Effect of charge ordering on crossplane thermal conductivity in correlated perovskite oxide superlattices

Lattice thermal conductivity is a sensitive probe of distortions of the translational invariance of crystalline materials. Crossplane thermal conductivity, $\kappa \left(T\right)$, was studied by the 3ω technique in superlattices, comprising orthorhombic charge/orbital ordered manganite ${\mathrm{Pr}}_{0.7}{\mathrm{Ca}}_{0.3}{\mathrm{MnO}}_3$ and cubic ${\mathrm{SrTiO}}_3$. The manganite/titanite superlattices show a peak in $\kappa \left(T\right)$ just above the charge order transition temperature $T_{\mathrm{CO}}\sim 240\mathrm{K}$ with a pronounced thermal hysteresis. This ordering peak in $\kappa \left(T\right)$ is successively suppressed with decreasing layer thickness. Our results demonstrate a minor effect of the interface density on $\kappa$, primarily visible below $T_{\mathrm{CO}}$. In contrast, much higher changes in $\kappa \left(T\right)$ evolve close to $T_{\mathrm{CO}}$, due to interface-induced effects on charge and orbital ordering. Our results suggest that the ordering peak in $\kappa \left(T\right)$ is a result of lattice softening above the phase transition which is modified by thickness-dependent misfit strain.

Figure 1

(a) Schematic drawing of the measurement geometry for the 3ω method including the Pt heater structures. (b) Modulus of the temperature oscillation vs the logarithm of the frequency with a PLD PCMO/STO bilayer (measurement) and with the STO substrate only (calculated from the slope of $\left|T\left[\ln \left(\omega \right)\right]\right|$) for $T=100$ and 300 K.

Figure 2

XRD patterns in θ-2θ geometry of the PLD and MAD samples. For the PLD samples (a) exemplary patterns are shown for the bilayer and for the $8\times 3.1\mathrm{nm}/3.1\mathrm{nm}$ superlattice close to the (002) ${\mathrm{SrTiO}}_3$ substrate reflection. The green and blue vertical lines express the expected 2θ angles for unstrained STO and PCMO single crystals, respectively. (b) Comparison of a $10\times 3.3\mathrm{nm}/3.3\mathrm{nm}$ MAD SL with additional SrO monolayers inserted at all interfaces (black) and a $14\times 3.2\mathrm{nm}/3.2\mathrm{nm}$ MAD SL, where SrO is added at the surface of STO layers only (red). PLD and IBS layers are measured by using copper ${\alpha }_1=0.15406\mathrm{nm}$ radiation and MAD layers by copper ${\alpha }_1$ and ${\alpha }_2=0.15418\mathrm{nm}$.

Figure 3

STEM HAADF images of a SrO-interface-engineered PCMO/STO SL with individual PCMO and STO layer thickness $t=1.75\mathrm{nm}$. (a) Overview and (b) high-resolution images with SrO monolayers at interfaces.

Figure 4

Temperature dependences of thermal conductivity, $\kappa \left(T\right)$, of PLD- (a) and MAD- (c) grown SLs with different thicknesses of individual layers, $t=1.75-25\mathrm{nm}$, and almost constant total film thickness, $t_{\mathrm{\Sigma }}=50\mathrm{nm}$, for PLD samples and different $t_{\mathrm{\Sigma }}=50-112\mathrm{nm}$, for MAD SLs. One can see a peak feature in $\kappa \left(T\right)$ for PLD-grown SLs with $t\ge 6\mathrm{nm}$ and for MAD-SLs with $t\ge 3\mathrm{nm}$. Moreover, the peak temperature, $T_p$, shifts to lower temperatures with decreasing $t$. In contrast, for SLs with very thin layers, i.e., $8\times (3.1/3.1\mathrm{nm})$ for PLD and $32\times (1.75/1.75)$ and $16\times (1.85/1.85)$ for MAD, no peak in $\kappa \left(T\right)$ was observed. They show for $T>150\mathrm{K}$ an almost temperature-independent and small, $\kappa <1$, thermal conductivity.

Figure 5

Thermal hysteresis of $\kappa \left(T\right)$ of the MAD-grown PCMO/STO bilayer (a) and a $10\times (3.35\mathrm{nm}/3.35\mathrm{nm})$ SL (b).

Figure 6

Selected area electron diffraction plane-view images of the MAD-grown PCMO/STO bilayer for a charge disordered state at $T=300\mathrm{K}$ (a), a charged ordered state at $T=170\mathrm{K}$ after cooling down from $T=300\mathrm{K}$ (b), and charged ordered state at $T=220\mathrm{K}$, obtained after warming up from $T=80\mathrm{K}$ (c).

Figure 7

Thermal conductivity of the PCMO/STO and LSMO/STO superlattices as a function of the interface density at $T=140\mathrm{K}$. All data are corrected by the contribution of the STO insulation layer as described in the text.