Bandgap engineering in perovskite oxides: Al-doped SrTiO3Agham B. Posadas, Chungwei Lin, Alexander A. Demkov, and Stefan Zollner Citation: Applied Physics Letters 103, 142906 (2013); doi: 10.1063/1.4824023 View online: http://dx.doi.org/10.1063/1.4824023 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/103/14?ver=pdfcov Published by the AIP Publishing Articles you may be interested in On the variations of optical property and electronic structure in heavily Al-doped ZnO films during double-stepgrowth process Appl. Phys. Lett. 104, 021913 (2014); 10.1063/1.4862201 Band-engineered SrTiO3 nanowires for visible light photocatalysis J. Appl. Phys. 112, 104322 (2012); 10.1063/1.4767229 Dopant-induced bandgap shift in Al-doped ZnO thin films prepared by spray pyrolysis J. Appl. Phys. 112, 083708 (2012); 10.1063/1.4759208 Enhanced leakage current properties of Ni-doped Ba 0.6 Sr 0.4 TiO 3 thin films driven by modified band edgestate J. Appl. Phys. 107, 024109 (2010); 10.1063/1.3291124 Spectral-resolved microprobe cathodoluminescence investigations of Al-doped single-crystalline Ba 0.6 Sr 0.4TiO 3 thin films Appl. Phys. Lett. 87, 181914 (2005); 10.1063/1.2125109

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.123.181.145 On: Tue, 06 Oct 2015 23:01:43

Bandgap engineering in perovskite oxides: Al-doped SrTiO3

Agham B. Posadas,1 Chungwei Lin,1 Alexander A. Demkov,1,a) and Stefan Zollner2

1Department of Physics, The University of Texas, Austin, Texas 78712, USA2Department of Physics, New Mexico State University, Las Cruces, New Mexico 88003, USA

(Received 24 July 2013; accepted 13 September 2013; published online 1 October 2013)

The ability to modulate the bandgap of a material without altering its functional properties is

crucial for fabricating heterojunctions for device applications. Here, we explore experimentally and

theoretically the effect of the substitution of Ti with Al on the bandgap of perovskite oxide SrTiO3.

We grow Al-doped SrTiO3 films directly on Si(100) and show from electron energy loss spectra

that the bandgap is increased by approximately 0.3 eV over undoped SrTiO3. This bandgap

increase is confirmed by spectroscopic ellipsometry measurements on identical films grown on

LaAlO3 substrates, which show a 0.3 eV blue shift in the steep increase of the absorption edge.

Current vs. voltage measurements show a reduction in leakage current by six orders of magnitude

at a field of 1 MV/cm. We use density functional theory to explain how Al replacing Ti modifies

the conduction band edge density of states resulting in the experimentally observed increase in the

bandgap. VC 2013 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4824023]

The integration of functional perovskite oxides on semi-

conductors such as silicon and germanium presents possibil-

ities for achieving both improved performance in standard

field effect transistors, as well as obtaining hybrid electronic

devices where environmental sensitivity can be directly

coupled to the charge density in the semiconductor.1–5

SrTiO3 was originally envisioned as a replacement gate

dielectric for scaled CMOS technology because of its very

high dielectric constant of �300 at room temperature.6,7

However, one critical issue that prevented this technology to

be developed is the absence of the conduction band offset

between Si and SrTiO3 (STO), making SrTiO3 unsuitable for

use as a gate dielectric.8–10 Many workers have proposed

methods to increase the conduction band offset of STO with

Si, including insertion of high band offset buffer layers11–13

and modifying the interface oxygen stoichiometry.14,15 Here,

we report direct epitaxial growth of Al-doped SrTiO3 on Si

by molecular beam epitaxy (MBE), with Al concentrations

of 10% and 20%, and demonstrate using spectroscopic

ellipsometry (SE) and electron energy loss spectroscopy

(EELS) that the bandgap of Al-doped SrTiO3 is increased by

�0.3 eV over undoped STO. We further demonstrate using

leakage current measurements of capacitor structures that

Al-doping decreases the leakage current of STO on Si by

about six orders of magnitude at a field of 1 MV/cm. Using

density functional theory (DFT) and tight-binding (TB) anal-

ysis of the effect of Al substitution for Ti in STO, we attrib-

ute the bandgap increase to the narrowing of the conduction

band width, resulting in the bottom of the conduction band

being shifted higher in energy, away from the valence band

top.

Al-doped SrTiO3 was grown on both Si(001) and

LaAlO3(001) substrates using MBE. Sr, Ti, and Al are all de-

posited using effusion cells operated at approximately 425,

1780, and 1010 �C, respectively, resulting in a flux of about

1 monolayer per minute. The film stoichiometry is controlled

by shuttering the metal sources for the appropriate length of

time during the growth. The metal fluxes are calibrated using

a quartz crystal thickness monitor with feedback from in situx-ray photoelectron spectroscopy (XPS) measurements of

the film composition. During the growth, the samples are

monitored by reflection high energy electron diffraction

(RHEED) using an electron energy of 18 keV at a glancing

angle of �3�. Growth on Si is performed at a substrate tem-

perature of 500–550 �C under an oxygen partial pressure of

4–5� 10�7 Torr. These particular growth conditions have

previously been found to result in a SiOx layer thickness of

<1 nm for undoped STO as measured by cross-section trans-

mission electron microscopy. For the growth of Al-doped

STO, the Sr shutter is held constantly open while the Ti and

Al shutters are opened alternately to achieve the desired Al

doping concentration. The total growth rate is �4 A per min.

Prior to growth, prime n-type Si substrates (cut into 20� 20

mm2 pieces) are degreased and then exposed for 15 min to

UV/ozone. The native oxide is removed using Sr-assisted de-

oxidation,16 followed by deposition of 1=2 monolayer of Sr

Zintl template at 550 �C.17 For growth on LaAlO3, 5 mm sin-

gle crystal substrates from CrysTec Gmbh are used. The sub-

strates are degreased and then loaded into the growth

chamber for outgassing (650 �C in vacuum for 30 min)

before the growth of Al-doped STO.

After film growth, Al-doped STO samples are character-

ized using in situ XPS to determine their composition and

electron energy loss spectrum. For film composition analysis,

150 A-thick samples are used to ensure no photoelectron sig-

nal from the substrate is detectable. XPS is performed using

a monochromated Al Ka x-ray source and a VG Scienta

R3000 electron energy analyzer. The analyzer is calibrated

using the Ag 3d5/2 core level, which is defined to be

368.28 eV. This results in the Fermi level of Ag being

located at 0.00 eV. High resolution spectra of the Sr 3d, Ti

2p, O 1s, and Al 2p core levels are measured using a pass

energy of 100 eV and a slit width of 0.4 mm. The scan is per-

formed using 50 meV steps and a dwell time of 157 ms per

step, with each spectrum measured three times and summed.

For composition determination, the integrated intensities ofa)Electronic mail: demkov@physics.utexas.edu

0003-6951/2013/103(14)/142906/4/$30.00 VC 2013 AIP Publishing LLC103, 142906-1

APPLIED PHYSICS LETTERS 103, 142906 (2013)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.123.181.145 On: Tue, 06 Oct 2015 23:01:43

each region (after subtraction of a Shirley background) are

modified using an energy exponent of 0.78, to account for

the kinetic energy dependence of the sampling depth.18 This

energy exponent results in a 1:1 Sr/Ti ratio for commercial

single crystalline STO substrates using our instrument.

Wagner relative sensitivity factors are then used to calculate

the composition of the films.19 In addition to core level spec-

tra, O 1s energy loss spectra were also measured for the Al-

doped STO samples to determine zone-averaged bandgaps.

The ellipsometric angles of pure and Al-doped SrTiO3

films on LaAlO3 were measured from 0.8 to 6.5 eV at three

angles of incidence (65�, 70�, and 75�) on a J. A. Woollam

Co. variable-angle-of-incidence rotating-analyzer ellipsome-

ter, which was equipped with a computer-controlled Berek

waveplate compensator. The optical constants of the films

were then determined from the ellipsometric angles by solv-

ing Fresnel’s equations at each wavelength, using the meas-

ured film thickness and surface roughness obtained by x-ray

reflectivity (Philips X’Pert). A 10% error in thickness causes

about a 10% error of the refractive index and the absorption

coefficient; and also about a 20 meV error in the onset of the

strong absorption (better than electron energy loss spectros-

copy). To avoid complications due to substrate critical point

absorption effects, films grown on LaAlO3 were used for the

ellipsometric measurements. The dispersion of the LaAlO3

substrate was taken from the literature.20

Current vs. voltage (I-V) measurements of capacitor

structures were performed on both the undoped and 20% Al-

doped STO samples using a Keithley 4200 Semiconductor

Characterization System. 90 A-thick samples were grown on

heavily doped n-type Si. 500 A-thick gold contacts were de-

posited through a shadow mask using a sputter coater with

an array of circular holes of 500 lm in diameter. After con-

tact deposition, the samples were heated in air to 250 �C for

20 min. For the back contact, the backside of the Si sub-

strates were scraped with a scalpel and placed on a copper

block using a thin layer of liquid In-Ga eutectic alloy in

between. The copper block was grounded and the bias was

applied to the top electrode using a whisker probe tip.

The samples were characterized using RHEED during

and after the growth. RHEED patterns along the h120i zone

axis for both the 10% and 20% Al-doped STO samples on Si

after the growth of 20 unit cells (78 A) are shown in Fig. 1.

Both images show sharp, well-defined streaks indicating the

high degree of crystalline order in the films, similar to what

we have reported for undoped STO on Si.21 Immediately af-

ter growth, the films are transferred in situ to the XPS

analysis chamber where core level and electron energy loss

spectra are measured. The Al 2p binding energy is found to

be 74.68 eV and the O 1s binding energy is found to be

530.62 eV. An additional weak feature in the O 1s spectrum

appears at a binding energy of 532.66 eV, which is not found

in undoped STO. Both the Sr 3d and Ti 2p spectra of the Al-

doped films are similar to that obtained for undoped STO.

Analysis of the stoichiometry shows an Al concentration of

12% and 23%, respectively, for the nominally 10% and 20%

doped samples. There is also a measurable deficiency in oxy-

gen of �5–10% suggesting compensated doping. Sheet re-

sistance measurements at room temperature indicate that the

Al-doped samples are highly insulating with a sheet resist-

ance larger than 109 X/square (the measurement limit of our

instrument).

The bandgap was determined from the onset of the

energy loss spectra for O 1s photoelectrons.22 Because plas-

mon excitations for dielectrics have characteristic energies

larger than the lowest valence band to conduction band tran-

sitions, the bandgap value for dielectrics can be determined

by the onset of energy-loss spectrum of a core level peak. To

avoid complications due to spin-orbit pairs and multiplet sat-

ellites, the O 1s loss spectrum, which has neither, is typically

used for this type of measurement. Because of the relatively

large angular acceptance in transmission mode (30�) and the

large wave vector for O 1s photoelectrons, the bandgap value

measured by this method is averaged over the entire

Brillouin zone.23 The onset of the energy-loss spectrum was

defined by linearly extrapolating the segment of maximum

slope to the background level. Fig. 2(a) shows the O 1s

energy loss spectra for undoped and Al-doped STO. The

core level spectrum was fitted to either two (undoped) or

three (Al-doped) components and subtracted from the energy

loss spectra for clarity. A Tougaard background24 was used

to model the spectral background, which was also subtracted

from the measured signal. The overall features of the spectra

are consistent with published results for bulk single crystal

STO.25 For undoped SrTiO3, we find the onset of the loss

features to be at an energy of 3.7 eV. For both 10% and 20%

Al-doped samples, the onset is shifted to lower kinetic

energy (higher loss energy) by about 0.3 eV to a value of

4.0 eV with an uncertainty of about 0.1 eV. The energy loss

spectra show that substituting Al for Ti in STO results in an

increase in the zone-averaged bandgap.

Fig. 2(b) shows the calculated absorption coefficient

from the ellipsometry measurements as a function of photon

energy for the three samples. The data for bulk SrTiO3 are

also shown for comparison.26 The solid and dotted lines

show the absorption coefficient for bulk and thin-film

SrTiO3, respectively. The dashed lines show the Al-doped

SrTiO3 absorption coefficient. Ellipsometry confirms that the

Al doping causes a shift in the absorption edge of STO to

higher energies by about 0.3 eV. There is also a decrease in

the maximum of the absorption coefficient with Al doping.

The results of the I-V measurements are shown in Fig.

3. The undoped STO initially shows a moderate leakage of

4� 10�6 A/cm2 at þ1 V (1.1 MV/cm) bias but rapidly

increasing to 3.5� 10�2 A/cm2 at þ3 V. However, after the

first voltage sweep, the undoped STO changes to a higher

leakage condition with a leakage current of 8� 10�4 A/cm2

FIG. 1. RHEED patterns of 8-nm-thick Al-doped SrTiO3 grown on Si with a

nominal Al concentration of 10% (left) and 20% (right). Both images are

taken along the h120i zone axis of STO and show excellent crystallinity sim-

ilar to undoped STO on Si.

142906-2 Posadas et al. Appl. Phys. Lett. 103, 142906 (2013)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.123.181.145 On: Tue, 06 Oct 2015 23:01:43

at þ1 V (9� 10�2 A/cm2 at þ3 V). Additional sweeps do not

show any further changes in the leakage current of undoped

STO. The change in the leakage current in undoped STO has

been observed previously in relation to memristor applica-

tions, possibly resulting from voltage-driven movement of

oxygen vacancies.27,28 On the other hand, the 20% Al-doped

STO sample shows a very low leakage of 7� 10�10 A/cm2

at þ1 V and 8� 10�7 A/cm2 at þ3 V. This is a six orders of

magnitude reduction in leakage current at þ1 V due to the

Al-doping. The Al-doped sample also does not show any

change in the leakage current between the first sweep and

subsequent sweeps, indicating that oxygen vacancies are

strongly bound to the Al sites and are not mobile. The low

leakage and stability of the I-V measurements for Al-doped

STO shows its suitability as a current injection barrier on Si.

We have performed DFT calculations and have analyzed

the results using a TB model. The details are given in the

supplementary materials.29 Our main conclusion is that each

Al dopant effectively behaves like a Ti vacancy in the energy

range of valence and conduction bands, and its main effect

on the electronic structure is to block Ti-Ti hopping, which

reduces the conduction band width and therefore increases

the bandgap. A simple rule, based on TB analysis, to under-

stand the orbitally decomposed density of states (DOS) is

given in the supplementary materials.

From the band theory point of view, Al-doped STO is a

“hole” metal. Experimentally however, the Al-doped STO is

found to be insulating. The as-grown sample is also found to

have a significant amount of oxygen vacancies, as quantified

using in situ XPS.30 By comparing undoped STO with Al-

doped STO, we find an effective oxygen to metal ratio that is

�5–10% lower in the Al-doped films compared to undoped

films, suggesting compensation. Based on these observa-

tions, we consider theoretically configurations containing

two Al and one oxygen vacancy (OV), as shown in Fig. 4.

As discussed in Refs. 31 and 32, the two main effects of OV

are to provide two electrons and to lower one of the eg levels

of two adjacent Ti atoms. In the case of Al doping, the two

electrons provided by an OV fill two holes provided by two

Al atoms, so the whole system remains insulating. Also, the

FIG. 3. Current vs. voltage for 90 A-thick undoped and 20% Al-doped STO

on silicon. Undoped STO shows a change in state after the first voltage

sweep settling to a high leakage current of 8� 10�4 A/cm2 at þ1 V. The Al-

doped STO shows a low leakage of 7� 10�10 A/cm2 at þ1 V and does not

exhibit a change in state after the first voltage sweep.

FIG. 2. (a) O 1s photoelectron energy loss spectrum extracted from the low kinetic energy (high binding energy) side of the O 1s core level spectrum for both

undoped (dashed line) and 20% Al-doped (solid line) STO on Si. The O 1s core level components and the background (fit as a Tougaard function) were sub-

tracted from the data. The linearly extrapolated onset of the energy loss features (see inset) is located at a kinetic energy 3.7 eV below the main O 1s core level

for the undoped STO sample, and at 4.0 eV for the Al-doped sample. (b) Spectroscopic ellipsometry data for undoped and Al-doped STO plotted as absorption

coefficient versus photon energy. The solid and dotted lines show the absorption coefficient for undoped STO thin film and bulk STO, respectively. The two

dashed lines show the Al-doped STO absorption coefficient for 10% and 20% Al concentrations. The Al doping causes a shift of the absorption edge to higher

energies by about 0.3 eV. There is also a decrease in the maximum of the absorption coefficient with Al doping.

FIG. 4. (a) DOS for the configuration containing two Al with one OV in

between in a linear arrangement. The position of Al is represented by the

black spheres. Only a reduction in the density of states near the bottom of the

conduction band is observed in this case. (b) DOS for the configuration con-

taining two body-diagonal Al plus one OV. The oxygen vacancy is indicated

by the “X”. The DOS of two body-diagonal Al plus two La (replacing Sr) is

also provided for comparison. In both cases the system is insulating and the

Fermi energy is in the gap. Both cases also show an enlarged bandgap.

142906-3 Posadas et al. Appl. Phys. Lett. 103, 142906 (2013)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.123.181.145 On: Tue, 06 Oct 2015 23:01:43

t2g� t2g* bands are at most only mildly affected since OV

mainly changes the eg levels.

Similar to what was found in Co-doped STO,33 OV and

Al are energetically favored to be adjacent to each other

because this arrangement facilitates charge transfer between

OV and Al. We find that the configuration separating Al and

OV is at least 500 meV higher in energy. Two of the config-

urations we have considered are shown in Fig. 4(a) [Al-OV-

Al] and (b) [2 Al along the body diagonal and OV]. Note

that the OV is chosen to be next to at least one of the Al

atoms. In both cases, the system is insulating. The linear Al-

OV-Al configuration has the lowest energy in the calcula-

tions as the OV is adjacent to two Al atoms.33 In both cases,

the Al and Ti are relaxed away from the OV, consistent with

the previous OV calculations at the local density approxima-

tion level.32 The total DOS are very similar to those without

OV, leading to either a reduction of DOS near the bottom of

the conduction band [Fig. 4(a)] or an actual bandgap increase

[Fig. 4(b)].29 It is likely that both configurations are present

experimentally since the growth is done at elevated tempera-

tures, resulting in randomization of the Al and OV distribu-

tion, which on average, leads to an increase in the bandgap

(see also Fig. S2). From the theoretical point of view, the

higher the Al concentration, the larger the resulting increase

in the bandgap. We expect the optimum Al concentration

would be the highest that can be incorporated into STO with-

out compromising the crystallinity. In the present study, the

20% Al concentration still shows excellent crystallinity but

the optimum concentration could be slightly higher.

We have grown single crystalline, epitaxial Al-doped

SrTiO3 on Si with Al concentrations of �10% and 20% and

show that there is a bandgap increase of �0.3 eV using both

electron energy loss spectroscopy and spectroscopic ellips-

ometry. I-V measurements confirm the large reduction in the

leakage current of STO on Si (six orders of magnitude) as a

result of Al-doping. Theoretical analysis suggests that Al

dopants effectively reduce Ti-Ti hopping, behaving in many

respects like a Ti vacancy. Due to this reduction of hopping,

the Al doped STO has a narrower conduction band width,

leading to an enlarged bandgap. This work shows that Al

doping is a highly effective method of increasing the

bandgap of STO and is expected to be applicable to all tita-

nates, including ferroelectric BaTiO3 and PbTiO3, poten-

tially allowing for bandgap engineering in functional

perovskite titanates integrated in semiconductor devices.

The experimental work at UT Austin is supported by the

Office of Naval Research under grant N00014-10-1-0489,

the Air Force Office of Scientific Research under grant

FA9550-12-10494, and the National Science Foundation

under grant DMR-1207342. The theoretical work is sup-

ported through Scientific Discovery through Advanced

Computing (SciDAC) program funded by U.S. Department

of Energy, Office of Science, Advanced Scientific

Computing Research and Basic Energy Sciences under

award number DESC0008877. All calculations are done at

the Texas Advanced Computing Center. The work at NMSU

was supported by the National Science Foundation under

grant DMR-1104934.

1J. W. Fergus, Sens. Actuators B 123, 1169 (2007).2J. W. Reiner, A. M. Kolpak, Y. Segal, K. F. Garrity, S. Ismail-Beigi, C. H.

Ahn, and F. J. Walker, Adv. Mater. 22, 2919 (2010).3S. D. Ha and S. Ramanathan, J. Appl. Phys. 110, 071101 (2011).4S.-H. Baek and C.-B. Eom, Acta Mater. 61, 2734 (2013).5S. Abel, T. St€oferle, C. Marchiori, C. Rossel, M. D. Rossell, R. Erni, D.

Caimi, M. Sousa, A. Chelnokov, B. J. Offrein, and J. Fompeyrine, Nat.

Commun. 4, 1671 (2013).6S. Jeon, F. J. Walker, C. A. Billman, R. A. McKee, and H. Hwang, IEEE

Electron Device Lett. 24, 218 (2003).7K. Eisenbeiser, J. M. Finder, Z. Yu, J. Ramdani, J. A. Curless, J. A.

Hallmark, R. Droopad, W. J. Ooms, L. Salem, S. Bradshaw, and C. D.

Overgaard, Appl. Phys. Lett. 76, 1324 (2000).8S. A. Chambers, Y. Liang, Z. Yu, R. Droopad, and J. Ramdani, J. Vac.

Sci. Technol. A 19, 934 (2001).9X. Zhang, A. Demkov, H. Li, X. Hu, Y. Wei, and J. Kulik, Phys. Rev. B

68, 125323 (2003).10F. Amy, A. S. Wan, A. Kahn, F. J. Walker, and R. A. McKee, J. Appl.

Phys. 96, 1635 (2004).11J. Junquera, M. Zimmer, P. Ordej�on, and P. Ghosez, Phys. Rev. B 67,

155327 (2003).12R. A. McKee, F. J. Walker, and M. F. Chisholm, Science 293, 468 (2001).13T. Schimizu and T. Yamaguchi, Appl. Phys. Lett. 85, 1167 (2004).14C. J. F€orst, C. R. Ashman, K. Schwarz, and P. E. Bl€ochl, Nature 427, 53

(2004).15P. W. Peacock and J. Robertson, Appl. Phys. Lett. 83, 5497 (2003).16Y. Wei, X. Hu, Y. Liang, D. C. Jordan, B. Craigo, R. Droopad, Z. Yu, A.

Demkov, J. L. Edwards, and W. J. Ooms, J. Vac. Sci. Technol. B 20, 1402

(2002).17A. A. Demkov and X. Zhang, J. Appl. Phys. 103, 103710 (2008).18C. D. Wagner, L. E. Davis, and W. M. Riggs, Surf. Interface Anal. 2, 53

(1980).19C. D. Wagner, J. Electron Spectrosc. Relat. Phenom. 32, 99 (1983).20C. M. Nelson, M. Spies, L. S. Abdallah, S. Zollner, Y. Xu, and H. Luo,

J. Vac. Sci. Technol. A 30, 061404 (2012).21M. Choi, A. Posadas, R. Dargis, C.-K. Shih, A. A. Demkov, D. H.

Triyoso, N. David Theodore, C. Dubourdieu, J. Bruley, and J. Jordan-

Sweet, J. Appl. Phys. 111, 064112 (2012).22S. Miyazaki, J. Vac. Sci. Technol. B 19, 2212 (2001).23S. A. Chambers, T. Droubay, T. C. Kaspar, and M. Gutowski, J. Vac. Sci.

Technol. B 22, 2205 (2004).24N. Fairly, “XPS lineshapes and curve fitting,” in Surface Analysis by

Auger and X-ray Photoelectron Spectroscopy, edited by D. Briggs and J.

T. Grant (IM Publications, Chichester, UK, 2003).25S. Kohiki, M. Arai, H. Yoshikawa, S. Fukushima, M. Oku, and Y.

Waseda, Phys. Rev. B 62, 7964 (2000).26S. Zollner, A. A. Demkov, R. Liu, P. L. Fejes, R. B. Gregory, P. Alluri, J.

A. Curless, Z. Yu, J. Ramdani, R. Droopad, T. E. Tiwald, J. N. Hilfiker,

and J. A. Woollam, J. Vac. Sci. Technol. B 18, 2242 (2000).27H. St€ocker, M. Zschornak, J. Seibt, F. Hanzig, S. Wintz, B. Abendroth, J.

Kortus, and D. C. Meyer, Appl. Phys. A 100, 437 (2010).28R. Muenstermann, T. Menke, R. Dittmann, S. Mi, C.-L. Jia, D. Park, and

J. Mayer, J. Appl. Phys. 108, 124504 (2010).29See supplementary material at http://dx.doi.org/10.1063/1.4824023 for

details of the theoretical analysis.30Because of the in vacuum transfer between the MBE chamber and the

XPS chamber, the ratio of oxygen to total metal can be quantified without

interference from adsorbed oxygen species due to air exposure.31C. Lin, C. Mitra, and A. A. Demkov, Phys. Rev. B 86, 161102 (2012).32C. Mitra, C. Lin, J. Robertson, and A. A. Demkov, Phys. Rev. B 86,

155105 (2012).33A. B. Posadas, C. Mitra, C. Lin, A. Dhamdhere, D. J. Smith, M. Tsoi, and

A. A. Demkov, Phys. Rev. B 87, 144422 (2013).

142906-4 Posadas et al. Appl. Phys. Lett. 103, 142906 (2013)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.123.181.145 On: Tue, 06 Oct 2015 23:01:43