470 Analyst, April, 1983, Vol. 108, +PI 470-475 Fully Automatic Flow Injection System for the Determination of Uranium at Trace Levels in Ore Leachates Thomas P. Lynch, Arthur F. Taylor and John N. Wilson Chemical Analysis Branch, BP Research Centre, Chertsey Road, Sunbury-on-Thames, Middlesex, T W 16 7LN An automatic flow injection system is described for the determination of uranium in ore leachates. Following injection from an autosampler, the leachate is extracted with a solution of tributyl phosphate in heptane, which removes uranium, and the organic phase is separated. The extract is reacted with an ethanolic solution of 2- (5-bromo-2-pyridylazo) -5-diethylaminophenol (BrPADAP) and benzyldimethyltetradecylammonium chloride (zephiramine) and the resulting ternary complex with U(V1) is measured spectrophoto- metrically a t 579 nm.The lower limit of determination is 0.1 p.p.m. of uranium and up to 50 samples per hour can be analysed. In terms of speed and sensitivity this improves significantly on published procedures using segmented flow systems. The technique is ideal for process control and can be applied to the analysis of ores following mineralisation. Keywords ; Flow injection ; ore leachates ; spectrophotonzetry ; uranium determination The exploitation of low-grade ores involves processes in which it is required to control uranium a t trace levels in residues, leachates and effluents, thus necessitating the use of sensitive analytical procedures that ideally are also rapid and selective. A number of relevant methods have been published, predominantly colorimetric, the usual approach being to use a non- specific chromogen together with a process for the suppression or removal of interference~.~-~~ Segmented flow systems have been used to automate the steps of colour development and measurement.13J4 Uranium is first separated from the leachate and concentrated by a manually performed solvent extraction, which is relatively time consuming.We were faced with a situation in which it would be necessary to analyse large numbers of acid leachates containing from less than 1 and up to 100 p.p.m. of U,O, in order to control ore processing, initially on a pilot plant and later on a production scale. We therefore proposed to in- corporate the solvent extraction process into a fully automatic system and at the same time to increase sensitivity so that leachates containing 0.5pgml-1 of U30, or less could be determined with a precision and speed suitable for process control.Experimental Choice of Reagent Dibenzoylmethane has been used in several colorimetric procedures for the determination of uranium following a solvent extraction process.lp2 Although we successfully adapted this procedure the sensitivity of the reagent (molar absorptivity 1.8 x lo4 1 mol-l cm-l) was barely adequate for our needs. The uranium benzoate - Malachite Green complex has been reported to have a molar absorptivity of 8.3 x 104 1 mol-l cm-l.l0 We were able to produce intensely coloured complexes using this reagent, but the colour was unstable, and a double solvent extraction was necessary to avoid interferences.Other w ~ r k e r s ~ - ~ s ~ ~ have shown the pyridyl- azo dyestuff 2-(5-bromo-2-pyridylazo)-5-diethylaminophenol (BrPADAP) to be a highly sensitive and effective reagent for the determination of uranium. In using this reagent we were able to make two modifications to published procedures that lowered the minimum level of determination. We combined an ethanol solution of the reagent with benzyldimethyl- tetradecylammonium chloride (zephiramine ; Fluka) , which together form a ternary complex with uranium. We observed an increase in molar absorptivity from 5.8 x lo4 to 6.8 x lo4 1 mol-l cm-l and a small shift in the absorption peak from 575 to 579 nm, at which wavelength interference from the background colour of the reagent is somewhat less significant (Fig.1). Following extraction of uranium with tributyl phosphate in heptane we formed theLYNCH, TAYLOR AND WILSON 471 al K m e $ 0.4 n Q: 0.2 ' 0 500 540 580 620 Wavelengthlnm Fig. 1. Absorption spectra of 1, re- agent blank Venus ethanol; 2, uranium - BrPADAP - zephiramine complex zw- sus reagent blank, Amax. = 579 nm; and 3, uranium - BrPADAP complex veYsus reagent blank, Amax. = 575 nm. coloured complex under non-aqueous conditions, in the presence of pyridine. This avoids the need for a dilution with relatively large volumes of ethanol and consequent reduction in sensitivity, in order to maintain a single phase system when aqueous buffers are used. As a result of these modifications an over-all increase in sensitivity of between five and ten times was achieved.The segmented flow systems referred to13914 both use BrPADAP, but in view of our intention to include a solvent extraction step the diffusion dependent flow injection analysis (FIA) procedure developed by RfiZiEka and Hanse1-1~~9~7 appeared to us to be more appropriate. This simple and flexible technique has recently been reviewed18919 and Karlberg and co-workers have described FIA systems incorporating solvent extraction steps for the analysis of certain pharmaceuticals.20921 The extraction of uranium and formation of the ternary complex appeared ideal processes to incorporate in a continuously flowing system. Reagents Dissolve 700 g of A1(N0,),.9H20, 9 g of NaF and 2.6 g of Na2S0, in about 900 ml of de-ionised water.Add 18 ml of glacial acetic acid and dilute to 1 1 with de-ionised water. Dissolve 0.1 g of BrPADAP and 10 g of zephiramine in about 500 ml of absolute ethanol and add 50 ml of pyridine. Mix 100 ml of tributyl phosphate with 900 ml of heptane. Aluminium nitrate salting solution. BrPADAP. Organic extractant, 10% V/V tributylphosphate in heptane. Filter all reagents through a coarse filter-paper such as a Whatman No. 541. Dilute to 1 1 with absolute ethanol. Apparatus A Technicon Autosampler is interfaced with the lOO-pl capacity sample injection loop via a Rheodyne system as follows: the Auto- sampler probe operates a microswitch, which sequentially activates a pair of Rheodyne 7163 solenoid valves. These operate a 5001 pneumatic activator, which charges or discharges a 5020 rotary sample valve.The system flow is controlled by two Watson Marlow peristaltic pumps, one fixed speed (JIHRK 18) and one variable speed (501s). Colour is measured by a Unicam SP6 spectropliotometer with a Hellma 1-cm tubular flow cell, Type OS178.12, the A flow diagram of the system is shown in Fig. 2.472 Organic extractant Alum in ium nitrate reagent Water LYNCH et al. : AUTOMATIC FLOW INJECTION 2.0 L r 1.2 L r 2.6-m coil - 1.2 L 0.9-m coil 1100-pl sample r Aqueous Analyst, Vol. 108 Flow-rate/ BrPADAP ml min-' reagent 1.4 I 4.5-m coil Fig. 2. Flow diagram of the system. signal from which is connected to a single pen chart recorder. In the reagent pump manifold Tygon tubing is used for aqueous solutions and Solvaflex tubing for organic solutions.Trans- mission and mixing tubes are of Altex PTFE tubing of 0.8 mm i.d. The phase separator (Fig. 3) is a modification of a design suggested by Betteridge,22 the ports being threaded to take Altex fittings (Anachem Ltd., Luton). Correct operation of phase separators (or de-bubblers in air segmented streams) is always important for good base-line stability, but there is one factor in our system that makes it critical. When the alkaline BrPADAP reagent is added to the organic extractant any entrained aqueous phase becomes partially miscible and this causes precipitation of aluminium hydroxide (from the aluminium nitrate) in the tubing and spectrophotometer cell. The precipitate is also stained red by absorbed BrPADAP. To allow better control of the take-off to waste ratio a separate, variable speed pump is used for the waste line, rather than relying on the conventional approach of selection of the appropriate pump tube size, with a single pump.0 0 1 0 0 A 1 nil 0 0 B Bottom C 3 Fig. 3. Diagram of the phase separator showing A, Perspex block; B, Neoprene gasket; and C, PTFE block. The Neoprene gasket is sandwiched between the PTFE block and the Perspex block and the assembly is bolted together with eight screws. The mixed aqueous-organic phase enters a t port 1 and the aqueous phase plus a trace of the organic phase are pumped t o waste from port 2. The organic phase is removed a t port 3, after which it is mixed with reagent.April, 1983 FOR DETERMINATION O F u I N ORE LEACHATES Results and Discussion 473 Calibration graphs were prepared by processing solutions of uranyl nitrate either in water or in an aqueous solution containing 25 g 1-1 of sulphuric acid.They were linear up to 50 p.p.m. of U,O,, but such was the sensitivity of the method that we normally operated in the range 0-20 p.p.m. of U,O,, samples with higher concentrations being diluted before injection. A typical calibration graph is shown in Fig. 4. a) (D 0.1 2 s n Q ' 6 min ',5 a) I Timehin (u ; 0.1 5 5: a I I I 0 4 8 12 16 Amount of U308, p.p.m. Fig. 4. Absorption spectra for (a) typical calibration peaks where the values on the peaks represent p.p.m. of U,O,; and (b) corresponding calibration graph. Chart speed, 1 min cm-l. Table I shows results for the determination of uranium in a series of leach liquors containing nominally 25 g 1-1 of sulphate.The first set of results was obtained using aqueous calibration standards. The second and third sets show results for the same samples by standard additions and against calibration standards containing appropriate levels of sulphate. Standard TABLE I DETERMINATION OF URANIUM IN ACID LEACHATES Each result is the mean of triplicate determina- tions, with no difference between any two indi- vidual results within a set exceeding 0.1 p.p.m. U,O, content, p.p.m. - 1 1.6 1.7 1.8 2 6.8 7.4 7.6 3 8.4 9.2 9.6 4 13.3 15.3 15.3 5 9.2 10.5 10.6 6 9.1 10.3 10.4 7§ Sample No. a* b t c: 8.6 10.1 - * Against sulphate-free standards. t By standard additions using microlitre additions of strong uranium solutions. $ Against standards containing 25 g 1-l of sulphuric acid, the nominal concentration of the samples. 5 A 10 p.p.m.standard in 25 g 1-' of sulphuric acid.474 LYNCH et al. : AUTOMATIC FLOW INJECTION Analyst, Vol. 108 additions were made in microlitre amounts using strong solutions of uranium so that sample volumes were effectively unchanged. Whilst the second and third sets of results are in very good agreement, the relatively low results of the first set illustrate the suppression of uranium extraction by sulphate ion reported by Walker and Vita.23 This is confirmed by sample 7, which is a 10 p.p.m. uranium standard with 25 g 1-1 of sulphate present. In Table I1 the suppression is quantified by comparing sets of 5 and 10 p.p.m. standards containing from TABLE I1 SUPPRESSION OF THE EXTRACTION OF URANIUM IN THE PRESENCE OF SULPHATE ION Concentrations indicated compared with sulphate-free standards SOd2-/g 1-1 5 p.p.m.U,O, 0 5.0 6 4.8 10 4.4 16 4.4 20 4.3 26 4.4 30 4.3 10 p.p.m. U,O, 10.0 9.7 8.9 8.9 8.8 8.8 8.8 0 to 30 g 1-1 of sulphate with a sulphate-free calibration. The suppression of 12-14% is constant within experimental error for from 10 to 30 g 1-1 of sulphate, although it is increased if the sample is injected directly into the aluminium nitrate reagent rather than into a separate aqueous stream (Fig. 2). Table I11 shows short-term repeatabilities at nominal levels of 1, 10 and 20 p.p.m. of U,O, using model leachates containing 25 g 1-1 of sulphate. Neutralisation of free acid does not modify the suppression. TABLE I11 SHORT-TERM REPEATABILITY U,O, (nominal Repeatability (95%), * level), p.p.m.n %-I p.p.m. U,O, 1 11 0.016 0.05 10 11 0.069 0.22 20 11 0.176 0.66 * For repeatability values, duplicate results by the same operator should be considered suspect if they differ by more than the amount stated in more than one in twenty determinations. It is particularly important in the processing of low-grade uranium ores to be able to monitor accurately trace levels of uranium in leachates, effluents and residues. Precision, speed and specificity are desirable characteristics of the analytical control method. One of the outstand- ing advantages of flow analysis is that system constants and timing sequences are maintained within far closer limits than is possible with manual processes; and, as in the present procedure, this results in excellent repeatability (Table 111).Good recoveries are substantially confirmed by the agreement with standard additions measurements (Table I). The sensitivity of the system is adequate for the determination of 0.5 p.p.m. of U,O, and the determination limit may be lowered if necessary by increasing the optical path length and by optimising the dis- persion, which is dependent on sample volume, flow-rates and lengths of the various elements of the flow line. The simplex approach of Nelder and Mead24 is a means of achieving this optimisation and Wade25 has recently described the application of a modified version to FIA systems. A novel feature of our system is the use of a second variable speed pump to control the off-take of the aqueous phase from the separator, and although this may seem extravagant it has contributed significantly to the reliability and long-term stability of the system.April, 1983 FOR DETERMINATION OF u IN ORE LEACHATES 475 Selectivity always presents a problem with naturally occurring materials of complex and variable composition. We found a complexing solution based on that proposed by Francoisl was ideal for the materials we were analysing.Of the well known interferences, Ce(IV), which is extracted, was reduced to Ce(III), which is not extracted, by sulphite; titanium and zir- conium were complexed by fluoride; and thorium was complexed by acetic acid. Aluminium nitrate prevents interference from common cations such as Fe(II1) and maintains a distri- bution coefficient, which, subject to the effect of sulphate described, ensures virtually complete extraction of uranyl ion into the organic phase.There is no reason why the composition of this solution should not be modified to process leachates from ores with different composition and interfering species. The anion may be determined easily in unknown solutions by ion chromatography, but in practice the sulphate concentration in a given series of leachates will usually be known and of relatively constant level. As has been shown, calibration standards may be formulated accordingly. Finally, considering speed of analysis, incorporation of the extraction process into the automatic system is a substantial advantage, as it enables leachates from the production pro- cess to be analysed without any pre-treatment. We were able to analyse up to 50 such samples in 1 hour.The sulphate effect does not present a great problem. 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. References Francois, C. A., Anal. Chem., 1958, 30, 50. Hennisch, G. W., Mikrochim. Acta, 1970, 258. Perez-Bustamente, J. A., and Palomares Delgardo, F., Analyst, 1971, 96, 407. Johnson, D. A., and Florence, T. M., Anal. Chim. Acta, 1971, 53, 73. Pakalns, P., Anal. Chim. Acta, 1974, 69, 211. Johnson, D. A., and Florence, T. M., Talanta, 1975, 22, 253. “Method of Uranium Determination in Pregnant Solution,” Report No. 1841 of the South African National Institute for Metallurgy, Johannesburg, 1976. Fujinaga, T., Kuwamoto, T., and Ozaki, T., Nippon Kagaku Kaishi, 1976, 12, 1852; Chem. Abstr., 1977, 86, 164764. Pollock, E. M., Anal. Chim. Acta, 1977,88, 399. Dubey, S. C., and Nadkarni, M. N., Talanta, 1977, 24, 266. Lyle, S. J., and Tamizi, M., Anal. Chim. Acta, 1979, 108, 267. Strelow, F. W. E., and Van Der Walt, T. N., Talanta, 1979, 26, 537. Prall, J . R., “The Semi-Automatic Determination of Traces of Uranium,” US Government Report Lyle, S. J., and Tamizi, M., Anal. Chim. Acta, 1980, 121, 341. Hung, S.-C., Qu, C.-L., and Wu, S.-S., Talanta, 1982, 29, 629. RbZiCka, J., and Hansen, E. H., Anal. Chim. Acta, 1975, 78, 145. RbtiCka, J., and Hansen, E. H., Anal. Chim. Acta, 1979, 99, 37. RbiiCka, J., and Hansen, E. H., Anal. Chim., Acta, 1980, 114, 19. Betteridge, D., Anal. Chem., 1978, 50, 832A. Karlberg, B., and Thelander, S., Anal. Chim. Acta, 1978, 98, 1. Karlberg, B., Johansson, P.-A., and Thelander, S., Anal. Chim. Ada, 1979, 104, 21. Betteridge, D., personal communication. Walker, C. R., and Vita, 0. A,, Anal. Chim. Acta, 1973, 67, 119. Nelder, J. A., and Mead, R., Comput. J., 1965, 7, 308. Wade, A. P., Anal. Proc., 1983, 20, 108. NLCO 1091, 1972. Received November 8th, 1982 Accepted December 2nd, 1982