The proton-electron plot of known atomic nuclei shows some rather marked regularities among atoms of lower atomic number. Up to O16 a simple step-wise figure appears into which the nuclear species H2, H3 and He4 could be fitted very nicely. Birge and Menzel have shown that the discrepancy between the chemical atomic weight of hydrogen and Aston's value by the mass spectrograph could be accounted for by the assumption of a hydrogen isotope of mass 2 present to the extent of 1 part in 4500 parts of hydrogen of mass 1.
It is possible to calculate with confidence the vapor pressures of the pure substances H1H1, H1H2, H1H3, in equilibrium with the pure solid phases. It is only necessary to assume that in the Debye theory of the solid state, θ is inversely proportional to the square root of the masses of these molecules and that the rotational and vibrational energies of the molecules do not change in the process of vaporization. These assumptions are in accord with well-established experimental evidence. We find that the vapor pressures for these molecules in equilibrium with their solids should be in the ratio of p11:p12:p13 = 1:0.37:0.29. The theory of the liquid state is not so vell understood but it seems reasonable to believe that the differences in vapor pressure of these molecules in equilibrium with their liquids whould be rather large and should make possible a rapid concentration of the heavier isotopes, if they exist, in the residue from the simple evaporation of liquid hydrogen near its triple point.
Accordingly two samples of hydrogen were prepared by evaporating large quantities of liquid hydrogen and collecting the gas which evaporated from the last fraction of the last cubic centimeter. The first sample was collected from the end portion of six liters of liquid evaporated at atmospheric pressure, and the second sample from four liters evaporated at a pressure only a few millimeters above the triple point. The process of liquefaction has probably no effect in changing the concentration of the isotopes since no appreciable change was observed in the sample evaporated at atmospheric pressure.
These samples were investigated for the atomic spectra of H2 and H3 in a hydrogen discharge tube run in Wood's so-called "black stage" by using the second order of a 21 foot grating with a dispersion of 1.31 Å per mm. With the sample evaporated at the boiling point no concentration so high as had been estimated was detected. We then increased the exposures so that the ratio of the time of exposure to the minimum required to get the H1 lines on our plates was about 4500:1. Under these conditions we found in this sample as well as in ordinary hydrogen faint lines at the calculated positions for the lines of H2 accompanying Hβ, Hγ, Hδ. These lines do not agree in wavelength with any molecular lines reported in the literature. However they were so weak that it was difficult to be sure that they were not ghosts of the strongly overexposed atomic lines.
The sample of hydrogen evaporated near the triple point shows these lines greatly enhanced, relative to the lines of H1, over both those of ordinary hydrogen and of the first sample. The relative intensities can be judged by the number and intensity of the symmetrical ghosts on the plates. The wave-lengths of the H2 lines appearing on these plates could be easily measured within about 0.02 Å. The following table gives the mean of the observed displacements of these lines from those of H1 and the calculated displacements:
Line Hα Hβ Hγ Hδ Δλ calc. 1.793 1.326 1.185 1.119 Δλ obs. Ordinary hydrogen -- 1.346 1.206 1.145 1st sample -- 1.330 1.119 1.103 2nd sample 1.820 1.315 1.176 --
The H2 lines are broad, as is to be expected for close unresolved doublets, but they are not as broad and diffuse as the H1 lines probably due to the smaller Döppler broadening. Although their intensities relative to the ghosts of the respective H1 lines appear nearly constant for any one sample of hydrogen, they are not ghosts for their intensities relative to the known ghosts for their intensities are not the same in the case of ordinary hydrogen and of the 1st sample as they are in the case of the second sample. They are not molecular lines for they do not appear on a plate taken with the discharge tube in the "white stage" with the molecular spectrum enhanced (H2γ was found as a slight irregularity on a microphotometer curve of this plate). Finally the H2α line is resolved into a doublet with a separation of about 0.16 Å in agreement with the observed separation of the H1α line.
The relative abundance in ordinary hydrogen, judging from relative minimum exposure time is about 1:4000, or less, in agreement with Birge and Menzel's estimate. A similar estimate of the abundance in the second sample indicated a concentration of about 1 in 800. Thus an appreciable fractionation has been secured as expected from theory. No evidence for H3 has been secured, but its lines would fall on regions of our plates where the halation is bad.
The distillation was carried out at the Bureau of Standards by one of us (F.G.B.), who is continuing the fractionation to secure more highly concentrated samples. The spectroscopic work was done at Columbia University by the other two (H.C.U. and G.M.M.) who are working on the molecular spectrum.
Harold C. UreyColumbia University,
F. G. Brickwedde
G. M. Murphy
December 5, 1931.
Birge and Menzel, Phys. Rev. 37, 1669 (1931).
Gale, Monk and Lee, Astrophys. J. 57, 89 (1928); Finkelnburg, Z. Physik 52, 57 (1928); Connelly, Proc. Phys. Soc. 42, 28 (1929).
Keesom and van Dijk, Proc. Acad. Sci. Amsterdam 34, 52 (1931).