y = 0.1469x 00 (r = 0.9999) using all the above values.

2. y = 0.1433x - 0.88 (r = 1.0000) using values up to and including 10,ug Cr/L ( 0.42 jug/L in a sample). This range is more relevant to serum chromium determinations and was used for the regression line values in table 5.14 above. The standard error of the mean was 0.04 jug Cr/L, which is equivalent to 0.0013 jug/L in a sample giving 95% confidence limits of plus or minus 0.0026 pg Cr/L.

5.6. CONCLUSIONS.

The electrothermal atomisation conditions developed are not entirely satisfactory. The precision of the alpha-2-globulin-bound Cr determinations is significantly worse for pool I and highly significantly worse for pool II compared with that of unprocessed chromium solutions of similar concentrations. The precision of the total protein-bound chromium determination is not significantly different from that of unprocessed material. The AAS/ETA relative response discussed in the last chapter was also found to be less satisfactory for the alpha-2-globulin-bound Cr series (82% compared with 99.5% for the total protein-bound Cr stream). The poor precision and response tend to confirm that the extra acidity of the alpha-2-globulin-bound Cr precipitates is reducing the efficiency of the wash procedure in removing the excess hfacac. However the advantages of a similar wash treatment for the standards and both sample streams are obvious. The three injections for each sample reduce the contribution to the overall precision by the AAS/ETA stage to a level sufficient to produce useful analytical data. However the value of the determinations has certain limitations which will be discussed in chapter VIII.

Traces of hfacac derivatives cause a reduction in both the size of and precision of the chromium signal. The addition of ammonium acetate minimizes the former but does not appear to help with the. latter, and is only useful following a process which leaves only minimal traces of the fluorinated diketone. The disadvantages of adding ammonium acetate are the obligatory use of uncoated graphite tubes, and the restriction of the tube life to about 100 firings. The reduced tube life is not a major drawback in the light of the increasing memory effects reported earlier. The initial successful tests with ammonium acetate had been made with a pyrolytically coated tube but it was a much used one, and presumably much of its coating in the critical central zone had been lost.

The failure of coated tubes to cope with ammonium acetate solutions may be due to the less penetrable surface compared with standard electrographite tubes. The failure of the injected material to penetrate the surface may lead to more extensive redistribution at the necessarily high temperature drying stage. However this does not appear to occur on observation with the naked eye using a mirror. The use of

pyrolytically coated tubes with ammonium acetate solution may be possible with a suitably programmed drying ramp.

In general pyrolytically coated graphite tubes were found to be superior in terms of sensitivity and memory effects, although of course, they were not suitable for the methods developed in this study. The only significant carryover demonstrated, was with uncoated tubes after 100 firings, rising to highly significant after 200 cycles. Uncoated tubes do however have some advantages. Sensitivity decreases with use for both types of tube but the fall is steeper for coated ones. The standard electrographite tubes show minimal and steady changes in sensitivity over their useful life.

Manning et al (14) reported that pyrolytically coated tubes showed higher sensitivity than ordinary tubes for chromium in 1976, and this has been confirmed by many workers. Veillon et al (1) reported that pyrolytically coated, tubes retained less chromium than uncoated ones. However, Slavin (15) criticised Veillons work on retained chromium. The temperature gradient along a furnace tube exceeds 1000° C and Slavin claimed that much of the radio chromium measured by Veillon had participated in the measurement and then condensed on the cooler edges of the tube. Veillon in a later publication (16) confirmed that the 51 Cr had been concentrated near the tube ends. The work by Veillon on the retention of chromium by graphite tubes had involved urine samples, which have a high chloride content, and the apparent lack of retained radio chromium in the central zone suggests that "analyte occlusion" is not a factor in halide interference. The evidence against "volatile chromium" has been .discussed earlier and the formation of a non-atomic chromium at the atomisation stage, in the presence of the halide, appears to be the most probable mechanism for interference by chloride and fluoride.

References Chapter V.

1. Veillon,C., et al, Anal. Chem., 52 (1980) p.457. 2. L'vov,B.V., Spectrochim. Acta, 33B (1978) p.153. 3. Segar,D.A., et al, Anal. Chim. Acta, 58 (1972) p.7. 4. Aggett,J., et al, Anal. Chim. Acta, 72 (1974) p.49. 5. Fuller,C.W., Anal. Chim. Acta, 81 (1976) p.199. 6. Persson,J.A., et al, Anal. Chim. Acta, 92 (1977) p.85. 7. Freeh,W.,. et al, Anal. Chim. Acta, 82 (1976) p.83. 8. Czobik,P., et al, Anal. Chem., 50 (1978) p.2.

9. Cruz,R.B., et al, Anal. Chim Acta, 72 (1974) p.231. 10. Churella,J., et al, Anal. Chem., 50 (1978) p.309. 11. Krasowski,J.A., et al, Anal. Chem., 51 (1979) p.1843. 12. Goodfellow,G., et al, Anal. Chim. Acta, 126 (1981) p.147. 13. Matsusaki,K., et al, Anal. Chim. Acta, 124 (1981) p.163. 14. Manning,D.C., et al, At. Absorpt. Newsl., 15 (1976) p.42. 15. Slavin,W., Atomic Spectrosc., 2 (1981) p.8.

16. Veillon,C., et al, Anal. Chim. Acta, 164 (1984) p.67.

CHAPTER VI.

THE DEVELOPMENT OF A METHOD FOR THE DETERMINATION