Spectral Weight Redistribution in $({\mathrm{LaNiO}}_3{)}_n/({\mathrm{LaMnO}}_3{)}_2$ Superlattices from Optical Spectroscopy

We have studied the optical properties of four $({\mathrm{LaNiO}}_3{)}_n/({\mathrm{LaMnO}}_3{)}_2$ superlattices (SL) ($n=2,3,4,5$) on ${\mathrm{SrTiO}}_3$ substrates. We have measured the reflectivity at temperatures from 20 to 400 K, and extracted the optical conductivity through a fitting procedure based on a Kramers-Kronig consistent Lorentz-Drude model. With increasing ${\mathrm{LaNiO}}_3$ thickness, the SLs undergo an insulator-to-metal transition (IMT) that is accompanied by the transfer of spectral weight from high to low frequency. The presence of a broad midinfrared band, however, shows that the optical conductivity of the $({\mathrm{LaNiO}}_3{)}_n/({\mathrm{LaMnO}}_3{)}_2$ SLs is not a linear combination of the ${\mathrm{LaMnO}}_3$ and ${\mathrm{LaNiO}}_3$ conductivities. Our observations suggest that interfacial charge transfer leads to an IMT due to a change in valence at the Mn and Ni sites.

Figure 1

Reflectivity of the LNO/LMO superlattices. (a) Room temperature reflectivities of the $n=2,3,4,5$ samples together with the reflectivity of pure LNO and that of the STO substrate. (b) Temperature dependence of the reflectivity of the $n=3$ compound. (c)–(e) Same for the $n=4,5$ and for pure LNO. Dashed lines correspond to intermediate temperatures 100, 150, 200, 250, and 350 K. In the inset of panel (c) the data at 20 K and the relative fit are reported in the whole range.

Figure 2

Optical conductivity of the LNO/LMO superlattices as extracted from a Lorentz-Drude fitting. (a) Room temperature optical conductivities of the $n=2,3,4,5$ samples and pure LNO and LMO. (b) Temperature dependence of the optical conductivity of the $n=3$ compound. (c)–(e) Same for the $n=4,5$ and for pure LNO.

Figure 3

Temperature dependence of the spectral weight $W(T)$ integrated up to ${\omega }_p$, for the $n=3$ (a), 4 (b), 5 (c) and LNO (d) compounds. For $n=3$ and $n=4$, $W(T)$ increases with increasing temperature, thus signaling the presence of incoherent excitations at low energies. For both $n=5$ and LNO the spectral weight can be fitted (continuous line) according to the quadratic relation $W(T)=W_0-B{T}^2$.

Figure 4

Comparison between the room temperature ${\sigma }_{1,n}$, and ${\sigma }_{1,n}^{*}$ (see text) for $n=2$ (a), 3 (b), 4 (c), 5 (d). The blue (red) area corresponds to spectral weight lost (gained) by the SL in the far- (mid-) infrared range due to the presence of interfacial effects. The green area highlights the transfer of spectral weight from higher energies. The redistributed charge density $\mathrm{\Delta }{N}^{\mathrm{FIR}}$ (see text) is plotted in the inset of panel (c).

Figure 5

Schematics of the spectral weight redistribution in LNO/LMO. (a) Transfer of spectral weight from coherent (blue) to incoherent (red) excitations in LNO, due to the reduction in the Ni oxidation state. (b) Spectral weight transfer in LMO from the high energy Jahn-Teller band (green) to the mid- and far infrared (red), due to the increased Mn valence. (c) Sketch of the spectral weight transfer in LNO/LMO SLs, due to the valence mixing associated with interfacial effects.