Marius Comper RO

Element 87 · Comptes rendus of the French Academy of Sciences, 1936 and 1937

Moldavium the lines and the atoms

In 1936, in Paris, the Romanian physicist Horia Hulubei attributed two X-ray lines to element 87, in a concentrate rich in alkali metals obtained from the mineral pollucite, and proposed the name moldavium, after the Romanian region of Moldavia (Hulubei's story, in Romanian). A curve based on Moseley's law puts element 87's lines at precise wavelengths, and the lines he reported in 1937 fall within half an X unit of them. But element 87 has no stable isotope: at the uranium levels measured in such minerals, a gram would hold on average far less than one francium atom.

Modern analyses measured 0.007–0.2 parts per million (ppm) of uranium in lepidolite. No published value was found for pollucite; 10 ppm is an assumption that favours francium.

What Hulubei printed

In his 1936 note in the Comptes rendus of the French Academy of Sciences, Hulubei describes a Cauchois spectrograph with a curved quartz crystal; a millimetre on the plate corresponded to about 5 X units of wavelength. His sample was a concentrate rich in alkali metals, obtained from pollucite. Where element 87's Lα emission should appear, he reports two lines, at about 1032 and 1043 X units. He discusses the neighbouring lines too: mercury's Lβ₃ at 1031.4, which he rules out because mercury's much stronger Lβ₁ line does not appear on his plates, and tungsten's forbidden lines at 1043.5 and 1044.4. He concludes that element 87 would be "rare, but not fugitive as had been supposed", and proposes the name Moldavium.

In his note of 8 November 1937 he adds three lines: one at about 838 X units, which his measurements cannot distinguish from lead's Lγ₁ line (838.01); a weak but persistent one at about 856 X units, which he considers distinct from known lines at the precision of his measurements; and a very weak one towards 715 X units, which he tentatively assigns to element 87. His notes say pollucite; the 2005 history by Adloff and Kauffman says lepidolite. Lepidolite was the source of the sample of the American physicist Frederick Hirsh, who searched for the element without success.

Where Moseley's law puts it

Moseley's law says that an X-ray line's energy rises regularly with atomic number. A curve fitted on the measurements from tungsten to radon and checked on radium and thorium predicts for element 87: Lα₁ 1028.5 X units, Lα₂ 1040.3, Lβ₁ 837.9, Lβ₂ 855.7, Lγ₁ 715.3. The values tabulated today for francium are almost the same; two of them are themselves interpolated.

The two 1936 lines fall 3.5 and 2.7 X units from the predicted positions, about 0.5–0.7 mm on the plate. Hulubei himself wrote in 1937, of a line reported by other authors, that a deviation of more than 3 X units from the predictions "is too much for a line measured to 0.1 X unit"; he had given his own 1936 values as approximate. The 1937 lines fall within half an X unit of the positions predicted for element 87.

90010001100120013001400150075808587901032 (1936)atomic number Z · Lα₁ wavelength, in X units
Filled dots: the elements from tungsten to radon, on which the curve is fitted; circles: radium and thorium, held out as a check; yellow dot: element 87, at 1028.5 X units; short line: the value printed in 1936. On the plate, the 3.5 X unit gap is about 0.7 mm.

What other lines appear nearby

Other elements have measured lines close to the wavelengths Hulubei reported, within 1.5 X units. Near 1032: mercury L1M3 at 1031.5, uranium L3M2 at 1032.5, indium KL2 (second order) at 1031.0; near 1043: tungsten L1N5 at 1043.1, arsenic KN2 at 1042.9, arsenic KN3 at 1042.9; near 838: lead L2N4 at 838.0, bismuth L2N1 at 837.6, radium L1M2 at 839.0; near 856: lead L2N2 at 856.8, mercury L1N5 at 854.9; near 715: radium L2N1 at 716.3. Hulubei checked many of them: he showed, for example, that the 1032 line persisted with an oil diffusion pump, and that 838 appeared only with the sample on the anticathode while other lead lines were absent. A line in the right place does not by itself say which element it comes from.

How many atoms could be there

Element 87 has no stable isotope. Francium-223, the longest-lived isotope, has a half-life of 22 minutes and occurs in nature from the decay of uranium-235, through a rare branch of actinium-227. In a mineral at secular equilibrium its atoms form and decay at the same average rate. Modern analyses give 0.007–0.2 ppm uranium in lepidolite, on average 1.0 × 10⁻⁴–3.0 × 10⁻³ francium-223 atoms per gram. Even at an assumed uranium concentration of 10 ppm, a gram would contain on average 0.15 atoms, and a tonne about 1.5 × 10⁵.

The most favourable value published in 1930 for X-ray analysis of metals was a threshold of about 0.0001 per cent, a millionth of the mass. For francium that means 2.7 × 10¹⁵ atoms per gram: at 10 ppm uranium the mineral would hold about 1.8 × 10¹⁶ times fewer, however much of it is used. How far concentrating the alkalis raised francium's share is not known. And if no actinium stays in the separated fraction, francium-223 stops forming: after a day, 2.0 × 10⁻²⁰ of the original would remain.

10⁻⁴10²10⁸10¹⁴10²⁰10²⁶1 g1 kg1 t1000 t1930 threshold for metals (0.0001%)10 ppm U0.007 ppm Umass of mineral · francium-223 atoms present at one time
Green lines: francium-223 atoms in a mineral with 0.007, 0.2 and 10 ppm uranium; yellow line: how many the 0.0001 per cent threshold reported for metals in 1930 would need. The lines are parallel: the ratio stays the same however much the mass grows.

What remains open

In 1936 a stable or long-lived isotope of element 87 was still a possible hypothesis; Hulubei based his conclusion on his spectral observations and controls. The calculation here uses what was learnt after 1939, when Marguerite Perey identified francium-223 as a product of actinium. The uranium content of Hulubei's pollucite, how much actinium remained in his concentrate and his spectrograph's exact resolution are not known. The origin of the lines remains unclear; under the assumptions above, element 87's atoms would have been many orders of magnitude too few to produce them.

Share the page

The link keeps the selected element, mineral mass and uranium concentration.

Sources and method

The L-line energies come from NIST's X-ray transition database; the fit uses only measured values, leaving out interpolated ones and those flagged as doubtful. For each line, the square root of the energy is fitted with a quadratic in Z over the elements from tungsten (74) to radon (86), a form inspired by Moseley's law; for radium (88) and thorium (90), held out, the predicted energies differ from the measured ones by less than 0.1%. The X units of the 1930s are converted with the factor implied by the comparison lines Hulubei himself printed: 1.00202 mÅ per X unit (range 1.00166 to 1.00232). The francium-223 count assumes secular equilibrium in the mineral: natural uranium with 0.7204% uranium-235 (half-life 704 million years), 1.380% of actinium-227 decays producing francium-223, and francium-223's 22.00-minute half-life (ENSDF). The comparison threshold is the most favourable one reported in 1930 by Laby for metals, 0.0001 per cent; Hevesy disputed it, and about 0.1 per cent was given for non-metallic substances.

  1. H. Hulubei, „Recherches relatives à l’élément 87”, Comptes rendus 202 (1936) 1927–1929, p. 1928; „Nouvelles recherches sur l’élément 87 (Ml)”, Comptes rendus 205 (1937) 854–857, pp. 855–856 (Wikisource scans).
  2. NIST X-ray Transition Energies Database; ENSDF, Fr-223, Ac-227.
  3. J.-P. Adloff, G. B. Kauffman, „Francium (Atomic Number 87), the Last Discovered Natural Element”, The Chemical Educator 10 (2005) 387–394.
  4. Uranium in lepidolite: Alexandre & Salvi, Minerals 15 (2025) 221, tab. 2; Current Science 118 (2020) 809, tab. 1. Separation of francium from actinium: E. K. Hyde, The Radiochemistry of Francium. X-ray analysis sensitivity in 1930: T. H. Laby, Nature 125 (1930) 818; G. von Hevesy, Nature 125 (1930) 776.