Abstract

In high precision x-ray absorption spectrometry collective motion of the atomic system is revealed through sharp spectral features arising from multielectron photoexcitations [1]. These minute details can only be resolved in pure atomic absorption, i e on samples of monatomic gases. Of these, only noble gases are readily accessible to experiment, and have extensively been studied [2-5]. Metals are also known to form monatomic vapors: however, technical problems in the containment of the sample are formidable. A bulk quantity of the vapor is required, with the absorption thickness of the order of d ~ 1, equivalent to ~ 10 mg/cm2 for most metals. A stable, well-defined thickness of this order precludes the use of atomic beams or jets. The sample must be contained in a cell with low absorption windows. The cell, as well as the windows, must be vacuum-tight, chemically resistant to the vapor, and must withstand the high temperature needed to vaporize the metal to sufficient vapor pressure. An ingenious solution, using inert-gas cooling of the windows, has been proposed in an experiment on rare-earth metals [6-8]: the accessible thickness of the sample, however, has not been sufficient to resolve the multielectron details of the absorption spectrum. Similarly, measurements on several transition metals [9] were limited to the immediate vicinity of the K-edge. In an experiment on mercury, where the requirements are considerably milder due to the low boiling point, traces of multielectron spectral features are reported [10].

Alkaline elements with boiling points in the vicinity of 700oC, represented another case of the manageable requirements for cell construction [11]. The choice of the windows is least critical for rubidium with its K- edge energy of 15.2 keV where many of candidate materials absorb only weakly. Fused quartz windows with thickness of 0.3 mm which can be prepared in a glassblower shop, would be acceptable. However, at the temperature required for sealing the quartz cell (1100oC), and possibly even at the operating temperature near the Rb boiling point, the metal vapors readily combine with the glass.

The proper cell material was found to be stainless steel (AISI 304). The cell (Fig 1.) is made of 27 cm long tube (1) with inner diameter of 10 mm and outer diameter 12 mm. Windows (2), made of 45 m thick stainless steel foil 8 mm in diameter, are welded on end flanges (3). An inlet tube (4) with inner diameter of 4 mm and outer diameter of 6 mm is used for attachment to the vacuum system, introducing rubidium and evacuating the cell.

Before inserting the Rb sample in the cell, vacuum tightness of the cell was checked with a helium leak detector. The leak rate was smaller than 10-9 mbar l/s. The stiffness and vacuum tightness of the windows were tested at 700oC and pressure of 10-2mbar. The cell was baked out for 15 minutes at the pressure of 10-2 mbar and temperature of 150oC.

Rubidium sample (m=16.2 mg) was introduced into the cell through the inlet tube at room temperature in an inert atmosphere of argon to prevent oxidation, the cell was reevacuated to the pressure of 10-1 mbar and the inlet tube sealed by pinching off.

In the absorption experiment the cell was inserted into a tunnel oven (6) and aligned to the open-air monochromatic synchrotron radiation beam (8) of the ROEMO II station, equipped with a Si(311) double crystal monochromator, at the DORIS storage ring of DESY, Hamburg. The absorption in the cell was monitored as a function of the temperature in the oven. The working point was chosen slightly above the temperature where the metal was completely vaporized (vapor pressure 71,3 kPa at 690oC), to ensure stable vapor density. The absorption spectrum above the rubidium K edge, shown in Fig 2, is comparable in detail to the well-studied spectra of the preceding noble gas, krypton.

For absolute determination of the absorption coefficient, an auxiliary measurement without the sample is required. For that purpose, the oven was outfitted with an air duct (7), leading through the insulation into the midpoint of the tunnel, whereby cool air could be blown. In this way, the cell was cooled in situ, without moving. The metal condensed on the side walls of the cell, around its middle, away from the beam path. In several heating- cooling cycles, completely reproducible results were obtained, testifying that rubidium was not chemically transformed. However, a tiny, stable remainder (0.3%) of the rubidium K edge was regularly found in the auxiliary absorption spectra after the first cooling: it is attributed to the surface adsorption on the windows.

In analysis of the experimental data [12] the present attenuation in the windows was found rather high, d ~ 2. Although it did not corrupt critically the sensitivity of the experiment, a retry with thinner windows is planned. The improvement may prove of crucial importance for experiments with other alkaline elements. The regions of interest are the L edges of cesium (~ 5 keV) and the K edge of potassium (3.608 keV). For cesium, stainless steel may again prove satisfactory, owing to the lucky coincidence that the region of interest falls into the low-absorption domain below the K edge energy of iron (7.4 keV). For potassium, though, new materials will have to be introduced, possibly titanium or vanadium.

The research was supported by the Ministry of Science and Technology of Slovenia. The contribution of the technical staff of the Institute for Vacuum Technology in Ljubljana is acknowledged. Dr. Ronald Frahm from HASYLAB provided expert advice on the beamline operation.

REFERENCES