A giant leap in the development of a nuclear clock: the LMU group has directly observed the de-excitation of the Th-229 isomer via internal conversion. This is the first direct proof of the existence of the isomeric state. Today, this work has been published with Nature. Let’s look at the experiment more closely:

The naive way to prove the existence of the isomeric state in Th-229 would be a detection of the VUV gamma that is emitted as the isomer decays into the ground state. This approach has been followed by a dozen of past experiments, and by a handful of ongoing experiments. So far, all of these experiments could not observe a signal, or were not able to unambiguously attribute the observed signal to the isomeric decay. Various methods have been employed to populate the isomer in the first place: α-decay of U-233 (this is the most commonly used method), β-decay of Ac-229, optical excitation by means of various light sources, electron bridge processes, and a few more. Very recently, a number of very well-designed experiments, employing either the U-233 decay or optical excitation by synchrotron radiation, were unsuccessful in finding evidence of the isomer. Besides the trivial explanations for these null measurements (“The isomer does not exist.” and “Lifetime and/or energy of the isomer are very different from what is currently believed.”), it is the internal conversion (IC) channel that can inhibits the emission of an optical signal. In the IC process, the isomer would release its energy not via emission of a gamma particle, but would transfer its energy to electronic excitations of the thorium ion itself, or neighboring ions or atoms (such as atoms of the substrate or crystal material that the thorium ion is bound to). In such a process, a low-energy electron would be released. It is precisely this decay channel (and not the optical one) that the LMU group used in their experiment.

The LMU group chose the following strategy: isomer population via α-decay of U-233, combined with isomer detection via observation of the electron released during IC de-excitation. The isomer production part is well-established and quite robust: a very thin layer of U-233 is deposited onto a disk. As the U-233 undergoes α-decay, the Th-229 daughter nucleus gets a momentum kick equivalent to an energy of up to 80 keV, which propells it a few 10 nm through the uranium material. If the thickness of the overlaying material is smaller, the nucleus will reach the free space. It can then be trapped in an ion trap, deposited onto a catcher plate, or guided to a detector. The fraction of daughter nuclei that appear in the isomeric state Th-229m (as opposed to the nuclear ground state Th-229g) is about 2 percent.

The detector of choice for the observation of single electrons are multi-channel plates (MCPs). These devices amplify a single electron to an electronic signal containing an avalance of millions of electrons. Naively, one would proceed to simply place the U-233 source in close proximity of the MCP and count the electrons emitted during de-excitation of the isomer. Unfortunately, this approach does not work, as any ion striking the MCP with an energy of tens of keV will produce a “click” (it’s all radioactive material, after all). The ions thus need to be slowed before being deposited carefully onto the detector. Building a device that could slow down the Th-229 ions (and filter out all other ions) is no mean feat and took the LMU group half a decade.

The fate of a Th-229m ion is the following: After on average 160.000 years, it is born through α-decay of a parent U-233 nucleus. It travels through a few nm of uranium, reaches the vacuum of a large vessel, and is slowed and buffer-gas cooled by helium. The thermalized ions are extracted through a nozzle into an ion guide and further into a quadrupole mass-separator, where ions with different mass numbers are removed from the beam. The Th-229m ion is then gently deposited onto the surface of an MCP, where the residual impact energy is carried away by phonons. The ion quickly rips off electrons from the surface to neutralize. Now that the Th-229m atom has become neutral, the isomer energy is above the first ionization threshold: the isomer de-excites by transferring its energy to the least bound electron, which leaves the Th-229 atom with an excess kinetic energy of at most a few eV. While the Th-229 ion absorbs yet another electron from the surface to neutralize again, the emitted electron starts a signal cascade in the MCP that grows to form a mature avalanche of electrons. These impinge onto a phosphor screen, where they are converted into visible photons. These in turn are imaged by a CCD camera. The rate of such events is low, but integration over half an hour yields a sufficiently large signal.

The LMU group then performed a myriad of cross-checks to exclude literally all other possible origins of the observed signal. The signal appeared only for different charge states of Th-229 and for U-235, which is also known to possess an isomer. It did not appear for any other isotope, and it did not appear with the U-233 source replaced by a U-234 source. The signal could not be matched to any α- or β-decay, as these generate clearly different images on the detector.

This experiment thus adds a very valuable piece to the mosaic of investigations of the Th-229 isomer. While a number of past experiments have inferred the existence of the isomer from indirect measurements, this is the first direct observation of single property of the isomer: its de-excitation via IC. Many other properties are still to be determined, namely its energy, its lifetime, and the ability to drive the isomer transition optically. Concerning energy and lifetime, the LMU experiment was not aimed to improve existing values, but it is in agreement with them. The inferred energy falls between the first and third ionization threshold (between roughly 7 and 18 eV, where the consensus on the energy is currently 7.8(5) eV). By storing the Th-229 ions for a little while before detection, a lower bound of about one minute could be placed on the lifetime in vacuum (current estimate: about 15 minutes). Adaptations of the experiment will allow to measure both energy and lifetime more precisely.