularly when small traces of oxygen are involved in the analysis. Proceed to titrate electrometrically with 0.01 N sodium thiosulfate, using a certified buret with a special small tip to give not over 0.02 ml. per drop. Measure out exactly 500 ml. of the sample from the 500ml. bottle, add potassium biiodate solution in the exact amount used in the 250-ml. sample, and titrate as in the previous case. The oxygen content for a 250-ml. sample is determined by the difference between the 500-ml. and the 250-ml. sample. f 1 ml. of 0.01 N NalS9011 is eauivalent t o 0.0612 ml. of oxmen at 25' C. and 760 mm:)- As an optional procedure, a titration with 0.01 N potassium biiodate solution may be found to give slightly greater accuracy. If this method is used, a known excess of 0.01 N sodium thiosulfate should be added.
Electrometric Titration A 1 x 1 cm. sheet of platinum provided with a suitable lead wire is used as the indicator electrode, and a saturated calomel-saturated potassium chloride system is used as the reference electrode. The two electrodes are mounted in an 800-ml. beaker. Stirring should continue during the entire titration. A mechanical stirrer is desirable. A portable potentiometer with a range of 0 to 1100 millivolts is used to obtain the e. m. f. per ml. of thiosulfate relationship. It is not necessary to make a complete titration, since the significant abrupt e. m. f. change is easily detected. The
standard thiosulfate solution may be added rapidly until the e. m. f. is approximately 0.3100 volt, and the titration then continued drop by drop until the end point is reached. The end point occurs between 0.290 and 0.260 volt, depending upon the nature of the solution, the p H of the solution, and the concentration of the iodide present. In any instance, a t the exact end point one drop of 0.01 N thiosulfate solution results in an abrupt drop of over 30 millivolts. The author has found that the end point for the average high-purity water involved in power plant boiler practice occurs a t an e. m. f. value of 0.2900 volt for temperatures between 15" and 30" C. As the end point is approached in iodine solutions, the e. m.f. tends to drift slowly to a higher voltage value immediately after the addition of thiosulfate. At the end point there is no tendency to drift. With an excess of thiosulfate, there is a tendency for the e.m.f. to drift to lower voltage values. As a matter of fact, the use of a properly sensitive galvanometer is all that is required to detect the end point by the drift method; however, this is more critical to operate than the above potentiometer method. Figure 1 shows a portable unit which the author used to carry out dissolved oxygen tests in the field. Provision has been made for all the necessary pipets, buret, electrodes, sample bottles, beaker, and mechanical stirring equipment. An antimony electrode is also included so that when used with a suitable portable potentiometer, it is also possible to obtain pH measurements.
Influence of Various Ions The normal variations in the concentration of sulfate and chloride ions have little influence on the end point. The iodide-ion concentration changes the e. m. f . value of the inflection point, yet the same thiosulfate end point results, irrespective of the iodide concentration. The inflection point of the curve for the e. m. f . per ml. of thiosulfate occurs a t a lower voltage for high iodide concentrations than for low iodide concentrations. This is constant for the author's procedure and hence does not involve any error. The hydrogen-ion concentration must be controlled within limits. The author's method is sufficiently standardized so that the hydrogenion concentration is held within the desired limits. More thiosulfate is required to reduce all of the iodine to iodide as the concentration of hydrogen ion is increased. The more nearly the solution is held to 7.0 pH, the less is the effect upon the thiosulfate end point.
Precision of the Method
FIGURE 1. PORTABLE UNIT
The high sensitivity of the change a t the end point is noteworthy. One drop of thiosulfate from a specially small buret tip, which has been found to be equivalent to 0.02 ml., results in a 30millivolt change. Thus the author was able to detect the presence of 0.0000016 gram of oxygen in the 250-ml. sample. This is roughly 0.0064 part by weight of
INDUSTRIAL AND ENGINEERING CHEMISTRY
oxygen per million parts by weight of water (1 part in about 156,000,000). Tests indicate that the limit of error of the titration is *0.001 ml. of oxygen a t 25” C. and 760 mm.
Advantages of the Method The author has used this equipment extensively in connection with power plant tests and has found the following advantstges in making dissolved oxygen determinations: The use of the simplified electrometric titration procedure eliminates the temperature, starch quality, and personal equation errors of the starch-iodide method. The analysis is based upon a difference determination whereby the influence of dissolved oxygen in the reagents, small ionic variations of foreign substances, secondary reactions, loss of dissolved oxygen by displacement, possible
VOL. 11, NO. 5
contamination of the sample by air when the reagents are added, and temperature variations are minimized. The complete equipment can be made in portable form.
Literature Cited (1) (2) (3) (4) (5) (6) (7) (8) (9) (10j
Foulk and Bowden, J . Am. Chem. SOC., 48, 2045 (1926). Hewson and Rees, J . SOC. Chem. Ind., 54,254t (1935). Jones and Kaplan, J. Am. Chem. SOC.,50, 2066 (1928). Kolthoff, Rec. trav. chim., 41, 172 (1922). Kolthoff and Furman, ”Potentiometric Titrations”, 2nd ed., p. 69, New York, John Wiley & Sons, 1931. Schwartz, M. C., Louisiana State Univ., Univ. Studies No. 21 (1935). Swartz and Gurney, Proc. Am. SOC.Testing Materials, 34, 796 (1934). Theriault, E. J., Pub. Health Bull. 151; Suppl. 90 (1925). Van Name and Fenwick. J . Am. Chem. SOC..47. 9.19 (1925). White, Leland, and Button, Proc. Am. So;. Tes>ing’Materials, 36, 697 (1936). Willard and Fenwick, J . Am. Chem. SOC., 44,2504,2516 (1922).
Measurement of Plastic Properties of Bituminous Coals Comparison of Gieseler and Davis Plastometer and Agde-Damm Dilatometer Methods R. E. BREWER
J. E. TRIFF, Central Experiment Station, Bureau of Mines, Pittsburgh, Penna.
HE torsional principle-that is, the measurement of resistance to shear caused by movement of a stirring
device within the heated coal c h a r g e w a s first applied in 1931 by Davis (5) to the determination of the “plastic” properties of bituminous coking coals. Since then, this general principle has been employed in various forms of other instruments (9-13). Although the later designs of the Davis plastometer have incorporated a few minor changes to ensure smoothness in operation and improve general appearance, the original instrument, after more than seven years of continued use, has proved satisfactory. It has been found, however, that the procedure (5-8) gives more uniform test results when modified (S), especially by the use of a larger sample of representative coal and by operation of the retort a t a slower speed of rotation, and that “minor limitations lie in the difficulties of determining accurately small changes during the period of greatest fluidity and extremely high resistances, above 63.4 kg.-cm. (55 pound-inches) shown by certain coals.” These difficulties have been overcome by the use of tension springs with a sensitivity of less than 0.23 kg.-cm. (0.2 pound-inch), during the period of greatest fluidity, which permit accurate measurements of resistances up to 149.8 kg.-cm. (130 pound-inches). In the Davis plastometer method (6) the coal charge as a whole is rotated and stirred; the property measured is the resistance to shear of the partly fused coal adhering to the inner periphery of the retort. I n the Gieseler plastometer method (9) the coal is static a t the start, and later stirring is proportional to the fluidity of the coal. Accordingly, with increase in fluidity of the heated coal are noted (a) a decrease in resistance, or torque, measured in kilogram-centimeters, in the Davis rotary retort, and ( b ) an increase of the rate of rotation of the stirring shaft in the Gieseler stationary retort.
In (a) the resistance is created by the movement of the coal in the retort against the rabble arms on the inside shaft, which is prevented from free rotation by the tension springs; in (b) the rotation of the stirring shaft is caused by application of a constant force on the loading pan. Gieseler (9) criticized the method of Davis (5) because the coal is heated under conditions corresponding to those of a rotary retort, permitting volatile matter and tar to escape more freely than in a coke oven. Gieseler (9) made the broad assertion that all methods for the determination of plasticity in which the coal is not prevented from expanding are unsatisfactory, quoting a statement from Davies and Mott (4) that a “coal which is free to expand loses volatile matter readily and plasticity ends a t a comparatively low temperature.” Davies and Mott (4) showed, however, that the temperature of solidification, termed by them the “end of plasticity”, is higher than the temperature of final expansion as determined by the Sheffield laboratory coking test on a number of the better coking coals. At the beginning of the solidification of a melted coal mass into semicoke a coal rapidly loses its fluidity. At this stage the Gieseler instrument naturally shows with increasing temperature less and less movement of the stirring shaft per unit of time, and as semicoke formation proceeds this movement falls off rapidly to zero. The apparent discrepancy in the interpretation of the “end of plasticity” is, therefore, purely one of definition. The temperature limits of the plastic range for the Davis plastometer test are defined as the difference between the temperature a t which resistance develops and that a t which resistance ends. This latter temperature is some degrees higher, the magnitude varying with different coals, than that of maximum resistance or solidification, which, in turn, is higher than the temperature