Localized Triplons and Site Stuffing in the Quantum Dimer Magnet BiYbGeO$_5$

TL;DR

This study establishes BiYbGeO$_5$ as a structurally disordered yet magnetically isolated quantum dimer magnet by combining high-resolution neutron diffraction with inelastic neutron scattering. It detects significant Bi$^{3+}$ stuffing on Yb$^{3+}$ sites (~19%) that breaks Yb–Yb dimers and suppresses interdimer triplon hopping, while directly observing dispersionless triplon excitations consistent with isolated dimers. Insurance of an XXZ-type intradimer exchange, with $J_{XX} = 0.100(1)$ meV and $J_Z = 0.202(1)$ meV, and a dimer distance $d = 3.541(11)$ Å, provides a quantitative framework for disorder resilience in entangled spin systems. No magnetic order is detected down to 58 mK, highlighting BiYbGeO$_5$ as a valuable platform to study how structural disorder influences quantum entanglement in dimer magnets and to explore disorder-enabled quantum simulations.

Abstract

Thermodynamic and muon spin-relaxation measurements have recently highlighted BiYbGeO$_5$ as a new example of a rare-earth-based quantum dimer magnet with isolated Yb$^{3+}$ spin-$\frac{1}{2}$ dimers. However, direct spectroscopic evidence of the triplet excitations and measurements of the structural disorder are lacking. In this work, polycrystalline BiYbGeO$_5$ was synthesized using conventional high-temperature solid-state methods and investigated via high-resolution neutron powder diffraction and inelastic neutron scattering. Diffraction measurements down to 58 mK reveal no signatures of magnetic order and indicate that nearly 20\% of Yb$^{3+}$ sites are replaced by non-magnetic Bi$^{3+}$, introducing significant structural disorder. Inelastic neutron scattering shows dispersionless triplon excitations, consistent with localized, non-interacting spin dimers. Fits to the triplet excitation spectrum identify an XXZ-type anisotropic exchange with $J_{XX}$ = 0.11(2) meV and $J_Z = 0.15(1)$ meV. These findings establish BiYbGeO$_5$ as a structurally disordered but magnetically well-isolated quantum dimer system, providing a model platform for studying the resilience of entangled spin states to site dilution.

Figures (5)

• Figure 1: a) Crystal structure of BiYbGeO$_5$ viewed along the $a$-axis momma_vesta_2011. The interlayer separation alternates symmetrically between 8.064(3) Å and 8.008(3) Å, which is considerably larger than the intradimer bond length of 3.484(7) Å. The fraction of Yb spheres shaded in the Bi color corresponds to the proposed amount ($19.2(26)\%$) of Bi$^{3+}$ stuffing at the Yb$^{3+}$ site. b) The distorted honeycomb layers are viewed along the $ac$-plane with distortion along the $b$-axis. The dimers are labeled with blue ovals.
• Figure 2: A Rietveld refinement of the structure of BiYbGeO$_5$ obtained from neutron diffraction performed at 20 K using an incident wavelength of $\lambda$ = 1.54 Å. The peaks marked by black ticks come from the Cu canister used for the measurement. These Cu peaks were fitted using Le Bail extraction, not to influence the refinement of the BiYbGeO$_5$ parameters. A spurious peak visible at $\sim$ 2.71 Å$^{-1}$ (marked by a black arrow) corresponds to aluminum present in the sample environment due to inadequate collimation at 1.54 Å incident wavelength. The overall $\chi^2$ for the fit is 7.237.
• Figure 3: a-c) $\Delta$E vs |Q| color contour slices of INS data at a) 250 mK, b) 1.5 K, and c) 20 K showing a flat triplon mode visible above the bright elastic line. The scale factor for all panels has been kept the same to show the triplon band getting fainter on increasing temperature, signifying the depopulation of the ground state. d) Cuts of the INS data made along $\Delta$E integrating |Q| from 0.25 Å$^{-1}$ to 1.4 Å$^{-1}$ for 250 mK, 1.5 K, 20 K, and 100 K. The intensities of all scans have been normalized to allow comparison between the spectral weights of the triplon peak as a function of temperature.The large peak at 0.0 meV is the elastic line. A temperature-dependent triplon mode centered at 0.135 meV is visible for all three temperatures.
• Figure 4: A cut along $\Delta$E made by subtracting the 100 K INS data from the 250 mK (integrating |Q| from 0.25 Å to 1.4 Å). This excitation has been fit to the sum of two Gaussians, centered at 0.100(1) meV and 0.1510(4) meV, consistent with an XXZ-type intradimer exchange interaction. Using the Hamiltonian for an XXZ isolated dimer magnet, the exchange parameters come out to be 0.100(1) meV and 0.202(1) meV. The full widths at half maximum for the Gaussians are 0.06 meV and 0.07 meV, respectively, which are smaller than the resolution of HYSPEC ($\sim$ 0.1 meV).
• Figure 5: A cut of the INS data taken at 250 mK made along |Q| by integrating $\Delta$E symmetrically around the 0.135 meV maximum from 0.070 meV to 0.200 meV. Four data points between 1.87Å and 1.92 Å were removed from the fit as they were present due to the elastic tail extending into the triplon band. The fitted line represents the calculated structure factor of a dimer magnet system that yields an intradimer distance of 3.541(11) Å.