Analyst, January, 1969, Vol. 94, @. 43-48 43 The Determination of Sulphur Dioxide with Rosaniline Dyes BY H. G. C . KING AND G. PRUDEN (Rothamsted Experimental Station, Harpenden, Herts.) The compositions of commercial rosaniline hydrochloride (magenta) and pararosaniline hydrochloride samples have been re-investigated to improve the colorimetric determination of sulphur dioxide. The dyes, purified in small amounts by paper chromatography, each give a linear calibration up to 80 pg of sulphur dioxide, with the hydrochloric acid - formaldehyde reaction at pH 1.1. Of the two dyes, pararosaniline hydrochloride is preferred because of its smaller reagent blank value. Impurities can be readily removed from pararosaniline base by recrystallisation, and this reagent is, therefore, recommended for routine work.IN seeking methods to improve the colorimetric finish to the determination of sulphur dioxide from sulphur in soils by Bloomfield's vanadium pentoxide method,l we have re-investigated the relative merits of pararosaniline hydrochloride (Ia) and rosaniline hydrochloride (Ib, magenta or basic fuchsin) as colorimetric reagents in the sulphite-hydrochloric acid- formaldehyde reaction. c1- NH2 la (R=H), Pararosaniline hydrochloride I b (R=CH,), Rosaniline hydrochloride (magenta) Acid-bleached magenta, first used by Steigmann2 as a quantitative reagent for bisulphite, was later used by Urone, Boggs and Noyes3 to determine sulphur dioxide in air. The use of magenta was largely discontinued about 6 years after its introduction as a quantitative reagent, because of an alleged lack of purity, usually ascribed to admixed paarosaniline hydro- chloride and to other, unspecified, compounds.It is known that pararosaniline hydrochloride may be formed during the manufacture of magenta, although the cause of erratic results, which some authors attribute to the impurities, has not been shown conclusively to be the result of the use of mixed magenta - pararosaniline hydrochloride reagents. The substitution of pararosaniline hydrochloride for magenta was first made by West and Gaeke? who did not discuss the purity of the new reagent. Later workers,6s6 however, drew attention to the existence of a variety of commercial pararosaniline reagents of varying dye potency. Pate, Lodge and Wartburg6 assayed spectrophotometrically eighteen samples of pararosaniline hydrochloride and basic fuchsin, and accepted only those showing the spectral maximum of the purest pararosaniline then available.More recently Scaringelli, Saltzman and Frey7 made a detailed study of the conditions affecting the determination of sulphur dioxide colorirnetrically, and used a counter-current distribution procedure to purify pararosaniline hydrochloride. We have examined various samples of pararosaniline hydrochloride and magenta by paper chromatography, and have further obtained sufficient of each of the pure dyes by 0 SAC and the authors.4 4 KING AND PRUDEN: THE DETERMINATION OF [AIzalyst, VOl. 94 chromatography on thick paper to compare their usefulness in the colorimetric determination of sulphur dioxide.Standard sulphite solutions were stabilised as the disulphitomercurate ion: [Hg(SO,),]2-, and the colour of the complex with the acid-bleached dyes was deter- mined at pH 1-1 after reaction with formaldehyde. With either dye a linear calibration was found from 0 to 80 pg of sulphur dioxide, but the reagent blank (E = 0.024) for para- rosaniline hydrochloride at the wavelength of maximum optical density (572 nm) was about one tenth of that (E = 0.196) of magenta at the same wavelength. Although none of the magenta samples we examined contained pararosaniline hydro- chloride as an impurity, it is possible that there are such samples, the purification of which would be difficult on a large scale. Also, because of its small reagent blank pararosaniline hydrochloride would be preferred as a reagent for determining sulphur dioxide.However, the preparation of the pure dye on a small scale by paper chromatography is unsuitable for routine work, especially if the dye contains a large proportion of impurity. We find that pararosaniline base (11)- NH2 I I, Pararosaniline base prepared by treating an acid-bleached solution of the hydrochloride with sodium hydroxide, can be substituted with advantage for the parent dye. The base can be made on as large a scale as convenient and, when recrystallised from aqueous methanol, is a reproducibly pure and stable reagent. Variability in the quality of the pararosaniline hydrochloride is over- come by removing impurities when the base is recrystallised. Rosaniline base, although it can be prepared and recrystallised in the same way as pararosaniline base, suffers from the disadvantage of its parent dye in giving the same high blank value in the determination. EXPERIMENTAL PARAROSANILINE HYDROCHLORIDE- After a chromatographic and spectrophotometric examination of various samples, one that contained few impurities (Hopkin & Williams Ltd., Catalogue No.6496.6) was selected for the preparation of small amounts of the pure dye by preparative paper chromatography and for making pararosaniline base. MAGENTA- Samples, some many years old, from various sources were examined. No sample was found to contain pararosaniline hydrochloride as an impurity, although one, labelled “Basic fuchsine,” consisted almost entirely of pararosaniline hydrochloride, with no magenta.In- organic salts were absent from every dye sample, but all of the samples contained more or less organic impurity. Small amounts of the pure dye could readily be isolated by paper chromatography . A sample of magenta, “For the preparation of Schiff’s reagent’’ (British Drug Houses Ltd.), was selected for preparative paper chromatography. PAPER CHROMATOGRAPHY- From the many combinations of solvents tested, the simplest solvent that would separate pararosaniline hydrochloride and magenta, giving appreciable RF values, was 60 per cent. METHODSJanuary, 19691 SULPHUR DIOXIDE WITH ROSANILINE DYES 45 aqueous ethanol. Solutions of pararosaniline hydrochloride and magenta (0.5 per cent. w/v in methanol) were applied to Whatman No. 3 chromatographic paper in the form of bands about 2 cm long, by using a fine glass capillary.The use of thick paper, rather than the thinner Whatman No. 1 and No. 2 grades, helps to keep the bands as compact as possible during the run. At a temperature of 22" C the solvent front travelled 30 cm (descending) in 64 hours, after which time the bands had increased to about three times their starting width. RF values, to the front edge of the bands, were 0.78 for pararosaniline hydrochloride and 0.83 for magenta. Although mixtures of the dyes could be distinguished by their colours, definite separation of the bands was made difficult by their broadening during the run. Each dye contained a violet-coloured mobile impurity, RF value 0-70. Magenta samples also contained a second violet impurity that ran immediately in front of the main dye band.Each dye left at the origin a blue - violet band that could not subsequently be eluted. The mobile bands could all be eluted with methanol. For preparative work the dyes were banded on to several Whatman No. 3 papers, 25 em wide, cut to a point at their lower edge. The main dye bands were collected in beakers placed under the papers during prolonged elution with the chromatographic solvent. SPECTROPHOTOMETRY- Measurements of spectra in the visible region were made in 1-cm cells with a Hilger & Watts Spectrochem Mark 2 spectrophotometer. Purified pararosaniline hydrochloride and magenta were dissolved in water or ethanol and diluted to give a final concentration of 0.5 mg in 100 ml of solvent.Pararosaniline base, in aqueous suspension, was made just acid with hydrochloric acid to regenerate the hydrochloride, then diluted with water to give the same final concentration. A solution of the base in methanol, in which it is sparingly soluble was diluted to give a final concentration of 0.5 mg per 100 ml of methanol. 0 Wavelength, nm Fig. 1. Visible spectra of A, pararosaniline hydrochloride in water ; B, magenta in water; and C, para- rosaniline base in ethanol The wavelength of maximum optical density of pararosaniline hydrochloride was 536 nm in water and M5nm in ethanol. For magenta, the maxima were 540nm in water and 546 to 547 nm in ethanol. Pxarosaniline base, which is insoluble in water, showed a maxi- mum at 541 nm in ethanol. Fig. 1 shows the spectra of the dyes in water and of the base in ethanol.46 KING AND PRUDEN: THE DETERMINATION OF [Afldyst, VOl.94 The violet impurity, RF value 0-70, showed a maximum at 639 nm for water, and at fj49nm for ethanol. The fast-running impurity in magenta had maxima at MOnm for water and 649 nm for ethanol. None of the impurities, in butanol, showed a maximum as high as 660 nm, reported by Scaringelli, Saltzman and Frey.7 PREPARATION OF PARAROSANILINE BASE- One gram of pararosaniline hydrochloride was dissolved in 250 ml of 2.6 N hydrochloric acid. After standing for 2 hours the bleached solution was filtered through a Whatman No. 1 filter-paper. The base was precipitated as pale magenta-coloured shiny plates by adding a slight excess of 2 . 5 ~ sodium hydroxide solution to the filtrate, and the precipitate was collected on a sintered-glass plate, porosity 3; the filtrate was colourless.The crude base was washed thoroughly with water to remove sodium chloride and excess of sodium hydroxide, and was then recrystallised by dissolving in 70 ml of methanol at the boiling-point, adding 300 ml of water at 80" C and allowing the solution to cool at room temperature. The re- crystallised product, 64 per cent. of the weight of starting material, blackened in the range 200" to 205" C and decomposed at 286" C. Scaringelli, Saltzman and Frey7 prepared the base by recrystallisation from water as an intermediate in their subsequent preparation of pararosaniline hydrochloride. We found that recrystallisation from water gave a yield of only 12 per cent.Recrystallisation from aqueous methanol undoubtedly leads to some loss of product, which is retained in the filtrate, but impurities that are soluble in the solvent are completely removed, as shown by a chromatographic check on the hydrochloride prepared from the pure base. COLORIMETRIC DETERMINATION OF SULPHUR DIOXIDE- Bleached 9ararosaniline reagent solutiort-One gram of pararosaniline base is dissolved in 60ml of concentrated hydrochloric acid and the solution diluted to 1 litre with water. We have not had the opportunity to study the stability of the reagent solution over a long period, but have kept a solution unchanged for 6 weeks in the dark; the reagent solution of Scaringelli, Saltzman and Frey' is reported to be stable for at least 9 months. Potassium tetrachZoromercurate solutim-A solution of 27.2 g of mercury( 11) chloride and 14-9g of potassium chloride in water is diluted to 1 litre.Formaldehyde solution, 0.2 per cent.-Five millilitres of 40 per cent. formaldehyde solution are diluted to 1 litre with water. The solution should be freshly prepared. Standard sulphite solution-A solution of 0-4 g of anhydrous sodium sulphite in 500 ml of water is prepared, corresponding to between 300 and 400pg of sulphur dioxide per ml. The concentration of sulphite is determined iodimetrically. One millilitre of 0-1 N iodine solution is equivalent to 3.2033 mg of sulphur dioxide. Immediately after analysis the sulphite solution is stabilised by dilution with potassium tetrachloromercurate solution.Ten millilitres of sulphite solution diluted to 600 ml with the tetrachloromercurate give a solution corresponding to between 6 and 8 pg of sulphur dioxide per ml. Wavelength, nm Fig. 2. Visible spectrum of the sulphur dioxide (60 pg) - pararosaniline hydrochloride - formaldehyde complexJanuary, 19691 SULPHUR DIOXIDE WITH ROSANILINE DYES 47 CALIBRATION- Aliquots of the dilute sulphite solution, corresponding to not more than 80 pg of sulphur dioxide, are transferred to 50-ml graduated flasks, and 10 ml each of bleached pararosaniline reagent and formaldehyde are added. The solutions are diluted to 50 ml (final pH 1-1) and are then allowed to stand for 20 minutes before measuring the optical densities, in 1-cm glass cells, at 672 nm against a reagent blank.Fig. 2 shows the visible absorption spectrum for the developed colour, and Fig. 3 the calibration graph. Sulphur dioxide, mg Fig. 3. Calibration for sulphur dioxide with A, para- rosaniline hydrochloride; and B, magenta, against respective reagent blanks DISCUSSION In the various dye samples we examined, neither the amounts nor the nature of the impurities were such as to give variable results in the colorimetric determination of sulphur dioxide. It is difficult to compare our results directly with those of other workers, who used reagents of varying quality from many sources but, unless objection is made to the higher reagent-blank value of magenta, there seems to be no reason why either of the rosaniline dyes should not be used. Although none of our magenta samples was contaminated with para- rosaniline hydrochloride, we found that one sample, labelled “Magenta,” was pararosaniline hydrochloride, and that one sample of the latter was magenta.* If pararosaniline hydrochloride contains a large amount of impurities, its optical density will be much less than Ei& = 2600 at 536 nm (water), but small amounts of the pure com- pound can be obtained by preparative paper chromatography. Column chromatography on cellulose is unsatisfactory because the dye is adsorbed strongly on the cellulose; this does not happen on paper chromatograms unless the paper is dried thoroughly after a run and the chromatogram is then re-run.For gram-scale amounts of the reagent when only impure samples are available, or as a routine measure, it is preferable to use pararosWne base, which can be prepared in good yield as a pure, crystalline product.The base gives an extremely pale acid-bleached reagent solution and as good a calibration as the pure hydrochloride. The impurities in pararosaniline hydrochloride and magenta can be readily detected by the chromatographic system described above. The small difference in the structures of the dyes makes it more difficult to detect a mixture, but this can be done if the bands of dye applied to the starting line of the chromatogram are kept as narrow as possible. It is useful, but not always necessary, just to dry the chromatogram and re-run it in the same solvent to improve the separation. Much information about possible contamination of magenta with pararosaniline hydro- chloride can be gained by a study of the spectral maxima of mixtures of the dyes.The spectral assay by Pate, Lodge and Wartburg6 showed that it is possible to distinguish between the dyes; the grouped samples having a maximum optical density at 543 nm as para- rosaniline hydrochloride, and those with maximum at 549 nm were grouped as basic fuchsin * Large blank values were found by Bloomfield in the colorimetric finish to his determination of sulphur in soils1 (Dr. C. Bloomfield, personal communication). His “pararosaniline hydrochloride” was, in fact, magenta.48 KING AND PRUDEN (magenta). Pate, Lodge and Wartburg do not state specifically the nature of the solvent in which measurements were made, and we have found no solvent for which the maximum optical density of magenta is as high as 549 nm. Earlier workers failed to point out the importance in spectral assay of the lower optical density of magenta compared with that of an equal concentration of pararosaniline hydro- chloride.Magenta that contains more than 10 per cent. of pararosaniline hydrochloride shows a decrease in its wavelength of maximum optical density, accompanied by an increase in its original optical density. For a spectral assay it is necessary first to prepare small amounts of the pure dyes chromatographically. Table I shows the change in optical density and maximum wavelength of mixtures of the dyes in water. TABLE I CHANGES IN SPECTRAL MAXIMUM AND OPTICAL DENSITY OF MIXTURES OF MAGENTA 06mg IN 1ood OF WATER AND PARAROSANILINE HYDROCHLORIDE AT A TOTAL CONCENTRATION OF Pararosaniline, % 0 10 20 30 40 60 60 70 80 90 100 Emax.(nm) . . 640 640 639 639 638 638 637 637 637 636 636 Magenta, % , . 100 90 80 70 60 60 40 30 20 10 0 0.889 0.909 0.960 0.960 0.984 1.107 1.177 1.180 1.193 1.293 1-303 Variations in the conditions for the colorimetric determination of sulphur dioxide have been studied by Scaringelli, Saltzman and Frey,' who measured the developed colour at one of two pH values, 1.6 (Method A), Amax. = 548nm, giving a large reagent blank value, or at 1.2 (Method B), Amax. = 575 nm, giving a smaller reagent blank, the change from the higher to the lower pH value being made by adding 3 N orthophosphoric acid. Under our conditions, with 100 mg of pararosaniline base in 1 litre of 6 per cent.v/v hydrochloric acid, the pH of the reaction mixture is 1.1 and the wavelength of maximum optical density is 572 nm. We suggest that, to eliminate variations in the determination of sulphur dioxide when impure commercial pararosaniline hydrochloride has been used, recrystallised pararosaniline base should be substituted. A considerable improvement can be made at once to the determination of sulphur in soils (measured as sulphur dioxide) to improve the original calibration of 2 to 25 pg of sulphur, linear in the range 0 to 10pg.l The determination of sulphur dioxide from other sources, e.g., from air, or in foods and drinks, may require modifications to the initial stages of the method to inactivate interfering substances, e.g., heavy metals, nitrogen dioxide and ozone. These modifications are fully discussed by Scaringelli, Saltzman and Frey,' but the final colorimetric determination of sulphur dioxide is unaltered. It has been suggested8 that p-aminoazobenzene is a superior reagent to pararosaniline hydrochloride. We cannot confirm this ; the reagent blank for the 9-aminoazobenzene reagent has twice the optical density of that of magenta, and the calibration is linear only up to 60 pg of sulphur dioxide. REFERENCES 1. 2. 3. 4. 6. 6. 7. 8. Bloomfield, C., Analyst, 1962, 87, 686. Steigmann, A., Analyt. Chem., 1950,22, 492. Urone, P., Boggs, W. E., and Noyes, C. M., Ibid., 1961, 23, 1617. West, P. W., and Gaeke, G. C., Ibid., 1956, 28, 1816. Pate, J. B., Lodge, J. P., and Wartburg, A. F., Ibid., 1962, 34, 1660. Barabas, S., and Kaminski, J., Ibid., 1963, 35, 1702. Scaringelli, F. P., Saltzman, B. E., and Frey, S. A., Ibid., 1967, 39, 1709. Kniseley, S. J., and Throop, L. J., Ibid., 1966, 38, 1270. Received August lst, 1968