OCTOBER, I968 THE ANALYST Vol. 93, No. I I I I Micro-analysis of Silicate Rocks Part V. Spectrophotometric Determination of Alumina* BY KOBERT A. CHALMERS AND MOHAMMED ABDUL BASIT (Chemistry Depavtment, University of ,4 berdeen, Old Abevdeen, Scotland) Alumina is determined spectrophotometrically as the 8-hyclroxy- quinolinate after ligand exchange in the organic phase with aluminium acetylacetonate extracted into benzene from aqueous medium a t pH 6 to 7. Interfering elements such as iron, titanium, vanadium and zirconium are removed by extraction into o-dichlorobenzene from M hydrochloric acid, but beryllium is not removed by this procedure. THE direct determination of alumina in silicate rocks has long been a problem in rock analysis, and the several methods proposed tend to be rather lengthy.The British Ceramic Research Association methodl separates aluminium as the hydroxo complex, and deals with any residual iron by reducing it and forming its bipyridyl or phenanthroline complex before precipitating aluminium 8-hydroxyquinolinate. Milner and Woodhead2 make a chloroform extraction of the cupferron complexes of iron, titanium, etc., and determine aluminium by addition of excess of EDTA, boiling to complete complex formation, and back-titration of the excess with iron(II1); salicylic acid is used as indicator. Miller and Chalmer~,~ working on the micro scale, preferred o-dichlorobenzene as solvent for the cupferron extraction, and separated aluminium (and beryllium) by extraction of the acetylacetonate into ether, following this by back-extraction into 6 M hydrochloric acid and precipitation as aluminium 8-hydroxy- quinolinate at pH 5 to separate it from any beryllium.Saj6,4 who has proposed an EDTA titration of iron and aluminium in the same solution, says nothing about the behaviour of titanium. PEibil and Veself6 have proposed methods for the determination of iron, aluminium and titanium together, but do not deal with the problem of zirconium. Kiss6 proposed a DCTA titration of aluminicm after extraction of iron and titanium, but a correction was required for co-titrated eleinerits such as manganese. Voin~vitch,~ after making a critical examination of six methods, concluded that the most accurate results were obtained by removal of silica, reduction of iron(III), precipitation of aluminium and titanium as the basic benzoate, and correction for the titanium present.He considered that the colorimetric method with 8-hydroxy-7-iodoquinoline-5-sulphonic acid needed more study of interferences, and gave a rather larger error. None of the colorimetric methods so far proposed is particu- larly reliable, except for the 8-hydroxyquinoline method based on the work of Gentry and Sherrington,* and that developed by Riley and Williams,9 but a preliminary separation of iron would still be required. For routine use in the “rapid” methods of rock analysis, a colorimetric method would be desirable. When RtiZiEka and StarflO published their paper on ligand exchange in the organic phase, it occurred to us that the acetylacetonate extraction method3 could be con- verted into a colorimetric one if the aluminium acetylacetonate could be transformed into the 8-hydroxyquinolinate in the organic phase.The acetylacetonate itself is not suitable for colorimetric work because the absorption peaks of the reagent and complex are close together and occur in the ultraviolet region ; we therefore examined this ligand-exchange system. * Paper presented a t the Second SAC Conference 1968, Nottingham. For details of Part IV of this series, see reference list p. 632. 0 SAC and the authors. 629630 CHALRIEI<S AND EASIT: MICRO-ANALYSIS 01; SILICATE ROCK [A?Zat!$St, VOl. ‘33 EXPERIMENTAL CHOICE OF REACTION CONDITIONS- In the original gravimetric method3 diethyl ether had been used as solvent for tlie acetylacetonate extraction, but as this solvent is rather too volatile for use in colorimetric methods, we therefore sought other suitable extracting solvents.A standard 0.2 ;II aluminium solution was prepared from high-purity aluminium metal and standardised gravi- metrically with 8-hydro~yquinoline.~~ Aliquots were adjusted to about pH 7 after addition of acetylacetone, and extracted3 with various solvents. The extracts were stripped with 6 M hydrochloric acid and the aluminium determined.3 Carbon tetrachloride, chloroform, benzene, toluene and pentyl acetate all gave 99.9 to 100.1 per cent. recovery of 2.68mg of aluminium in replicate experiments. Of these, carbon tetrachloride and chloroform are inconvenient if the extractions are made in stoppered tubes and the upper layer is withdrawn with the aid of Witt’s apparatus, and were not considered further.The number of extractions and the duration of shaking needed were found by determining the amount of aluminium extracted in a single extraction (to ensui-e complete removal of the organic phase, the walls of the tube were rinsed with small amounts of solvent, and the washings were added to the extract, but there was no intimate mixing of the wash liquid with the aqueous phase). I t was found that 3 ml of benzene extracted 95.3 to 95.6 per cent. of the aluminium present in 2 ml of aqueous phase, whether the shaking time was 1 or 5 minutes. The amounts of aluminium taken ranged from 2-68 to 13.40 mg. When three extractions were made (3 ml, 3 ml, 1 ml) with 2 minutes of shaking, recovery of 2-68 to 13-40 mg of aluminium was 99.9 to 100-1 per cent.The compatibility of the acetylacetonate extract and the 8-hydroxyquinoline exchange solution was next examined. Toluene seemed a more desirable solvent to use than benzene because of its lower volatility but, when a toluene solution of 8-hydroxyquinoline was mixed with a toluene extract, a turbidity often appeared before the exchange reaction was complete; it was later found that the turbidity depended on the grade of toluene used, and that toluene could be satisfactorily used as solvent. No turbidity appeared when benzene was used as solvent, but the exchange reaction was slow, taking 48 hours to reach completion. The slow rate of exchange could be accounted for if the exchange reaction proceeds via an S N , mechanism, as steric hindrance would prevent access of 8-hydroxyquinoline to the aluminium ion and use of the empty 3d orbitals of the latter for further co-ordination, and a further retarding effect might occur if the aluminium acetylacetone were polynuclear.The reaction will be slowed by the presence of the excess of acetylacetone used in the extraction step. The stability constants of the two complexes are sufficiently different (log fI3 = 22.3 for the acetylacetonate,12 log l/Ksp = 36 for the 8-hydro~yquinolinate~~) for exchange to be complete. The rate of reaction was accelerated by the classic expedient of heating the solution, and equilibrium was reached (constant absorbance) after heating for 15 minutes on a water-bath. The absorbance remained unchanged even after 2 hours’ heating, and on cooling the solution to room temperature remained stable for several days.The necessary volume of 8-hydroxyquinoline solution for complete exchange was found by applying the procedure to a fixed amount of aluminium (160 pg) and acetylacetone and by using various amounts of exchange solution. A plot of absorbance against volume of exchange solution added showed that the absorbance became constant when a d-iold molar excess of 8-hydroxyquinoline had been added. The excess of reagent did not interfere with the spectrophotometric measurement of the aluminium complex at the absorption maximum 395nm if an accurately measured amount was added. A Beer’s law plot was linear over the range 0 to 200 pg of aluminium in 50 ml of solution, and was highly reproducible.The slope (measured under the conditions described in Procedure) was 0.208, ml pg-l. INTERFERENCES- It was known from previous work3 that of the elements commonly found in rocks, oldy beryllium was co-extracted with aluminium as the acetylacetonate, and beryllium occurs usually only in traces. As might have been expected, beryllium was also found to give an exchange reaction with 8-hydroxyquinoline. Attempts to mask either aluminium or beryllium in the aqueous phase, or to strip one of them preferentially from the organic phase, by use of fluoride, hydroxide or EDTA, all failed, and if beryllium is present in significant amounts a correction must be applied for it.October, 19681 PART v. SPECTROPHOTOMETRIC DETERMINATION OF ALUMINA 631 Freedom from other interferences was established by preparing synthetic silica-free “rock-sample” solutions (solution B in the Shapiro and Brannock scheme14) containing the equivalent of 12 per cent.of MgO, 16 per cent. of CaO, 2.8 per cent. each of TiO,, P,O,, Coo, MnO, NiO and Cr,O,, and amounts of aluminiuni ranging from 10 to 23 per cent. The recovery of aluminium was 99.8 to 100.0 per cent. METHOD REAGENTS AND APPARATUS- n benzene, and benzene is used instead of ether for the extraction. As described in Part IV,3 except that the 8-hydroxyquinoline solution is 2 per cent. w/v PROCEDURE- Prepare a Shapiro and Brannockl4 “Solution B” and take an aliquot corresponding to not more than 200 pg of aluminium (400 pg of alumina). This will usually be equivalent to a 1-mg rock sample.If a sufficiently precise microbalance is available (and a sample ground finely enough) a sample can be weighed out and decomposed as described earlier,3 and the solution used direct. Proceed with the cupferron and acetylacetone extractions as described in Part IV, but use benzene instead of diethyl ether. Transfer the benzene extract of the aluminium acetylacetonate into a 50-ml standard flask and add by safety-pipette 10 ml of 8-hydroxyquinoline solution in benzene. Mix, then heat the flasks on a steam-bath for 30 minutes, cool them to room temperature, dilute to volume with benzene, mix, and measure the absorbance at 395 nm in 10-mni cells against a reagent blank. RE s ULTS The method was applied first to synthetic samples as already described, and then to various rock samples and standards.Because of the notorious uncertainty of the results of alumina determinations in silicate rocks,15 916917 recovery was checked by adding known amounts of standard aluminium solution to the aliquots taken for analysis. Results for the samples, spiked and unspiked, are shown in Table I, together with such comparison results TABLE I ALIJMINA I N STANDARD ROCK SAMPLES A1203 Weight used, Found, Sample mg per cent. Olivine basalt . . . . 1.045 12.4, 12.6 Felspar 70* . . . . 1.038 18.1 S1 Syenitet . , . . 1.205 9.4, 9.4 BCRl Basalt: . . . . 1.459 13.1, 13.2 T1 Tonalites . . . . 1.038 16.3, 16.3 G2 Granite: . . . . 1.044 14.9, 14.9 GSPl Granodiorite: . . 1.002 15.5, 15.5 GR Granite11 . . . . 1.006 14.2, 14.2 AGV1 Andesite: .. 1.169 16.0, 16.1 16.0, 16.0 I’CC1 Peridotitc . . 11.22 0-73 DTSl Dunite$ . . . . 14.26 0.21 * U.S. Bureau of Standards. Certificate or other value, per cent. 17.88 18*0* 13.8$§ 1 6 ~ 3 ’ ~ 16.511 15-5$§ 15.4: 14-59 14.757 14*51° 12*63,18 9.019 9 q 17.2’;s. 0*69$ 0.913 0.28: 0.39s 7 Added, Pg 30.9 46.4 30.9 30.9 61.7 46.2 15.4 15.4 15.5 53.6 - - t Canadian Association of Applied Spectroscopy. $ US. Geological Survey. Geology Department, University of Aberdeen. )I Tanganyika Survey. 7 Centre de Recherchcs Petrographiques et Geochimiques, Nancy. AlS+ -Jc--7 Found, tcg 30.9 46.3 31.0 30.9 61.8 46.3 15.4 15.4 15.4 83.6, 53.6 - - as were available. The recovery in spiked samples (amounts of alumina added ranged from 2.5 to 11.0 per cent.) is good, and the results agree reasonably well with those obtained by other methods, excepttfor the Andesite AGVl. Analysis of AGVl by the Ceramic Research Association methodl gave a still lower result than that obtained by the method described632 CHALMERS AND BASIT many 1.2. 3. 4. 5. 6. 7. 8 . 9. 10. 11. 12. 13. 14. 147. 16. 17. 18. 19. here. We offer no explanation of this, although, as we had only about 1 g of sample to work with, the possibility of segregation or sampling error cannot be ruled out. We wish to thank the Pakistan Council for Scientific and Industrial Research for granting study leave to M.A.B., and Dr. Lappin, of the Geology Department in this University, for of the rock samples. REFERENCES Rennett, I-I., Ilawley, W. G., and Eardley, R. P., Trans. Brit.Ceram. Soc., 1958, 57, 1. Milner, G. W. C., and Woodhead, J. L., Analytica Chim. Acta, 1955, 12, 127. Miller, C. C., and Chalmers, R. A., Analyst, 1953, 78, 686. Saj6, I., Acta Chim. Hung., 1955, 6, 251. P?ibil, R., and Veselp, V., Talanta, 1963, 10, 233. Kiss, E., Analytica Chim. Acta, 1967, 39, 223. Voinovitch, I. A., and Debras, J., Chirn. Analyt., 1957, 39, 418. Gentry, C. H. R., and Sherrington, L. G., Analyst, 1946, 71, 432. Riley, J. P., and Williams, H. P., Afihrochim. Acta, 1959, 525. RfiiiCka, J.. and Starf, J., lalanta, 1967, 14, 909. Chalmers, R. A., and Basit, M. A., Analyst, 1967, 92, 680. SillCn, L. G., and Martell, A. E., “Stability Constants of Metal-Ion Complexes,” Special Publication Hollingshead, R. G. W., “Oxine and its Derivatives,” Volume I, Butterworth, London: 1954, p. 59. Shapiro, L., and Brannock, W. W., Bull. U.S. Geol. Surv., 1956, No. 1036-C. Fairbairn, H. W., Schlecht, W. G., Stevens, R. E., Dennen, \?’. H., Ahrens, L. H., and Chaycs, F., Fairbairn, H. W., and Schairer, J. F., Amer. Miner, 1952, 37, 744. Schlecht, W. G., Analyt. Chew., 1951, 23, 1568. Guthrie, W. C. A., and Miller, C. C., Min. Mag. Lond., 1933, 23, 405. Ingamells, C. L., and Suhr, N. H., Geochim. Cosmochim. A d a , 1963, 27, 897. 17, Chemical Society, London, 1964. Ibid., 1951, No. 980. NOTE-Referencc 3 is to Part IV of this series. Received March 29th, 1968