used. One milliliter of aqueous sample, 5.00 nil. of acid, and 0.20 ml. of reagent, corresponding to a n acid content of 80.6 volume %, were chosen. Time and Temperature of Color Development. The complex between ninhydrin and the dehydration products of propyltne glycol is heatsensitive. Attempts t o speed color development a t elevated temperatures resulted in loss of color intensity; therefore the ninhydrin reagent must be added after the reaction mixture is cooled at 25’. Addition of 0.20 ml. of reagent to the cooled mixture does not raise the temperature sufficiently to affect color stability. Total color develops in ap: proximately 50 minutes at 26” and remains constant for 30 minutes more. The color development isunusual, in that after the samples stand for 50 to 60 minutes and are diluted t o 12.5 ml. with concentrated acid, the color rapidly gains in intensity for 3 to 5 minutes. A 1-hour period for color development at 25O, followed by dilution with concentrated acid and a n additional 5minute waiting period, was chosen as optimum. Reagent Concentration. The concentration and stability of the ninhydrin reagent have been discussed for propionaldehyde (9). T h e same conditions prevailed for the mixture of propionaldehyde and allyl alcohol. INTERFERENCE A N D APPLICATION T O OTHER COMPOUNDS
Additional glycols were tested to determine the specificity of the ninhydrin reaction toward propylene glycol; ethylene glycol, diethylene glycol, triethylene glycol, 1,2-propylene glycol, dipropylene glycol, 1,3-propylene g1: col, 1,2,3-propylene glycol (glycerol), 1,3, butylene glycol, 1,4-butylene glycol, 2,3butylene glycol, and Z-ethyl-l,3-hexandiol. Only 1,Bpropylene glycol and dipropylene glycol gave a positive test,
indicating that the test is specific for propylene glycol and its polymers in a mixture of glycols. Because the Komaroa-sky (4, 11) test reacts n ith aldehydes, ketones, and monohydric compounds a s well as glycols, these types of compounds were tested. Previous work (9)showed that propionaldehyde was the only aldehyde that gave a color reaction, while ketones gave no color. The monohydric compounds tested included methanol, ethanol, 1-propanol, 2-propanol, 2methyl-1 -propanol, 1-butanol, 2-butanol. 3-methyl-1-butanol, 3-methyl-1-butanol, 2-methyl-2-butano1, I-pentanol, 3pentanol, and allyl alcohol (2-propen-l01). Only the latter gave the violet complex with ninhydrin. This method has been used for propionaldehyde (9) and could be adapted to the determination of allyl alcohol itself or any compounds that would yield either allyl alcohol or propionaldehyde. Compounds tested included polymers and derivatives of propylene glycols and condensation products of ethylene oxide or propylene oxide and alcohols such as Cellosolves, Carbitols, and Dowanols: propylene oxide, dipropylene glycol, polypropylene glycol P-750, polypropylene glycol P1200, propylene glycol monobutyl etber, Pluronic L-60 (a propylene oxide-ethylene oxide polyether), ethylene oxide, ethylene glycol monomethyl ether, ethylene glycol monoethyl ether, ethylene glycol monobutpl ether, ethylene gl!-c.ol diethyl ether, diethylene-glycol diethyl ether, and diethylene glycol monobutyl ether. Compounds containing the propylene molecule reacted quantit7tively. The others did not react. Apparently the polymers and derivatiL es of propylene oxide are dehydrated as readily as the simple glycol in strong sulfuric acid. They gave the coloi intensity expected when compared to propylene glycol-example, 50 y of dipropylene glycol yielded a color in-
tensity equivalent to 56.6 y of propylene glycol, compared to 56.7 y calculated. Several miscellaneous compounds were investigated: propane, propene, 3chloro- 1,2-propanediol, 3-chloro- 1,2epoxypropane, 1,2-dichloropropane, 1chloro-2-nitropropane, 2-chloro-1-nitropropane, I-chloro-3-nitropropane, 1amino - 2 - propanol, and 2 - amino - 1propanol. Only 2-amino-1-propanol yielded a positive reaction. LITERATURE CITED
(1) Braumel, I. &I., Ax.4~.CHEII. 26. 930 (1954). (2) Cannon, IT. -4.,Jackson, I,. C., Ibid.. 24. 1053 (1952). (3) Curme, G. O., Johnston, F., “Gl~‘cols,’’ p. 3, Reinhold, Sew York, 1952. (4) Dal Sogare, S., Mitchell, J., Jr., ASAL. CHEJI.2 5 , 1376 (1953). (5) I h l Nogare, S., Norris, T. D., l‘litchell, J., Jr., Ibid., 2 3 , 1473 (1951). (6) Desnuelle, P., Saudet, M., BiiII. SOC. chim. France 12,871 (1945). ( 7 ) Fromageot, C., Heitz, P., Mikrochim. Acta 3, 52 (1928). (8) Hoepe, G., Treadwell, IT. D., Helv.Chim. Acta 2 5 , 353 (1912). (9) Jones, 1,. R., Riddick, J. A , , AXAL. CIIEM.26, 1035 (1954). 10) Jordan, C. B., Hatch, V. 0 . : I b i d . 25,636 (1953). 11) Komarowsky, A , , Chem.-Zfg. 27, 807 (1903).
12) RIalaprade, L. A., Bull. ~ O C . chim. France 43, 683 (1928). 13) Reinke, R. C., Luce, E. S., IND. ESG.CHEX.,ASAL. ED. 18, 244 (1946). 14) Schryver, S. B., Proc. Roy. SOC. (London) 82B, 226 (1909). 15) Riggia, S., “Quantitative Organic Analysis via Functional Groups,” p. 16, Wiley, New York, 1954. 16) Warshowsky, B., Elving, 1.’ J., IND. ESG. CHEX.,ASAL. ED. 18, 253 (1946). RECEIVEDfor reviex Xovember 29, 1956. hccepted January 30, 1957.
Removal of Interferences in the Scott-Sanchis Fluoride Determination R. E. SHOUP Ohio Department o f Health, Division o f laboratories, Columbus
b Aluminum, carbonates, organic colorproducing material, hydroxides, iron, manganese, phosphates, and SUIfates-interfering substances commonly found in public water supplies-are removed by precipitation with cadmium and mercury in the presence of sufficient boric acid-sodium hydroxide buffer to maintain a pH of 8.0.
1216
ANALYTICAL CHEMISTRY
A
IO, Ohio
of ions cause error in fluoride determination. Although distillation ( I ) is efficient, it is time-consuming, and must be manipulated with precision to avoid contamination of the distillate. By precipitating the interfering substances with cadmium and mercury, excellent recovery of fluorides has been obtained by the Scott-Sanchis SUMBER
( 2 ) method. Sufficient boric acidsodium hydroxide buffer is used to maintain a pH of 8.0, or over, after addition of about 0.25 gram of cadmium nitrate per 150 ml. of sample. The precipitate thus formed removes aluminum, color from organic matter, iron, manganese, and phosphates, and reduces the carbonates and hydroxide
alkalinity below the level of interference. The use of about 0.2 gram of mercuric nitrate per 150 ml. of sample reduces sulfate concentration to 50 to 100 mg. per liter, a t which it does not interfere. Of the substances t h a t interfere in fluoride determination by the ScottSanchis (2, 3) method, phosphate is becoming more prominent because of the use of phosphate compounds in water treatment. Aluminum is often found in public water supplies, following coagulation procedures. Sulfates interfere in many waters. Occasionally alkalinity exceeds 300 mg. per liter, and causes a decrease in the fluoride reading. Color from organic decomposition produces hues that cause difficulty in comparing the sample with standards. Varying the concentration of fluorides and interfering substances produces nonlinear factors of error (6). Therefore, even though complete analyses are made, the correction factors are difficult to calculate. Few water works laboratories are equipped, or staffed, to make complete analyses on their samples, nor is it feasible for many of them to attempt distillation ( 1 ) . The following method has been used, with excellent rcsults on a number of trade wastes, although it was devised primarily to overcome interferences in public water supplies from aluminum, carbonates, color from organic matter, hydroxides, iron, manganese, phosphates, and sulfates. REAGENTS
Specid Buffer Solution. Dissolve 20 grams of reagent grade boric acid (H3B0,) in about 75 ml. of warm distilled water. Carefully add 20 grams of reagent grade sodium hydroxide, cool. and make up to 100 ml. with distilled water. Cadmium nitrate [Cd(NOs)z.4H201, reagent grade. Mcrcuric nitrate [Hg(KOdzl, reagent grade. PROCEDURE
1 . To 150 ml. of sample, add about 10 drops of special buffer solution. Mix w l l . 2. Add 0.25 to 0.30 gram of cadmium nitrate and mix gently until a fluocculent precipitate is completely formed. (If the resulting pH is below 8.0, take a new sample and use a few mor? drops of buffer.) 3. Aldtlabout 0.2 gram of mercuric nitrate and mix gently until a ycllolv precipitate is completely formed. (If the sulfates in the sample are less than 200 mg. per liter, step 3 may be eliminated. If sulfates are known to be the only interfering substance present, steps 1 and 2 may be eliminated.) 4. Centrifuge or filter, making sure that all precipitate is removed, and
determine fluorides on a suitable aliquot of the filtrate in the usual manner. DISCUSSION
At pH 10.0, or above, cadmium forms insoluble compounds with phosphates, which will not go back into solution a t p H above 8.0. Bell (4) has shown how a number of phosphates precipitate under various p H conditions. Experimental work indicates that, if the p H is first adjusted to 10.0, metaphosphates with molecular complexes of 1 to 100 will all precipitate with cadmium, or be reduced below 1 mg. per liter. They will not go back into solution until the treated sample is neutralized to below p H 8.0. Under these conditions, aluminum, color from organic material, iron, and manganese also precipitate and are centrifuged, or filtered off, on retentive paper. The first 15 to 40 ml. of the filtrate are discarded, or returned to the filter, because some of the precipitate may penetrate the paper a t first. I n 0.25 gram of cadmium nitrate, there are about 92 ma. of cadmium. This reacts with 55 to 992 mg. of phosphate, according to the form of the phosphates and valence combinations. I n 150 ml. of sample, the range is 357 to 6550 nig. per liter of phosphate. The excess cadmium, which does not combine with phosphates, forms insoluble hydroxides and carbonates, reducing alkalinity. Cadmium chloride may be used if the chlorides in the sample are in the range of 0 to 200 mg. per liter. If chlorides are high, considerable excess of cadmium chloride may introduce additional interference. Yitrates in the range of 2000 mg. per liter, or lesser amounts in the presence of high chlorides, interfere. Evperimental work indicates that in the range of 200 mg. per liter of nitrates with 2000 mg. per liter of chlorides, the nitrates nullify the effect of the Chlorides. When the chlorides are \Tithin the range of concentration acceptable in the U. S. Public Health Service standards ( 7 ) , the nitrates introduced by this method do not interfere. I n 0.2 gram of mercury nitrate, there are about 134 mg. of mercury. This will react with about 250 mg. of the sulfate ion. I n 150 nil. of sample, 0.2 gram of mercuric nitrate theoretically precipitate 1666 nig. per liter of sulfates. I n public water supplies, complex ionization a l l o w more dissociation; as much as 100 mg. per liter of sulfates may remain in solution, well below the concentration that will cause error in the fluoride readings. With higher concentrations of sulfates in the original sample, correspondingly more mercuric nitrate must be used. Mercuric acetate is satisfactory unless the alkalinity is high, in which case
the acetate will decrease the apparent fluoride reading. I n the presence of sodium, the acetate introduced will form alkalinity beyond the range that can be tolerated by the Scott-Sanchis (2) reagent. Mercuric nitrate does not interfere when used in amounts necessary to reduce the sulfates e0 the solubility limit of mercury sulfate. Mercuric sulfate precipitates equally well at p H 3.0, 7.0, and 10.0; no adjustment is necessary ahead of the addition of mercuric nitrate. Although barium is commonly used to precipitate sulfates, excess barium combines with the sulfates in the fluoride reagent. Mercuric sulfate, being more soluble than barium sulfate, does not interfere in moderate concentrations. Therefore, less precision is required if a compound of mercury is used. EXPERIMENTAL DATA
A portion of a test solution containing 0.5 mg. per liter of fluoride and 10 mg. per liter of sodium hexametaphosphate mas treated with cadmium nitrate a t p H 9.2. After mixing, the pH had dropped to 7.25. Sodium hydroxide solution mas then added until the p H had returned to 8.0. As cadmium reacts almost instantly with hydroxide, the volume of sodium hydroxide solution required introduced a dilution factor that could only be estimated in the final analysis. The aniount of cadmium hydroxide floc that developed made filtration difficult The recovery of fluorides indicated that all the phosphate was eliminated. This parallels the work of Bell (4) to show that hexametaphosphates precipitate a t pH 8.0. Solutions, containing 10 mg. per liter of sodium hexametaphosphatc, w r e adjusted to pH 2.2, 8.0, and 10.0 with hydrochloric acid and sodium hydroxide, respectively. After being swirled gently in a flask to enhance flocculation, they were filtered. Analyses by a modification of DenigBs’ (6) method for phosphate determination and the Scott-Sanchis (2) method for fluoride determination, are shown in Table I. Work with other metaphosphates shows that the sample must be adjusted to p H 10.0, or above, to contain less than 1 mg. per liter of combined phosphate. If the pH, after addition of cadmium nitrate, drops below 8.0, some forms of phosphates may go back into solution. A municipal water supply that had been treated with sodium hexametaphosphate showed 0.75 mg. per liter of fluorides, pH 7.6, and 15 mg. per liter of phosphate. Upon distillation ( 1 ) it was found to contain only 0.1 mg. per liter of fluoride. This sample VOL. 29, NO. 8, AUGUST 1957
1217
was treated with cadmium nitrate at its natural pH, and a t p H 1 0 0 Analyses of the filtrates are shown in Table I, samples 7 and 8. From 765 to 4900 mg. of cadmium nitrate per liter introduced no interference, but complete removal of phosphate was certain only when the pH of the filtrate was above 8.0. As the alkalinity in the original sample governs the resulting pH, after treatment with a given amount of cadmium, the boric acidsodium hydroxide buffer n-as developed. About 0.5 ml. of the buffer per 150 ml. of sample held the pH and alkalinity within the prescribed limits, even where concentrations of interferences were far greater than are expected in routine samples. Extreme conditions may exist where more or less buffer will be required. The pH curve of the special buffer solution, as titrated with acid, is shown in Figure 1.
11.0 10.0
9.0
I,
Table
NO. 1 2 3
4 5
6 7
8
Adjusted PH
18
After Cd Treatment, Mg./L. Observed Deviation (NaPO& 1.0 +1.0 10 4 0.3 +0.3 7 0.8 $0.3
2.2
8.0 8.0 10.0 10.0 10.0 7.6 10.0
0 0 0
0.0 0.5
0.0 0.0 0.0
i.O
1.5 0
+0.3
0.4
0.1
0.0
EfFect of Phosphates and Aluminum on Fluoride Determination
Before Preparation, Mg./L. F recovered Deviation
Sample Added, Mg./L. No. (NaPOl)e A1 F 1 0 2 0.0 2 0 2 0.2 3 0 2 0.7 4 0 5 0.0 5 0 5 0.2 6 0 5 0.7 7 2 2 0.0 8 2 2 0.2 9 2 2 0.7 10 2 5 0.0 11 2 5 0.2 12 2 5 0.7 13 7 2 0.0 14 7 2 0.2 15 7c 2 0.7 17
:I 1
Effect of pH on Phosphate Removal
Added, Ml./L. (NBPO,)~ F 10 0.0 0.0 10 0.5 10 0.0 10 10 0.5 ~. 10 1 .o 15 0 1 15 0.1
TaMe II.
16
8.0
I.
sample
7 7
5
0.0
5
0.7
5
0.0
-0.05 -0.28
0.06
-0.14 -0.42 + O . 07
0.0
0.28
0.07 0.2
0.47 0.03 0.1 0.34 0.56 0.64
0.88
0.55 0.6 0.78
0.2
0.0
0.15 0.42
0.0
0.0
-0.23
+O. 03
-0.1 -0,36 + O , 56 +O. 44 $0.18 + O . 55 +0.4
+O. 08
After Preparation, Mg./L. F recovered Deviation 0.01 +O.Ol 0.2
0.69
0.0 0 2
0 7 0 0
02
0 68 0.0
0.2 0.7 0.0 0.2 0.68 0.0 0.2 0.69
0.0
-0.01 0.0
0 0 0 0 0 0 0 0 -0 02 0.0 0.0 0.0 0.0 0.0
-0.02
0.0 0.0 -0.01
7.0
Table 111.
l7
4.0
1
9
1 4 6 Acid, MI.
6
7
8
Sample No.
Alk.
1 2
400
Figure 1. pH curve of 1 ml. of special buffer solution, titrated with 0.8N sulfuric acid
A well water, typical of undergound water in about 20% of the state of Ohio, contained 1409 mg. per liter of total solids, 179 alkalinity, 798 hardness, 15 silicon dioxide, 0 chlorides, and 900 sulfates; p H 8.0. It showed 2.3 mg. per liter of fluorides. A 1 to 1 dilution, which reduced the sulfate interference, showed 1.05 mg. per liter of fluorides; multiplied by the dilution factor, it indicated 2.1 mg. of fluorides per liter. One portion was distilled, and another portion was treated with mercuric nitrate; both showed 1.8 mg. of fluorides per liter. When alkalinity, aluminum, sodium hexametaphosphate, and sulfate were varied, each modified sample showed 1.8 mg. of fluorides per liter by the proposed procedure. Distillations rvere made on natural and synthetic samples containing a wide variety of concentrations of interferences well beyond the range that could be anticipated in public water supplies. Recovery of fluorides, after treatment by the described method, matched (in every case except 12 18
Recovery of Fluorides in Presence of Alkalinity, Aluminum, Phosphates, and Sulfates
ANALYTICAL CHEMISTRY
400 400 0 0
0
8 9 10 11
12 i3
‘14
15 16
400 400
400 400 400 400 400 400 400 0
Added, Mg./L. SO4 (NaPO& A1 0 0 0 0 0 0 0
500
500
500
500 500
500 500 500
500
500 500 500 0
0 0 0 0 0 0 0
10 10 10 10 10 10
100
one) the analysis of the parallel distillate. In one specific combination, No. 15 in Table 111, the cause for the decrease of 0.1 mg. per liter of fluoride has not been found. Table I1 shows the effect of combinations of phosphates and aluminum on various fluoride concentrations, before and after preparation by the proposed method. I n treating these samples, step 3 in the procedure was eliminated. Table I11 shows the recovery of fluorides with various combinations of alkalinity, sulfates, phosphates, aluminum, and fluorides. Cadmium nitrate and mercuric nitrate, when added separately, and combined, to standards containing var-
0 0 0 0 0 0 0 0 0 0
5 5 5
0
F 0.0 0.2 0.7 0.0 0.2 0.7 0.0 0.2 0.7 0.0 0.2 0.7 0.0 0.2 0.7 0.2
Recovered, Deviation, Mg./L. Mg./L. 0: 0 0.0 0.2 0.0 0.7 0.0 0.0
0.0
0.18 0.7
-0.02 0.0
0.0
0.0 0.0 0.0
0.2 0.7 0.02 0.23 0.7
+0.02 +0.03 0.0 0.0 0.0
0.0
0.2 0.6
-0.1
0.0
0.2
ious concentrations of fluorides, did not interfere with the fluoride determination. LITERATURE CITED
(1) Am. Public Health Assoc., Xew Pork, “Standard Methods for the Examination of Water, Sewage, and Industrial Wastes,” 10th ed., p. 99, 1956.
(2) Ibid.,-p. 103. (3) Ibid., p. 104. (4) Bell, R. N., IXD. ENG.CHEM.,=\SAT,. ED.19, 97 (1947). ( 5 ) Denigits, Merck Index, Merck & Co.. Inc., 5th ed.. I). 690. N o . 886. 1940. ’ Lamar, W.L., Drake, P. G., J . Am. M e d . Assoc. 47, 563 (1985). U.S . Public Health Repts. 61, 371-84 (1946). RECEIVEDfor review May 31, 1956 Accepted January 23, 1957. I
_
