Buttersack et al. have recently published what seems to be the first photoemission experiments on solvated electrons.¹ Solvated electrons occur when group 1 metals are dissolved in certain solvents, the most popular is liquid ammonia. Once a standard undergraduate chemistry experiment, dropping a lump of sodium into liquid NH₃, cooled below its boiling point of 240 K, leads to an intense blue solution. The sodium metal ionises, losing its outer electron, but this electron doesn’t reduce the ammonia, instead it exists in a cavity of NH₃ molecules, just as an anion would. This ‘solvated electron’ isn’t bound to an atom, but is confined within the cavity, and the colour comes from transitions between energy levels of the electron in a box. If more sodium is added, eventually the entire solution becomes bronze in colour with a metallic lustre. At this point the concentration of electrons has increased enough to form a metal.

While these electron solutions have been known for two centuries, they seem about the least promising samples for photoemission studies possible. Highly volatile, toxic, flammable and that’s just the electrolyte, before we get to the alkali metal. The team at U49/2-PGM-1 beamline at BESSY II overcame these difficulties to measure the photoemission spectra of these solvated electrons for the first time. Using the SOL3 setup², They sprayed a micro jet (10-25 microns in diameter) of the alkali metal / ammonia solution through the interaction chamber, and focused the X-rays of the synchrotron on the jet. Emitted electrons then pass into the electron analyser, which is held at much lower pressure. After passing through the X-ray beam, the jet is caught in a cold trap.

The results show a distinct evolution of the spectrum with increasing metal concentration. To start, a peak appears at around 2 eV binding energy, which grows in size – this is the localised solvated electron peak. As the concentration of metal rises further, the peak shape evolves: a Fermi edge appears as well as plasmon satellites – both characteristics of a metal with delocalised electrons. The experimental data match well with calculations carried out by the team. The fact that the peak is identical no matter what group 1 metal is used proves that the electrons are truly detached from their metal ions.

It is fascinating to see photoemission spectra from electrons that are bound (and so of course have a binding energy) but yet are not in atoms. However, other types of materials do come to mind that have similar features, and a topic not covered by the authors in this paper was a comparison to these other classes of materials. The most obvious analogy to the solvated electron is the solid state electrides. Solid electrides are compounds where an electron acts in the role of an anion in a crystal structure. Notoriously difficult to make, and usually very air sensitive, both organic and inorganic versions of these compounds are known,³ and some have great potential as catalysts. XPS has been measured of some electrides and these show Fermi edges similar to that seen in liquid ammonia.

A second, and far more commonplace, analogy to the solvated electrons is the humble semiconductor. It’s been known for decades that by doping a semiconductor, we can go from an insulating material to a conducting one. For example, as the amount of n-type dopant is increased, the number of electrons increases (this is sometimes even called ‘electron doping’) To start with these electrons are localised, but get enough of them into the material and they get close enough to become delocalised – a degenerate semiconductor is formed. The evolution of the photoemission spectrum on increasing doping of a semiconductor is very similar to that seen in liquid ammonia.⁴

There are likely some very interesting differences between the liquid state solvated electrons, and solid state electrides and degenerate semiconductors. Now that the technical barriers to measuring ammonia solutions have been overcome, it remains to be seen whether further photoemission experiments will be able to give us more insight.

1. Science  05 Jun 2020: Vol. 368, Issue 6495, pp. 1086-1091
2. Review of Scientific Instruments 88, 073107 (2017)
3. Acc. Chem. Res. 2009 20;42(10):1564-72. doi: 10.1021/ar9000857.
4. Journal of Applied Physics 88, 5180 (2000)