Digital modulation of the nickel valence state in a cuprate-nickelate heterostructure

Layer-by-layer oxide molecular-beam epitaxy has been used to synthesize cuprate-nickelate multilayer structures of composition ${\left({\mathrm{La}}_2{\mathrm{CuO}}_4\right)}_m/\mathrm{LaO}/{\left({\mathrm{LaNiO}}_3\right)}_n$. In a combined experimental and theoretical study, we show that these structures allow a clean separation of dopant and doped layers. Specifically, the LaO layer separating cuprate and nickelate blocks provides an additional charge that, according to density-functional theory calculations, is predominantly accommodated in the interfacial nickelate layers. This is reflected in an elongation of bond distances and changes in valence state, as observed by scanning transmission electron microscopy and x-ray absorption spectroscopy. Moreover, the predicted charge disproportionation in the nickelate interface layers leads to a metal-to-insulator transition when the thickness is reduced to $n=2$, as observed in electrical transport measurements. The results exemplify the perspectives of charge transfer in metal-oxide multilayers to induce doping without introducing chemical and structural disorder.

Figure 1

Structural details of ${\left(\mathrm{LCO}\right)}_3/\mathrm{LaO}/{\left(\mathrm{LNO}\right)}_4$. (a) Basal (in-plane, perpendicular to the growth direction) and apical (out-of-plane, parallel to the growth direction) distances between La atoms neighboring a transition metal site. (b) HAADF-STEM image. The black arrow indicates the growth direction [parallel to (001) in pseudocubic (pc) notation]. (c) XRD scan along the ${\left(00\mathrm{l}\right)}_{\mathrm{pc}}$ direction. The substrate peak is marked with a red asterisk. The blue numbers indicate the $l$ indices of the heterostructure and result in a $c$-axis parameter of 37.24 $\pm$ 0.01Å. (d) Side view of the relaxed structure of the column marked in (b) from $\mathrm{DFT}+U$ calculations. (e) Spatial elemental distribution extracted from the EELS spectrum image with color code: La—green, Cu—blue, and Ni—red, respectively. The bottom panel shows the averaged line profile from the EELS spectrum imaging. The dotted lines indicate the nominal interfaces.

Figure 2

(a) Superimposed annular bright field (ABF) (red) and high-angle annular dark field (HAADF) (green) images of the sample shown in Fig. 1. While ABF mode is sensitive to light atoms, HAADF mode is sensitive to heavy atoms. La atoms appear bright green, Cu and Ni atoms dark green, and O atoms black. The black dashed lines serve as guide to the eye to distinguish cuprate from nickelate layers. The corresponding measured O-(Cu,Ni)-O distances (index: exp) for basal and apical O atoms are shown in (b) together with the calculated ones (index: th).

Figure 3

$\mathrm{DFT}+U$ results for the local valence electron differences relative to bulk as function of the ion position in [001] direction of 3/4. Each layer contains two inequivalent transition metal ions. The split Ni1 and Ni2 values reflect the charge disproportionation.

Figure 4

(a) Layer-, spin-, and site-resolved calculated electronic density of states of 3/4. Only inequivalent layers are shown. Blue and grey areas depict the Cu and Ni $3d$ states in total, whereas orange areas and black lines depict the projections onto the $d_{x^2-y^2}$ and $d_{3z^2-r^2}$ orbitals, respectively. The two columns represent the two inequivalent transition metal sites per layer. The printed numbers are the local magnetic moments. The images of the $d$ orbitals represent the number and anisotropy of the electrons. (b) and (c) XAS spectra measured with the electric field parallel to (perpendicular to) the sample surface (orange and black lines) together with the normalized dichroic signal (grey line) defined by $(I_x\left(E\right)-I_z\left(E\right))/((2I_x+I_z)/3)$ at the Cu (b) and Ni (c) $L$ edges. As the La $M_4$ line overlaps with the Ni $L_3$ line, we fitted a Lorentzian to the La peak, which we then subtracted from the spectrum.

Figure 5

Temperature-dependent resistivity of ${[{\left({\mathrm{La}}_2{\mathrm{CuO}}_4\right)}_m/\mathrm{LaO}/{\left({\mathrm{LaNiO}}_3\right)}_n]}_l(m=1$, $n=2,3,4$, and $l=7,10,9$) with average formal Ni valences of 2.5+, 2.67+, and 2.75+ for $n=2,3,4$, respectively, compared to ${\left[{\left({\mathrm{LaNiO}}_3\right)}_n{\left({\mathrm{LaAlO}}_3\right)}_n\right]}_k$ superlattices with $n=2,4$ and $k=6,3$ as indicated in the legends.

Figure 6

Layer-, spin-, and site-resolved calculated electronic density of states of 3/2. Only relevant layers are shown. Blue and grey areas depict the Cu and Ni $3d$ states in total, whereas orange areas and black lines depict the projections onto the $d_{x^2-y^2}$ and $d_{3z^2-r^2}$ orbitals, respectively. The two columns represent the two inequivalent transition metal sites per layer. The printed numbers are the local magnetic moments. The charge disproportionation in conjunction with the layerwise AFM order in the nickelate opens a band gap of 0.28 eV.

Figure 7

Dependence of the band gap width $E_{\text{gap}}$ and the Ni magnetic moments for 3/2 (top and middle panel) and of the Ni magnetic moments [interfacial (IF) and central (C) layer] for 2/4 (bottom panel) on $U_{\text{Ni}}$. The magnetic moment difference between Ni1 and Ni2 sites reflects the disproportionation.