Evidence for Atomic-Scale Inhomogeneity in Superconducting Cuprate NMR

Jamil Tahir-Kheli

2024-11-13

Abstract:In 1990, the Millis, Monien, and Pines (MMP) model and its improvement, the Zha, Barzykin, and Pines (ZBP) model in 1996, emerged as a realistic explanation of the cuprate NMR. These two models assume a single electronic component, translational symmetry, and that the electrons simultaneously have aspects of localized antiferromagnetic (AF) spins and delocalized Cu $d_{x^2-y^2}$ band states. NMR experiments were routinely fit to these models in the 1990s and early 2000s until they finally failed as NMR experiments developed further. It appears that cuprate theorists have given up on explaining the NMR and the NMR data is forgotten. Here, we assume a two-component model of electrons where the electrons reside in two regions, one metallic with delocalized band states, and the other antiferromagnetic with localized spins. This model breaks translational symmetry. We show that the normal state spin relaxation for the planar Cu, O, and Y atoms in $\mathrm{YBa_2Cu_3O_{7-\delta}}$ and their Knight shifts are explained by this two-component model. The temperature dependence of the Cu spin relaxation rate anisotropy in the superconducting state is also explained qualitatively.

Superconductivity,Strongly Correlated Electrons

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the difficult problem of interpreting nuclear magnetic resonance (NMR) data in copper - based superconductors. Specifically, the paper aims to propose a new model to explain the NMR phenomena in copper - based superconducting materials, especially in view of the failures encountered by the early Millis, Monien, and Pines (MMP) model and the Zha, Barzykin, and Pines (ZBP) model when interpreting NMR experimental data.

### Background and Problems
1. **Limitations of Early Models**:
- **MMP and ZBP Models**: These two models assume that electrons have a single component and that the system has translational symmetry. These models could well explain the NMR experimental data in the 1990s and early 2000s, but with the development of experimental techniques, these models gradually became ineffective.
- **Main Problems**:
- **AF Spin Correlation Length**: The AF spin correlation length estimated by the MMP and ZBP models is about 2 lattice spacings, which leads to the nuclear - nuclear spin coupling strength between O and Cu atoms not conforming to the experimental results.
- **Temperature Dependence**: The temperature - dependent change of the spin relaxation rate of the Y nucleus in YBa₂Cu₃O₇₋δ is inconsistent with the temperature dependence of the Knight shift and the static spin susceptibility.

2. **Evidence from Modern Experiments**:
- Many experiments have found that there is an inherent inhomogeneity in the electronic structure of copper - based superconductors, and these experiments are usually interpreted as evidence of some spin - density wave (SDW) or charge - density wave (CDW).

### Solutions
The paper proposes a new two - component model, assuming that copper - based superconductors are inhomogeneous on an atomic scale and contain two different regions:
- **Insulating AF Region**: It has localized electron spins and is located at the planar Cu sites.
- **Metallic Region**: It has delocalized electronic states and mainly exists within this region.

### Model Features
- **Nuclear Spin Relaxation and Knight Shift**:
- The total spin relaxation of each specific nuclide is the average value of all its isotopic nuclei in the crystal.
- The Knight shift in each region is determined by the static susceptibility of that region.
- **Temperature Dependence**:
- NMR in the normal state can be explained by this model.
- The anisotropy of the Cu spin relaxation rate in the superconducting state has also been qualitatively explained.

### Mathematical Descriptions
- **Dynamic Susceptibility**:
- The relationship between the dynamic susceptibility \(\chi(q, \omega)\) and nuclear spin relaxation and Knight shift is given by the Moriya expression:
\[
\frac{1}{kT_1} = \left( \frac{k_B T}{(g\mu_B)^2 \hbar^2} \right) \left( \frac{1}{N} \right) \sum_{q, \beta \neq \alpha} |k_F^\beta(q)|^2 \lim_{\omega \to 0} \left( \frac{\text{Im} \chi(q, \omega)}{\omega} \right)
\]
- Here, \(k\) is the atomic number, \(\alpha\) is the magnetic field direction, \(\beta\) is one of the two directions perpendicular to the magnetic field direction, and \(k_F^\beta(q)\) is the form factor of the wave vector \(q\) in the Brillouin zone.

- **Spin Correlation Time**:
- The expression for the spin correlation time \(\tau_{AF}(T)\) in the AF region:
\[
\tau_{AF}(T) = \left( \frac{T}{T + T_x} \right) \tau_0
\]
- The expression for the spin correlation time \(\tau_{M}(T)\) in the metal region:
\[
\tau_{M}(T) = 2 \pi \hbar N(0)^2 (k_B T) \cdot \lambda_{\text{dwell}}