Built-in Electric Field Induced Mechanical Property Change at the Lanthanum Nickelate/Nb-doped Strontium Titanate Interfaces

The interactions between electric field and the mechanical properties of materials are important for the applications of microelectromechanical and nanoelectromechanical systems, but relatively unexplored for nanoscale materials. Here, we observe an apparent correlation between the change of the fractured topography of Nb-doped SrTiO3 (Nb:STO) within the presence of a built-in electric field resulting from the Schottky contact at the interface of a metallic LaNiO3 thin film utilizing cross-sectional scanning tunneling microscopy and spectroscopy. The change of the inter-atomic bond length mechanism is argued to be the most plausible origin. This picture is supported by the strong-electric-field-dependent permittivity in STO and the existence of the dielectric dead layer at the interfaces of STO with metallic films. These results provided direct evidence and a possible mechanism for the interplay between the electric field and the mechanical properties on the nanoscale for perovskite materials.


Fracturing Method and Procedure
For cross-sectional scanning tunneling microscopy and spectroscopy (XSTM/S) measurements, a reliable and reproducible fracturing method and procedure is needed for high quality interfaces imaging. [1][2][3][4][5][6][7][8] Figure S1(a) shows the schematics of the sample holder for XSTM/S measurements. The sample holder is modified from a regular STM sample plate by adding two metal (stainless steel) blocks as clamps. The clamps are fixed by screws on the sample plate and the samples are held by another set of screws to hold the other metal block to the one mounted on the sample plate. After the oxide thin film were deposited on Nb:STO substrate (usually with dimension of 10 × 10 × 1 3 ), sample was further cut into desired dimension of 8 × 8 × 1 3 . The controllable fracture process is done with an initial notch served as the weak point for controlling the location of the cracking front. The notch is created by cutting the sample half way through the sample. The key for creating high quality interfaces for XSTM/S measurements is to fracture from the side of the film, as shown in Fig. S1(b). The samples are then mounted sideway for side fracturing, and the fracturing process is done by holding the bottom half of the sample by the clamps while having the top portion is moving against a sturdy metal cleaver. Although we expect to see hump on the top portion of the sample if a trench is observed in the bottom portion (the one that is measured by XSTM/S), after the fracture, the top portion of the sample dropped into the UHV chamber and cannot be recovered for further study.

Comparison with LCMO/Nb:STO and (YBCO/LCMO)n/Nb:STO
The fractured interfaces of LaNiO3/Nb:STO (LNO/Nb:STO) are very different from that of La2/3Ca1/3MnO3/Nb:STO (LCMO/Nb:STO), [6] in both topography and electronic band bending. The topography in LCMO/Nb:STO exhibits no trench; [6] while that in LNO/Nb:STO, a trench is clearly shown with ~6 nm width and ~0.6 nm depth (see Fig. 2 in main text). On the other hand, the band bending in LNO/Nb:STO was revealed by measuring dI/dV spectra point-by-point across the interfaces (see Fig. 3 in main text); while the spatial evolution of the dI/dV spectra across the LCMO/Nb:STO interfaces does not show any sign of band bending. [6] Same conclusion could be drawn with the dI/dV mapping at tunneling bias of 3.0 V for both LCMO/Nb:STO and LNO/Nb:STO interfaces that a gradual contrast change is seen in LNO/Nb:STO interfaces and no contrast change in LCMO/Nb:STO interfaces were observed. [6] Furthermore, the results for YBCO/LCMO superlattices on Nb:STO showed clear topography difference between YBCO and LCMO. [1] The topographic height difference is ~2 nm and is argued to be caused by the different fracture toughness in the main text.

Definition and the Meaning of the Effective CBM
To quantitatively analyze the observed dI/dV spectra across the interfaces, we defined the effective CBM, CBMeff, for further discussion.
Here is how to determine the CBMeff: (1) find the statistical average conductance, (dI/dV)ave, and the standard deviation of the conductance, (dI/dV)sdv, in the flat noise region (blue regions in Fig. 3(a)-(c)); (2) at each position, the CBMeff is defined to the onset bias where the dI/dV signal averaged over a vicinity of ±0.3 V region (which equivalent to averaging over 21 data points in our data set) is higher than (dI/dV)ave + (dI/dV)sdv. To clearly explain this procedure, we created a simulated dI/dV spectra with white noise and used them to discuss the effects of the true CBM shifting and change of the conductance near Fermi energy. The simulated dI/dV spectrum in Fig. S2(a) is created with the following equation plus a white noise: Where is set to be 2.10 eV. The white noise is simulated with a random number within (-0.5, 0.5) range. Following the procedure described above, the values of (dI/dV)ave and (dI/dV)sdv are shown in the figures with dashed lines. The CBMeff is determined to be at ~1.28 V, which is ~0.23 V higher than the CBMtrue. The location of the CBMtrue cannot be determined due to the probing limitation of the dI/dV signal. Now, let's discuss two different scenario: (1) shift of CBMtrue; and (2) change of the conductance (dI/dV) without shifting of the CBMtrue. For first scenario, we created a simulated dI/dV spectrum using Eq. S-1 with just changing the to 2.92 eV. Following the same procedure, the CBMeff is found to be ~0.22 eV higher than the CBMtrue. Note that the deviation between CBMeff and CBMtrue in this case is the same as that of the original one ( Fig. S2(a)).
On the other hand, for the second scenario, the simulated dI/dV spectrum was created by multiply 0.5 to the Eq. S-1 followed by adding the white noise. In this case, remains the same as 2.10 eV. Following the same procedure, the CBMeff is determined to be 1.49 V, which is ~0.44 V higher than the CBMtrue, which is not shifted compared to the case in Fig. S2(a).
In short, the CBMeff is different from the CBMtrue due to the signal limitation of the STS measurements. Two scenarios are discussed to have the effect of shifting the CBMeff: (1) the location of the CBMtrue is shifting; (2) the conductance (dI/dV) near the measuring limitation is changing.