Ultraviolet Spectrophotometric Determination of Sulfate, Chloride, and Fluoride with Chloranilic Acid R. J. BERTOLACINI and J. E. BARNEY II Research Department, Standard Oil Co. (Indiana), Whiting, Ind.
b The intense absorption of solutions of chloranilic acid in the ultraviolet region provides the basis for the spectrophotometric determination of trace amounts of sulfate, chloride, and fluoride. Sulfate is determined by reaction with barium chloranilate in 50% ethyl alcohol, buffered a t pH 4 (absorption measured a t 332 mF); chloride, with mercuric chloranilate in 50% methyl Cellosolve, 0.05N in nitric acid (absorption measured a t 305 mp); fluoride, with strontium chloranilate in 50% isopropyl alcohol, buffered at pH 4 (absorption measured at 332 mp). Limit of detection is 0.06 p . p m for sulfate, 0.05 p.p.m. for chloride, and 5 p.p.m. for fluoride. The procedures for sulfate and chloride provide greater sensitivity, with fewer interferences, than currently available methods. The procedure for fluoride is subject to interference from most anions but rapid in limited routine uses. The method can b e applied to other anions.
C
SALTS of chloranilic acid (2,5-dichloro-3,6dihydroxy-p-benzoquinone) have proved useful for determining sulfate (3) and chloride in aqueous solution ( 1 ) . The reaction hletal chloranilate anion hydrogen ion -+ acid chloranilate ion metal-anion salt produces reddish purple acid chloranilate ion, which is determined colorimetrically. The metal chloranilate and the solvent system were chosen so that the metal-anion salt is so much less soluble than the metal chloranilate that the reaction is quantitative. Sensitivity and selectivity are enhanced by proper selection of metal chloranilate, solvent, and pH. Some cations react with chloranilic acid and must be removed by ion exchange resins prior to analysis. Interference from anions is not serious in either of the methods developed (1, 3). Both methods are most useful in the 5 to 100 p.p.m. range ; however, more-sensitive methods for determining both sulfate and chloride are needed. In the previous studies, sensitivities were limited by the absorption of the acid chloranilate ion a t the wave length of maximum absorption in the visible ERTAIN
+
202
+
ANALYTICAL CHEMISTRY
region. However, pbenzoquinones, especially 2,5-dihydroxy-p-benzoquinonej absorb strongly in the ultraviolet region (8). Preliminary investigations revealed that aqueous solutions of chloranilic acid-a substituted 2,5dihydroxy - p - benzoquinone-also absorb strongly in the ultraviolet region. Figure 1 shows that absorption of chloranilic acid solutions is about 30 times higher a t 330 than a t 530 mp. A logical approach to increasing the sensitivity of the methods was to study the absorption of chloranilic acid solutions in the ultraviolet region a t various pH’s in suitable solvents. The principles employed for determining sulfate and chloride (1) can be used in methods for determining other anions, especially other halides. if suitable metal chloranilates and solvents can be found. For determining fluoride, the halide selected for study, a metal chloranilate which was insoluble, yet more soluble than the metal fluoride, was required. Calcium chloranilate was unsuitable because it LTas less soluble than calcium fluoride in all the solvents tested which were transparent in the 1.5
+
1.0
4
0.5
0
WAVELENGTH,
mp
Figure 1 . Absorption spectrum of chloranilic acid in 50% ethyl alcohol 0.024 mg. of chloranilic acid per ml. Cary Model 11 Spectrophotometer
ultraviolet. Strontium chloranilate proved to be satisfactory in 50% isopropyl alcohol. Fluoride was selected because of the general need of a colorimetric method to determine this anion. Most available colorimetric methods for fluoride depend on the bleaching of some colored metal complex by the formation of stable complexes between the metal ion and fluoride (6, 7 , 9); these methods and others (4, 6) require frequent calibration, lack sensitivity, or are subject to interferences. h method that does not suffer those limitations v a s sought. ULTRAVIOLET ABSORPTION OF CHLORANILIC ACID SOLUTIONS
ilbsorption of chloranilic acid solutions in the ultraviolet region of the spectrum was studied over a wide range of pH in three solvents: 50% ethyl alcohol, 50% methyl Cellosolve (Union Carbide Chemicals Co.), and 50% isopropyl alcohol. Ethyl alcohol has been used for the colorimetric determination of sulfate, and methyl Cellosolve for chloride. Isopropyl alcohol was the solvent in which strontium chloranilate reacted with fluoride to give a reddish purple solution. All three solvents are transparent above 280 mp. Figure 2 s h o w that, a t a given pH, a change in solvent has little effect on the absorption spectra. All these spectra were obtained on a Cary LIodel 11 spectrophotometer with a reagent blank in the reference cell. The major absorption peaks in alkaline and weakly acidic solutions occur a t 332 and 320 to 325 mp; those in strongly acid solutions occur a t 305 mp. Schmarzenbach (8) attributes the peaks of 2,sdihydroxy-pbenzoquinones in the ultraviolet region to the double enol structure; he states that addition of a proton causes a slight shift in, , ,A to shorter wave lengths. The authors’ spectra substantiate these observations. The peaks at 332 and 320 to 325 mp arise from absorption by the chloranilate and acid chloranilate ions, and the peak at 305 mp by chloranilic acid. Furthermore, absorption spectra in 1 0 0 ~ o ethyl alcohol and isopropyl alcohol show a peak only a t 305 mp;
'
:::w 0.2
7!cu ;;m t " " " 1
0.4
0.2
p 8
t
1
0.4
0.2
6!!u 1.:
0.6
0.4 0.2
0
I .o
0.6 0.4
0.2
l-6r.I
0
0.6
0.4
0.2
b
0
280 290 300 3 0 320 330 340 350 eM)290 300 310 320 330 340 350 LEO 290 300 310 320 330 340 350
LmJL Figure 2.
Effect of p H on absorption spectra of solutions of chloranilic acid
0.025 & 0.003 mg. recrystallized chloranilic acid per ml. 50% solvent
1. 2. 3. 4.
Curves 2 through 7 ,
Solvent alone
5 . pH 4 (0.005M potassium acid phthalate) 1 S H,"O. 6 . pH 9 (Beckman buffer No. 14049) 0.05.k HA03 7 . 1N KOH 10.5N KOH in isonronvl alcohol) pH 2 (no added acid or alkali)
-
& .
in these solvents chloranilic acid is not ionized and no absorption from chloranilate is observed. If only a few per cent of mater is added to either solution, peaks a t 325 mp appear. In undiluted methyl Cellosolve, the peak a t 325 mb was obtained; this solvent contained a small amount of water, and ionization undoubtedly occurred. Lack of absorption in 50% isopropyl alcohol 0.5N in potassium hydroxide cannot be explained. Another feature in the spectra is the relatively high absorption below 305 m p in solutions buffered a t pH 4. Because this absorption was not observed in the other solutions, some unknown interaction among chloranilic acid, the buffer, and perhaps the solvents may be the cause. However, the shape of the spectra, the location of the absorption maximum a t 332 mp, and the absorption above 305 mp are not affected by the buffer. These curves show that greatest sensitivity in all three solvent systems is obtained in alkaline solution; the best appears to be one buffered at pH 9. Because of the wide availability of the Beckman DU spectrophotometer and its suitability for quantitative ultraviolet spectrophotometry, further studies of analytical procedures for sulfate, chloride, and fluoride were carried out with this instrument. However, adding barium chloranilate to 50% ethyl alcohol or strontium chloranilate to 50% isopropyl alcohol, both buffered at pH 9, or mercuric chloranilate to 50% methyl Cellosolve, buffered a t pH 4, gave much higher blank absorption than corresponding amounts of chloranilic acid in these solvent systems. These reagent blanks absorbed so strongly that when they were used in 1-cm. cells the meter of the Beckman spectrophotometer could not be brought to zero. However, a change in pH from 9 to 4 in the first two solvent systems, and from 4 to 0.05N in nitric acid in the third, halved absorption and the Beckman instrument could be used. Because the differences in absorption of chloranilic acid solutions of these pH's in 50% ethyl alcohol, 50% isopropyl alcohol, and 50% methyl Cellosolve are not so large, the barium chloranilate and strontium chloranilate probably react with the pH 9 buffer, and the mercuric chloranilate with the pH 4 buffer, to produce some substance that absorbs strongly at 300 to 335 mp, Therefore, for determining sulfate and fluoride, a pH of 4 was selected; for chloride, a solution 0.05N in nitric acid. Only slight sensitivity is sacrificed, and the readily available Beckman DC spectrophotometer may be used. PROCEDURES
Barium chloranilate ( 1 ) and mercuric VOL. 30, NO. 2, FEBRUARY 1958
203
chloranilate (3) were prepared by addition of an aqueous solution of the metal chloride and nitrate, respectively, to a stirred aqueous solution of chloranilic acid. Barium chloranilate may also be purchased from the Fisher Scientific Co. or J. T. Baker Chemical Co. To prepare strontium chloranilate, a 50% aqueous solution of reagent-grade strontium nitrate was added dropwise to a stirred 0.1% aqueous solution of chloranilic acid maintained a t 50" C. After this mixture had stood overnight, the supernatant liquid was decanted, and the precipitate was washed by decantation with distilled water and filtered through a fritted-glass filter of medium porosity. The water-wash and filtration steps were repeated twice. The precipitate was then washed with ethyl alcohol, filtered as before, and dried for 1 hour a t 110' C. Similar procedures were devised for sulfate, chloride, and fluoride. I n each procedure the solution to be analyzed is passed through a column 1.5 cm. in diameter and 15 cm. long containing Dowex 50 X 8 resin, hydrogen form, to remove interfering cations. The pH of the effluent is adjusted, and a suitable aliquot (less than 40 ml.) is transferred to a 100-ml. flask. Ten milliliters of buffer and 50 ml. of solvent are added, and the mixture is diluted to volume with ion-exchanged distilled water. About 0.2 gram of the appropriate metal chloranilate is added, and the flask is shaken intermittently for 15 minutes. The excess metal chloranilate and metalanion salt is removed by filtration or centrifugation, and the absorbance of the filtrate is read in a Beckman DU spectrophotometer a t 332 mp in 1-cm. cells against a blank treated in the same manner. Concentration of the anion is obtained from a calibration curve prepared from standard ammonium salt solutions. For sulfate, the p H of the effluent is adjusted to 7 with dilute nitric acid or ammonium hydroxide and pH paper of range 6 to 8, an aliquot containing not more than 0.5 mg. of sulfate is taken, a 0.05JI solution of reagent-grade potassium acid phthalate (final pH = 4) is used as buffer, barium chloranilate is added, and the sample is filtered through Whatman No. 42 filter paper. For chloride, the pH is adjusted to 7, an aliquot containing not more than 0.5 mg. of chloride is taken, 0.5N nitric acid is used as buffer, mercuric chloranilate is added, and the sample is centrifuged. For fluoride, the pH is adjusted t o 7, an aliquot containing not more than 5 mg. of fluoride is taken, 0.05M potassium acid phthalate is used as buffer, strontium chloranilate is added, and the sample is filtered through Khatman No. 42 filter paper. DISCUSSION
Accuracy and precision of the three procedures were determined by analyzing dilute solutions prepared from standard aqueous solutions of ammonium salts of the three ions. Sulfate and chloride solutions were standardized 204
ANALYTICAL CHEMISTRY
gravimetrically; the fluoride solution was standardized by adding a measured excess of standard calcium nitrate to precipitate calcium fluoride from a 1% acetic acid solution, and titrating the excess calcium with sodium Versenate. The results in Table I show a standard deviation of 0.2 p.p.m. for sulfate, 0.03 p.p.m. for chloride, and 1.2 p.p.m. for fluoride; average errors are of 0.1, 0.03, and 1.6 p.p.m., respectively. Table I. Analysis of Standard Sulfate, Chloride, and Fluoride Solutions
Taken 0.96 1.93 3.86 0.10 0.50 1.00 3.00 5.0 30.0 40.0
(Parts per million) Found Sulfate 0.91,0.85, 1.10 Av. 0.95 1.80, 2.10, 1.85 Av. 1.92 3.80, 3.50, 4.00 Av. 3.77 Chloride 0.12, 0.11, 0.11 Av. 0.11 0.46, 0.49, 0.51 Av. 0.49 0.90, 1.00, 0.95 Av. 0.95 3.05, 3.00, 3.02 Av. 3.02 Fluoride 3 . 5 , 4 . 5 , 3.5 Av. 3 . 8 27.5, 32.0, 32.5 Av. 30.7 39.0, 38.5, 40.5 Av. 39.3
acid chloranilate ion to give negative errors, by reacting with the metal chloranilate and releasing additional acid chloranilate to give positive errors, or by absorbing between 300 and 335 mp to give positive errors. Errors produced by the first two reactions are not serious for sulfate and chloride ( I , 3) but are for fluoride. Ten solutions, each containing one anion as the ammonium salt a t a concentration of 10 or 100 p.p.m., were analyzed by the procedure for fluoride. Results, expressed as apparent parts per million of fluoride, {Tere: Anion NosHPOd--
so,-c1-
10 P.P.M. 0 1
3
>60
100 P.P.M. 1
>60 >60 >60
Bromide, iodide, carbonate, and oxalate interfere to the same extent a8 chloride. Hence, only nitrate may be present when the procedure is used. Absorption of the common anions in the region between 300 and 335 mp is not a problem. Bromide, carbonate, chloride] citrate, fluoride, iodide, oxalate, phosphate, sulfate, sulfide, sulfite, and tartrate do not absorb in this region. Nitrate absorbs slightly. At the wave length used, 2500 p.p.m. of nitrate produces an absorbance roughly equivalent to 1 p.p.m. of chloride, 0.3 p.p.m. of sulfate, and 0.4 p.p.m. of fluoride. CONCLUSION
The procedures for sulfate and chloride were tested on t v o municipal waters. Sulfate was also determined by the conventional turbidimetric method, and chloride by the ferric thiocyanate colorimetric method (2). Table I1 shows that agreement b e h e e n the methods was within experimental error. These waters also contained fluoride, which was not determined because of interference from sulfate and chloride. All glassware should be cleaned before use with 1 to 1 nitric acid (or hydrochloric acid for sulfate), then rinsed thoroughly with distilled water. Most detergents absorb strongly between 300 and 335 mp, and the last traces of them are difficult to remove from laboratory glassmre. Anions may interfere in all three procedures in three ways: by precipitating or otherwise reacting with the Table II.
Source Gary, Ind. Chicago, Ill.
Sulfate, chloride, or fluoride in aqueous solution can be determined in 30 minutes. Based on the assumption that a difference of 0.005 absorbance unit can be detected with certainty, limit of detection for 1-em. cells is 0.06 p.p.m. of sulfate, 0.05 p.p.m. of chloride, and 0.5 p.p.m. of fluoride in the original sample. Hon-ever, the method for fluoride is not usable below 5 p.p.m. because the calibration curve is not reproducible in this range. The detection limits for sulfate and chloride have probably been reached, but that for fluoride might be improved by the use of other metal chloranilates and other solvents. Wide application of the methods for sulfate and chloride is suggested because of the relative lack of interferences from other anions. The method for fluoride is much less useful because of
Determination of Sulfate and Chloride in Water
(Parts per million) Chloride Sulfate Chloranilate Turbidimetric Chloranilate Thiocyanate 7.9 7.6 19.1 20.4 6.9 7.2 23.1 23.4
the many interferences. It is rapid, however, and frequent calibration is not required. It should be useful for determining fluoride after distillation of silicon tetrafluoride and subsequent hydrolysis. Extension of the principles used in these studies to the ultraviolet spectrophotometric determination of other anions in the concentration range of 1 to 100 p.p.m. should be feasible. A useful technique for further studr should be the measurement of the ultraviolet absorption of organic acids or anions Produced by reaction of metal salt of the acid to be determined.
Studies of other organic acids, particularly other substituted 2,5-dihydroxy-pbenxoquinones, may lead to more sensitive. more specific methods. Salts of bromanilic acid, nitranilic acid, polyporic acid, and atromentin are being studied in these laboratories. ACKNOWLEDGMENT
Assistance of R. W. Fish and Betty Ellen Ries is gratefully acknowledged. LITERATURE CITED
( I ) Barney, J. E., Bertolacini,-$. J., -4NAL. CHEM. 29, 1187 (1901).
Bergmann, J. G., Sanik, J., Ibid., 29, 241 (1957). Bertolacini, R. J., Barney, J. E., Ibid., 29, 281 (1957). Curry, R. P. Mellon, PVI. G., Ibid., 28, 1567 (1956). Lambert, J. L., Ibid., 26, 558 (1954). Lothe, J. J., Ibzd., 28, 949 (1956). Sanchis, J. AI., IND.ENQ.CHEM., A N A L . ED. 6 , 134 (1934). Schnwzenbach, G., Suter, H., Helv. Chim. Acta 24,617 (1941). Taltuitie, S . A., IND.Eso. CHEM., ANAL.ED. 15, 620 (1943). RECEIVED July 26, 1957. Accepted October 24, 1957. Eighth Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, AIarch 1957.
Spectral Absorption of Asphaltic Materials H. E. SCHWEYER Department of Chemical Engineering, University o f Florida, Gainesville, Fla.
b An exploratory study of the absorption of ultraviolet and infrared energy b y several asphaltic materials included consideration of the applicability of the Beer-Lambert law to asphalts dissolved in a mixture of iso-octane and 1 -butanol for ultraviolet absorption, and in carbon tetrachloride and thin films for infrared absorption. Typical spectrograms are illustrated for several asphalts.
F
commercial products are as compleu in chemical nature as those defined by the generic term asphalt or asphaltic bitumen. In the hope that ultraviolet and infrared spectrophotometry might explain differences in asphaltic materials, an exploratory study has been carried out t o establish techniques. The literature on the subject is limited and when this paper was submitted only the publications of Eilers ( 2 ) , and Killiford (12) using x-rays were available to the author. The recent papers of Stewart (10) and Linnig and Stewart ( 7 ) have provided additioiinl information on asphalts separated into fractions by chromatographic techniques. However, the numerous publications on spectral absorption of lubricating oils are pertinent where the oils in asphalts may be considered as higher-boiling hydrocarbons of the same general nature. Pertinent studies on high-boiling petroleum oils are those of Charlet, Lanneau, and Johnson ( I ) , Hastings and associates ( d ) , Hersh, and associates (6)> and Lillard, ,Tones, and Anderson (6). EW
ULTRAVIOLET SPECTROGRAMS
The technique for asphalt requires
the use of solutions. While iso-octane (Phillips Petroleum "spectral grade") is a suitable diluent for most petroleum oils, it is not a good solvent generally for asphaltic materials. A mixture of 15% (volume) of 1-butanol (synthetic, with a boiling range of 1.5" C. including 117.7" C.) and 85% iso-octane is a satisfactory solvent a t asphalt concentrations of 0.02 gram per liter or less and provides suitable transmittance in a Beckman spectrophotometer Model DU, employing the mixed solvent as the blank for a cell thickness of 1 em. Approximately 0.1 gram of asphalt is x%-eighed directly into a tared 50-ml. volumetric flask and the proper amount of premixed stock-solvent solution added to fill the flask. After solution is complete, a 1-ml. aliquot of the concentrated solution is made up to 100 ml. in a volumetric flask with mixed solvent t o give a final concentration of about 0.02 gram per liter (which can be computed exactly from the R-eight of asphalt used). Smaller flasks and a larger number of dilution steps may be used to conserve solvent or sample (10-ml. flasks recommended). As observed bjeye, this technique gave complete solution for the soft residual asphalts studied.
As solvents for asphaltic materials benzene, carbon tetrachloride, and carbon disulfide were discarded either because their absorption was too high in the spectral ranges of interest or for other reasons. The use of 1-butanol is subject t o some objection from a purity standpoint, but when used in small proportions (15%), the type of impurities likely to be present were not as critical as for the other solvents. The mixed 1-butanol and iso-octane has been found satisfactory for most soft asphalts (up to 130" to 150' F., ring and ball softening points). How-
ever, certain components from harder air-blown asphalts and from soft catalytic-blown asphalts have been found to be incompletely soluble in the mixed solvent. The absorbance for the final dilute solution in comparison with the blank is measured directly on the instrument a t different wave lengths starting a t 400 mp and a t decreasing values by increments of 5 mp until 240 is reached. Then readings are taken a t 2-mp increments until no reading can be made, The absorbance per unit of concentration (grams per liter) and per unit of thickness (centimeters) is the absorptivity as computed from a = log (Io/I)/(cb) = A / ( & ) ,absorptivity
where lais the transmittance of the blank, I is the transmittance of the unknown in solution, and A is the absorbance at a given wave length as read directly on the instrument; if the product cb has a value of 1, then a and A are equal. c is the experimental concentration in grams per liter and b is the cell thickness in centimeters. The validity of this equation for asphaltic materials was checked on an extracted fraction of one asphalt using different concentrations and different cell thicknesses (Table I). The acetonesoluble portion from a butanol extract of a Texas residuum S-160 (see Table 111) by the Trader-Schweyer method (11) was used for these data. Data in Table I indicate that the absorptivity in 1-cm. cells is not affected by concentration within average limits of =t2 units, which the author considers acceptable based on experiments with different spectrophotometers, different operators, and different samples. With identical starting samples, the ultraVOL. 30, NO. 2, FEBRUARY 1958
* 205