January 15, 1934 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 61

most favorable range. In solutions of pH above 10, the sul- fide electrodes showed a maximum variation among them- selves of 11.5 millivolts. Table I shows that the sulfide elec- trode is not reliable in this range even when average values of several electrodes are taken.

Sugar, starch, and nitrates have no deleterious effect upon the electrodes between pH 2.2 and 10. However, hydroxy acids, such as tartaric, citric, and lactic, render the electrode useless. This has been found to be true of the antimony oxide electrode as well (1) . The sulfide electrodes were used in the back-titration of alcoholic potassium hydroxide in determining the saponification number of oils. As would be expected, the strongly alkaline solution quickly removes the sulfide film, but the electrode still functions. The end point as determined by finding the maximum value of AE/AV agrees with the phenolphthalein end point. In this respect the sulfide elec- trode offers no advantage over the ordinary antimony- antimony oxide electrode.

SUMMARY 1. Antimony electrodes coated with antimony sulfide

have been prepared by five different methods.

2. Electrodes prepared by suspension in hot 0.30 N nitric acid for 1 hour, followed by saturation with hydrogen sulfide, may be used to determine the pH of solutions in the range from 2 to 10.

3. Electrodes so prepared agree among themselves within about 3 millivolts if the pH is 10 or less, but may differ by as much as 11.5 millivolts in more alkaline solutions.

4. Starch, sugar, and nitrates have no deleterious effect in the range over which the electrode functions in their ab- sence.

5. The electrode, like the oxide electrode, is useless in the presence of hydroxy acids.

6. The electrode should be useful in determining the saponification of highly colored oils, but offers no advantage over the ordinary antimony electrode.

LITERATURE CITED (1) Barnes and Simon, J. Am. Soc. Agron., 24, 156 (1932). (2) Kolthoff and Furman, “Potentiometric Titrations,” 3rd ed.,

p. 446, Wiley, 1931. (3) Parks and Beard, J. Am. Chem. Soc., 54, 586 (1932).

R ~ C E I V E D September 12 , 1933.

Colorimetric Determination of Fluorine 0. M. SMITH AND HARRIS A. DUTCHER, Oklahoma Agricultural and Mechanical College, Stillwater, Okla.

HE.fluoride content of natural water has taken on new significance, since the researches of Smith, Lantz, and T Smith (4), Churchill (2), and Kehr (3) have shown

that it may be the cause of the tooth defect known as mottled enamel. Of the numerous methods which have been de- vised for the detection and determination of fluorides, few find satisfactory application in the field of water analysis, where a very sensitive method is required.

In the modified Casares-DeBoer method used by Thompson and Taylor (5) , the fluorides are determined by the degree of fading of a zirconium-alizarin lake. Willard and Winter (6) suggest the use of quinalizarin (1, 2, 5, 8-tetrahydroxyan- thraquinone) as an indicator in their method. Quinalizarin seems to the writers to have advantages over alizarin when used as in the Casares-DeBoer method, in that it is more sensitive to small changes in fluoride content and the change in color is easier to distinguish. For example, a difference in color between samples containing 0.2 and 0.4 part per million is greater in the case of zirconium-quinalizarin than zirconium-alizarin reagent. The best range of the colorimet- ric standards is from 0 to 2 p. p. m. or 0.0 to 0.1 mg. of fluoride per 50 ml. Above this concentration the fading is too great, and comparisons are not easily made.

RECOMMENDED METHOD

The zirconium-quinalizarin reagent is prepared by mixing equal parts of a 0.14 per cent solution of quinalizarin (1, 2, 5 , 8-tetrahydroxyanthraquinone) and an 0.87 per cent solution of zirconium nitrate, and diluting the mixture 1 to 40. The quinalizarin is dissolved in a 0.30 per cent sodium hydroxide solution, as it is insoluble in water.

Precipitate the sulfates by the addi- tion of 5 ml. of 2 per cent barium chloride solution to 100 ml. of the sample. After settling several hours, draw off a 50-ml. portion for the test. The barium sulfate may be filtered off if desired. Add 3 ml. of 1 to 1 hydrochloric acid and 5 ml. of the zirconium nitrate-quinalizarin reagent and mix thoroughly. After 20 minutes compare with standards made at the same time and in the same manner. Comparisons are easily made in

METHOD OF ANALYSIS.

American Public Health Association tubes with standard fluoride solution containing from 0 to 2 p. p. m. in steps of 0.2 part.

The fading of the color of the zirconium-quinalizarin or zirconium-alizarin lake is a function of time, temperature, and acidity. After 15 minutes a t room temperature the change is very slow and the color remains sufficiently constant for comparisons. Care must be taken that exact amounts of indicator (zirconium-quinalizarin reagent) and acid are added to the sample and to the standards and that the time of fading is the same for unknown and standard.

Of the commonly occurring ions, none in the quantities occurring in natural or treated water affect the results except aluminum, iron, sulfates, and phosphates. Less than 20 p. p. m. of sulfates have no effect and may easily be removed by barium chloride, since the Ba++ ion has no effect. Iron above 10 p. p. m. changes the color, making comparisons im- possible. Aluminum has no effect up to 0.2 p. p. m. as Al; above this amount and up to 0.6 p. p. m. as A1 the results will be low by 0.1 p. p. m. of fluorine. Aluminum is rarely present in water in amounts greater than 1 p. p. m. expressed as A1203 or 0.5 p. p. m. as Al, and may thereby be neglected in most cases. This is further confirmed in the comparative results between those obtained by direct colorimetric and distillation methods. Phosphates affect the color when from 0.3 to 0.4 p. p. m. or more are present. The color is different and is easily recognized by one experienced in the determina- tion. Fortunately these two substances rarely occur in water in amounts greater than 0.5 p. p. m. except in certain localities. When they do occur the distillation method seems to be the reliable procedure.

In case phosphates and aluminum are present, distillation with perchloric acid as recommended by Boruff and Abbott (1) will be necessary.

Place the sample containing approximately 0.2 ml. of fluoride in a 125-ml. distilling flask, and add a few glass beads and suf- ficient dilute sodium hydroxide to make it just alkaline to litmus. Reduce the volume to 10 to 15 ml. by distilling off the water, and obtain a 50-cc. distillate according to the procedure out- lined by Boruff and Abbott (1).

62 A N A L Y T I C A L E D I T I O N Vol. 6, No. 1

In carrying out the distillation, care must be exercised to avoid bumping, as the presence of much perchloric acid in the distillate will seriously interfere with the colorimetric test. Its presence can be readily detected by the off color of the sample when the reagents are added. The temperature should not be allowed to rise above 150' C., in order to avoid decomposition of the perchloric acid. Sulfuric acid cannot be used, as sufficient passes over into the distillate to affect the color developed on adding the quinalizarin reagent. In routine water analysis the distillation method is time-con- suming and it becomes desirable to determine how necessary is this precaution.

In a study of 201 Oklahoma waters, 177 samples had a fluoride content of 1 or less. Of these, 59 gave the same value when determined by both methods; in 88 samples the re- sults obtained by the direct colorimetric method were below those obtained by distillation by an average of 0.13 p. p. m. of fluorine, and in 30 cases they were above by an average of 0.11 p. p. m. The 15 samples containing 1.1 to 2.0 p. p. m. and 9 containing 2.1 and above were too few in number to draw definite conclusions. These 201 cases indicate that the

direct colorimetric method may give results which are about 10 per cent lower than that obtained by distillation.

The authors' experience confirms that of previous investi- gators that fluorine is usually found in natural waters in quantities less than 2 p. p. m., 95 per cent of the samples containing less than 2.0 p. p. m.; hence a method that is accurate to within 10 per cent of the numerical value, when the fluoride content is from 1 to 2 p. p. m., and that is rapid and reliable should be a valuable acquisition to water analysis and worthy of trial by other investigators.

LITERATURE CITED (1) Boruff and Abbott, IND. EKQ. CHEW, Anal. Ed., 5, 236-8

(2) Churchill, IND. ENO. C H ~ M . , 23,996 (1931). (3) Kehr, J. Am. Water Worlcs Assoc., 23, 214-29 (1931). (4) Smith, Lantz, and Smith, Ariz. Expt. Sta., Tech. Bull. 32, 253

(5 ) Thompson and Taylor, IND. ENQ. CHEW, Anal. Ed., 5, 87 (1933). (6) Willard and Winter, IND. ENQ. C H ~ M . , 5, 7 (1933). RECEIVED October 3, 1933. Presented before the Division of Water, Sew- age, and Sanitation Chemiatry a t the 86th Meeting of the American Chemi- cal Society, Chicago, I11 , September 10 to 15, 1933.

(1933).

(1931).

Determination of Potash in Fertilizers F. B. CARPENTER AND R. 0. POWELL, Virginia-Carolina Chemical Corporation, Richmond, Va.

UGH work has been done by the Association of Official M Agricultural Chemists to overcome the inaccuracies in the present method, which specifies the determination of only the water-soluble potash, but no attempt has been made to determine the available potash, as in the case of other fertilizer constituents. This method has been in use for a long time and is incorporated in practically all state laws, but there is no logical reason why it should be continued if it does not show the correct results in available potash.

A series of tests was made on a number of mixed fertilizers, including the five samples sent out by the Chemical Control Committee of the National Fertilizer Association, to deter- mine the effect on the percentage of potash produced by washing with larger quantities of water than are specified by the official method. Four successive leachings were made on each sample with approximately 225 cc. of water, and the potash was determined in the separate solutions and in the final residue. Potash was also determined in solutions made with 1 per cent hydrochloric acid. Blank tests were made on all reagents, and the same method was employed in all analyses.

TABLE I. POTASH DEVERMINATIONS IN SOLUTIONS FROM

The results are shown in Table I.

SUCCESSIVE LEACHINGB WITH WATER, FROM RESIDUE, AND FROM HYDROCHLORIC ACID SOLUTIONS

BOILED 1 PER CENT

-LIACHINQS~ - IN HCl SAYPLH~ First Second Third Fourth RESIDUE TOTAL SOLUTION

1 2 3 4 5 6 7 8 9 10 11

17.24 3.90 3.88 4.04 3.96 5.32 4.18 3.10 3.92 5.38 6.15

4 226 CC. each.

0.10 0.11 0.12 0.07 0.10 0.06 0.04 0.03 0.09 0.10 0.02

0.04 0.04 0.05 0.03 0.02 0.05 0.05 0.04 0.05 0.05 0.02

These analyses show that small amounts in the fourth

0.02 0.02 0.02 0.03 0.02 0.04 0.03 0.03 0.03 0.03 0.01

0.16 0.40 0.40 0.07 0.04 0.09 0.14 0.17 0.13 0.13 0.14

17.56 4.47 4.47 4.24 4.14 5.56 4.44 3.37 4.22 5.69 6.34

17.50 4.40 4.20 4.22 4.17 5.65 4.4s 3.27 4.24 5.66 6.36

potash is being recovered in washing, after approximately

900 cc. of water have passed through the material, and there still remains from 0.04 to 0.40 per cent in the residues. If the washings had been continued, it is probable that most of the potash would finally have been recovered. If the sum of potash obtained in the different leachings, plus that in the residues, is compared with that obtained in the acid solutions, the results are in close agreement.

The two samples showing the largest amount in the residue, 0.40 per cent, were also tested by washing with water and digesting in ammonium citrate, as in the determination of insoluble phosphoric acid. The results showed 0.06 per cent of insoluble potash in each case.

Solutions made with 1 per cent hydrochloric acid give very good results, especially if neutralized with caustic soda instead of ammonia. There should be no objection to this procedure, as insoluble silicates are not decomposed in acid solutions of this strength.

In all samples the potash was supplied from a soluble source, and the results indicated that some of the potash is slightly fixed, so as to yield slowly to solution in water, but is in a form which is assumed to be readily available as a plant food. This is supported by the fact that the insoluble portion is soluble in ammonium citrate, the standard test for available phosphoric acid. As these tests were made on regular grades of mixed fertilizers, no data are available concerning the amount of potash in the different components, but in previous work similar losses have been confirmed on carefully formulated theoretical mixtures.

CONCLUSIO?: Much work has been done to perfect the official method

so as to obtain theoretical results, but it would appear that the only remedy for correcting the inaccuracies is to change the method so as to determine available instead of water- soluble potash.

RECEIVED September 20, 1933. Presented before the Division of Fertilizer Chemistry a t the 86th Meeting of the American Chemical Society, Chicago, 111, September 10 t o 15, 1933.