1. IntroductionThe discovery of a two-dimensional electron gas (2DEG) in LaAlO3/SrTiO3 (LAO/STO) heterostructure has aroused significant interest in recent years due to a plethora of novel physical phenomena observed at the interface.[1–7] Vast scientific efforts have been made to understand the fundamental physics and apply the LAO/STO heterostructure to the next-generation electronics (see, for example, Refs. [8–11]). The discovery of this 2DEG has also facilitated the explorations of other conducting oxide interfaces, such as amorphous LAO/STO,[12,13] GdTiO3/STO,[14] LaGaO3/STO,[15] and amorphous LAO/La1−xSrxMnO3/STO.[16]
At another polar/non-polar interface, the LaCrO3/SrTiO3 (001) (LCO/STO) is reported to be insulating even though a band bending structure is observed in LCO film, which stems from the built-in field in LCO.[17] The analyses of the interfacial structure and composition suggest substantial cation intermixing and partially reducing Ti and Cr ions near the interface, leading to the deviation of interfacial metallicity.[18] In addition, it is found that the LaAl1−xCrxO3/STO undergoes a metal-insulator transition when x reaches 0.6, which is speculated to be due to the fact that electrons are trapped in LCO before being transferred to STO due to the multivalence of Cr ions.[19]
Polar-discontinuity also exists in the LCO/STO (111) heterostructure ((LaO3)3− and Cr3+ in LCO while (SrO3)4− and Ti4+ in STO). Whether a 2DEG exists in this system is not only a verification of the validity of the polar catastrophe model, but also important from the viewpoint of material design. In the (111) crystalline orientation, the Cr3+ ions in two layers of LCO can form a honeycomb lattice, in which the magnetic moments of adjacent Cr3+ ions in the two sublattices of the buckled honeycomb lattice align anti-ferromagnetically (Fig. 1(a)). Therefore, this system promises to realize both the spinless Haldane model[20,21] and the Kane–Mele model.[22] Theoretical study[23] shows that one unit-cell of another perovskite-type transition-metal oxide grown between LCO (111) honeycomb lattices is possible to realize the quantum spin Hall (QSH) effect or quantum anomalous Hall (QAH) effect, where the spin–orbit coupling is tuned by an electric field. So it is of great scientific significance to study the LCO/STO interface in the (111) direction.
In this work, we fabricate LCO film on STO (111) and investigate the electrical transport property and electronic structure of this heterostructure. It is found that when the oxygen partial pressure
used during the deposition of LCO is low, the LCO/STO (111) becomes metallic after 11 unit cells (uc) of the LCO have grown. The band alignment is investigated by x-ray photoelectron spectroscopy (XPS) and the doping mechanism is discussed properly.
2. Experimental procedureAtomically flat and Ti-terminated STO (111) substrates were prepared by chemical etching for 30 s through using buffered hydrogen fluoride (pH = 5.1) and subsequent annealing in one atmosphere oxygen for 2 h at 910 °C.[24] The LCO film was grown on an STO (111) substrate by the pulsed laser deposition (PLD) technique (KrF laser, 248 nm) from a ceramic target of LCO at
varying from 1.0×10−4 Torr (1 Torr = 1.33322×102 Pa) to 1.0×10−7 Torr. During the film deposition, the laser energy density was 1 J/cm2 with a repetition rate of 1 Hz and the substrate temperature was kept at 820 °C. After growth, the surface morphology and crystal structure of LCO film was examined by atomic force microscopy (AFM) and x-ray diffraction (XRD), respectively.
Sheet resistance was measured by the four-point method. For the Hall effect measurement, the sample was mechanically-patterned in hall-bar geometry with
in length and
in width. The contacts were made of 25-micron diameter Al wires through ultrasonic bonding. The transport properties were measured at temperatures ranging from 300 K to 13 K and magnetic field up to 1 T.
Without any surface treatment, the as-grown samples are sent ex situ to the XPS chamber with a base pressure of about 1×10−10 torr. The x-ray source is monochromatic 150-W Al Kα radiation with an energy resolution of 0.45 eV. The XPS spectra were collected in the direction normal to the surface at room temperature. During measurement, an electron gun was used to help us avoid the charging effect due to the insulating nature of some samples. In addition, the binding energy obtained was corrected by using the C-1s (284.6 eV). For all samples, a Shirley background has been subtracted from the raw data.
3. Results and discussion3.1. Morphology and structureThe LCO is an insulator with a pseudo-cubic lattice parameter of 3.885 Å, excellently matching to STO (3.905 Å). A typical surface morphology of an 18-uc (∼4.5 nm) LCO film grown on STO (111) under a
of 5×10−6 Torr is shown in Fig. 1(b). It shows an atomically flat surface that inherits from the STO substrate, indicating the layer-by-layer growth model is achieved. In Fig. 1(b), the height of the step edge is measured to be 0.25 nm (not shown here), which is equal to the lattice constant of LCO in the (111) direction. The crystal structure of LCO film is examined by XRD, which is shown in Figs. 1(c) and 1(d). The films are epitaxial with the (111) single phase character with the lattice constant of the LCO film expanding to a larger value (∼3.954 Å) due to the low
[25,26] as can be seen clearly in Fig. 1(d). In our work, the thickness of the thin film is examined by x-ray reflection (XRR) (inset of Fig. 1(d)).
3.3. Ti-2p3/2 core levelAs previously stated, for LCO/STO (111), the onset of metallicity comes when the thickness of LCO reaches 11 uc. This criticality is also observed in the XPS Ti-2p3/2 core-level spectrum (Fig. 3). For all LCO thicknesses measured, the spectrum is mainly composed of the Ti4+ peak (peak 1 in Fig. 3). However, when the LCO thickness is 11 uc, a peak (green areas in Fig. 3(f)) appears at about 2-eV lower in energy than the Ti4+ main peak and becomes more and more obvious when LCO gets thicker (Figs. 3(g) and 3(h)). This low energy peak is an unambiguous signature of the Ti3+ signal, which is an indication of electron doping of STO.[27,28] The relative content of Ti3+ is calculated as the ratio between the area under the Ti3+ peak and the area under the entire fitting curve. Results together with the peaks-fitting parameters are listed in Table 1. It is shown that while the Ti3+ relative content is very small for LCO thinner than 11 uc, it increases abruptly to 5.02% at 11-uc LCO and further increases to 7.38% and 8.55% for 14 uc and 18-uc LCO, respectively. A similar magnitude of the Ti3+ relative content is also reported in the literature for the conducting LAO/STO interface.[18,29] Note that except for the two peaks mentioned above, there is one additional peak that is about 0.9 eV lower than the Ti4+ peak (peak 2 in Fig. 3). The signals of 2-uc and 3-uc samples are invisible or small, but they become evident when the thickness of LCO is 5 uc, well below the critical thickness of 11 uc, and nearly keeps constant upon increasing the LCO thickness. This peak is also seen in LAO/STO[29] but its origin is unspecific so far.
Table 1.
Table 1.
Table 1.
Parameters used for fitting the Ti-2p3/2 core-level spectrum of LCO/STO (111) in Fig. 3. The FWHMs for the Ti4+ main peak (peak 1 in Fig. 3), peak 2 and Ti3+ peak are listed. ΔBE1 is the energy separation between Ti4+ and peak 2. ΔBE2 is the energy separation between Ti4+ and Ti3+. The percentage of corresponding peak is the ratio between the area under the peak and that under the entire fitting curve with uncertainty estimated at ±5% of the base value.
.
LCO thickness/uc |
Peak 1 (Ti4+) FWHM |
Peak 2 FWHM |
Ti3+ FWHM |
ΔBE1/eV |
ΔBE2/eV |
Relative of peak 2/% |
Relative of Ti3+/% |
2 |
1.15 |
1.25 |
1.25 |
0.9 |
2 |
1.19 |
0.62 |
3 |
1.15 |
1.25 |
1.25 |
0.9 |
2 |
4.35 |
0.64 |
5 |
1.15 |
1.25 |
1.25 |
0.89 |
1.99 |
10.68 |
1.78 |
7 |
1.15 |
1.25 |
1.25 |
0.95 |
1.98 |
17.33 |
1.51 |
9 |
1.15 |
1.25 |
1.25 |
0.9 |
2 |
15.88 |
1.83 |
11 |
1.15 |
1.25 |
1.25 |
0.9 |
2 |
18.27 |
5.02 |
14 |
1.15 |
1.25 |
1.25 |
0.9 |
2 |
11.85 |
7.38 |
18 |
1.15 |
1.25 |
1.25 |
0.9 |
2.2 |
17.43 |
8.55 |
| Table 1.
Parameters used for fitting the Ti-2p3/2 core-level spectrum of LCO/STO (111) in Fig. 3. The FWHMs for the Ti4+ main peak (peak 1 in Fig. 3), peak 2 and Ti3+ peak are listed. ΔBE1 is the energy separation between Ti4+ and peak 2. ΔBE2 is the energy separation between Ti4+ and Ti3+. The percentage of corresponding peak is the ratio between the area under the peak and that under the entire fitting curve with uncertainty estimated at ±5% of the base value.
. |
3.4. Valence band offset and built-in fieldTo further understand the mechanism of the conductivity in LCO/STO (111), we investigate the electronic structure of the heterostructure by XPS. The valence band offset (VBO),
, can be calculated from the following formula:[30]
where the
(
is the binding energy difference between core level of La-4d
5/2 (Sr-3d
5/2) and valence band maximum (VBM) in bulk LCO (STO). In our work,
is the binding energy of the La-4d
5/2 core level measured in LCO/STO (111) with thick LCO film (18 uc) and the
is the binding energy of VBM of LCO in the same sample, which is determined by extrapolating the leading edge of the VB spectrum to the energy axis in order to account for the instrumental broadening (Figs.
4(a) and
4(b)). The VBM of LCO (
is measured to be 0.52 ± 0.05 eV (Fig.
4(a)) and
is 101.92 ± 0.05 eV. To deduce the VBM of STO (111) (
, an Nb-doped (0.7 wt%) STO (111) substrate is used after the same chemical etching and annealing procedure as that of un-doped STO. By adopting the same linear method used for
, the
and the
are measured to be 2.8 ± 0.05 eV (Fig.
4(b)) and 130.65 ± 0.05 eV, respectively. Here the value of the later is a little larger than that of the (001) direction, which is 130.54 eV from Chambers.
[31]
is the difference in binding energy between Sr-3
and La-4d
5/2 core level measured at the LCO/STO (111) heterostructure and it is an interfacial property. The value of
first increases from 27.82 ± 0.05 eV to 28.31 ± 0.05 eV when the thickness of LCO increases from 2 uc to 11 uc and then reaches to a plateau of 28.23 ± 0.05 eV at 18 uc.
Using the above formula, the VBO of LCO/STO (111) heterostructure can be calculated. The result is shown in Fig. 4(d) for various thickness values of LCO. It can be seen that the VBO increases with the thickness of LCO increasing, and goes to a plateau after 11 uc, where it peaks at 2.45 ± 0.05 eV. The increase of VBO with LCO thickness increasing indicates the existence of a built-in field in LCO, which is also evidenced by the increasing of the full width at half maximum (FWHM) of La-4d core-level peak with LCO thickness increasing (Fig. 4(e)). The increase of FWHM with LCO thickness increasing results from peak-broadening of the La-4d core-level spectrum due to the built-in field in LCO, which will lead to the increasing of electron energy in LCO. Our results on the tendency and magnitude of VBO and FWHM are comparable to the results of Chambers.[17]
3.5. DiscussionAt a polar-discontinuous interface, a potential energy divergence problem usually arises and is deemed to trigger the interfacial metallicity at LAO/STO (001) according to the polarization catastrophe model, in which it is believed that the electrons at VBM of LAO transfer to the bottom of the conduction band of STO at the critical thickness (4 uc).[9] Indeed, the observed increase of VBO and FWHM with LCO thickness increasing as well as the existence of a built-in field in LCO is a reflection of energy divergence. However, the observed magnitude of VBO at a critical thickness of 11 uc LCO (2.45 ± 0.05 eV) is less than the energy gap of STO (3.2 eV) (Fig. 4(d)). Small VBO is also reported in LAO/STO[32] and it is now recognized in the literature that this discrepancy can be reasonably explained when the surface states in LAO, such as in-gap states induced by oxygen vacancies, are considered.[33–36] In the present study, due to the low
used for growing a conducting LCO/STO (111) interface, it is possible that electrons donated by oxygen vacancies in the film reside in the in-gap states of LCO that is higher in energy than LCOʼs VBM. Due to the increase of potential with LCO thickness increasing, electrons in these in-gap states can be transferred to STO at a critical thickness of 11 uc, at which the energy of in-gap states in LCO equals the energy of STOʼs conduction band (Fig. 5(a)).
Another explanation of the interfacial metallicity can be made after inspecting the electron structure on the STO side. Figure 4(f) displays the FWHM of Ti-2p3/2 core-level spectrum as a function of LCO thickness. The FWHM of the Ti-2p3/2 core level has the same tendency as VBO and FWHM of La-4d5/2 (Figs. 4(d) and 4(e)). The variation of the FWHM of Ti-2p3/2 we have observed is large compared with Chambersʼs results,[17] in which the FWHMs of Ti-2p3/2 and Sr-3d5/2 core level change to a less extent than the FWHMs of Cr-3p and La-4d. In Ref. [17], the small variation in FWHM of Ti-2p3/2 is explained as the consequence of out-diffusion of Ti ions into LCO film, leading to the presence of
. Although the variation in FWHM of the Sr-3d5/2 core level is also small in our work (Fig. 4(c)), the relatively large change in the FWHM of the Ti-2p3/2 core level should be mainly ascribed to the built-in field in the STO substrate, not just to the Ti out-diffusion during growth. From the band-diagram viewpoint, this built-in field in STO indicates a band bending downward and a notched structure forming in STO near the interface (Fig. 5(b)), which is widely observed at the conducting LAO/STO interface.[37,38] This notched structure in STO can lead to the formation of a quantum well at an interface at the critical thickness of 11 uc, at which the STO conduction band passes through the Fermi level (Fig. 5(b)).
4. ConclusionsIn this work, we fabricate a high-quality LCO/STO (111) heterostructure by the PLD method and investigate the electrical transport property and electronic structure. The LCO/STO (111) heterostructure grown in
lower than
is metallic when the thickness of LCO is no less than 11 uc. From the observation of XPS, the Ti3+ signal in the Ti-2p3/2 core-level spectrum shows an abrupt increase at 11 uc. The magnitude of VBO, and the FWHMs of La-4d5/2 and Ti-2p3/2 all increase with LCO thickness increasing, indicating bands bending in LCO film and STO substrate. While the oxygen vacancy forming in the substrate during growth can be ruled out as a major source of interfacial conductivity, the interfacial metallicity can be explained as a consequence of the charge-transfer process in which electrons in LCO in-gap state transfer to STO due to potential energy divergence. Alternatively, the metallicity may be related to the quantum well forming at STO. Our work provides an insight into the origin of 2DEG found at LAO/STO and other polar/non-polar interface. Moreover, due to the interesting lattice and spin structure of LCO in the (111) direction, our work provides a basis for further exploring the novel topological quantum phenomena in this system.