July, 19661 BRAID, HUNTER, MASSIE, NICHOLSON AND PEARCE 439 An Examination of Some of the Factors Affecting the Determination of Carbon Dioxide by Non-aqueous Ti trime t ry BY P. BRAID, J. A. HUKTER, W. H. S. MASSIE, J. D. NICHOLSON AND B. E. PEARCE (Heviot- Watt Univevsity, Chambers Street, Edinburgh 1) Some of the factors, including choice of absorbent, indicator and titrant, that affect the determination of carbon dioxide by non-aqueous titrimetry are examined experimentally. The results of this examination are used to give a procedure that is recommended for the determination of milligram amounts of carbon dioxide. The method involves absorption of the carbon dioxide in a 5 per cent. v/v solution of ethanolamine in formdimethylamide, followed by titration with standard tetrabutyl ammonium hydroxide in benzene - methanol solution t o a visual end-point with thymolphthalein indicator.THE direct titrimetric determination of carbon dioxide with visual indications is possible when carried out in non-aqueous solvents. Attempts to increase the accuracy of the method have usually involved changing the combination of reagents or conditions of absorption or titration in order to improve retention of the carbon dioxide, or to facilitate observation of the equivalence point. Changes that have been made appear to have resulted from trial and error rather than from any systematic examination of the various factors involved. Grant et aZ.1 used formdimethylamide as the absorbing solvent instead of acetone or pyridine as used by Blom and Edelhausen.2 They also reported more satisfactory end-point detection with thymolphthalein as indicator in place of thymol blue.LTarious titrants have been tried, and Grant et aZ.l found that potassium methoxide in benzene - methanol solution was superior to sodium methoxide. The method described by White3 involves continuous titration of the carbon dioxide as it is introduced into the solvent to prevent its removal by the carrier gas, whereas Patchornick and Shalitin4 overcame this difficulty by the use of a solution containing benzyl- amine to retain the carbon dioxide, probably as benzylamine N-benzyl carbamate. The Steel Company of Wales (private communication, 1964) found that retention in formdimethyl- amide could be improved by the addition of an unspecified amount of ethanolamine.Indicator fading has been described by White3 who applied a time-dependent “fade correction” to the volume of titrant used; a procedure that was later adopted by Boyle, Stevens and Sunderland.5 This paper describes experiments that have been carried out to examine the absorption of carbon dioxide by various organic amines in order to select the most suitable for addition to formdimethylamide to improve its carbon dioxide retention characteristics. A semi- quantitative comparison of different indicators in the selected solvent system has also been made, and the conclusions resulting from experience gained with various non-aqueous titrants during the course of this section of the work are given. The procedure finally chosen is suitable for the determination of carbon dioxide produced in various ways, and some results for the micro determination of carbon in O.A.S.compounds by rapid combustion are included to show how the titrimetric method could be applied to a standard system. EXPERIMENTAL ABSORPTION AND RETENTION OF CARBON DIOXIDE- The reactions of a number of amines with carbon dioxide were examined by passing the gas sequentially into undiluted samples of diethylamine, diamino-ethane, butylamine, benzylamine, cyclohexylamine and ethanolamine.440 BRAID et al. : AN EXAMINATION OF SOME OF THE FACTORS AFFECTING [Analyst, Vol. 91 A vigorous exothermic reaction occurred each time, producing a syrupy liquid with ethanolamine and white solid products with each of the other amines. The compounds formed differed in stability ; those from diethylamine and diamino-ethane decomposed rapidly but the others were more stable, the ethanolamine compound appearing unchanged after 24 hours.Infrared spectra of the 4 relatively stable products were obtained by using the potas- sium bromide disc techniques with a Perkin-Elmer infracord spectrophotometer. For each product the twin-SH, peaks (3500 to 3300 per cm) were replaced by a single peak correspond- ing to a =NH group. The -NH, peaks at 1650 and 1580 per cm vanished, and new peaks were observed in the region of 1600, 1300 and 800 per cm, which would indicate the presence of a -?r'H,' group, and at 1610 to 1550 and 1420 to 1300 per cm as would be expected for a COO- group. It therefore seems likely that substituted ammonium carbamates are formed. There are two possible reactions for the titration of carbamates by a strong base such as potassium methoxide- (a) H,NR.oOC.NHR + KOCH, = RHN.COOK + RNH, + CH,OH or (b) H,hR.OOC.NHR + KOCH, = SRNH, + KCH,CO,.Patchornick and Shalitin4 favour (a), but either gives the equivalence relationship- 1 litre of N base = 44-01 g of carbon dioxide. The absorption efficiencies of the different amines for carbon dioxide were next examined for solutions of the amines of different concentrations in formdimethylamide. The initial proportion of amine to formdimethylamide used was 3 per cent. v/v, but if good recovery of carbon dioxide was obtained the concentration was lowered to 1 per cent. v/v, whereas if poor recovery resulted with the 3 per cent.solution the concentration was raised to 5 per cent. v/v. Known amounts of carbon dioxide, obtained by heating weighed milligram amounts of sodium hydrogen carbonate, were swept into 10-ml samples of the chosen absorbent contained in a small flask by a stream of purified nitrogen maintained a t a flow-rate of about 50 ml per minute. After 15 minutes the absorbed carbon dioxide was titrated with approxi- mately 0.02 N standard tetrabutyl ammonium hydroxide in benzene - methanol. The titrant was introduced from a 5-ml microburette until a faint blue colour was obtained with thymol- phthalein indicator. The results are summarised in Table I. TABLE I RECOVERY OF CARBON DIOXIDE ABSORBED Carbon dioxide range calculated froin weights of Proportion sodium hydrogen of solvent in carbonate, formdimethylamide mg 3*& v / \ ~ of cyclohexylamine.. 1.691 to 3-676 5 O / O v/v of cyclohexylamine. . 2-302 to 3.550 30,, x ~ / v of ethanolamine . . 2.061 to 2.919 lo/, \-/v of ethanolamine . . 1.855 to 2.601 3"/, v/v of benzylaminc . . 2.320 to 2.874 l:,, v/v of benzylaminc . . 2.288 to 2.807 3*, v/v of butylaniinc . , 2.405 to 3-119 l o b v/v of butylamine . . 2.579 to 2.818 3 O ; v/v of diethylamine . . 2-546 to 2.800 5:& v/v of diethylamine , . 2-583 to 2.736 Number of deter- minations 3 3 5 9 5 9 7 7 9 9 BY DIFFERENT SOLVENTS Carbon dioxide range found, mg 0.595 to 1.330 0.63 1 to 0.95 1 2.062 to 2.905 1.825 to 2.565 2-297 to 2.813 2.321 to 2.804 2-353 to 3-145 2.572 to 2.795 2.380 to 2.687 2.556 to 2.717 Error range, per cent.-69.6 to -57.5 -74.1 to -70.5 - 0.75 to + 1.45 - 1.62 t o + 1.83 - 3.95 to - 0.99 - 2.79 to -+ 1.44 - 2.17 to + 1.95 - 0.82 to + 0.04 - 8.25 to + 0.04 - 4.39 to + 2.12 The most accurate determinations were those with 1 per cent. butylamine or 3 per cent. ethanolamine. Those with 3 per cent. butylamine were less satisfactory, but this is attributed to increased difficulty in discerning the end-point of the titration rather than to any decrease in the efficiency of absorption of the carbon dioxide. Tndeed, all of the amines tended to make the end-point colour changes less sharp than when they were absent. This effect was least evident with ethanolamine, with which satisfactory end-points could be obtained a t a 5 per cent. concentration in the fonndimethylamide, although they became less satisfactory if the amine concentration were further increased.July, 19661 DETERMINATION OF CARBON DIOXIDE BY YON-AQUEOUS TITRIMETRY 441 SPECTROPHOTOMETRIC EXAMINATION AND COMPARISON OF DIFFERENT INDICATORS- Titrations of carbon dioxide with alkali in the presence of various indicators with visual observation of the colour changes had revealed that in some instances the changes of colour were rather gradual.To permit the selection of the most suitable indicator the end-point characteristics of a number of indicators were compared by spectrophotometric titration of carbon dioxide solutions. A Unicam SP600 spectrophotometer was used with a specially constructed glass cell (see Fig. 1) suitably supported on a wooden block. The cell was fitted with a 5-ml microburette with drawn-out jet, and with a suitably bent polythene tube for entry of nitrogen to stir the solution.Burette I V I Good en block Fig. 1 . Titration cell for spectrophotometric examination of indicators Indicators whose colour change in aqueous solution occurred at pH values above 7 were chosen, as in aqueous solution these would be likely to be suitable for the titration of a relatively weak acid with a strong base. Azo violet, which has been recommended by Beckett and Tinley6 for similar titrations in non-aqueous solvents, was also examined. The absorption spectra of suitable solutions containing equal concentrations of the chosen indicators in the “acid” and “base” forms, respectively, were examined, and from the results the wavelengths of greatest increase in absorbance on passing from “acid” to “base” forms were found.Several 1-ml portions of an approximately 0-3 M solution of carbon dioxide in ethanol- amine - formdimethylamide were diluted to 40 ml with the solvent mixture in the titration cell. After the addition of the amount of indicator that had been found by trial and error to give a change in optical density of about 0.8 units during the titration, the carbon dioxide was titrated with an approximately 0.1 N non-aqueous base. The end-point colour change was followed spectrophotometrically by a procedure similar to that described by Hunter and Miller.7 The results obtained with different indicators in 3 and 5 per cent. v/v solutions of ethanolamine in formdimethylamide are shown in Fig.2. Similar graphs were obtained when benzoic acid (used for standardisation of basic non-aqueous titrants) was titrated instead of carbon dioxide. With either acid the sharpest change was found to be that from colourless to blue with thymolphthalein. This change corresponds to the easily discernible first appearance of a blue colour in a previously colourless solution, and confirms the qualitative conclusion of Grant et aZ.l that thymolphthalein gives end-points superior to those obtained with thymol blue.442 BRAID et al.: AN EXAMINATION O F SOME OF THE FACTORS AFFECTING [Ana&St, YO]. 91 Volume of basic titrant I division = I ml Fig. 2. Spectrophotometric titration curves for different indicators in carbon dioxide - ethanolamine - formdimethylamide solutions: curve I\, phenolphthalein in 3 per cent.ethanolamine - formdimethylamide ; curve B, phenolphthalein in 5 per cent. ethanolamine - formdimethylamide ; curve C , thymolphthalein in 3 per cent. ethanolamine - formdimethylamide; curve D, thymolphthalein in 5 per cent. ethanolamine - formdimethyl- amide; curve E, m-cresol purple in 3 and 5 per cent. ethanol- amine - formdimethylamide; curve I;, azo violet in 3 and 5 per cent. ethanolamine - formdimethylamide ; curve G, thymol blue in 3 per cent. ethanolamine - formdimethylamide; curve H, thymol blue in 5 per cent. ethanolamine - formdimethylamide The possible difference between the observed end-point and the true equivalence point is, in practice, rendered less significant by titrating always to about the same intensity of the blue colour, and by the use of the same end-point in titrations of carbon dioxide and in standardisations.CHOICE OF TITRANT- In the course of the investigations several different titrants were tried, including solutions of the alkali methoxides in benzene - methanol, alkali hydroxides in various alcohols and tetrabutyl ammonium hydroxide in benzene - methanol. Difficulties were variously encountered due to precipitation of (presumably) alkali alkyl carbonates, to marked day- to-day variations in the results of standardisations even when maximum precautions were taken to preserve the freshness of the solutions (suggesting the occurrence of decomposition reactions), and to “fading” of the end-points. The phenomenon of “fading” of end-points in non-aqueous titrations of carbon dioxide has been reported by White,3 who attributed it to the presence of water and found that an empirical correction factor could be applied.Boyle et aZ.5 have similarly applied an empirical correction factor to allow for “fading.” The occurrence of fading in the various experiments of this work was not found to be in any way reproducible, and no satisfactory explanation for it could be found despite numerous experiments involving the alteration of a variety of conditions. The effect seemed to be connected in some way with the presence of methanol in the titrated system, but reproducible results could never be obtained. Standard solutions of tetrabutyl ammonium hydroxide in benzene - methanol were, however, essentially stable and did not react with carbon dioxide contamination to give rise to precipitates.More benzene (and consequently less methanol) could be present and still maintain a clearer solution than with the alkali methoxides, and fading effects were found to be essentially absent with this titrant.July, 19661 DETERMINATION OF CARBON DIOXIDE BY NON-AQUEOUS TITRIMETRY 443 In some of the preliminary experiments use was made of solutions obtained by dilution of the commercial tetrabutyl ammonium hydroxide of about 0.1 N to about 0.02 N by the addition of benzene, but it was later found that equally satisfactory results could be obtained with the 0.1 N solution in a 1-ml microburette. By virtue of the smaller increase in volume of liquid in the absorber, this facilitated the performance of a greater number of titrations (about 10) with the one sample of absorbent.PROCEDURE FOR THE DETERMINATION OF CARBON DIOXIDE APPARATUS- Titrationjask-This is essentially similar to that used by Grant et aZ.l and consists of a 50-ml Pyrex conical flask with side-tubes to lead the gas stream to the bottom of the flask and to allow the carrier gas to escape after absorption of the carbon dioxide. Stirring is done magnetically, and the burette tip is immersed in the absorbant to facilitate the addition of small increments of titrant. Burette-A grade A burette, of 1-ml capacity, fitted with a suitable reservoir, is used. Polytetrafluoroethylene stop-cocks are used ; these greatly facilitate the prolonged use of the burette by eliminating the sticking that results from the washing of lubricant from glass taps by the non-aqueous titrant.Soda-lime tubes are attached to prevent ingress of atmospheric carbon dioxide. REAGENTS- A bsorbent-Prepare a 5 per cent. v/v solution of ethanolamine in formdimethylamide ; ordinary-reagent grade. A 20-ml sample is adequate for up to about 10 carbon dioxide deter- minations of 5 to 10mg each, after which the titration flask is almost full and the ethanol- amine content of the absorbant has been reduced to 3 to 2.5 per cent. Titrant-This is a 0.1 x tetrabutyl ammonium hydroxide in benzene - methanol solution. I t is an approximately standard solution supplied by British Drug Houses, Poole, Dorset. Indicator-Prepare a 0.05 per cent. w/v solution of thymolphthalein in formdimethyl- amide; use 5 drops per 20 ml of absorbent.AZkaZimetric standard-Fuse O.A.S. benzoic acid on a platinum lid and break into small pieces to facilitate the addition of milligram amounts to the titration flask for standardisations. PROCEDURE- Standardisation of the 0.1 N tetrabutyl ammonium hydroxide-Milligram amounts of the fused O.A.S. benzoic acid are dissolved in 20-ml portions of the absorbent (previously brought to the first observable blue of the indicator) in the titration flask in a carbon dioxide- free atmosphere, and titrated with the titrant to the same blue end-point. C,H,COOH + C,H,NOH = C,H,COo + C,HgG + H,O The reaction is- whence 1 litre of N C,H,NOH = 122-1 g of C,H,COOH. Several standardisations may be performed without renewing the solvent, although the end-point deteriorates after a time, probably owing to the accumulation of water formed in the reaction.Titration of carbon dioxide-The procedure is similar to that of the standardisation. The carbon dioxide is flushed into the absorbent (previously titrated to pale-blue and stirred by the magnetic stirrer) by the carrier-gas stream, and then titrated to the same pale-blue end-point. The reactions are probably- kH,(CH,),(OH)(CO~)(OH)(CH,),NH + C,H,NOH = 2KH,(CH,),OH + C,H,NHCO,, 2NH2(CH,),0H + CO, = ~H,(CH,),(OH)(CO~)(OH)(CH,),NH i.e., effectively CO, + C,H,NOH = C,H,NHCO, whence 1 litre of N C,H,NOH = 44-01 g of CO,. Results obtained for carbon dioxide produced by heating milligram amounts of sodium hydro- gen carbonate are given in Table 11.Table I11 summarises the results obtained by applying the method to the determination of carbon in O.A.S. compounds by the rapid combustion process.444 BRAID, HUNTER, MASSIE, NICHOLSON AND PEARCE [A%dySt, 1:Ol. 91 TABLE I1 SOX-AQUEOUS TITRIMETRIC DETERMINATION OF CARBON DIOXIDE PRODUCED BY HEATING WEIGHED AMOUNTS OF SODIUM HYDROGEN CARBOXATE AND ABSORBED I N 5 PER CENT. V/V ETHANOLAMINE IN FORMDIMETHYLAMIDE Sodium hydrogen carbonate, mg 3.268 7.958 6.116 4.928 8-979 3-701 4.903 4.862 5.369 5-802 6-250 5.332 20.439 15-129 8.992 7.303 Carbon dioxide calculated, mg 0.854 2.084 1.601 1.290 2.351 0.969 1.283 1.273 1.406 1.519 1.636 1.396 5.348 3.961 2.355 1.912 Carbon dioxide found, mg 0.859 2.128 1.609 1.286 2.342 0.964 1.282 1.277 1-404 1.546 1-634 1.404 5-340 3-951 2.367 1.932 Error, mg + 0.006 + 0.044 + 0.008 - 0.004 - 0.009 - 0.005 - 0.00 1 + 0.004 - 0.002 + 0.027 - 0.002 + 0.008 - 0.008 -0~010 +0*012 + 0.020 TABLE I11 NON-AQUEOUS TITRIMETRIC ANALYSIS OF CARBON IN O.A.S. COMPOUNDS Number Weight Carbon Standard of range (found), Carbon deviation, deter- taken, percentage Mean, (calculated), percentage Compound minations mg range per cent. per cent. units Benzoic acid . . 9 3.6 to 4.7 67-45 to 68.44 67.91 68.84 0.36 Sucrose . . . . 10 4.3 to 5-8 41.70 to 42.84 42-15 42.10 0.34 Saphthalene . . 10 2.5 to 4.5 92.45 to 94.45 93.36 93.71 0.52 REFERENCES 1. 2. 3. iihite, D. C., Talanta, 1963, 10, 737. 4. 5. 6. 7. Grant, J. A., Hunter, J. A., and Massie, W. H. S., Analyst, 1963, 88, 134. Blom, L., and Edelhausen, L., Analytica Chim. Acta, 1955, 13, 120. Patchornik, A, and Shalitin, Y., Analyt. Chem., 1961, 33, 1887. Boyle, W. C., Stephens, F. B., and Sunderland, M’., Ibid., 1965, 37, 935; Beckett, A. H., and Tinley, E. H., “Titration in Non-aqueous Solvents, Hunter, J. A., and Miller, C. C., Analyst, 1956, 81, 79. Third Edition, British Drug Houses Ltd., Poole, Dorset, p. 35. Received September 9th, 1965