Magnetism modulation of Fe/ZnO heterostructure by interface oxidationWen-Chin Lin, Po-Chun Chang, Cheng-Jui Tsai, Tsung-Chun Hsieh, and Fang-Yuh Lo Citation: Applied Physics Letters 103, 212405 (2013); doi: 10.1063/1.4834699 View online: http://dx.doi.org/10.1063/1.4834699 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/103/21?ver=pdfcov Published by the AIP Publishing

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06

Magnetism modulation of Fe/ZnO heterostructure by interface oxidation

Wen-Chin Lin,a) Po-Chun Chang, Cheng-Jui Tsai, Tsung-Chun Hsieh, and Fang-Yuh Lob)

Department of Physics, National Taiwan Normal University, Taipei 116, Taiwan

(Received 8 October 2013; accepted 12 November 2013; published online 22 November 2013)

In this study, the magnetic coercivity (Hc) of Fe/ZnO heterostructure was significantly enhanced by

2–3 times after applying a suitable current. This Hc enhancement originates from the Fe-oxidation at

the Fe/ZnO interface induced by direct current heating. Depth-profiling X-ray photoemission

spectroscopy analysis confirmed the formation of FeO, Fe3O4, and Fe2O3 close to the interface region,

depending on the Fe thickness and annealing process. This study demonstrates that direct current

heating can moderately change the local interface oxidation and modulate the magnetic properties.

These results clearly reveal the correlation between magnetism and interface properties in the Fe/ZnO

heterostructure and provide valuable information for future applications. VC 2013 AIP Publishing LLC.

[http://dx.doi.org/10.1063/1.4834699]

The metal/oxide hetero-structure has attracted much

attention in recent decades because of its potential in various

applications such as heterogeneous catalysis.1,2 In particular,

the combination of magnetic metal with an oxide thin film,

e.g., the magneto-tunneling junction, has been widely stud-

ied and applied in data storage and spintronics. In order to

reduce the energy consumption in technical applications,

electrical control of magnetism has long been sought, and in

the past few years, several groups have proposed ideas for

the electric field control of magnetism. In 2012, Shiota et al.reported the induction of coherent magnetization switching

in a few atomic layers of FeCo using voltage pulses.3 In

addition, Wang et al. also reported the electric-field-assisted

switching in CoFeB/MgO/CoFeB magnetic tunnel junc-

tions.4 But until now, most of studies have focused on the

MgO systems, and there is still quite limited knowledge

about magnetic metals/ZnO.

ZnO has been an important semiconducting material,

not only for photoluminescence but also more importantly

for its promising applications in spintronics. Transition

metal-doped ZnO is predicted to be useful as a magnetic

semiconductor for room-temperature (RT) application.

Furthermore, ZnO exhibits the largest electromechanical

response, among the known tetrahedral semiconductors,

which makes it suitable for devices in microelectromechani-

cal and communication systems.5,6 In 2010, Luo et al.reported on the enhanced electromechanical response of Fe-

doped ZnO films by modulating the chemical state and ionic

size of the Fe dopant.

In view of the various applicable properties of ZnO, the

combination of ZnO with other functional materials, like

magnetic thin films, is of fundamental interest to explore.

But until now, few studies have reported on the magnetic

metal/ZnO systems. The natural oxidation of transition metal

in the proximity of ZnO is always unavoidable. Su et al.reported the thickness-dependent transition of Co chemical

state from the initial Co mixed oxidation to metallic state in

Co/ZnO(10�10).7 Results of Wett et al. show that, at ambient

temperature, Fe grows on ZnO(0001) in the pseudo layer-by-

layer mode with the FeO formation at the interface.8

Annealing leads to a stepwise oxidation of the Fe to FeO

(670 K) and Fe2O3 (820 K).8 However, the corresponding

evolution of the magnetic behavior is still lacking. Study of

Zhou et al. finds enhanced coercivity in the FePt/ZnO core/

shell nanoparticles after annealing at 773 K, but without fur-

ther chemical state analysis.9 In our experiment, the Fe/ZnO

heterostructures were prepared on Al2O3(0001) substrates by

e-beam evaporation and pulsed laser deposition (PLD). The

Fe-oxidation at the interface induced by direct current heat-

ing and the corresponding evolution in magnetism were

investigated.

Multiple layers of Au/Fe/ZnO/Au were sequentially de-

posited on an Al2O3(0001) substrate. The Au layers were pre-

pared for both the protection of Fe and also as the electrodes

for applying a current. The Au and Fe layers were deposited

by e-beam heated evaporators in an ultra-high vacuum (UHV)

chamber with a base pressure of 3� 10�9 mbar. The ZnO thin

films were prepared at RT in an oxygen ambient pressure of

8� 10�2 mbar by a PLD technique. The laser wavelength was

266 nm from a Nd:YAG Q-switch laser with an energy den-

sity of �4 J/cm2. Measurement of photoluminescence (PL)

spectroscopy was carried out at RT with an excitation wave-

length of 325 nm from a He-Cd laser. The magnetic hysteresis

loops were investigated at RT using the magneto-optical Kerr

effect (MOKE).10 The size of the junction area for applying

voltage is 1 mm� 1.5 mm. The MOKE laser spot is smaller

than 0.5 mm� 0.5 mm. Therefore, the laser spot can be fully

focused within the junction area in the MOKE measurements.

To study the Fe/ZnO interface property, depth-profiling X-ray

photoemission spectroscopy (XPS) investigation was carried

out by combining an in-house XPS with Arþ ion sputtering.

As shown in Fig. 1(a), the Au/Fe/ZnO/Au layers were

vertically stacked on a single crystalline Al2O3(0001) by

e-beam evaporation and PLD. The top and buffer Au layers

were used not only as the protection for Fe from contamina-

tion and oxidation but also as the electrodes for current

application. Fig. 1(b) shows the PL spectra of 2 nm

Au/2.5 nm Fe/2 nm ZnO before and after 500 K-annealing.

These two PL spectra differ in the intensity and reveal simi-

lar features. Four emission peaks were identified from the

spectra at around 2.65, 3.00, 3.20, and 3.35 eV, representing

a)E-mail: wclin@ntnu.edu.twb)E-mail: fangyuhlo@ntnu.edu.tw

0003-6951/2013/103(21)/212405/5/$30.00 VC 2013 AIP Publishing LLC103, 212405-1

APPLIED PHYSICS LETTERS 103, 212405 (2013)

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06

oxygen-defect related, donor-acceptor pair, bound exciton,

and free exciton emission lines.11–13 After 500 K-annealing,

the PL intensity is significantly reduced. In the heated

Fe/ZnO, the relative intensity of 3.35 eV emission decreases

while that of 2.65 eV emission becomes more pronounced,

indicating the increase of oxygen defects after the direct

heating process. The above observations could be due to cap-

ture of oxygen or oxidation of iron at the interface during

heating.

Fig. 2 shows the normalized magnetic hysteresis loops

of 5 nm Fe/2 nm ZnO as measured by in-plane MOKE after

applying different bias voltages. The bias voltage was gradu-

ally increased to 12 V by steps of 0.5 V. The MOKE mea-

surement was performed after applying the voltage for 30 s.

With the increasing bias voltage, the ratio of remanence to

saturation remained at 100%, but the shape of the hysteresis

loop became tilted. Meanwhile, the magnetic coercivity (Hc)

significantly increased from 55 Oe to 180 Oe. Fig. 3 shows

the more detailed information on this effect. The Hc, current,

resistance, and sample temperature were plotted as a func-

tion of applied voltage. The sample temperature was the

maximum temperature recorded by a k-type thermocouple

attached to the junction area when the voltage was applied. When the bias voltage was below 5 V, the Hc remained

invariant and the current increased linearly with bias voltage,

indicating the constant resistance and the ohmic conducting

behavior in the Fe/ZnO heterostructure. Similar ohmic con-

ducting behavior also has been reported for other metal/ZnO

systems.14,15 When the bias voltage was increased above

5 V, significant irreversible effects on magnetism and con-

ducting property were observed. In addition, the Hc was

enhanced to more than 2 times the original condition. With

the increase of bias voltage, as well as Hc, the sample current

first increased and then reached a maximum. Subsequently,

the current decreased monotonically and gradually became

stable and invariant after 9 V. Correspondingly, the electric

resistance slightly decreased during the Hc transition and

then increased monotonically. The conducting current heated

the Fe/ZnO junction to a maximum temperature of �500 K.

Interestingly, in the transition region of 6–9 V, there was a

turning point of the conducting current and sample tempera-

ture at �7 V due to the resistance change. After 9 V, the cur-

rent reduced to 150 mA, while the Hc was saturated at

�180 Oe and then remained invariant, although the sample

annealing temperature remained above 400 K.

FIG. 1. (a) Schematic illustration of the Au/Fe/ZnO/Au/Al2O3(0001) sample

structure. (b) PL spectra of Fe/ZnO heterostructures measured before and af-

ter applying a current to induce the coercivity enhancement.

FIG. 2. Normalized magnetic hysteresis loops of 5 nm Fe/2 nm ZnO meas-

ured by in-plane MOKE after applying different voltages. The coercivity

gradually increased from 48 Oe after 5.5 V and saturates at 180 Oe after

10 V.

FIG. 3. (a) Magnetic coercivity, (b) electric current, (c) electric resistance,

and (d) sample temperature are plotted as a function of applied bias voltage.

212405-2 Lin et al. Appl. Phys. Lett. 103, 212405 (2013)

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06

As indicated by the shadow area in Fig. 3, the Hc starts

to increase while the temperature is above �400 K. During

7–9 V, the temperature starts to decrease, but still sustains

above the critical temperature of �400 K; meanwhile, the Hc

keeps increasing, indicating that the oxidation process keeps

going during 7–9 V. After 9 V, the temperature decreases

nearly to the critical temperature of 400 K; meanwhile, the

Hc becomes invariant. However, the resistance keeps

increasing after 9 V. The invariant Hc implies the stop of oxi-

dation process, which is because of the limited diffusion

length, the finite amount of oxygen from ZnO, and the rela-

tive low temperature. On the other hand, even the tempera-

ture is relative low after 9 V, it is still above the critical

temperature of 400 K, which cannot sustain further oxida-

tion, but might be enough to trigger some effect on the 2 nm

ZnO layer and to increase the resistance.

Fig. 4 summarizes the Hc enhancement in various

Fe/ZnO hetero-junctions. With the variation of Fe and ZnO

thickness, similar Hc enhancement always occurred when the

applied voltage or current density was large enough. In Fig.

4(a), for 2 nm ZnO, the minimum voltage for triggering the

Hc enhancement needs to be larger than 6–9 V, while for

3 nm ZnO, the critical voltage must be larger than 12–14 V.

The thicker ZnO layer leads to a larger resistance and thus a

higher voltage is required to reach the same heating power.

In Fig. 4(b), the Hc evolution for the various samples is plot-

ted with the x-axis as the applied current density. A mini-

mum current density of 10–13 A/cm2 is always required for

inducing the Hc enhancement.

In order to investigate the interface properties of Fe/Zn

junction, the depth-profiling XPS investigation was carried

out after each cycle of ion sputtering. Fig. 5 shows the XPS

spectra of 2.5 nm and 10 nm Fe/ZnO with the depths indi-

cated. In this experiment, two junctions on a single sample

plate were simultaneously prepared with exactly the same

parameters. In this way, although the depth-profiling XPS is

a destructive investigation, we can still perform a compara-

tive experiment with each pair of samples, since one had

been modified by applying current and the other one

remained unchanged. Many previous reports provide the

characteristic XPS peak positions of Fe (707 eV 6 0.2 eV),

Feþ2 (709.9 eV 6 0.2 eV), and Feþ3 (711.4 eV 6 0.2 eV).8,16

The peak positions of Fe, Feþ2, and Feþ3 are indicated by

dashed lines in Fig. 5. For 2.5 nm Fe/ZnO, the top Fe layer is

mostly pure metallic Fe, but the interface Fe layer has been

clearly oxidized. In the as-deposited sample, as shown in

Fig. 5(b), the peaks of Fe and Feþ2 co-exist and the depth is

close to the Fe/ZnO interface. Since 400–500 K annealing is

sufficient to induce a significant Hc enhancement, in Fig.

5(a), after annealing, a peak located at 710.5 eV between

Feþ2 and Feþ3 gradually dominates when the depth is close

to the Fe/ZnO interface. This indicates the mixture of Feþ3

FIG. 4. Magnetic coercivity Hc of various Fe/ZnO heterostructures plotted

as a function of (a) bias voltage and (b) current density. In (b), the arrow

indicates the evolution direction of Hc with the variation of current density.

FIG. 5. Depth-profiling XPS of Fe measured at the indicated depth, close to

the Fe/ZnO interface. For comparison, two samples simultaneously prepared

with the same parameters are investigated before and after direct current

heating, respectively. The peaks of Fe, Feþ2, and Feþ3 are indicated by the

arrows.

212405-3 Lin et al. Appl. Phys. Lett. 103, 212405 (2013)

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06

and Feþ2, implying the formation of Fe3O4.17–19 In the case

of as-deposited 10 nm Fe/ZnO, as shown in Fig. 5(d), the

pure Fe peak dominates with a shoulder of Fe-oxide in

as-deposited sample. After annealing-induced Hc enhance-

ment, the Feþ2 gradually dominates the XPS spectrum with

increasing depth. In comparison with Fig. 5(a), the thicker

Fe film leads to a shortage of oxygen for Feþ3 formation

because of the limited ZnO thickness. Thus, only FeO is

formed in 10 nm Fe/ZnO, rather than the Fe3O4. As shown in

Fig. 5, at the Fe/ZnO interface, the chemical state of Fe

transfers from Fe2O3 to Fe3O4/FeO, and then to pure Fe.

Fe2O3 and Fe3O4 can be ferromagnetic, while FeO can be

antiferromagnetic, depending on their crystalline structure.

The oxidation of the Fe will decrease the saturation moment

even it is ferromagnetic Fe3O4, not to mention antiferromag-

netic FeO. Thus, the average saturation moment of Fe is sup-

posed to decrease after annealing. In the study of Xie et al.,it is found that even the annealed intrinsic ZnO crystal shows

weak ferromagnetism.20 The mixed magnetic signals from

Fe, Fe-oxides and ZnO are complicated to analyze individu-

ally and worth further study. Besides the Fe-oxidation, direc-

t-annealing may also trigger the Fe diffusion into ZnO. But

the possibility strongly depends on the annealing tempera-

ture. The diffusion of O into Fe overlayer is expected to

occur at a temperature lower than that for the Fe diffusion

into ZnO. Actually, the annealing temperatures for the Co

and Mn diffusion into ZnO (800–970 K) are much higher

than the critical temperature (500 K) for Hc-enhancement

observed in our experiment. Thus, the possibility of Fe diffu-

sion into ZnO is supposed to be relatively low and

negligible.20–23

As compared to the method of indirect-heating by an

additional heater, direct current heating can change the mag-

netism more locally through the interface oxidation, because

most of the heating power is loaded in the highly resistive

ZnO layer. In the study of Wett et al. on the Fe/ZnO(0001)

system, at ambient temperature, FeO was formed at the inter-

face. Further annealing led to a stepwise oxidation of Fe to

FeO (670 K) and Fe2O3 (820 K).8 Their experiment was

mostly performed on 2.4 nm Fe/ZnO(0001) and the XPS

measurement monitored only the surface top layers. Our

study, as shown in Fig. 5, explored the depth-profiling

Fe-oxide formation close to the interface. Annealing at

400–500 K leaves the top layer as pure metallic Fe, but sig-

nificantly transfers the underlayer, which is in proximity to

ZnO into FeOx. A proper choice of Fe thickness and anneal-

ing process can help control the different Fe-oxide forma-

tion. As reported, the magnetic coercivity of Fe3O4/ZnO thin

film is about 600 Oe at room temperature, which is much

larger than the Hc of pure Fe films.19,24,25 Study of Zhou

et al. also shows the enhanced coercivity in the FePt/ZnO

core/shell nanoparticles after annealing at 773 K, implying

the further oxidation of Fe by annealing.9 Thus, the

annealing-induced Hc enhancement is strongly correlated to

the Fe-oxide formation at interface. The direct annealing

process can moderately change the combination of highly

coercive Fe3O4 and soft magnetic Fe and thus modulate the

collective magnetic behavior. Interesting effects on the

spin-dependent transport property are also expected and wor-

thy of further study.

Fe/ZnO heterostrutures were prepared for the study of

the correlation between magnetism and interface properties.

Stepwise increase of the applied current through Fe/ZnO

junction significantly enhanced the magnetic coercivity by

2–3 times. The direct current heating induced the oxidation

of Fe layers adjacent to ZnO. Depth-profiling XPS revealed

that 400–500 K annealing was sufficient to induce serious

oxidation of Fe layers at the interface, while the top Fe layers

remained unchanged. The different phases of Fe-oxidation,

FeO, Fe3O4, and Fe2O3 can be moderately changed by direct

heating temperature and Fe thickness. Our study demon-

strates that the direct current heating of an Fe/ZnO hetero-

struture is an efficient method to modulate the interface

oxidation of Fe, as well as the magnetic property. These

observations will be valuable in future study and the applica-

tion of magnetic metal/oxide systems.

The authors acknowledge Professor Wang-Chi Vincent

Yeh for valuable discussion and technical support. The XPS

spectra were taken at the Nanotechnology Research Center,

National Dong Hwa University, Hualien, Taiwan. This work

was financially sponsored by National Science Council of

Taiwan under Grant Nos. NSC 99-2112-M-003-009-MY3,

NSC 99-2923-M-003-001-MY2, NSC 102-2112-M-003-

003-MY3, and NSC 100-2112-M-003-007-MY3.

1C. R. Henry, Surf. Sci. Rep. 31, 231 (1998).2C. T. Campbell, Surf. Sci. Rep. 27, 1 (1997).3Y. Shiota, T. Nozaki, F. Bonell, S. Murakami, T. Shinjo, and Y. Suzuki,

Nature Mater. 11, 39 (2012).4W.-G. Wang, M. Li, S. Hageman, and C. L. Chien, Nature Mater. 11, 64

(2012).5J. T. Luo, Y. C. Yang, X. Y. Zhu, G. Chen, F. Zeng, and F. Pan, Phys.

Rev. B 82, 014116 (2010).6A. Dal Corso, M. Posternak, R. Resta, and A. Baldereschi, Phys. Rev. B

50, 10715 (1994).7S. H. Su, J. H. Lai, H. H. Chen, T. H. Lee, Y. J. Hsu, R. L. Wang, and J. C.

A. Huang, J. Phys. Chem. C 116, 9917 (2012).8D. Wett, A. Demund, H. Schmidt, and R. Szargan, Appl. Surf. Sci. 254,

2309 (2008).9T. Zhou, M. Lu, Z. Zhang, H. Gong, W. S. Chin, and B. Liu, Adv. Mater.

22, 403 (2010).10C. S. Chi, B. Y. Wang, W. F. Pong, T. Y. Ho, C. J. Tsai, F. Y. Lo, M. Y.

Chern, and W. C. Lin, J. Appl. Phys. 111, 123918 (2012).11A. Teke, €U. €Ozg€ur, S. Do�gan, X. Gu, H. Morkoc, B. Nemeth, J. Nause,

and H. O. Everitt, Phys. Rev. B 70, 195207 (2004).12B. K. Meyer, H. Alves, D. M. Hofmann, W. Kriegseis, D. Forster, F.

Bertram, J. Christen, A. Hoffmann, M. Strasburg, M. Dworzak, U.

Haboeck, and A. V. Rodina, Phys. Status Solidi B 241, 231 (2004).13P. Ding, X. Pan, Z. Ye, H. He, H. Zhang, W. Chen, C. Zhu, and J. Huang,

Appl. Phys. A 112, 1051 (2013).14J. O. Song, K.-K. Kim, S.-J. Park, and T.-Y. Seong, Appl. Phys. Lett. 83,

479 (2003).15H. Sheng, N. W. Emanetoglu, S. Muthukumar, B. V. Yakshinskiy, S.

Feng, and Y. Lu, J. Electron Mater. 32, 935 (2003).16A. P. Grosvenor, B. A. Kobe, M. C. Biesinger, and N. S. Mclntyre, Surf.

Interface Anal. 36, 1564 (2004).17R. Master, S. Tiwari, R. J. Choudhary, U. P. Deshpande, T. Shripathi, and

D. M. Phase, J. Appl. Phys. 109, 043502 (2011).18Y.-J. Chen, F. Zhang, G.-g. Zhao, X.-y. Fang, H.-B. Jin, P. Gao, C.-L.

Zhu, M.-S. Cao, and G. Xiao, J. Phys. Chem. C 114, 9239 (2010).19A. M€uller, A. Ruff, M. Paul, A. Wetscherek, G. Berner, U. Bauer, C.

Praetorius, K. Fauth, M. Przybylski, M. Gorgoi, M. Sing, and R. Claessen,

Thin Solid Films 520, 368 (2011).20W. G. Xie, F. Y. Xie, X. L. Yu, K. Xue, J. B. Xu, J. Chen, and R. Zhang,

Appl. Phys. Lett. 95, 262506 (2009).21M. C. Mugumaoderha, R. Sporken, J. Ghijsen, and J. A. Dumont, J. Phys.

Chem. C 115, 20603 (2011).

212405-4 Lin et al. Appl. Phys. Lett. 103, 212405 (2013)

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06

22J. A. Dumont, M. C. Mugumaoderha, J. Ghijsen, S. Thiess, W. Drube, B.

Walz, M. Tolkiehn, D. Novikov, F. M. F. de Groot, and R. Sporken,

J. Phys. Chem. C 115, 7411 (2011).23M. P. Hyman, E. Martono, and J. M. Vohs, J. Phys. Chem. C 114, 16892

(2010).

24M. Paul, D. Kufer, A. M€uller, S. Br€uck, E. Georing, M. Kamp, J.

Verbeeck, H. Tian, G. Van Tendeloo, N. J. C. Ingle, M. Sing, and R.

Claessen, Appl. Phys. Lett. 98, 012512 (2011).25M. Paul, D. Kufer, A. M€uller, A. Ruff, M. Kamp, N. J. C. Ingle, M. Sing,

and R. Claessen, J. Appl. Phys. 110, 073519 (2011).

212405-5 Lin et al. Appl. Phys. Lett. 103, 212405 (2013)

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

122.100.68.88 On: Sat, 23 Nov 2013 02:08:06