Thorium-229's clock transition shows up in absorption under a sub-nanowatt continuous laser

Absorption only works if nearly all the light can be absorbed. In the high-harmonic experiments, a femtosecond frequency comb put on the order of 100,000 modes onto the crystal and a single mode was resonant with the nucleus ^[7]. About one part in 100,000 of the incident light could interact ^[2]. The four-wave-mixing sources failed the same test in a different way, with spectral widths several orders of magnitude wider than the nuclear linewidth in the host crystals ^[8]. So detection went the long way round: block the beam, then count the photons the isomer emits as it decays, with a time constant of about 600 seconds ^[6].

Six hundred seconds is ten minutes ^[3], and a measurement that reads that decay has to sit through it. The new readout is the attenuation of the transmitted laser power ^[18]. "This eliminates the slow nuclear fluorescence decay from the detection process, allowing for clock operation with fast signal acquisition," the authors write in Nature ^[5][20]. Earlier in the paper they set out the requirement plainly: a solid-state nuclear clock needs a laser whose linewidth is comparable with the crystal-field-broadened nuclear linewidth, plus a fast, sensitive and robust detection method ^[19].

The light starts as an infrared diode laser at 1,187 nanometres and passes through three consecutive second-harmonic steps ^[10]. Each step halves the wavelength, so 1,187 becomes 593.5, then 296.9, then 148.4 nanometres ^[1], which is the thorium resonance ^[2]. The final doubling stage uses a strontium tetraborate crystal ^[11], one of the few materials that are transparent in the vacuum ultraviolet and also phase-match there ^[13]. Power delivered to the nucleus is below a nanowatt ^[4]. A competing continuous-wave source, four-wave mixing in cadmium vapour, produces about 100 times more vacuum-ultraviolet power ^[12].

Inside the calcium fluoride host, the group resolved two distinct thorium centres and measured the isomeric shift between them ^[14]. One centre has a static electric crystal field gradient below 0.1 volts per square angstrom, against gradients around 100 volts per square angstrom in earlier work, more than a thousandfold smaller ^[15][4]. The authors read that as a high-symmetry arrangement of the ions around the nucleus, and expect nuclear resonance lines nearly independent of the lattice spacing ^[16]. In a solid, thorium-229 doubles as a test case for laser Mössbauer spectroscopy, sensitive to how the nucleus couples to its surroundings ^[17].

The paper does not report a measured instability, so it leaves open how well a clock built this way would keep time. It does show both laser-side requirements met in one apparatus: a source narrow enough to put its whole power inside the line ^[9], and a readout that works without waiting on the isomer ^[5]. For anyone building the source, the solid-state chain needs frequency stabilisation of only the single laser at 1,187 nanometres, a wavelength that can be linked to other optical clocks ^[12].