We present structural and electrical characterization of SrZrO3 that has been epitaxially grown on Ge(001) by oxide molecular beam epitaxy. Single crystalline SrZrO3 can be nucleated on Ge via deposition at low temperatures followed by annealing at 550 °C in ultra-high vacuum. Photoemission spectroscopy measurements reveal that SrZrO3 exhibits a type-I band arrangement with respect to Ge, with conduction and valence band offsets of 1.4 eV and 3.66 eV, respectively. Capacitance-voltage and current-voltage measurements on 4 nm thick films reveal low leakage current densities and an unpinned Fermi level at the interface that allows modulation of the surface potential of Ge. Ultra-thin films of epitaxial SrZrO3 can thus be explored as a potential gate dielectric for Ge.

Epitaxial thin films of perovskite structured multifunctional oxides have attracted tremendous attention for the electronic, magnetic, and optical properties they exhibit. The ability to epitaxially integrate multifunctional oxides on semiconductors has opened a pathway to exploit these properties in a variety of device applications.1–3 Presently, epitaxial integration involves the initial growth of a perovskite structured alkaline earth titanate, AeTiO3 (Ae =Ca, Sr, and Ba), directly on the (001) surfaces of Si, Ge, or GaAs.4,5 The initial growth of AeTiO3 serves as an epitaxial platform for the subsequent integration of other multifunctional oxides.

For many envisioned applications, electrical coupling between the oxide and semiconductor is essential. In this regard, AeTiO3 is ideal for charge transport between the oxide and semiconductor,6 given the type-II band arrangement titanates have with respect to Ge, GaAs, and Si.7–10 However, for other envisioned applications, charge transport between the oxide and semiconductor is undesirable, and thus, a type-I band alignment is necessary, in which the conduction (valence) band of the oxide is above (below) the conduction (valence) band of the semiconductor.11 For a type-I alignment, SrHfO3 has been epitaxially grown on Si and Ge in place of AeTiO3.12,13 Recently, it has been demonstrated that a type-I band alignment can also be achieved through the growth of solid-solution SrZrxTi1-xO3 on Ge. Films of the Zr content up to x = 0.7 were characterized.14 The control of the Zr content in SrZrxTi1-xO3 enables the band offset to be tuned continuously from type-II to type-I, analogous to how band alignments are controlled through the composition at heterojunctions between compound semiconductors. The tuning of the band alignment allows the spin, charge, and polarization degrees of freedom of oxides to be electrically coupled to semiconductors.15 

For gate dielectrics, maximizing the conduction and valence band (VB) offsets in a type-I arrangement between the oxide and semiconductor is essential. Furthermore, minimizing the physical thickness of the oxide is also necessary. Here, we present structural and electrical characterization of SrZrO3 (SZO) grown on Ge, the solid-solution end member of SrZrxTi1-xO3. Aside from acting as an epitaxial platform that can facilitate electrical coupling between multifunctional oxides and semiconductors, SZO per se could be exploited as a gate oxide for complementary metal-oxide semiconductor (CMOS) devices. Recent interest has focused on exploring compound III-V semiconductors and Ge as potential channel materials to replace Si. Germanium is of particular interest for p-type metal oxide semiconductor devices (PMOS) given that it exhibits the highest hole mobility among all semiconductors (∼1900 cm2 V−1 s−1). We find that SZO exhibits several characteristics that are ideal for Ge based PMOS, namely, large conduction and valence band offsets as revealed by X-ray photoemission spectroscopy (XPS) measurements, low leakage currents as revealed by current-voltage (I-V) measurements, and the ability to modulate the surface potential of Ge as demonstrated in capacitance-voltage (C-V) measurements for 4 nm thick films.

Crystalline SZO films were grown on both p-type (Ga) and n-type (As) 2″ diameter Ge (001) wafers (AXT Inc., ρ  ≈  0.02 Ω cm) using oxide molecular beam epitaxy (MBE) in a custom-built ultra-high vacuum (UHV) chamber operating at a base pressure of < 2 × 10−10 Torr. The Ge wafers were introduced into the growth chamber after an etch and oxidation process that involved repeated dips in diluted HCl and H2O2.16 A clean dimerized Ge surface was obtained by thermally desorbing the resulting GeOx from the surface in ultra-high vacuum at ∼600 °C, as shown in the reflection high energy electron diffraction (RHEED) image of Fig. 1(a). The Ge was then cooled to ∼400 °C at which a half monolayer of the Sr metal evaporated from a thermal effusion cell (Veeco) was deposited to passivate the clean Ge surface. All thermally and e-beam evaporated metal fluxes were calibrated using a quartz crystal microbalance (Inficon). The Ge was then cooled to room temperature at which an additional 1 monolayer of Sr was deposited in a background O2 pressure of 3 × 10−7 Torr to form SrO. While the substrate remained at room temperature, 2 monolayers of Zr evaporated through an electron-beam evaporator (Thermionics), and an additional 1.5 monolayers of Sr were deposited separately through a shuttered process at 3 × 10−7 Torr of O2 (i.e., 1 monolayer ZrO2, then 1 monolayer of SrO, then 1 monolayer ZrO2, and finally 0.5 monolayer SrO). The amorphous 2.5 unit-cell (u.c.) thick layer of SZO was heated in UHV up to 590 °C, with crystallization commencing near 510 °C as shown in the RHEED image of Fig. 1(b) taken along the [10] direction. Subsequent SZO layers were grown through co-deposition of Sr and Zr fluxes at a substrate temperature of ∼590 °C and background oxygen pressure of 3 × 10−7 Torr at a typical growth rate of ∼1 u.c. per minute. As the thickness of the films increased beyond 22 u.c., a faint 2 × reconstruction emerges in the RHEED along the [10] direction, as shown in Fig. 1(e). The 2 × reconstruction is consistent with a superstructure found in bulk SZO that is characterized by tilting of the ZrO6 octahedra.17 Also shown in Figs. 1(d) and 1(f) are RHEED images taken along the [11] direction for the 10 u.c. and 37 u.c. films, respectively.

FIG. 1.

RHEED of SZO on Ge. (a) Clean Ge taken along the [11] direction. (b) After crystallization of 2.5 unit-cells of SZO on Ge, taken along the [10] direction. (c) 10 unit-cells of SZO on Ge taken along [10] and (d) [11] directions. (e) 37 unit-cell thick SZO on Ge taken along [10] and (f) [11] directions. Note the faint 2×structures denoted by the red arrows in (e). The specular reflection is indicated by the white arrow.

FIG. 1.

RHEED of SZO on Ge. (a) Clean Ge taken along the [11] direction. (b) After crystallization of 2.5 unit-cells of SZO on Ge, taken along the [10] direction. (c) 10 unit-cells of SZO on Ge taken along [10] and (d) [11] directions. (e) 37 unit-cell thick SZO on Ge taken along [10] and (f) [11] directions. Note the faint 2×structures denoted by the red arrows in (e). The specular reflection is indicated by the white arrow.

Close modal

X-ray diffraction (XRD) measurements were taken using a Panalytical Materials Research Diffractometer.

XPS measurements were performed using a Scienta Omicron R3000 energy analyzer with a monochromatic Al Kα x-ray source. No charge compensation was needed because all films were sufficiently conductive that no measurable positive charge built up on the surface during XPS measurements. The binding energy scale was calibrated using the Au 4f7/2 peak (84.00 eV) from a polycrystalline Au foil. A Shirley background subtraction was used for fitting all spectra. Spin-orbit (SO) splittings are generally different for oxides than for elements due to the existence of angular momentum coupled multiplets and charge transfer satellites in oxide spectra which redistribute intensity among the various final states.18 Therefore, with the exception of the Ge 3d spectrum, we did not constrain the SO splittings to elemental values in our fits but rather allowed them to vary. The values that emerged from the fits are 1.76 eV for Sr 3d and 2.39 eV for Zr 3d. For the Ge 3d spectrum, we constrained the SO splitting to the value measured for clean Ge(001), 0.59 eV [Fig. 3(c)].19 

High resolution scanning transmission electron microscopy (STEM) was performed on the post-annealed 4 nm thick films used to make MOS capacitors to examine the effect of annealing on the interfacial structure. Samples were prepared for STEM using conventional wedge polishing (Allied Multiprep), followed by Ar+ ion milling (Fischione Model 1050) while cooling with liquid nitrogen. A probe-corrected FEI Titan G2 60–300 kV operated at 200 kV was used for high-angle annular dark-field (HAADF) STEM imaging. The detector inner semi-angle was 77 mrad, and the probe convergence semi-angle was 19.6 mrad. Images were recorded with the RevSTEM method to remove drift related scan distortion,20 using 20 image frames acquired with a 2 μs/pixel dwell time and a 90° scan rotation between successive images.

For the C-V and I-V measurements, circular Ni electrodes of 30 nm thickness and 80 μm diameter were deposited through a shadow mask using electron beam evaporation. The backsides of the Ge wafers were mechanically scratched, and InGa eutectic was applied to form a counter electrode. The C-V characterizations were performed using an Agilent 4284A LCR meter on a Micromanipulator 8060 probe station using flexible, 10 μm radii tungsten probe whiskers. The I-V measurements were taken with an Agilent 4155C meter.

XRD measurements provide a quantitative analysis of the structure of the 37 u.c. (∼15 nm) thick SZO films. Figure 2(a) shows a survey scan of a 37 u.c. SZO film. The analysis of the (002) peak, shown more clearly in the inset of Fig. 2(a), indicates that the 37 u.c. thick film has an out-of-plane lattice constant of 4.132 Å. The finite thickness fringes corroborate that the film thickness is ∼15 nm and also indicate that the SZO-Ge interface and SZO surface are abrupt on the nanoscale. The rocking curve, which was measured using an analyzer crystal detector, exhibits a full-width at half-maximum (FWHM) of ∼0.63° for the 37 u.c. thick film, as shown in Fig. 2(b). The FWHM was determined by fitting a pseudo-Voigt profile to the entire dataset, as shown in Fig. S1 found in the supplementary material. To obtain the real-space map shown in Fig. 2(c), the Ge (224) and perovskite (103) peaks were analyzed. Both peaks can be measured at the same φ angle (rotation about the sample surface normal), thereby circumventing errors arising from a lack of alignment between the [001] direction and the φ-axis. The scattering vector in real space was resolved into in-plane and out-of-plane components and multiplied by h2+k2 and l, respectively, to give dimensions matching the lattice parameters. To facilitate comparison on the same plot in Fig. 2(c), the lattice constants of the perovskite SZO were multiplied by 2 to account for the 45° rotation of the perovskite unit cell with respect to the diamond cubic Ge substrate. The real-space map indicates that the ∼15 nm thick SZO film is mostly relaxed with respect to the Ge substrate, as expected from the 2.5% lattice mismatch. The SZO film also exhibits a non-cubic structure, as indicated by the diagonal dashed line which represents dz = dxy. The elongation in the spectral intensity about dxy for Ge (blue) is attributed to the geometry of the slit detector used for the measurement. This instrumental effect also elongates the spectral intensity for the SZO film (red), which is also intrinsically broadened due to the mosaicity and finite thickness of the film. The RSM indicates that the in-plane and out-of-plane lattice parameters for the film are 4.08 Å and 4.13 Å, respectively, resulting in a unit-cell volume of 68.8 Å3. The unit-cell volume is slightly smaller than that found for bulk SrZrO3 by Wong et al. (69.1 Å3) and is consistent with a small degree of strain to the substrate, resulting in a 0.6% in-plane contraction and a 0.5% out-of-plane expansion of the lattice.17 The difference in the unit-cell volume between the film and bulk could also be due to residual cation or anion non-stoichiometry.21 

FIG. 2.

(a) Survey scan of the SZO-Ge heterojunction. The inset shows the SZO (002) peak in greater detail. (b) Rocking curve of the SZO (002) peak, showing a FWHM of ∼0.63°. (c) Direct-space map of the SZO-Ge heterojunction. The lattice constants of SZO have been multiplied by 2 to enable direct comparison with Ge. The diagonal dashed line represents dz = dxy.

FIG. 2.

(a) Survey scan of the SZO-Ge heterojunction. The inset shows the SZO (002) peak in greater detail. (b) Rocking curve of the SZO (002) peak, showing a FWHM of ∼0.63°. (c) Direct-space map of the SZO-Ge heterojunction. The lattice constants of SZO have been multiplied by 2 to enable direct comparison with Ge. The diagonal dashed line represents dz = dxy.

Close modal

XPS spectra were collected on 37 u.c. and 6 u.c. thick SZO films to determine the valence band (VB) offset with respect to Ge using methodology described elsewhere.14,22 The 37 u.c. film was used to determine the energy difference between shallow core levels chosen for analysis (Zr 3d5/2 and Sr 3d5/2) and the top of the VB for a bulk-like sample. These spectra are shown in Fig. 3. The Zr and Sr 3d spectra shown in Figs. 3(a) and 3(b), respectively, show the evidence of surface hydroxylation due to air exposure. As a result, these spectra were fit using two pairs of spin-orbit doublets, one for the lattice (green) and one for the surface hydroxide (brown). The leading edge of the VB [Fig. 3(d)] intersects the energy axis at a binding energy of 4.22(3) eV. Assuming a band gap of 5.6 eV for bulk SZO,23 the Fermi level falls ∼1.4 eV below the conduction band minimum. Combining the j = 5/2 lattice peak energies with the VB maximum, we determine that ESr3d5/2 – Ev = 130.25(4) eV and that EZr3d5/2 – Ev = 178.53(4) eV for 37 u.c. SZO.

FIG. 3.

XPS spectra for SZO on Ge. (a) Zr 3d, (b) Sr 3d, and valence band (d) spectra for bulk-like 37 u.c. SZO/p-Ge(001), along with (c) Ge 3d for clean p-Ge(001). The lattice spin-orbit peaks are shown in green, and those from surface hydroxyl species are shown in brown. The sums of the individual peaks are shown in red. The valence band maximum is determined from the intersection of the leading edge and the energy axis, as determined by linear regressions (red). (e)–(h) XPS spectra for 6 u.c. SZO/n-Ge(001). The Ge 3s peak partially overlaps the Zr 3d spin-orbit doublet but can be fit to a single Gaussian, as shown in panel (e). Likewise, the Zr 4p doublet partially overlaps the Gr 3d spectrum for the 6 u.c. film and must be removed in order to accurately fit the Ge 3d doublet. To do so, the Zr 4p spectrum from the 37 u.c. film was used, as seen in panel (g). The VB spectrum for the heterojunction is well reproduced by the sum of the VB spectra (red) for clean p-Ge(001) (green) and 37 u.c. SZO/p-Ge(001) (brown), each being positioned in energy based on the measured core-level binding energies, as seen in panel (h).

FIG. 3.

XPS spectra for SZO on Ge. (a) Zr 3d, (b) Sr 3d, and valence band (d) spectra for bulk-like 37 u.c. SZO/p-Ge(001), along with (c) Ge 3d for clean p-Ge(001). The lattice spin-orbit peaks are shown in green, and those from surface hydroxyl species are shown in brown. The sums of the individual peaks are shown in red. The valence band maximum is determined from the intersection of the leading edge and the energy axis, as determined by linear regressions (red). (e)–(h) XPS spectra for 6 u.c. SZO/n-Ge(001). The Ge 3s peak partially overlaps the Zr 3d spin-orbit doublet but can be fit to a single Gaussian, as shown in panel (e). Likewise, the Zr 4p doublet partially overlaps the Gr 3d spectrum for the 6 u.c. film and must be removed in order to accurately fit the Ge 3d doublet. To do so, the Zr 4p spectrum from the 37 u.c. film was used, as seen in panel (g). The VB spectrum for the heterojunction is well reproduced by the sum of the VB spectra (red) for clean p-Ge(001) (green) and 37 u.c. SZO/p-Ge(001) (brown), each being positioned in energy based on the measured core-level binding energies, as seen in panel (h).

Close modal

The Zr 3d and Sr 3d spectra for the 6 u.c. film shown in Figs. 3(e) and 3(f) also show the evidence of surface hydroxylation and were fit accordingly. The Ge 3d and Zr 4p spectra shown in Fig. 3(g) strongly overlap. Therefore, the Zr 4p spin-orbit doublet measured for the 37 u.c. film was subtracted after appropriate shifting and scaling in order to isolate the pure Ge 3d spectrum (green). After this procedure, there is very little residual intensity characteristic of GeOx. The Ge 3d peak area ratio for GeOx to clean Ge is ∼0.05. Using the Sr 3d5/2 and Ge 3d5/2 binding energies, the VB offset is determined to be 3.65(7) eV. Likewise, the Zr 3d5/2 and Ge 3d5/2 binding energies yield a VB offset value of 3.66(7) eV. The energy difference between the Ge 3d5/2 peak energy and the valence band maximum in pure Ge(001), which is needed for the VB offset determination, is 29.34(4) eV.22 Assuming a value of 5.6 eV for the SZO bandgap, the band alignment is type I (nested), and the conduction band offset is 1.4(1) eV, as summarized in the band diagram of Fig. 4.

FIG. 4.

Band diagram showing the type-I band alignment between SZO and Ge obtained through XPS measurements.

FIG. 4.

Band diagram showing the type-I band alignment between SZO and Ge obtained through XPS measurements.

Close modal

The VB offset can be checked using the measured valence band spectrum for the thin-film heterojunction, along with those for clean p-Ge(001) and the 37 u.c. SZO film. Based on the measured Ge 3d5/2, Sr 3d5/2, and Zr 3d5/2 binding energies for 6 u.c. SZO-Ge, along with the energy differences between the core-level binding energies and the VB maxima for p-type Ge and thick-film SZO, the VB spectra for pure reference materials can be accurately positioned along the energy scale. After scaling the intensities to account for the SZO film thickness, the Ge and SZO VB spectra can be added, and the sum should overlap well with the measured heterojunction spectrum. The match is quite good, as seen in Fig. 3(h), confirming that the VB offset is 3.66 eV.

The type-I band offset enables SZO to act as a gate dielectric on Ge for PMOS. To explore SZO as a gate dielectric, we grew 10 u.c. (∼4 nm) thick SZO films on n-type Ge and performed electrical and structural characterizations. We find that as-grown films are electrically leaky due to residual oxygen vacancies that are introduced in the relatively low oxygen background pressure of the MBE chamber. To minimize oxygen vacancies, the heterojunctions were annealed in a tube furnace (MTI) for 5 min at 400 °C in flowing wet oxygen. Due to the thinner nature of the 4 nm thick films and post-growth annealing in oxygen, both their physical structure and interface with Ge are quite different from those of the 15 nm thick films.

To elucidate the effect of annealing on the interfacial structure, high-resolution STEM imaging was performed. The thin film is mosaic, arising from substrate/film misfit and substrate steps,24 as shown in Fig. 5. The 4 nm thick SZO in-plane and out-of-plane lattice parameters measured from the STEM images are 4.052 ± 0.003 Å and 4.157 ± 0.005 Å respectively, consistent with pseudomorphic growth. The values are presented as  ± the standard error of the mean. The in-plane and out-of-plane lattice constants of the 4 nm thick SZO are smaller and larger, respectively, than those of the 15 nm thick films, which are more relaxed due to the thickness. The in-plane parameter for SrZrO3 is about 5 pm larger than that for Ge (4.005 Å), indicating partial relaxation. Lattice parameters were extracted using previously described methods.25 Furthermore, Fig. 4 also shows interface regions that remain free of interfacial oxide, while GeOx forms non-uniformly in others, which arises from the post-growth annealing. In regions without oxide, Fig. 5 indicates that the interface also exhibits regions with and without a Zintl reconstruction, which has been observed in SrZrO3 thin films on Ge grown by atomic-layer deposition.26 

FIG. 5.

HAADF STEM image of the SZO-Ge interface showing good crystallinity of the 4 nm thick film. A non-uniform GeOx layer at the interface in the left transitions gradually to a coherent interface at the right. White arrows indicate a region exhibiting a Zintl reconstruction.

FIG. 5.

HAADF STEM image of the SZO-Ge interface showing good crystallinity of the 4 nm thick film. A non-uniform GeOx layer at the interface in the left transitions gradually to a coherent interface at the right. White arrows indicate a region exhibiting a Zintl reconstruction.

Close modal

Figure 6 shows representative C-V characteristics of an annealed 4 nm thick film, with the corresponding leakage current shown in the inset, indicating the densities of 3 × 10−3 A cm−2 and 6 × 10−3 A cm−2 at -1 V and +1 V, respectively . Dispersion in the capacitance as a function of the applied bias is observed for all frequencies, indicating the modulation of the surface potential of Ge. The upturn in capacitance at negative bias observed for 500 Hz indicates that inversion can be achieved in our SZO-Ge capacitors. The dispersive peak-like structure observed at negative bias for frequencies in the range of 5 kHz to 1 MHz arises from the so-called weak inversion response due to thermally generated minority carriers.27 Measurements taken from several junctions show minor differences in the C-V characteristics, as shown in Fig. S2, which likely arises in part from the variation in the interfacial structure observed in the film. Similarly, leakage current measurements taken from several junctions also show minor to negligible variations, as summarized in Fig. S3. The I-V measurements exhibit negligible hysteresis with sweeping bias, as shown in Fig. S4.

FIG. 6.

Capacitance-voltage characteristic of SZO-Ge heterojunctions, showing modulation of the surface potential of Ge. The bias voltage is applied to a nickel electrode on top of SZO. The inset shows the corresponding leakage current measurements for this junction.

FIG. 6.

Capacitance-voltage characteristic of SZO-Ge heterojunctions, showing modulation of the surface potential of Ge. The bias voltage is applied to a nickel electrode on top of SZO. The inset shows the corresponding leakage current measurements for this junction.

Close modal

The measured band offsets and leakage currents of our SZO-Ge capacitors are well comparable with or are better than other candidate gate dielectric materials for Ge, which include materials such as amorphous Al2O3, ZrO2, LaAlO3, Y2O3, HfO2, TiO2, and epitaxial SrHfO3.13,28–34 Although modulation of the surface potential of Ge is clearly achieved, interface trap states are present, as can be seen from the separation in capacitance between the low and high frequency C-V characteristics taken at 500 Hz and 1 MHz, respectively, within the depletion regime (∼0.5 V to ∼1 V). Accurately quantifying trap state densities, however, is challenging with room temperature C-V measurements alone, given the small band gap of Ge.35 With regard to the origin(s) of interface trap states, we note that the film is partially relaxed even at 4 nm thickness, which relaxation can create dangling bonds at the interface. Also, the formation of GeOx at the interface and growth using e-beam can contribute to trap states. More advanced techniques such as rapid thermal annealing in combination with the control of the oxygen reactivity should minimize the formation of GeOx. Postgrowth annealing in forming gas may also help minimize interface trap states, although this should be done with care as to not introduce oxygen vacancies in SZO.

In summary, we have presented structural and electrical characterization of epitaxial SZO on Ge, grown using oxide MBE. Single crystalline SZO can be nucleated on Ge through deposition at low temperatures and subsequent annealing in UHV, similar to SrTiO3. As a standalone material, ultra-thin films of SZO exhibit characteristics that are ideal for PMOS devices, including low leakage currents and large conduction and valence band offsets. The type-I band offset also enables SZO to act as both epitaxial and electrical platforms to couple the properties of multifunctional oxides to semiconductors.

See supplementary material for additional structural and electrical characterization of the SZO films.

This work was supported by the National Science Foundation (NSF) under Award No. DMR-1508530. XPS measurements performed at the Pacific Northwest National Laboratory was supported by the U.S. Department of Energy, Office of Science, Division of Materials Sciences and Engineering under Award No. 10122. The PNNL work was performed at the Environmental Molecular Sciences Laboratory, a national scientific user facility sponsored by the Department of Energy's Office of Biological and Environmental Research and located at PNNL. E.D.G. and J.M.L. gratefully acknowledge funding from the National Science Foundation (Award No. DMR-1350273). E.D.G. acknowledges support for this work through a National Science Foundation Graduate Research Fellowship (Grant No. DGE-1252376). This work was performed in part at the Analytical Instrumentation Facility (AIF) at North Carolina State University, which is supported by the State of North Carolina and the National Science Foundation (Award No. ECCS-1542015). The AIF is a member of the North Carolina Research Triangle Nanotechnology Network (RTNN), a site in the National Nanotechnology Coordinated Infrastructure (NNCI).

1.
S.-H.
Baek
and
C. B.
Eom
,
Acta Mater.
61
,
2734
(
2013
).
2.
J.
Hoffman
,
X.
Pan
,
J. W.
Reiner
,
F. J.
Walker
,
J. P.
Han
,
C. H.
Ahn
, and
T. P.
Ma
,
Adv. Mater.
22
,
2957
(
2010
).
3.
J. H.
Ngai
,
F. J.
Walker
, and
C. H.
Ahn
,
Annu. Rev. Mater. Res.
44
,
1
(
2014
).
4.
R. A.
McKee
,
F. J.
Walker
, and
M. F.
Chisholm
,
Phys. Rev. Lett.
81
,
3014
(
1998
).
5.
L.
Zhang
and
R.
Engel-Herbert
,
Phys. Status Solidi RRL
8
,
917
(
2014
).
6.
L.
Kornblum
,
D. P.
Fenning
,
J.
Faucher
,
J.
Hwang
,
A.
Boni
,
M. G.
Han
,
M. D.
Morales-Acosta
,
Y.
Zhu
,
E. I.
Altman
,
M. L.
Lee
,
C. H.
Ahn
,
F. J.
Walker
, and
Y.
Shao-Horn
,
Energy Environ. Sci.
10
,
377
(
2016
).
7.
F.
Amy
,
A.
Wan
,
A.
Kahn
,
F. J.
Walker
, and
R. A.
McKee
,
J. Appl. Phys.
96
,
1601
(
2004
).
8.
S. A.
Chambers
,
Y.
Liang
,
Z.
Yu
,
R.
Droopad
, and
J.
Ramdani
,
J. Vac. Sci. Technol. A
19
,
934
(
2001
).
9.
L.
Kornblum
,
M. D.
Morales-Acosta
,
E. N.
Jin
,
C. H.
Ahn
, and
F. J.
Walker
,
Adv. Mater. Int.
2
,
1500193
(
2015
).
10.
L.
Ji
,
M. D.
McDaniel
,
S.
Wang
,
A. B.
Posadas
,
X.
Li
,
H.
Huang
,
J. C.
Lee
,
A. A.
Demkov
,
A. J.
Bard
,
J. G.
Ekerdt
, and
E. T.
Yu
,
Nat. Nanotechnol.
10
,
84
(
2015
).
11.
R.
McKee
,
F.
Walker
, and
M.
Chisholm
,
Science
293
,
468
(
2001
).
12.
C.
Rossel
,
B.
Mereu
,
C.
Marchiori
,
D.
Caimi
,
M.
Sousa
,
A.
Guiller
,
H.
Siegwart
,
R.
Germann
,
J.-P.
Locquet
,
J.
Fompeyrine
,
D. J.
Webb
,
C.
Dieker
, and
J. W.
Seo
,
Appl. Phys. Lett.
89
,
053506
(
2006
).
13.
M. D.
McDaniel
,
C.
Hu
,
S.
Lu
,
T. Q.
Ngo
,
A.
Posadas
,
A.
Jiang
,
D. J.
Smith
,
E. T.
Yu
,
A. A.
Demkov
, and
J. G.
Ekerdt
,
J. Appl. Phys.
117
,
054101
(
2015
).
14.
J.
Moghadam
,
K.
Ahmadi-Majlan
,
X.
Shen
,
T.
Droubay
,
M.
Bowden
,
M.
Chrysler
,
D.
Su
,
S. A.
Chambers
, and
J. H.
Ngai
,
Adv. Mater. Interfaces
2
,
1400497
(
2015
).
15.
J. H.
Ngai
,
K.
Ahmadi-Majlan
,
J.
Moghadam
,
M.
Chrysler
,
D. P.
Kumah
,
C. H.
Ahn
,
F. J.
Walker
,
T.
Droubay
,
M.
Bowden
,
S. A.
Chambers
,
X.
Shen
, and
D.
Su
,
Journal of Materials Research
32
(2),
249
259
(
2017
).
16.
P.
Ponath
,
A. B.
Posadas
,
R. C.
Hatch
, and
A. A.
Demkov
,
J. Vac. Sci. Technol. B
31
,
031201
(
2013
).
17.
T. K.-Y.
Wong
,
B. J.
Kennedy
,
C. J.
Howard
,
B. A.
Hunter
, and
T.
Vogt
,
J. Solid State Chem.
156
,
255
(
2001
).
18.
S. A.
Chambers
, “
Hard X-ray Photoelectron Spectroscopy (HAXPES)
,” Springer Series in Surface Sciences Vol. 59 (
Springer International Publishing
,
New York
,
2016
), p.
341
380
.
19.
E.
Landernark
,
C. J.
Karlsson
,
L. S. O.
Johansson
, and
R. I. G.
Uhrberg
,
Phys. Rev. B
49
,
16523
(
1994
).
20.
X.
Sang
and
J. M.
LeBeau
,
Ultramicroscopy
138
,
28
(
2014
).
21.
C. M.
Brooks
,
L. F.
Kourkoutis
,
T.
Heeg
,
J.
Schubert
,
D. A.
Muller
, and
D. G.
Schlom
,
Appl. Phys. Lett.
94
,
162905
(
2009
).
22.
S. A.
Chambers
,
Y.
Du
,
R. B.
Comes
,
S. R.
Spurgeon
, and
P. V.
Sushko
,
Appl. Phys. Lett.
110
,
082104
(
2017
).
23.
R.
Schafranek
,
J.
Baniecki
,
M.
Ishii
,
Y.
Kotaka
,
K.
Yamanka
, and
K.
Kurihara
,
J. Phys. D: Appl. Phys.
45
,
055303
(
2012
).
24.
X.
Shen
,
K.
Ahmadi-Majlan
,
J. H.
Ngai
,
D.
Wu
, and
D.
Su
,
Appl. Phys. Lett.
106
,
032903
(
2015
).
25.
X.
Sang
,
A. A.
Oni
, and
J. M.
LeBeau
,
Microsc. Microanal.
20
,
1764
(
2014
).
26.
S.
Hu
,
E. L.
Lin
,
A. K.
Hamze
,
A.
Posadas
,
H.
Wu
,
D. J.
Smith
,
A. A.
Demkov
, and
J. G.
Ekerdt
,
J. Chem. Phys.
146
,
52817
(
2017
).
27.
K.
Martens
,
C. O.
Chui
,
G.
Brammertz
,
B.
De Jaegger
,
D.
Kuzum
,
M.
Meuris
,
M.
Heyns
,
T.
Krishnamohan
,
K.
Saraswat
,
H. E.
Maes
, and
G.
Groeseneken
,
IEEE Trans. Electron. Devices
55
,
547
(
2008
).
28.
S.
Swaminathan
,
M.
Shandalov
,
Y.
Oshima
, and
P. C.
McIntyre
,
Appl. Phys. Lett.
96
,
082904
(
2010
).
29.
C. O.
Chui
,
S.
Ramanathan
,
B. B.
Triplett
,
P. C.
McIntyre
, and
K. C.
Saraswat
,
IEEE Electron Device Lett.
23
,
473
(
2002
).
30.
Y.
Kamata
,
K.
Ikeda
,
Y.
Kamimuta
, and
T.
Tezuka
,
in 2010 Symposium on VLSI Technology
(
IEEE, Honolulu, HI, USA
,
2010
).
31.
C. H.
Lee
,
C.
Lu
,
T.
Tabata
,
W. F.
Zhang
,
T.
Nishimura
,
K.
Nagashio
, and
A.
Toriumi
, in
International Electron Devices Meeting
(
IEEE, Washington, DC, USA
,
2013
).
32.
C. H.
Lee
,
C.
Lu
,
T.
Tabata
,
T.
Nishimura
,
K.
Nagashio
, and
A.
Toriumi
, in
2013 Symposium on VLSI Technology
(IEEE, Kyoto, Japan,
2013
).
33.
R.
Zhang
,
P. C.
Huang
,
N.
Taoka
,
M.
Takenaka
, and
S.
Takagi
, in
2012 Symposium on VLSI Technology
(
IEEE, Honolulu, HI, USA
,
2012
).
34.
C.
Hu
,
M. D.
McDaniel
,
A.
Jiang
,
A.
Posadas
,
A. A.
Demkov
,
J. G.
Ekerdt
, and
E. T.
Yu
,
ACS Appl. Mater. Interfaces
8
,
5416
(
2016
).
35.
D.
Kuzum
, Ph.D. thesis,
Stanford University
,
2009
.

Supplementary Material