Atomic Weight of Arsenic

The earliest attempts to determine the atomic weight of arsenic were made prior to 1826 by Berzelius. By oxidising the element to arsenious oxide he first obtained the value 80.2 (O = 16), and later by reduction of the oxide by means of sulphur he arrived at the figure 75.02. In 1845 Pelouze, and in 1859 Dumas, applied the method of titration of known quantities of pure silver to the analysis of arsenic trichloride and both obtained values slightly lower than that of Berzelius, the mean value from the two series of experiments being 74.91. A lower figure, however, was obtained in 1859 by Wallace, who used the same method of titration on the bromide, his value being 74.19. Kessler in 1861, after investigating the oxidation of arsenious oxide by means of potassium chlorate and potassium dichromate, deduced the molecular weight of the oxide and calculated the atomic weight of arsenic to be 75.13. That the value should be near to 75 was in accord with the periodic law and with the physical and chemical properties of the element, and the generally accepted value up to 1896 was 74.9.

In the latter year the problem was again attacked with much improved experimental methods. Hibbs, by heating sodium pyroarsenate in hydrogen chloride at a temperature well below the fusion point of sodium chloride, obtained the latter as a pure product, and from the ratio Na₄As₂O₇:4NaCl obtained as the mean value from ten experiments 74.876. Ebaugh made three series of somewhat similar experiments; in the first, silver arsenate was heated in hydrogen chloride and the silver chloride formed, after weighing, was reduced to silver in a stream of hydrogen; in the second and third series lead arsenate was converted similarly into lead chloride and lead bromide respectively. The following are the mean values obtained and the ratios from which they were derived:

Ag3AsO4:3AgCl	As = 74.90
Ag3AsO4:3Ag	As = 75.06
Pb3(AsO4)2:3PbCl2	As = 74.93
Pb3(AsO4)2:3PbBr2	As = 74.84

Baxter and Coffin, in 1909, also heated silver arsenate in hydrogen chloride. Samples of different origin were used, with slightly different results. They also dissolved the arsenate in nitric acid and precipitated the silver as chloride or bromide. The following mean values were obtained:

Ag3AsO4:3AgCl (1st series)	As = 75.05
Ag3AsO4:3AgCl (2nd series)	As = 74.93
Ag3AsO4:3AgBr	As = 74.92

The value internationally accepted up to 1928 was 74.96. However, in 1929 the British Sub-committee on Atomic Weights proposed that 74.934, the value obtained by Aston from calculations based upon mass spectrograms, should be accepted, it being considered that Aston's method was less liable to error than any other. Krepelka, from analyses of especially purified arsenic trichloride, obtained as the mean of 13 determinations 74.936 ± 0.001, a result which nearly coincides with that of Aston. The value 74.93 was adopted internationally in 1930.

Later work by Aston on the mass spectrum of gaseous arsenic hydride led to the lower figure 74.92, and in 1933 Baxter and his co-workers analysed arsenic trichloride and arsenic tribromide by comparison with silver. The compounds were prepared from arsenic and the pure halogens, and were subjected to prolonged fractionation in exhausted glass apparatus. They were then converted to ammonium halide and arsenite and the former precipitated with silver nitrate. The trichloride was also compared with iodine pentoxide by hydrolysing weighed quantities of the former with sodium hydroxide in a vacuum, then after neutralisation adding a weighed, nearly equivalent, quantity of iodine pentoxide and finding the end-point in the presence of starch by neutralising. with phosphate and adding standard dilute iodine or arsenite solutions. The weights of iodine pentoxide were corrected for retained moisture and adsorbed air.

The average value for the atomic weight of arsenic from all the above experiments is 74.908 and, in view of the concordance of the results, the value adopted by the International Union of Chemistry in 1934 was 74.91, this value remaining unchanged in 1938.

In 1935, Baxter and Frizzell determined the ratio of arsenic trichloride to iodine, using the value 126.917 for the atomic weight of iodine. Weighed quantities of arsenic trichloride, purified as described above, were hydrolysed with disodium hydrogen phosphate, and the arsenious acid formed allowed to react with weighed, very nearly equivalent, quantities of pure iodine in nearly neutral solution. A slight deficiency of one or the other was then made up with dilute solutions of arsenious acid or iodine. The reactions were carried out without contact with oxygen by working in exhausted vessels up to the final determination of the end-point.

These results suggest a slightly higher figure than the International value, 74.91, and more recent work also points in the same direction. Krepelka and Kocnar in 1936 synthesised arsenic tribromide from highly purified bromine and arsenic and from determinations of the ratios AsBr₃:3Ag and AsBr₃:3AgBr arrived at the value 74.923, which closely approaches Aston's value derived from the mass spectrum of arsine.

The evidence would justify, therefore, the value 74.92 for the atomic weight of arsenic.