Determination of beryllium and selenium in human urine and of selenium in human serum by graphite-furnace atomic absorption spectrophotometry.

For human urine beryllium (Be), each sample (500 microl) was diluted (1+1) with Nash reagent (containing 0.2% (v/v) acetylacetone and 2.0 M ammonium acetate buffer at pH 6.0) and then a 20-microl volume of Triton X-100 (0.4%, v/v) aqueous solution was added. An aliquot (10 microl) of the diluted urine mixture was introduced into a graphite cuvette and was atomized according to a temperature program. The method detection limit (MDL, 3sigma) for Be was 0.37 microg/l in the undiluted urine sample and the calibration graph was linear up to 65.0 microg/l. Calibration graphs were prepared by the standard addition method. Accuracies of 98.6-102% were obtained when testing standard reference material (SRM 2670) freeze dried human urine samples. Precision (relative standard deviation, RSD) for urine Be was < or = 2.3% (withinrun, n = 5) and was < or = 3.0% (between-run, n = 3). For human urine and serum selenium (Se), samples (100 microl) were diluted with HNO3 (0.2%, v/v) to make a (1+1) dilution for urine analysis or a (1+4) dilution for serum analysis. An additional aliquot (10 microl) of Triton X-100 (0.1%, v/v) was added to each 200 microl of (1+1) diluted urine (or 20 microl of the Triton X-100 was added to each 500 microl of (1+4) diluted serum) sample. After the diluted sample mixture (10 microl) was introduced into a graphite cuvette, the corresponding chemical modifier (10 microl, containing Ni2+ + Pd + NH4NO3 in HNO3 (0.2%, v/v)) was added to it and the mixture was atomized. The MDL (3sigma) for Se in urine and in serum was 4.4 and 21.4 microg/l in undiluted sample, respectively, and the calibration graphs were linear up to 150 and 400 microg/l. Accuracies of urine Se were 98.9 - 99.4% by testing SRM 2670 (NIST) urine standards with RSD (between-run, n = 3) within 2.9%; and that of serum Se was 97.2% when testing a certified second-generation human serum (No. 29, #664) with RSD (between-run, n = 3) of 1.4%. The proposed method can be applied easily, directly, and accurately to the measurement of Be and Se in real samples (including six urine Se and four serum Se from patients of Blackfoot Disease in Taiwan).