April, 19661 GREGORY 25 1 The Determination of Residual Reagents in Mineral Flotation Liquors Anionic Surface-active BY G. R. E. C. GREGORY ( JVarren Spritzg Laboratory, Stevenage, Herts.) A method is described for the determination of residual amounts of anionic surface-active reagents, used as flotation promoters, in mineral flotation liquors. The method is equally effective for the long carbon chain carboxylates and the anionic non-soapy surface-active agents. The reagent used is the cationic copper(1r) triethylenctetramine complex, which reacts in alkaline solution with anionic surface-active agents to give an adduct that can be extracted into an isobutanol - cyclohexane mixture. The copper asso- ciated with the surface-active anion in the extract is determined photo- metrically as the coloured complex using diethylammonium diethyldithio- carbamate. Long chain carboxylates with carbon numbers from C,, to C,,, as well as anionic non-soapy surface-active reagents, can be determined in the range 0-2 to 10 p.p.m.The effect of a number of likely interferences has been investigated. ANIONIC surface-active agents, particularly the carboxylate soaps, are extensively used in the mining industry as promoters for the separation of minerals by flotation. A simple and rapid method of determining residual amounts of these reagents is required for plant control, and also for investigating their adsorption on to mineral surfaces. The usual method of analysis for traces of anionic surface-active agents is that of Jones,l modified by Longwell and Maniece,2 in which a cationic dye, methylene blue, is added and the coloured adduct extracted into chloroform solution for photometric measurement.Many similar dye-adduct extraction systems have been reported3 but these generally fail when applied to the carboxylic acid soaps. The quaternary dye pinacyanol has been suggested as a reagent for the determination of laurates in weakly alkaline solution using bromobenzene as extractant for the a d d ~ c t , ~ but the reagent is too unstable for routine use. Tomlinson and Sebba5 determined traces of soaps in neutral solution with the cationic dye crystal violet. The adduct formed was removed by a process described as “ion flotation” and the optical density of the residual dye solution measured. The reaction is non-stoicheiometric, requires precise control of conditions and is very susceptible to interference from dissolved inorganic ions.Milligram amounts of long chain fatty acids in aqueous solution were determined photo- metrically by Aye& and by Iwayama7 by extracting the coloured cobalt or copper(I1) soaps into chloroform. Metal cations are more attractive as reagents than organic dyes that are difficult to obtain in an adequate state of purity, may not be completely stable and frequently tend to adsorb on to the surfaces of vessels used in the analysis. On the other hand, the extinction coefficients of the cobalt or coppcr(r1) soaps in organic solution, as in the methods of Ayers and of Twayama, are too low to be used for the present purpose. However, it has been found that the sensitivity can be considerably increased by adding a sensitive colorimetric reagent for the metal concerned to the metal-soap extract.This procedure has the advantage that slight differences in the extinction coefficients of the metal soaps of the various fatty acids used do not affect the optical density of the extract. A suitable reagent for copper(I1) soaps is dimethylammonium diethyldithiocarbamate. The extinction coefficient of the copper(1r) complex is about 11,000, and both the reagent and the complex are soluble in organic solvents. Attempts to extract microgram amounts of fatty acids from neutral or near neutral aqueous solution into organic solvents have not been successful. At this concentration level no extraction occurs when cupric nitrate is added, as in the method of Ayers, but some extraction does take place when an acetate buffered triethanolamine solution is used as described by Iwayama.Here, however, the reagent blank is very high. A better extraction occurs when alkaline solutions of complex copper ammines are used in place of the simpler copper(I1) salts. Thus, for example, traces of soaps extract into chloroform with copper(I1)252 GREGORY: DETERMINATION OF RESIDUAL ANIONIC [Analyst, Vol. 91 ions in the presence of alkali hydroxide and an excess of ammonia. Recoveries using this method tend to be low and erratic, but can be improved by replacing the cuprammonium ion by the complex triethylenetetramine copper(I1) cation. The necessary alkaline pH can conveniently be achieved by incorporating the required base in the copper reagent solution.EXPERIMENTAL PREPARATION OF THE REAGENT- The reagent is prepared with a slight excess of triethylenetetramine in addition to that necessary to complex the copper, thus avoiding precipitation of copper as the hydroxide on adding alkali. There is a tendency for copper(1) oxide to precipitate slowly from the reagent when either sodium or potassium hydroxide is used, but a completely stable reagent can be prepared by using monoethanolamine as the base. This forms a complex with copper(I1) ions which does not react with soaps to give extractable products, and is much less stables (log K = 16-48) than that of trieth~lenetetramine~ (log K = 20.6). The separation of the organic phase is also much cleaner when monoethanolamine is used as the base. CHOICE OF EXTRACTANT- Of a wide range of organic liquids tried, only chloroform and the immiscible aliphatic alcohols give anything approaching complete extraction of the soap adducts.Chloroform gives a lower extraction than the alcohols and this is obtained only at very high pH values. The alcohols, however, tend to give cloudy extracts containing aqueous entrained phase ; also the reagent blanks are high and increase as the ionic strength of the aqueous phase is increased. The entrainment effect is least with isobutanol, although particularly high reagent blanks are obtained. The addition to the isobutanol of either benzene or cyclohexane, which do not themselves extract the soap adducts, not only reduces the reagent blanks but also prevents entrainment of the aqueous phase without affecting the recovery.Because it has a much lower toxicity, the use of cyclohexane is preferred to that of benzene. REAGENTS- Copper - triethylenetetramine reagent-Dissolve 25 g of copper( 11) nitrate trihydrate in 125 ml of water and stir slowly into a solution containing 16.25 g of triethylenetetramine in 125 ml of water. Add 250 ml of monoethanolamine in 250 ml of water and dilute with water to 1 litre. Isobutanol - cyclohexane extractant-Mix 200 ml of isobutanol with 800 ml of cyclohexane. Diethylammonium diethyldithiocarbamate solution-Dissolve 2 g of diethylammonium diethyldithiocarbamate in 100 ml of isobutanol. Standard soap solution-Dissolve 0.250 g of the required pure fatty acid in 200 ml of methylated spirit, add 1 ml of 0.88 sp.gr.ammonia solution and dilute with water to 1 litre. Standard solutions of non-soapy surface-active agents may be prepared in water or alcohol - water mixture. Prepare freshly every 2 days. PROCEDURE- Perform the extractions in 75-ml boiling tubes fitted with size B24 interchangeable ground-glass stoppers. Centrifuge samples which are cloudy, or which contain particulate matter, until clear. Transfer a 25-ml sample by pipette to an extraction tube and add 5 ml of the copper - triethylenetetramine reagent, followed by exactly 10 ml of the isobutanol - cyclohexane extractant. Stopper the tubes and invert rapidly 100 times. After the phases have separated, transfer the organic layer to a dry test-tube by means of a dropping pipette fitted with a rubber bulb.Mix the extract with two drops of the diethylammonium diethyl- dithiocarbamate solution and allow to stand in a dark place for 15 minutes. Measure the optical density relative to the extractant in 2-cm cells at a wavelength of 435 mp. A blank determination must be made. For calibration purposes, take aliquots of a standard solution in a series of extraction tubes and dilute each to 25ml with water to give a range of concentrations from zero to 10 p.p.m. Complete the determinations as described above.April, 19661 SURFACE-ACTIVE REAGENTS IN MINERAL FLOTATION LIQUORS 253 DISCUSSION The presence of traces of surface-active material in reagents or on glassware must be avoided. One batch of triethylenetetramine, which is only readily available in technical quality, gave a 0.3 p.p.m.intercept on the calibration graph. This can be overcome by adding an equivalent amount of sodium lauryl sulphate to the reagent, but at the cost of a slightly higher blank. Distilled water from an all-glass still has been found satisfactory for the determination. Glassware should be treated with chromic acid, rinsed with water, alcohol and chloroform and allowed to drain dry. COMPOSITION OF THE EXTRACTANT- The addition of cyclohexane to the isobutanol extractant has two effects. These are the reduction of the reagent blank and the prevention of entrainment of aqueous copper solution when soaps are present. With too much cyclohexane the recovery is reduced and cloudy extracts are obtained. These effects are shown in Fig.1. The best extractant contains 80 per cent, by volume of cyclohexane, and it can be seen that the recovery is not critically dependent upon its exact composition. Identical results are obtained when the cyclohexane is replaced by benzene, but hexane almost completely suppresses extraction. EFFICIENCY OF EXTKACTION- No further improvement in extraction occurs after shaking the tubes 40 times, but vigorous agitation sometimes produces a fairly stable emulsion when more than 5 p.p.m. of a soap is present. By rapidly inverting the tubes 100 times, in place of shaking, maximum extraction is obtained and the phases usually separate completely within 10 minutes. The volume of the organic phase after extraction is 8.5 ml, that of the aqueous phase having increased by 1-5 ml.Successive extractions of the aqueous phase with more of the same extractant result in an increasing isobutanol concentration in the aqueous phase, thus giving increased blanks and entrainment of aqueous copper solution. By performing second and subsequent extractions with 8-5 ml of (5 + 80) isobutanol - cyclohexane mixture, the original extraction conditions are maintained. Under these conditions no further recovery can be detected in a second extraction. A = Ll r-l Y Ll Cyclohexane, per cent. I I I 70 80 90 0 I Fig. 1 . Effect of addition of cyclo- Fig. 2. Effect of pI-1 on the determina- hexane to the isobutanol extractant: curve A, blank; curve B, sample corrected for blank tion of 4.74 p.p.m. of oleic acid EFFECT OF pH- From Fig. 2 it can be seen that a large increase in extraction occurs as the pH is raised above a value of 9, and that it reaches a maximum between pH 11 and pH 12.6.When the pH is raised above 13 there is a noticeable lightening in the blue colour of the aqueous phase, and the extraction falls to a low value. With the recommended reagent, the aqueous phase after extraction has a pH of 11.6.254 GREGORY: DETERMINATION OF RESIDUAL ANIONIC [Analyst, Vol. 91 DEVELOPMENT OF COPPER(II) DITHIOCARBAMATE COLOUR- JanssenlO has determined the stability constant for the copper(I1) diethyldithiocarbamate system and gives log K = 28.8. This is sufficiently high for the coloured complex to be expected to form readily in the presence of triethylenetetramine. Whereas, in practice, the addition of a considerable excess of triethylenetetramine has no effect on colour development, the reaction is slow under the conditions used.This can be seen from Fig. 3, which also shows that the dithiocarbamate complex is not stable to light. After addition of the reagent the extract should be allowed to stand in the dark for 15 minutes. Fig. 3. Development of copper(I1) dithiocarbamate colour : 0, protected from light; X, exposed to light I I 1 500 I000 I500 Mole r a t i o of copper to f a t t y acid Fig. 4. Effect of rcagent concentration on the extractior REAGENT CONCENTRATION- The relationship between the optical density of the extract and the mole ratio of copper(I1) triethylenetetramine complex to oleic acid is shown in Fig. 4. The very large excess of copper reagent required for the reaction to approach completion suggests that the adduct formed between the fatty acid anion and the complexed copper cation must be highly dissociated. For a high recovery, the mole ratio of copper to fatty acid must be at least of the order of 1000.COMPOSITION OF THE EXTRACTED SPECIES- Chaberek and hlartellll state that triethylenetetramine forms with divalent copper a square planar tetraco-ordinated (1 + 1) ion, and Schwar~enbach~ has determined the stability constant log K = 20.6. It is very likely that it is this stable chelate which reacts with fatty acid anions to give a neutral adduct which can be extracted into organic solvents. Because of the high dissociation of the adduct, the Job method of continuous variation12 and the mole ratio method13 are not suitable for determining its composition.The Harvey and Manning slope-ratio method14 is usually effective in such cases, but cannot be applied here as very stable emulsions are formed when the necessary very large excess of surface-active reactant is present. However, on making the assumption that extraction is sensibly complete, a comparison of the measured molar extinction coefficients for copper(I1) diethyldithiocarbamate in the equilibrated extractant, and for oleic acid taken through the full procedure provides an approximate figure for the combining ratio of oleic acid with the copper complex. The measured molar extinction coefficient for copper( 11) diethyldithiocarbarnate, prepared from copper(I1) oleate in (5 + 80) isobutanol - cyclohexane mixture, is 10,700. For oleic acid taken through the procedure the value found is 5280, after correction for the reduction in volume of the extractant.The ratio between these figures is 2.03, which is close to the expected ratio of 2 molecules of oleic acid for each molecule of copper, supporting the view that the extracted compound is probably a neutral ion-association system.April, 19661 SURFACE-ACTIVE REAGENTS IN MINERAL FLOTATION LIQUORS 255 APPLICATION TO OTHER SURFACE-ACTIVE ANIONS- A calibration for oleic acid is given in Fig. 5 and shows a close adherence to the Beer - Lambert law. Similar calibrations are obtained with other fatty acids from myristic (C,,H,,COOH) to behenic (C,,H,,COOH) . Lauric acid (C,,H,,COOH) gives low recoveries and no extraction occurs with capric acid (C,H,,COOH).Very low and erratic recoveries are obtained with the almost insoluble montanic acid (C,,H,,COOH). The method is also effective for non-soapy anionic surface-active agents of the alkyl sulphate, alkyl phosphate and alkyl - aryl sulphonate types. Fig. 5. Calibration for oleic acid When plotted on a molar basis, the calibrations for various anionic surface-active agents all lie fairly close to one another. Comparative optical densities for a number of these materials are given in Table I. The optical density figures quoted are those given by a molar solution of the surface-active agent, under the conditions of the determination. TABLE I RELATIVE MOLAR SENSITIVITY OF THE METHOD FOR A NUMBER OF ANIONIC SURFACE-ACTIVE AGENTS Surface-active agent Lauric acid .. . . . . . . . . . . .. *Palmitic acid .. . . . . . . . . .. *Stearic acid . . . . . . . . . . . . . . *Oleic acid . . . . . . . . . . . . . . Behenic acid . . . . . . . . .. . . *Myristic acid. . . . . . . . . . . . .. *Sodium lauryl sulphate . . . . . . . . . . Sodium cetpl sulphate . . . . . . . . . . Sodium dodecylbenzene sulphonate . . . . . . Sodium di-2-ethylhexyl sulphosuccinate . . . . . . Di-n-nonyl phosphoric acid . . . . . . . . Optical density for a 10-5 molar solution 0.016 0.305 0.316 0.314 0.3 12 0.287 0.333 0-278 0.266 0.328 0.3 34 * Purity of materials checked by analysis, other substances are of reagent chemical grade. REPRODUCIBILITY OF THE METHOD- The mean was 4.76 p.p.m. and the standard deviation 0.05.INTERFEHENCES- Flotation systems with soaps are frequently adversely affected by the presence of dissolved cations that precipitate insoluble metal soaps. The soap remaining in solution is determined in this procedure after spinning the sample liquor in a centrifuge to remove allinsoluble material. If it is required to determine the total amount of soap present, both in soluble and insoluble form, the effect of some of the precipitating cations can be overcome by the addition Sixteen determinations were made of the oleic acid concentration in a 4.74 p.p.m. solution.256 GREGORY: DETERMINATION OF RESIDUAL ANIONIC [Ana&St, VOl. 91 of EDTA. This liberates carboxylate ions from many soaps, while forming a copper(r1) complex of lower stability 15(10g K = 18-86) than that of triethylenetetramine.However, a new calibration must be made with EDTA added as the extraction is slightly reduced in its presence. The addition of 1 ml of a 6-25 per cent. solution of disodium ethylenediamine tetra- acetate dihydrate before the copper - triethylenetetramine reagent, is sufficient to complex the dissolved cations present in most hard waters in Great Britain. The effect on the determination of oleic acid, of the presence of a number of common cations, and also of some anions which might be present in flotation liquors, is shown in Table 11. TABLE I1 EFFECT OF VARIOUS IONS ON THE DETERMINATION OF 4-74 PARTS PER MILLION OF OLEIC ACID Ion . . K+ . . NH,+ . . Mg2+ . . Ca2+ . . .4P+ . . Mn2+ , . Fe3+ . . Co2+ . . Ni2+ .. Zn2+ . . F- . . c1- . . SO,2- . . PO,,- . . CO,Z- . . B,0,2-. . CN- . . CH,COO- CrO,2- Silicate . . . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . Concen- t ra tion present, p.p.m. 240 240 48 120 240 240 . . 120 240 240 . . 120 240 120 240 120 240 240 240 240 240 240 240 100 240 240 240 Added as : KNO, NH,NO, Ni(NO,), ZnSO, NaF NaCl Na,SO, Na,H PO, Na,C03 KCN CH,COONa K,CrO, Water glass N a,B@ 7 No EDTA present 7.- Oleic acid found, Error, p.p.m. per cent. 4.60 - 2.9 3.84 - 19.0 4.60 - 2.9 3.35 - 29.3 2-02 - 57.4 2.58 - 45.5 3.38 - 28.7 0*59* - 87.5 3*02* - 36.3 3.63* - 23.4 2.67* - 43.8 4-05 - 14.5 3.35 - 29.3 3.98 - 16.0 3.7 1 - 21.7 4.86 -+ 2.5 4.74 0 4.54 - 4.3 4.49 - 5.3 4.41 - 7.0 4-60 - 2.9 4-60 - 2.9 4-60 - 2.9 4.57 - 3.5 4.6 1 - 2.7 EDTA added , Oleic acid found, p.p.m.- - 4.58 4.83 2.92 3.87 3*72* 3.84* 3.97* 4-58 4-78 4.63 4.95 4-76 - - - - - - - - - - 5 Error, per cent. - - - 3.3 + 2.0 - 38.4 - 18.4 - 21.6 - 18.9 - 16.2 - 3.3 + 0.9 - 2.2 + 4.4 + 0-4 - - - - - - - - - - * Some precipitation of metal compounds occurred in these cases, Non-soapy surface-active agents, which do not generally form insoluble metal salts, are less affected by cationic interference. Thus for a 5 p.p.m. solution of sodium lauryl sulphate in the presence of 240 p.p.m. of calcium as calcium chloride, the concentration found was 4.86 p.p.m., an error of -2.8 per cent., whereas the error caused for oleic acid is over 50 per cent., although this difference was not apparent in the presence of EDTA. Traces of long chain alkyl amines do not interfere seriously with the determination of fatty acids, but the surface-active quaternary amines completcly inactivate an equivalent amount of soap.APPLICATION TO FLOTATION SYSTEMS- In practice flotation liquors may contain considerable amounts of residual solids and slimes. Although these generally remain in the aqueous phase after the extraction, the adsorbed surface-active agent tends to be stripped by the reagent and is determined. Most filter materials adsorb anionic surface-active agents to a considerable extent, even from pure solutions. Nylon filter- cloth is free from this defect but has a relatively high porosity. Particulate matter can readily be removed, however, by centrifuging the sample before analysis. Much of the preceding part of this paper has dealt with ideal systems.April, 19661 SURFACE-ACTIVE REAGENTS IN MINERAL FLOTATION LIQUORS 257 A number of anionic surface-active agents have been successfully recovered from laboratory test flotations.Continuous analysis records of anionic surface-active agents are being obtained by the use of the Technicon AutoAnalyzer. The application of the method to this system, and the results obtained in flotation experiments will be the subject of a future publication. CONCLUSION The determination of anionic surface-active agents, including the fatty-acid soaps, in mineral flotation liquors can be done by extraction of the adduct with the copper - triethylene- tetramine complex, followed by photometric determination of the coloured copper - diethyl- dithiocarbamate complex in the organic phase. The method should also find application in fields other than in the mineral flotation industry. 1. 2. 3. 4. 5. 6. 7 . 8. 9. 10. 11. 12. 13. 14. 15. REFERENCES Jones, J. H., J . Ass. Off. Agric. Chem., 1945, 28, 398. Longwell, J., and Maniece, W. D., Analyst, 1955, 80, 167. Rosen, J . M., and Goldsmith, H. A., “Systematic Analysis of Surface-active Agents,” Interscience - _ _ _ , op. cit., p. 54. Tomiinson, H. S., and Sebba, F., Analytica Chim. Acta, 1962, 27, 596. Ayers, C. W., Ibid., 1956, 15, 7 7 . Iwayama, Y., J . Pharm. SOG. Japan, 1959, 79, 552; Analyt. Abstr., 1960, 7, 3519. Flannery, J., Ke, B., Grieb, M. W., and Trivich, D., J . Amer. Chem. SOL, 1955, 77, 2996. Schwarzenbach, G., Helv. Chim. Acta, 1950, 33, 974. Janssen, M. J . , R e d . Trav. Chim. Pays-Bas Belg., 1956, 75, 1411. Chaberek, S., and Martell, A. E., “Organic Sequestering Agents,” John Wiley & Sons, Inc., New Job, P., Annls Chim., 1928, 9, 133. Kolthoff, I . M., and Elving, P. J ., “Treatise on Analytical Chemistry,” Interscience Publishers Harvey, A. E., jun., and Manning, D. L., J . Amer. Chem. Soc., 1951, 70, 4488. Schwarzenbach, G., and Freitag, E., Helv. Chim. Acta, 1951, 34, 1503. Publishers Inc., New York, p. 16. York, 1959, p. 149. Inc., New York, Part 1, Volume 5, p. 2981. Received June 18t?z, 1965