Figure 1

(Color online) Jahn-Teller domain wall at $T=0$, as obtained by the classical variable minimization process described in the text. The cluster size used is $4\times 28$. (a) Variation of the order parameter perpendicular to the domain wall (black squares are results for $\lambda =1.25$; red circles are for $\lambda =1$). (b) Geometry for the calculation of the conductance; see text. (c) $2\times 12$ portion of the simulation cluster near a domain wall showing octahedral distortions in real space, for $\lambda =1.25$. The oxygen displacements obtained by energy minimization have been rescaled to clearly show the two different lattice configurations in left and right of the domain wall.Reuse & Permissions

Figure 2

(Color online) Conductance as a function of temperature for different system sizes. DW stands for conductance along domain walls (see Fig. 1) and Bulk stands for conductance in the bulk limit. Details about the Monte Carlo procedure are provided in the text. Despite some small-size effects, the conductance is clearly qualitatively different when domain walls are present. For the bulk case, data for all sizes fall within the same curve in the scale of the figure. Note that the conductance is roughly independent of the dimensions along the $b$ axis (perpendicular to the walls), consistent with the hypothesis that the conductance takes place along the domain walls. Also note that assuming a hopping amplitude $t$ of approximately 0.5 eV, then $T=0.06t$ is approximately room $T$.Reuse & Permissions

Figure 3

(Color online) Density of states a $T=0$. States appear inside the bulk gap energy region when close to a domain wall [more specifically DW refers to the projected DOS at the line of manganese sites, the closest to the domain wall in Fig. 1c]. The states in this band are localized in the direction perpendicular to the domain wall (their weight decays exponentially in that direction), but a small dispersion exists parallel to the domain wall. The band arising from the two $e_{\text{g}}$ orbitals is split due to the periodicity of the JT distortions in the direction parallel to the domain wall. For $\lambda =1.25$ the gap is on the order of $t$ $\left(\approx 0.5\text{eV}\right)$, even in the presence of a domain wall. For the smoother domain wall corresponding to $\lambda =1$ the gap is very small and it cannot be resolved in this scale (inset).Reuse & Permissions

Figure 4

(Color online) JT domain-wall profile for $T=0.01$ and $T=0.06$ obtained via Monte Carlo simulations, at $\lambda =1.25$ and for an $a=3$, $b=20$ system. The error bars show standard deviation due to thermal fluctuations. The temperature makes the domain walls smoother, reducing the energy gap.Reuse & Permissions

Figure 5

(Color online) Thermal averaged electronic density of states for $T=0.06$, $\lambda =1.25$, projected over different atomic planes parallel to the domain wall (in practice these are atomic chains since we use a 2D geometry). The zero of reference for each case is provided by the result at the far left or right in the figure. At finite temperature the domain walls are not as abrupt as at $T=0$; the bands associated with it (see Fig. 3) are wider and some spectral weight is shifted to the Fermi energy $E_{\text{F}}$. The eigenenergies arising from the Monet Carlo simulation have been given a small width $\left(0.01t\right)$, and the spectrum has been symmetrized with respect to $E_{\text{F}}$ to reduce noise.Reuse & Permissions