Nearly degenerate, opposite-parity levels of atomic dysprosium are highly sensitive to variation of the fine-structure constant, α, due to large relativistic corrections of opposite sign for the opposite-parity levels. In this unique system, in contrast to atomic-clock comparisons which require optical frequency combs to compare energy levels in different atoms, the difference of the electronic energies of the opposite-parity levels can be monitored directly utilizing a radio-frequency (rf), electric-dipole transition between them. The precision spectroscopy of this energy difference over a period of about a year can be used to constrain both a temporal variation and a gravitational-potential dependence of α since, during the data acquisition period, the Earth is located at different values of the gravitational potential of the Sun.

Because dysprosium has many naturally abundant stable isotopes, many rf transitions can be utilized for this search. Our measurements for the frequency variation of the 3.1-MHz (F = 10.5 → F = 10.5) transition in ¹⁶³Dy and the 235-MHz transition in ¹⁶²Dy, monitored over an eight-month period with a first-generation apparatus, provide a rate of fractional temporal variation of α of (−2.4±2.3) × 10⁻¹⁵ yr⁻¹ or a value of (−7.8±5.9) × 10⁻⁶ for k _α, the linear-variation coefficient for α in a changing gravitational potential. The latter result has been combined with other experimental constraints to extract the first limits on k _μ and k_q, which characterize the variation of m_e/m_p and m_q/m_p in a changing gravitational potential, where m_e, m_p, and m_q are electron, proton, and quark masses.

Systematic effects such as shifts due to collisions with the background gases, residual magnetic fields, and rf electric-field inhomogeneities have been investigated. These investigations form the basis for the design of a second-generation apparatus that has been constructed. The new design incorporates a differentially pumped vacuum system to reduce collisional shifts, new rf electrodes to reduce rf inhomogeneities, and better magnetic shielding to suppress Zeeman shifts. Systematic evaluations show an order of magnitude improvement in shifts related to rf inhomogeneities, and two orders of magnitude improvement in collisional shifts. In addition, a protocol to minimize magnetic field shifts has been developed. With these improvements, two orders of magnitude improvement in constraints for α variation are expected.